Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 
... " 
Informal Relationships in the W orkplace: Associations with Job 
Satisfaction, OrganIsational Commitment and Turnover Intentions 
/ 
A thesis presented in partial fulfilment of the requirements for the degree of 
Doctor of Philosophy 
In 
Psychology 
at Massey University, Albany, New Zealand. 
Rache1 Kathleen Lopes Morrison 
/ . . -- . ---_ . .  -- _ . . . . -- " 'iiliil l ii� 1 il iil ill /111 iiil� 1061756820 
2005 
1 
Abstract 
Abstract 
The aim of this thesis was to develop and test a theoretical model of friendships in the 
workplace. Friendships within organisations may have it profound effect on an employee's 
experience of work, pofentially either hindering or facilitating organisational functioning, yet 
friendships have seldom been studied in an organisational context. The association between 
friendships at work and organisational outcomes such as job satisfaction, organisational 
commitment and turnover intentions were investigated in three studies, assessing support for a 
theoretical model. 
In the first study, employees of a large Auckland hospital I (n = 124) were surveyed using a 
written questioIUlaire. Results indicated that cohesiveness and opportunities for friendship were 
related to increased job satisfaction, leading to increased organisational commitment and 
decreased turnover intentions. The actual prevalence of friendships was primarily related to 
decreased turnover intentions. Overall there was good support for the proposed model. The 
need for further research to ascertain the generalis ability of the findings was highlighted. 
A second study was conducted using an Internet based questionnaire, accessed both from within 
New Zealand and worldwide. A diverse sample of employees responded (n = 412). The 
analysis (structural equation modelling) indicated further support for most aspects of the model, 
suggesting that the fmdings are generalisable and the model is robust. 
The model of workplace relationships was cross-validated in a third study, confmning linkages 
between friendships at work and organisational outcomes. The model was then tested for inter­
group invariance. The model was invariant across groups reporting differing needs for 
affiliation, autonomy and achievement, but non-invariant across groups occupying relatively 
less or more interdependent jobs. Results suggest that the interdependence of individuals' jobs 
affects the salience of work friendships more than respondents' subjective needs. 
Overall, the research suggests that the presence of workplace friendships has a significant effect 
on several workplace variables, with the effect of friendships being more salient for individuals 
in interdependent work roles. The implications of the research findings are potentially far 
reaching. Not only do workplace friendships improve employees' experiences of work, but they 
also have the potential to affect the fmancial "bottom line" through factors such as enhanced 
organisational commitment, job satisfaction and reduced intentions to leave. 
) Waitemata District Health Board, Auckland (NZ) 
11 
Acknowledgements 
Acknowledgements 
I would like to thank my supervisors, Stuart Carr and Richard Fletcher for their help, 
support and invaluable feedback during this process. 
I am also grateful to Massey University for awarding the doctoral scholarship that made 
this project possible. In addition, the Massey University psychology department 
provided a grant to assist in travel to London in 2004, where parts of this project were 
presented at the Fourth Annual Conference on Emotions in the Workplace (PhD forum). 
I would like to acknowledge Waitemata DHB, for providing access to staff, many of 
whom took the time to fill out the initial survey, which began this entire project. In 
addition, the thesis could not have continued without those who generously gave their 
time to respond to the Internet based survey. 
Finally I would like to thank my family and friends, particularly my husband Justin 
Lopes, for listening, supporting and understanding. Thanks also to Karren Towgood 
and Rhiannon Herrick for proofreading. 
Note: 
Parts of this thesis have been published in the New Zealand Journal of Psychology: 
Morrison, R. (2004). Informal relationships in the workplace: Associations with job 
satisfaction, organisational commitment and turnover intentions. New Zealand Journal 
of Psychology, 33(3), 1 1 4-1 28. 
Parts of this thesis have been published in The Bulletin: 
Morrison, R. (2004). Gender comparisons in workplace friendships. The Bulletin(102), 
3 5-40. 
111 
Table 01' Contents 
Table of Contents 
Chapter 1: Introduction .....•.................
.
.........•....................•...............•.......................... 1 
Chapter 2: Literature Review ............................••... � ...........•.......................................... 6 
2.1 Where does this Project Sit Within Psychology? ........................................... 6 
2 . 1 . 1  Social Psychology: A History of Friendship Research .. . . . . ... . . . . ....... . .. .. . . ... 6 
2 . 1 .2 Organisational Psychology: A History of Human Factors Research .... . . . . . .  9 
2 . 1 .3  The Current Research . . . . . . . . . . . ... . . . . . . . .. . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . .. . 1 2  
2.2 What is it to be a Friend? ............................................................................... 14 
2 .2 . 1 Friends Compared to Other Relationships .. . . ... ... . . . . . . . . . . . .. . . . . .. . . . . . . . . . . . ... . . . . .  1 5  
2.2.2 The Nature of Friendship . . . ... . . . . . . . . . . . ..... . . . . . . . . . ... .. . . . . . . . . . . .. . . . . . . . . . . . .. . . . .. . . . . . . . .  1 7  
2.2.3 Gender Differences in Friendship . . . . . . . . . ... . . . . . . . . . ... ... . . . ...... .. ...... . . .... . . . . . . .. . . .  1 9  
2.2.4 Why Do People Engage in Friendships? ... . . .. . . . . . . . . . . . . . .. . . . . . . . ... . . . . . . . . . . . . . . . . . . .  2 1  
2.3 Relationships in a Workplace Context .......................................................... 24 
2.3 . 1  Types of  Organisational Peer Relationships . . . ..... . . . .. . . . ....... .... . . . . . .. . . . . . . . . .. . 25 
2.3.2 The Informal Structure of Organisations .... . . . . . . . . . ... . . . . ... . . . . . .. . . . . . . . . . . . .. . . .. . . . .  29 
2.3 .3 The Impact of Friendships in the Workplace . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . . .  3 1  
2.3 .4 The Impact of the Organisation on Friendships ... . . . . . . . . . . . . ..... . . .. . . . . . . . . . . . . . ... 50 
2.3 .5 Friendship Prevalence ........ . . . . . . . . . . . . . . . . . . ....... . . ..... . ... ...... ....... ... . . ....... . . . . . . . ... 57 
2.3 .6 Gender Differences in Organisational Relationships .. .. . . ... . . . . .. . . . . ... . . . . . . . . .. 6 1  
2.3 .7 Negative Relationships . . . . . . . . ... . . ...... . . . . . . . . . . . .. . . . ..... . . . . . . . .. . .. . . . . . . . ... . . ... . . . . . ...... 63 
2 .3 .8 Blended Relationships: Friends as Work Associates ... . ......... . . . . .... . . . . . . . . . . .  66 
Chapter �: Testing the Model in a New Zealand Hospital ......................•.......•......•.. 70 
3.1 Introduction ..................................................................................................... 70 
3.1 . 1  A New Zealand Hospital as a Setting for Organisational Research . . . . . . . . . .  70 
3 . 1 .2 Research Questions .... . ................ . .... . ....... . .  .-. . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . ......... 72 
.3 . 1 .3 Research Methodology . . . ..... . . .. . . . . . . ........ . . . . .. . . ... . ..... . . . ... . . ..... . ...... . .... . . ... .. . .. 73 
3.2 Method .............................................................................................................. 74 
3 .2.1 Participants .... . . . . . . . . . ...... ..... . . .. . . .. . ...... . .. . . . .. . . . .. .. . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . .. . . .  74 
3.2.2 Materials . . . . ...... .. . . ... .. . . . .. . . . .... � . . . . . ......... . . . . . ..... ... .. . ..... . ... . . . . . . ... . ... . .. . . . . . . . . . . . . .  75 
3.2.3 Procedure ... . . . . . . . . . . .. . . ...... . ... . . . . ... . .. . . . ... ... ... . . . . ..... . . . . . . . . . . .. . .. . . . ... . . ... . ... . . . . . . . .. . .  82 
iv 
Table of Contents 
3.3 Analysis and Results ................................................................................•....... 83 
3.3 . 1  Imputing Missing Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . ..... . . .  83  
3.3.2 
3.3.3 
3.3.4 
Factor Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . ... . . . . . . . . . . . . .  83 
Path Analysis . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . ... ... . . .... . . .. . . . . . . . .  95 
Peer Types . . . . . . . . . . . . . . . . . . . . . . . . ... . ........... . ...... . ...... . ... . . . . . . . ... . . . ... . . . . . . . . .... . . . .. . . . . .  1 03 
3.4 
3.4. 1 
Discussion ...
.
.................................................................................................... 1 04 
3.4.2 
3.4.3 
3.4.4 
3.4.5 
3.5 
3.6 . 
Research Question 1 . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .. ... . .  1 04 
Research Question 2 . . . . . . . . . . . . . . . . . . . . . . . .  : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... . . .  1 09 
Research Question 3 . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 1 2 
Job Satisfaction and Commitment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . .. . . . . . ... . . . . . . ... .. 1 1 4 
Satisfaction/Commitment and Turnover . . . . .... . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . .... . .  1 1 5 
Limitations of the Study and Directions for Future Research ........... ....... 115 
Conclusions and Implications ...............................................................•.....•. 117 
Chapter 4: Internet Study ......................................................................................... 120 
4.1 Introduction ................................................................................................... 120 
4. 1 . 1  Gender Differences in Friendships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . ... 1 2 1  
4.2 Method ....................................................................................................•....... 122 
4.2 . 1  Participants . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . .. . . . . .. . . . . .. . . . . . . ........ .. . . . . . .... . .. . . . . . . .. 1 22 
4.2.2 Materials . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . .................... ........... . . . .... . . . . . . . . .... . . ... . . . . . . . . ... 1 24 
4.2.3 Procedure . . . . . . . . . . . . . . . . . . . .. . . . . . . . .. . . . . . . . . . ... . . . . . . . . . . . . . ... . .. ... . . . . . . . . . . . .. .. . . . . . ... ....... ... 1 24 
4.3 
. 
Analysis and Results ...........•...........................................•................•....•.......• 125 
4.3.1 Factor Analysis .. . ... . . . .... . . . . . . .... ·. . . . . . . ...... . . . ... ... . .... ... . . . ... . . . .......... .. . . . . .......... 1 25 
4.3.2 Building the Causal Model.. .. . . ... . . . . . . . .... . . . . . ... . . . . . ........ .. . . ..... . ... ..... . . ....... . .  1 43 
4.3 .3 Statistical Significance of the Model.. . ... ... . . . . . ... . . . ... .... ... ......... . . .... ..... .. . . .  1 45 
4.3 .4 Other Research Questions .. ..... ... . . .... . .. .. . . . . . . . . ........ . . . ... ... . . ........... ... . .... . . .. 1 49 
4.4 Discussion ....................................................................................................... 152 
4.4.1 Resume of Findings . . . . . . ... . . . . ... . . . . . . . . .. . . . ... ... ... . . . . .. . . . . . .... .. . . . . . ...... ... . . ...... . ... 1 52 
4.4.2 Links to Existing Research . . . . . . . . . . . . ... . . . . . . . . . . . ...... . .... . ... . . . . . ..... . . .............. ... 1 57 
4.5 Limitations of the Study and Directions for Future Research ................•. 162 
v 
Table of Contents 
4.6 Conclusions and Implications .............................. � ........................................ l63 
Chapter 5: Cross-Validating the Causal Model and Testing for Invariance Between 
Sub-groups Differing in Need Strength and Interdependence of Job ......................... 165 
5.1 
5 .1.1 
5 . 1 .2 
5.2 
5 .2.1 
Introduction ................................................................................................... 165 
Needs for Affilia�ion, Autonomy and Achievement . . . . . .. . . . . .. . . . .... . . . . ... . ... 166 
Interdependence of Work Role . . . .. . ..... . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . .... . . . ... . . .  1 71 
Method ............................................................................................................ . 174 
Participants . . . . ..... ..... ... . . . . . . . . . . ... . . .. . . . . ... .. . . . . ... . . . ..... . . ... . . . . . . . . . . . . . .... . . . . . . .... . . . . 174 
5 .2.2 Materials . . . . . . . . . . . ... . . .. . . ... .. . ... . . . . . . . . . .. . . . . . . ... . . ... . .. . . . . . . . .... . . . . . ........ . . . . . . . ... . . . . ... 175 
5 .2 .3 Procedure .. . .. . . . . . . . . . . . . . . ... . . ... . . . .  � . . ... . . . . .. . . .. .. . . . . .. . . .. . .. . . . .. . . . . . . ... . . . . . . .. . . . ... . . . . . . .  178 
5.3 Analysis and Results ...................................................................................... 179 
5 .3 . 1  Measurement Models of the Scales . . . . . . ... . . . . . .... .. . . . .. . . ... . . . . . . . . ... . .. . . . . . . . . . . . . . 180 
5 . 3 .2 Model ofWorkplace Relationships . . ... . ..... . . . . . . . . . . . . . . .. ..... . . . . . . . . . . . . . . . ...... . . . .  180 
5 .3 . 3  Comparing Groups . . .. . . . . . . . . . .. ... . . .. . . . ... . . . . . .. . . . . . . ..... . . ... . .. . . . .... .. . . . . . .... . .. . . . .... 182 
5.4 Discussion ....................................................................................................... 189 
5.5 Conclusion ...................................................................................................... 190 
Chapter 6: Synthesis and Implications ..................................................................... 192 
6.1 Summary of Major Findings ........................................................................ 192 
6. 1 . 1  Hospital Study ... . . . .... ... ..... . . . . . . .. .. . . . . . . . . . . . . . . . . . . . .. . . . ... . . . . ... . . . ... . ... . . .. . . . . .. . . . . . . .  192 
6. 1 .2 Internet Study . .. . ... .. . .... . . . . ... . ... ... . . . ... . .... . .. . .... ... . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . .... . . ... 194 
6.1 .3  Invariance Testing Study . . ... . . . . . .. . .. ..... . . . . . . ... . . . . . .... . . . . . ... . . . . . . . . . . ... .. .... . . ..... . 195 
6.2 Synthesis ......................................................................................................... 196 
6.2 . 1  Friendship Opportunities versus Friendship Prevalence . . . . .. . . . . . . . . . . .... . .... 196 
6.2.2 Interdependence of Work Role versus Individual Differences in Needs 198 
.6.2.3 Gender Differences . . . . . . . . . . . . . .. . . . . . .... . . . . . . . . ... .... . ... .. . . ... ... . . . .... ........ . . . . . ...... ... 199 
6.3 Future Directions ........................................................................................... 202 
6.3 .1 Performance ... ...... . . . .. . .......... . .. . . . . . . . . . . . ..... . . . . .... ... . . ... . . . . . . . . ... . . .... . . . . . . . . . .. .. ... 202 
6.3 .2 Negative Relationships . . . ... ... . . . . . . . . .. . . . .. .... . . . . .... . . . ... . . . .. . .. . . . .  : . . . . . . . . . . .. . . . ... . . .  202 
6.3.3 Organisational Identity and Attributions of Responsibility .. . ... . . . . . . . ... .... 203 
VI 
Table of Contents 
6.4 Implications .......................................................... � ......................................... 207 
6.4. 1 Enhancing the Quality of Work Life . . . .. . . . . ... . . . . ..... . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . .  207 
6.4.2 Situations Where the Findings Might have More Relevance . . . . . . . . . . . . . . . . . .  209 
6.4.3 Possible Negative Implications . . . . . ... . . . . . . . .... . . . . . . . . ... . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . .  2 1 0  
6.5 Conclusion ......................................................................... . . . . . . . . . .................... 2 12  
References ........................................................................................................................ 213 
Appendices ...................................................................................................................... 232 
Appendices for the Literature Review (Chapter 2) ................. . . . . . . . . . . . . . . .. .. . ............ 232 
Appendices for the Hospital Study (Chapter 3) . ...... .. . . . ............ . . . . . . . . . . ................... 236 
Appendices for the Internet Study (Chapter 4) . .. .... . . . . . . ............. . . . . . . . . .. .. . . . . . . . . . . ...... 266 
Appendices for the Invariance Testing Study (Chapter 5) .................................... 294 
vu 
List of Tables 
List of Tables 
Table 1 :  Demographics of Respondents of the Hospital Study ...................................... 75 
Table 2: First Pattern Matrix of the Factor Analysis of the Friendship, Cohesion and 
Dual Role Tension Measures (items to be removed are shaded in grey) ................ 86 
Table 3: Final Pattern Matrix of the Factor Analysis of the Friendship, Cohesion and 
Dual Role Tension Measures .................................................................................. 88 
Table 4:  First Pattern Matrix of the Factor Analysis of the Satisfaction, Commitment 
and Leaving Intention Measures (items to be removed are shaded in grey) ........... 9 1  
Table 5: Final Pattern Matrix of the Factor Analysis of the Satisfaction, Commitment 
and Leaving Intention Measures .................................. ........................................... 93 
Table 6: Correlations between Measured Variables: Hospital Study ............................. 94 
Table 7: Showing Critical Ratio (C.R.) Values (parameter estimate divided by standard 
error) of the Regression Paths in Model 1 ............................................................... 96 
Table 8: Showing Critical Ratio Values (parameter estimate divided by standard error) 
of the Regression Paths in Model 1 When Specified with Non-Significant 
Regression Paths Removed ..................................................................................... 97 
Table 9: Demographic Data for the Internet Study ...................................................... 1 22 
Table 1 0: Industry of Respondents ........................................................................ ....... 1 23 
Table 11: Initial Pattern Matrix for the Workplace Friendship Scale (items to be 
removed are shaded in grey) ................................................................................. 1 28  
Table 12 :  Final Pattern Matrix for the Workplace Friendship Scale ........................... 1 28 
Table 1 3 :  Initial Pattern Matrix for the Job Satisfaction Scale (items to be removed are 
shaded in grey) ...................................................................................................... 1 3  1 
Table 14: Final Pattern Matrix for the Job Satisfaction Scale ...................................... 1 3 1  
Table 1 5 :  Pattern Matrix. for the Cohesion Scale ......................................................... 1 3  7 
Table 16: Initial Pattern Matrix for the Needs Assessment Questionnaire (items to be 
removed are shaded in grey) ................................................................................. 1 40 
Table 1 7: Final Pattern Matrix for the Needs Assessment Questionnaire ................... 1 40 
Table 1 8 : Fit Indices for the Measurement Models: Internet Study ............................. 1 42 
Table 1 9: Correlat�ons between Measured Variables: Internet Study .......................... 1 43 
Table 20: Alpha Levels of the Sub-Scales ................................................................... 1 43 
viii 
List of Tables 
Table 2 1 :  Showing Critical Ratio Values (parameter estimate divided by standard error) 
ofthe Regression Paths in Model 2 When Specified with Non-Significant 
Regression Paths Removed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  147 
Table 22: Fit Indices for the Causal Model: Internet Study ........... .. . . .. . . . . . . . . . . . . . . . . . . . . . . . .  148 
Table 23 : Correlations between Subscales, Showing those where Measured Variables 
are Differentially Correlated for Men and Women . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . .. ... . .. .. .. .  149 
Table 24: Independent Samples T-test, Comparing Respondents With and Without 
Negative Relationships at Work . . . . . . . . .. . . . . . . . . .. . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  15 1  
Table 25 :  Hypothesised Relationship between Occupation, Affiliation Opportunities 
and Opportunities to Acquire Career Capital. . . . . . .. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 7 1  
Table 26: Breakdown o f  Demographic Data for Respondents to Cross-validation / 
Invariance study .................................................................................................... 175  
Table 27:  Needs Assessment Questionnaire Items Used in  the Current Study . . . . . . . . . . . .  1 77 
Table 28 :  Fit Indices for the Measurement Models (n = 445) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 80 
Tcibie 29: Fit Indices for the Causal Model : Based on the Full Sample (n = 445) . . . . . .  1 8 1  
Table 30: Chi-Square Statistics for Tests of Invariance across Sub-groups of the Sample 
.......................... � ... ................... ...... ..................... . .......................... ... ... .......... ........ 1 82 
Table 3 1 : Goodness of Fit Statistics for Tests of Invariance Across Those in Relatively 
Less and More Interdependent Jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 84 
Table 32:  Goodness of Fit Statistics for Tests of Measurement Model Invariance Across 
Those in Relatively Less and More Interdependent Jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 86 
Table 33:  Showing Correlation Coefficients and Critical Ratio Values (parameter 
estimate divided by standard error) of the Correlation Between the Cohesion 
(Social Support) and Friendship Opportunity Variables for High and Low 
Interdependence Groups . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1 88 
Table 34: Goodness of Fit Statistics for the High and Low Interdependence Groups. 1 89 
ix 
List of Figures 
List of Figures 
Figure 1: Showing the hypothesised relationship between friendship opportunities and 
job satisfaction . . .  : . .. . . . . . . ........ . . . . .. . . .. . . . .. . . . . . .. . ..... .. ... ... . . .. . .. . ..... . ....... ................ ........... 3 8  
Figure 2: Showing the hypothesised relationship between job satisfaction and 
organisational commitment . . . . . . .. . . . . . . . . . .. . . ... . . . . .. . . . . . . .. . . . . . . . . . . .. . .. . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. .. 42 
Figure 3: Showing the hypothesised relationship between friendship opportunities, job 
satisfaction and organisational commitment .. . . . . . . . . . . . . .. . . .. . . . . . . ... ... . . . .. . .. . . . . . . . . . . . . . . . . . .  . 46 
Figure 4: Showing the hypothesised relationship between job satisfaction, commitment 
and intention to leave . .. .. . . . . . .. . .. . . .. . . . . . .. . . . . . . . .. ... .. .. .. ........ . . .... ... . . . . . . . . .. . .. .. ... . .. . .. . .. . . . . . .  48 
Figure 5: Showing the hypothesised relationship between friendship opportunities and 
cohesion ... .. . . . ......... ...... . . . . . . . . . . . . . . . . ..... . .. .. .. . . . . . . . . . . . . . . . . . . .. . . . . . .. . . . . .. . . .. . .. . . .. .. . . . . . . . . . . . .. . . . . .  54 
Figure 6: Showing the hypothesised relationship between cohesion and intention to 
leave . . . . ... . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . .... . .  56 
Figure 7: Showing the hypothesised relationship between cohesion and organisational 
commitment and job satisfaction . ..... ... ... . ... .. . . .. . . . . . . . . . . . . . . . . .. . . .. . . .. . . . ... ...... . . . . . . . . . . . . . . . . .  57 
Figure 8: Showing friendship prevalence as the result of increased friendship 
opportunities and cohesion . ..... .... . . ..... . ... ....... ...................... .................... ..... . .. . .. . .... 58 
Figure 9: Showing the hypothesised relationship between friendship prevalence and 
intention to leave . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . .... . . .. .. . . . . . . . . . . . .. ... . . . . . . . . . . ... . . . .  60 
Figure 1 0: Showing negative relationships added to the model. . . . . ... ... . . . . . . . .. . . . . . . . . . . . . . . . . .  66 
Figure 1 1: Showing the hypothesised moderating effect of dual role tension (Model 1 )  
................................................................................................................................. 69 
Figure 1 2: Path analysis of Model 1 .. .. . ... ...... . .. . . . . . . ... . ..... .. . . . .. .. . . ... .. . . ..... . .... . . .. . . . . . . . . .. . . . . .  98 
Figure 1 3: Path from friendship opportunities and cohesion to organisational 
commitment . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  99 
Figure 14: Path from job satisfaction and organisational commitment to intention to 
leave . ... ... .. . . . . .. .... . . . . . . . . . . . . . . . . .. ... .. . . . .. . .. .. . .. ..... .. .. . . ... . . . . . . . . . . . .. . . ... . ... .. ..... .... ... . . . . . . . . . . . . . .  1 00 
Figure 1 5: Path from friendship opportunities and cohesion to friendship prevalence 1 00 
Figure 1 6: Path from cohesion to intention to leave .. . . . . . . . . .. . . . .. .. . ... .. .. . . . ... . .. . . . . ... . . . . . . .. .  1 0 1  
Figure 1 7: Path from friendship opportunities to intention to leave . . . . . . . . . . .. . . . . .. .. . .... ... 1 02 
x 
List of Figures 
Figure 18: Model 1 redrawn showing only the significant regression weights (Hospital 
Study) .................................. .................................................................................. 103 
Figure 19: AMOS diagram for friendship scale .......... : .............................................. .. 130 
Figure 20: AMOS diagram for the job satisfaction scale ... . . ........... .. .. .................... . .... 133 
Figure 21: Scree plot for Organisational Commitment Questionnaire ........... ........... . .. 134 
Figure 22: AMOS diagram for the Organisational Commitment Questionnaire ......... 136 
Figure 23:  Scree plot for the cohesion scale ......................... ....................... ....... .......... 136 
Figure 24: AMOS diagram for cohesion scale ............ ................................. ..... ........ . .. 13 8 
Figure 25: AMOS diagram for the Needs Assessment Questionnaire ........... .............. 141 
Figure 26: The causal model tested for the Internet Study (latent variables only) . ..... 144 
Figure 27: Model 2, SEM results of Internet Study .......................................... ...... ..... 146 
Figure 28: Results of SEM analysis of the theoretical model. Values represent 
standardised estimates and correlations based on the full sample (n = 445) . .. . .... 181 
Figure 29: The hypothesised mediating effect of identification and attribution, on the 
relationship between friendship prevalence and job satisfaction . .................. .. . . . .. 205 
Figure 30: Alternative hypothesised mediating effect of identification and attribution, 
on the relationship between friendship prevalence and job satisfaction . ........ . . . . .. 206 
.. . '\.! ' , ' 
Xl 
Introduction 
Chapter 1 :  Introduction 
"Wit!iout friends, no one wouU c!ioose to five. " 
-JIristotCe-
The above quote suggests that, for many people, friendships are essential in creating a 
sense of well-being and happiness .  Friends who share our interests, enjoy our company 
and value our ideas, enrich our lives; they provide comfort, support and help in times of 
need (Venieg� & Peplau, 1 997). 
Close friendships often evolve fr�m existing formal relationships in work places, and 
for many people, these relationships are maintained within organisational settings. Yet, 
despite the frequency of these relationships, we know very little about how dual 
friendship/work relationships function, and how the blurring of relational boundaries 
might affect organisational functioning, the enjoyment of work, and perhaps even 
performance. 
This research focus was developed partly as an outcome of conversations with friends 
and acquaintances who had experienced a great deal of tension in the workplace as a 
result of the breakdown of personal relationships. Others mentioned stress as a result of 
"being on the out," being excluded by a dyad-with whom they still had to work. On the 
other hand, people spoke of how absolutely fundamental their friendships at work were 
to their enjoyment of work, to their satisfaction with their job, and to their desire to 
remain in their present employment. 
1 
Introduction 
Workplace relationships can profoundly impact people's experience of work, both 
negatively and positively. This impression was reinforced by the response received to 
an article about my research into workplace friendships, which appeared in the New 
Zealand Herald (Middleton, 2002). The following are excerpts of emails from people 
who wrote to me. 
This was from a woman who told her employers she was leaving her place of work 
because she felt limited and had no room to grow or develop in her job: 
" . . .  but truth be told I no longer wanted to work for a company where 
I couldn' t make friends. To spend 8 hours a day (minimum) at a place 
where you do not have close relationships with any. of the staff 
(regardless of whether you enjoy your actual job or not) is a very sad 
thing in my opinion. I am now looking for a role with a smaller more 
stable company where I. can not only continue to enjoy the actual 
work I do but also hopefully enjoy the company of my workmates in a 
more trusting friendly environment" (T. D. Feb, 2002). 
The following quote was from an employee who described difficulties associated with 
her recent change of job: 
"The first year was a horrendous time. I hated - not so much the 
actual job, but the feeling of isolation from not having any friends in 
my new job. I actually longed to return to my previous job, just to be 
with my previous co-workers, whom I missed terribly. This was in 
spite of the fact that I was earning less money in the previous job" 
(D.C. Feb, 2002). 
The woman who wrote the quote below chose to leave her job primarily because of the 
lack of informal relationships and friends in her former workplace: 
2 
Introduction 
"I have recently left a job where the staff were not encouraged to talk 
to each other while at work. They did not sit together at lunchtime 
and did not have tea breaks. The atmosphere was just terrible" (D.D. 
Feb, 2002) . .  
A Human Resource manager wrote as well, commenting that: 
" . . .  those who develop a friendship with equal colleagues and interact 
with them socially out of work have tendency to remain loyal to the 
company" (B.O. Feb, 2002). 
In addition to this anecdotal evidence, which indicates that individuals make 
employment decisions based on the quality of their relationships at work, research has 
also indicated that organisations too, prefer that their employees work well together. An 
American study by Lozada ( 1996) found that 90 percent of dismissals are the result of 
poor attitudes, inappropriate behaviour and difficulties with interpersonal relationships 
rather than deficient technical skills. The finding that people are so often dismissed for 
reasons other than being unable to do their jobs, highlights the importance of informal 
interpersonal relationships at work; being good at your job is not sufficient if you cannot 
work with people. Thus, the informal relationships employees have at work not only 
seem to have a significant effect on turnover; on whether employees choose to stay in 
their jobs and on their subjective enjoyment of their jobs, but also on whether 
organisations want them to remain or decide to end their employment. 
Intuitively, it is likely that the informal relationships individuals have in the workplace 
will affect their experience of work. Correspondence with members of the public, in 
response to the article mentioned above, and with friends and colleagues seems to 
support this. However it is possible that the people who felt strongly enough about 
these issues to write in response to a newspaper article are not representative of the 
general working population. As there is currently little empirical research investigating 
the phenomenon of informal workplace relationships published in the literature, it is the 
aim of this thesis to address this gap. Thus, a focus of this project is to examine the 
3 
Introduction 
possible effects of informal workplace relationships, both on the well-being of 
employees, and on organisational outcomes such as job satisfaction, organisational 
commitment and intention to leave an organisation . . To examine the phenomenon of 
friendships at work, a theoretical model is proposed and tested. 
In the studies described in this thesis, informal interpersonal relationships other than 
romantic relationships are examined. Some previous research has looked at romantic 
relationships between work-mates (e.g., Anderson & Hunsaker, 1 985; Quinn, 1 977), but 
sufficient research exists to suggest that friendship and romantic relationships are 
distinct relational types, and should be investigated separately (Bridge & Baxter, 1 992). 
Although there is an enormous literature on formal dyadic units (e.g., supervisor­
subordinate, mentor/protege), very little research has examined the role of friendships as 
they relate to organisational effectiveness (Dillard & Fritz, 1 995). A contribution this 
thesis will make is addressing a lack in the literature of research directly focusing on 
friendships, both in a New Zealand context, and internationally. 
Because there is evidence to suggest that different work roles attract and retain people 
with corresponding values and or needs (Schneider, 1 987; Schneider, Goldstein, & 
Smith, 1 995; Schneider & Reichers, 1 983), it is not unreasonable to make some 
assumptions about role occupants. For example it is possible that employees with 
highly autonomous jobs, which require very little social interaction, will be less affected 
by friendships (or lack of friendships) in their workplace than those who work very 
closely with others in the course of their day. Similarly it is possible that people with 
different needs might be differently affected by social opportunities at work. This thesis 
will also examine whether these differences do, in fact, exist. To address this question, 
this thesis will examine whether or not the relationships between workplace friendships 
and organisational outcome variables (commitment, intention to leave, etc.) differ 
between employees from different types of jobs, and between people with different 
needs. Thus, a second focus of this project relates to the notion that the presence of 
satisfying relationships at work is likely to have more salience for some people than 
others. 
4 
Introduction 
Aristotle believed that without friends no one would choose to live. In New Zealand' s  
increasingly secular society (of primarily nuclear families), there may be  a growing 
reliance on the workplace to give a sense of belonging that people formerly drew from 
their church, community, or extended family. The day-to-day living of the populace is, 
more than ever, focused on working life. Perhaps, today, Aristotle would say instead . . .  
"Without friends no one would choose to work. " 
5 
Literature Review 
Chapter 2 :  Literature Review 
2.1 Where does this Project Sit Within Psychology? 
The focus of this project, infonnal relationships in organisations, sits within the realm of 
both social and organisational psychology. Friendship fonnation and maintenance has 
long been studied within the field of Social Psychology (e.g., Adams, 1 963 ; Baxter, 
1 990; Baxter, 1 988;  Bridge & Baxter, 1 992; Buunk & Prins, 1 998; Duck & Periman, 
1 985; Duck & Wright, 1 993; Fine, 1 986; Wright, 1 969, 1 974, 1 984). The study of 
social needs at work has been a feature of Organisational Psychology (or, historically, 
Management Studies) since Elton Mayo first associated the fulfilment of employees' 
social needs to productivity and efficiency in the workplace in the 1 940s (Mayo, 1 945). 
2.1 .1 Social Psychology: A History of Friendship Research 
Since the time of the ancient Greek and Roman philosophers, and no doubt the Eastern 
philosophers prior to that, friends have been recognised as important sources of 
affection, pleasure, companionship and support. Predating the scientific method, the 
works of Aristotle and Plato were philosophical rather than empirical, but represent 
some of the earliest conceptualisations of friendship, as distinct from other relationships 
in people's lives. For example, according to Plato, true friendship is derived from basic 
human needs and desires, such as wanting to strive toward goodness, to love and be 
loved, and to seek self-understanding (Blieszner & Adams, 1 992). Aristotle, on the 
other hand, described a multi-dimensional conceptualisation of friendship, outlining 
three functions, or characteristics of friendship; the quality of goodness (two people 
having mutual admiration for each others loyalty and justness), the quality of 
enjoyableness (a friendship that is valued for the pleasure taken in the others company), 
• and the quality of utility (a friendship that is valued for the benefits the friendship 
6 
Literature Review 
provides). Aristotle believed that perfect friendships would be characterised more by  
goodness, than by either utility or enjoyableness (Bukowski, Nappi, & Hoza, 1 987). 
Interestingly, there has been empirical research in Social Psychology that examines the 
validity of the multi-dimensional model first proposed by Aristotle. Murstein and Spitz 
( 1 974) conducted research with college women, testing Aristotle's theory. In Murstein 
and Spitz' s  study respondents were asked to rate seven target individuals2 on an 80-item 
bipolar adjective checklist. The authors examined the factor structure resulting from 
these ratings and found that the three dominant factors related closely to the three 
components of friendship proposed by Aristotle, thereby supporting his model. 
More recently Bukowski et al. ( 1 987) also assessed the validity of Aristotle' s model, 
addressing some of the limitations of Murstein and Spitz's study by researching both 
men and women. Bukowski et al. asked respondents to complete two bi-polar adjective 
rating scales, one for their best male friend and the other for their best female friend. It 
was expected that subjects' ratings of the items representing the constructs of goodness, 
utility and enjoyableness would reveal goodness to be the most central component of 
friendship (the items chosen were those that had the highest factor loadings in the 
Murstein and Spitz (1 974) data) . Like Murstein and Spitz, Bukowski et al. found that 
Aristotle' s three friendship components could be measured reliably. In addition, 
Aristotle' s notion that goodness is the most central component of friendship was 
supported, as goodness was more highly correlated with the other two scales, than they 
were with each other. This implies that a friendship characterised by goodness will also 
feature qualities of utility and enjoyableness, but friendships characterised by utility will 
not necessarily feature enjoyableness, and those that are characterised by enjoyableness 
will not necessarily feature utility. Thus both the Bukowski et al. ( 1 987) and the 
Murstein and Spitz (1 974) studies found support for Aristotle's conceptualisation of 
friendship, establishing that the three components could be measured reliably, and also 
supporting the validity of the model as a means of understanding friendships. 
2 Most admired friend, most useful friend, most enjoyable friend, best female friend, best male friend, the 
self and the ideal self. 
7 
Literature Review 
Despite the very long history of observations and commentary on friendships in the 
writings of philosophers and social histonans, friendship relationships have received 
relatively little attention from behavioural scientists compared to that given to romantic, 
family or neighbour relationships (Blieszner & Adams, 1 992). Nonetheless, looking 
back over past literature, a growing interest in the study of friendship relationships is 
apparent. Blieszner and Adams outline six trends in friendship research, which have 
occurred over the past few decades. They are: 
1 .  Expansion of the early (pre 1 970's) focus on child peer interaction to the study 
of friendship across the lifespan. 
2 .  Movement towards studying friendship as a distinct category of relationship. 
3 .  Broadening of  adult friendship research from a disciplinary (social psychology) 
to a multi-disciplinary (sociology, anthropology, management studies) focus. 
4 .  Change from studying friendship as a collection of attributes of individuals to 
studying it as a relationship. 
5. Greater recognition of the importance of studying the quality of relationships 
(i.e., intimacy and multiplexity) as well as the quantity of interaction (i .e., 
propinquity and proximity). 
6. Expansion of the range of research methods used in friendship studies (e.g., 
critical incidents, quantitative research, statistical procedures, content analysis, 
discourse analysis). 
It is possible to add one final trend to Blieszner and Adams' list: 
7. The study of friendships within different contexts, and the impact of context on 
friendship. 
The current thesis focuses on item six, in that it makes the most of the expansion of the 
range of research methods used in friendship studies; both qualitative and quantitative 
techniques will be used to collect data, and much of the data collection takes place 
online. Additionally, path analysis and structural equation modelling (SEM) (Byme, 
200 1 )  will be used to analyse data, in order to establish if the data collected supports the 
8 
Literature Review 
proposed models. This thesis also addresses item seven, plaCing the research within the 
workplace as a context, and investigating the impact aspects of the workplace may have 
on friendship relationships. 
In sum, scholars of Social Psychology have long been concerned with defining 
friendship and examining the functions of this relationship, both for individuals and 
within society as a whole (Blieszner & Adams, 1 992). Over the course of history, 
conceptions of friendship have moved from abstract philosophies about friendship, to 
the recognition of the vital role that friends play in providing the social support essential 
for coping with daily life. Recent conceptual and methodological advances in the field 
of social psychology have opened up new areas of enquiry. Path analysis and SEM, for 
example, allow researchers to draw inferences about possible antecedents and 
consequences of friendships. 
2.1 .2 Organisational Psychology: A History of Human Factors Research . 
The importance of social relationships in the workplace has been recognised by scholars 
for most of the last century. Elton Mayo (1 945) brought workplace friendships to 
attention when he wrote the first management book focusing on the. social needs of 
employees. Mayo advocated the role of emotional factors in determining employee 
behaviour, contending that the extent to which employees received social satisfaction in 
the workplace was the most powerful influence on productivity (Mayo, 1 945). In 
addition, Mayo argued that the key determinant of job satisfaction was group 
interaction, and highlighted the importance of good leadership and satisfying personal 
relations in the workplace. Moving on in time, the work of Maslow ( 1 954), developed a 
five-level hierarchy of human needs, ranging from basic physiological needs, to safety, 
'belongingness' and love, esteem and self-actualisation. Maslow's theory contends that; 
as lower level needs (such as physiological and safety needs) are satisfied, higher level 
needs (such as esteem and self-actualisation needs) emerge as motivators. Because of 
its intuitive appeal, and because no measuring instruments existed which could test it, 
Maslow's  theory was accepted for many years. However, more recent research has 
provided only partial support for Maslow's theory, and has not provided strong evidence 
that the theory predicts performance (e.g., Betz, 1 982; Betz, 1 984; Wahba & Bridwell, 
9 
Literature Review 
1 976). For example, nearly thirty years ago Wahba and Bridwell ( 1 976) reviewed 
thirteen studies testing Maslow's theory arid found only partial support for the concept 
of a need hierarchy and no support for Maslow' s gratification/activation proposition3 • 
More recently, a study by Betz (1 982) compared female managers to female 
homemakers. It was found that neither group of women ranked their needs in 
accordance with Maslow's hierarchy. 
In spite of the lack of direct empirical support, Maslow's theories can be applied 
broadly to Industrial-Organisational Psychology. For example, the opportunities to 
satisfy needs are still studied in organisational contexts, and needs have been related to 
organisational commitment (Steers, 1 977; Steers & Braunstein, 1 976), job satisfaction 
and tendency to leave (Zinovieva, ten Horn, & Roe, 1 993). In a New Zealand based 
study, Shouksmith (1 994) administered the Multidimensional Job Satisfaction Scale to a 
sample of 1 1 2 1  health professionals to ascertain which organisational factors were 
associated with commitment. Opportunity for personal growth (or self-actualisation) 
was found to be related to all three forms of commitment. Thus, the needs themselves 
have been shown to influence performance in certain jobs, and although the simple 
hierarchy that Maslow envisioned may not adequately reflect the complexity of human 
motivation, his theory still has use. 
Not long after Maslow published his Need Hierarchy theory, Herzberg, Mausner and 
Snyderman (1 959) wrote The Motivation to Work, which describes how employee 
satisfaction stems from factors such as achievement, recognition for accomplishment, 
challenging work, increased responsibility and opportunities for growth and 
development. Herzberg's ( 1 966) two-factor theory of job satisfaction was influential in 
the development of many subsequent measurement tools (Tovey & Adams, 1 999). 
Herzberg et al. postulated that satisfaction and dissatisfaction are two separate 
phenomena. Intrinsic factors, i.e., factors intrinsic to the nature and experience of doing 
work, which they found to be job satisfiers, include; achievement, recognition, work 
itself and responsibility. They named these factors 'motivators. '  Extrinsic factors, 
3 The gratification/activation proposition holds that gratification of a given need submerges it, and 
"activates" the next higher need in the hierarchy. 
10  
Literature Review 
which they found to be job 'dissatisfiers, ' include; company policy, administration, 
supervision, salary, interpersonal relations and working conditions. They named these 
'hygiene' factors. Herzberg maintained that hygiene factors counteract physical needs 
and can avoid discomfort, but cannot produce pleasure. On the other hand, the satisfiers 
of psychological needs (motivators) can produce pleasure, but their absence does not 
produce discomfort. Herzberg viewed the two dimensions of satisfaction and 
dissatisfaction as independent of each other. Consequently, it is possible to be 
simultaneously satisfied and dissatisfied with different aspects of the same job. 
There has been criticism of the Herzberg edifice; Hackman and Lawler ( 1 97 1 )  state that 
a number researchers have been unable to find support for some of the major tenets of 
the theory. For example, King (1 970) evaluated five versions of the two-factor theory, 
finding two to be entirely invalid, another to be supported only by experimenter coding 
bias and two versions to have indeterminate validity. Hulin and Smith ( 1 967) 
empirically tested the implications of the two-factor theory and found no support for the 
theory at all (finding instead that any variable in a job that makes it desirable, will make 
the job undesirable by its absence). In spite of historical criticisms, Herzberg's two­
factor theory is often used to provide a framework within which to interpret job 
satisfaction research. For example, Knoop (1 994) examined the relationship between 
the satisfaction and work values of Canadian teachers, finding support for Herzberg' s  
theory. Adigun and Stephenson (1 992) also found support for Herzberg' s  theory with a 
British, but not a Nigerian, sample of employees. Fumham Forde, and Ferrari ( 1 999) 
used Herzberg' s  two-factors to provide a framework within which to examine 
personality and work performance. Maidani ( 1 99 1 )  used Herzberg's classification 
scheme to compare employees working in the public with those working in the private 
sector. Maidani found further support for Herzberg's two factors with employees' 
motives for work emphasising intrinsic / motivator factors of employment, and the 
satisfaction of employees not being attributable to hygiene factors. Finally, Yamashita 
( 1 995) investigated the job satisfaction of nurses in a large, acute-care hospital in 
southern Japan and also found support for Herzberg's  theory, with the diss�tisfaction of 
nurses being related to extrinsic (hygiene) aspects of work such as having few 
opportunities for promotion or less favourable working conditions. In his review of 
1 1  
Literature Review 
influential Management books, Bedeian (2001 )  states that it was The Motivation to 
Work which " . . .  set the stage for the ensuing job redesign movement and quality-of­
working life revolution" (p. 225). It is the progression of this working-life revolution 
that has set the scene for the current study. 
Another early theorist whose work is relevant to the current study is McClelland ( 1 96 1 ), 
who initiated Achievement-Motivation theory. Central to McClelland's  theory are 
needs for achievement, power and affiliation. Medcof ( 1 995, p. 1 94) cites the 
Personality Research Form Manual, compiled by Jackson ( 1 989) which describes a 
person with a high need for achievement as follows: "Aspires to accomplish difficult 
tasks; maintains high standards and is willing to work toward distant goals; responds 
positively to competition; willing to put forth effort to attain excellence. " The 
Personality Research Form Manual describes a person with a high need for power 
(dominance) as follows: "Attempts to control environment, and to influence or direct 
other people; expresses opinions forcefully; enjoys the role of leader and may assume it 
spontaneously." and describes a person with a high need for affiliation as follows: 
"Enjoys being with friends and people in general; accepts people readily; makes efforts 
to win friendships and maintain associations with people." 
The research on needs, though historically focused on managerial and organisational 
success, does highlight the importance of the fulfilment of social needs at work. We 
live in a society in which people derive much of their identity from their occupation and 
where there is an increasing reliance on the workplace to fulfil social needs. Yet, 
despite the prevalence of friendships at work, and despite the historical acceptance of 
the influence of social factors in organisations, there has been surprisingly little 
empirical research or discussion in New Zealand, or overseas, focusing specifically on 
friendships in the workplace. 
2.1.3 The Current Research 
Several organisational researchers in the past decade have commented on the dearth of 
literature in the area of workplace friendships. For example, Nielsen, Jex and Adams 
(2000) state that the area of workplace friendship has remained relatively under-
1 2  
Literature Review 
examined, both empirically and theoretically. Similarly, Zom ( 1 995) states that 
workplace friendship is a phenomenon that is almost never studied; Winstead, Derlega, 
Montgomery and Pilkington (1 995) comment that there is relatively little research on 
the relationship between friendship and job satisfaction; and Fritz ( 1 997) states that 
within the organisational literature there remains a large gap regarding the comparative 
attributes of men's  and women's peer relationships (although this has been studied 
extensively in non-work contexts). More recently, Raabe and Beehr (2003) also state 
that there is little research on co-worker (peer) influences at work, compared to the 
amount which has focused on leaders and supervisors. 
One of the few studies which has directly examined workplace friendships was by 
Riordan and Griffeth ( 1 995) (refer section 2 .3 .3  for a full description of this study). 
These authors also comment on the gaps in the literature, stating that, " . . .  the 
interpersonal job dimensions of dealing with others and friendship opportunities have 
received only cursory, if any, examination in many of the existing job characteristic 
studies." They go on to comment that the " . .  . lack of research evidence is surprising 
since improved interpersonal relationships may influence a variety of performance and 
attitudinal outcomes" (Riordan & Griffeth, 1 995, p. 1 4 1 ). 
In order to begin examining the impact of friendships at work, "Friendship Opportunity" 
as a dimension of perceived job characteristics was developed by Hackman and Lawler 
( 1 97 1 )  to examine the degree to which a job allowed employees to talk to one another 
and establish informal relationships at work. Although, historically, there has been a 
great deal of research on the effect of social support in the workplace (e.g., Chapman, 
1 993 ; Gant et aI. ,  1 993 ; LaRocco & Jones, 1 978;  Loscocco & Spitze, 1 990; Vinokur, 
Schul, & Caplan, 1 987; Williams, Ware, & Donald, 1 98 1 )  relatively little work has 
attempted to specify what the attitudinal consequences are of perceived friendship 
opportunities within the job. Social support can be distinguished from friendship in that 
social support can be obtained from many sources in an employee's life; in addition 
relationships with individuals from whom one simply receives social support will likely 
lack the intimacy and sense of uniqueness associated with close friendships. Although 
most of the research on social support suggests that support in an individual's work 
1 3  
Literature Review 
environment is related to the achievement of vanous organisationally desirable 
outcomes such as stress reduction, satisfaction, self-esteem and retention; friendship 
opportunities, as a variable has seldom been examined. The current study will go some 
way towards addressing this shortfalL 
As well as addressing the lack in the literature, of research focusing on friendships in 
the workplace, another focus of this project is turnover. Turnover represents one of the 
most important issues for any organisation. The money and time invested in hiring and 
training an individual who leaves the organisation is lost forever. These costs are 
considerable, and increase further up the organisational hierarchy, i .e., replacing a 
senior manager represents a more significant cost than replacing a more junior member 
of staff (Richer, Blanchard, & Vallerand, 2002). Thus, in addition to workplace 
friendships, employee turnover also deserves the scientific attention it receives in the 
current study. 
2.2 What is it to be a Friend? 
Friendships are voluntary relationships that exist primarily for enjoyment and 
satisfaction, rather than for the fulfilment of a particular function or role (Sapadin, 
1 988). Adams ( 1 994) states that; " . .  . in contrast to other forms of intimate relationships 
in our society, friendship is uniquely voluntary" (p. 163) .  Blood and legal ties designate 
relatives, colleagues are designated by one's employer or organisation, neighbours are 
designated by proximity but friends, particularly adult friends, are selected. There is no 
formal ceremony to mark the beginning or the end of a friendship, as there is for other 
relationships (such as marriage), and even definitions of friendship are subjectively 
variable. This is a complicating factor in the study of friendship, as one person may 
consider their hairdresser with whom they occasionally socialise a friend; while another 
may use the term only for people with whom they are very intimate (Sapadin, 1 988). 
Thus, it is important to first define friendships for the purposes of this study, and to 
distinguish friendship relationships from other relationships that may exist in an 
individual's life. 
14  
2.2.1 Friends Compared to Other Relationships 
Friendship versus love relationships 
Literature Review 
Rubin (1 970) conducted research to differentiate between romantic love and liking of 
friends; his results indicated that liking and loving are two distinct constructs, associated 
with different behaviours and emotions. More recent evidence that love relationships 
are distinct from friendships comes from a survey of German students (Lamm & 
Wiesmann, 1 997). Respondents were asked to write down how they could tell if they 
liked someone, loved someone, or were in love with someone. The authors found that 
the most distinctive characteristic of "liking" was wanting to interact socially with the 
other; the most distinctive characteristic of "love" was trusting the other; and the most 
distinctive characteristic of being "in love" was arousal. Both liking and love were 
associated with a positive mood in the other' s  presence. Bridge and Baxter (1 992) also 
state that sufficient research exists to suggest that friendship and love relationships are 
distinct relational types. The current study will focus on friendships rather than love . 
relationships. 
Formal versus informal relationships 
The workplace relationships which have received by far the most research attention are 
those of superior-subordinate (or supervisor-supervisee) (e.g., Allen, 1 995 ;  Gant et al. ,  
1 993; Largent, 1 987; Murphy & Ensher, 1 999; Sias & Jablin, 1 995;  Vecchio & Bullis, 
2001 ; Wesolowski & Mossholder, 1 997) and mentor-protege (e.g., Beans, 1 999; Ensher 
& Murphy, 1 997; Riggins & Kram, 200 1 ;  Kram & Isabella, 1 985; Kram, 1 983). These 
relationships lack the voluntary aspect of friendships, as they are prescribed by the 
organisation; they are an example of formal relationships and are therefore not a direct 
focus of the current study. This is not to say, however that a true friendship cannot 
evolve from an existing formal organisational relationship (refer section 2.3 . 8  on 
blended relationships). This study focuses on informal organisational relationships; 
those not prescribed by the organisation. 
1 5  
Literature Review 
Friendship versus acquaintanceship 
Interpersonal relationship literature, both empirical and theoretical, suggests that people 
distinguish between friendship and acquaintance relationships, and that different rules 
govern people's interaction in the two relationships. 
"Friendship groups are defined as groups with close interpersonal ties 
and positive, amiable, pre-existing relationships between members. 
Acquaintance groups are defined as groups with limited familiarity 
and contact among members. Members may know each other through 
casual encounters, however. . .  a strong bond is lacking" (Jebn, 1 997, 
p. 776). 
Similarly "friendly relations", which are characterised by social/personal interactions, 
lack the intimacy, sense of uniqueness, strength of affective bond, and felt obligations 
associated with personal friendship. (Bridge & Baxter, 1 992)�. 
In contrast to the acquaintance relationship; Wright (1 984) states that close friendships; 
" . . .  involve shared interests and activities, various kinds of intimacy, 
including self-disclosure and the sharing of confidences, emotional 
support, small talk, shop talk and the exchange of tangible favours" 
(Wright, 1 988, p. 370) 
Kenny (1 994) defined friendship as a " . . .  mutual, voluntary relationship from which 
members expect intimacy, companionship, and responsiveness" (p. 1 024). 
Thus friendship, by these definitions, is voluntary and reciprocal. It is a relationship 
which is seen as unique and special by the participants, and which enhances their lives 
(i.e., is rewarding). Further, a definition by Wright (1 974) incorporates the context in 
which friendships occur. He defines friendship as a relationship involving voluntary 
interaction, in which " . . .  the commitment of the individuals to one another usually takes 
precedence over their commitment to the contexts in which the interaction takes place" 
1 6  
Literature Review 
(p.94). This is important when conceptualising workplace relationships (section 2 .3)  as 
it implies that the boundaries of genuine friendships supersede the role boundaries that 
may exist in a particular context. Wright's definition relates closely to the definition 
given in the section on workplace relationships (2.3 . 1 )  of the "special peer" in which 
formal workplace roles are ignored or downplayed. 
2.2.2 The Nature of Friendship 
The structure of friendship 
Blieszner and Adams ( 1 992) outline aspects of the structure of friend pairs (dyads), 
describing aspects of the structure of friendships. The structure of friendships include 
the power hierarchy (the probability that one member of the dyad will be able to "get 
their own way" if disagreement were to arise, it is a measure of the influence one 
individual has over another), the status hierarchy (the distance between members in 
terms of prestige or moral worth), solidarity (the degree of intimacy) and homogeneity 
(the similarity of participants in terms of factors such as gender, ethnicity, occupational 
status, age, etc) (Adams & Blieszner, 1 994; Blieszner & Adams, 1 992). It is likely that 
all of these aspects of friendships will be relevant in organisational contexts. Status and 
power particularly are very salient in many workplaces because work roles are usually 
prescribed by the organisation and clearly defined status hierarchies often exist. 
Phases of friendship 
Relationships are dynamic, and friendships, like other relationships, will change over 
time; have beginnings when partners become acquainted, middles (when other features 
increase/decrease) and sometimes endings. Movement from one phase to another might 
be deliberate or might occur by chance (Adams & Blieszner, 1 994; Blieszner & Adams, 
1 992). Adams and Blieszner (1 994) outline the three phases of friendship on the basis 
of how they evolve over time: 
Formation involves movement from stranger to acquaintance to friend. It involves 
identification of, and attraction to, a potential friend and then getting to know that 
person. Social factors which have been found to influence attraction include: proximity, 
1 7  
Literature Review 
adversity and prevailing mood, these circumstances facilitate the communication of 
sources of attraction such as similarity, familiarity and physical attractiveness (Carr, 
2003). There are behavioural changes as partners .move from acquaintanceship to 
friendship. For example, as friendships become closer they tend to involve more 
interactions over a broader range of activities. More intimate friendships become 
increasingly dependent on affection rather than on frequency of contact. 
Maintenance is the most variable phase in terms of processes. Friends may evaluate the 
relationship and decide to retain it at its current level, make it more or less intense or 
change the types of activities they and their friend engage in. Some individuals might 
consciously employ friendship development strategies in this phase. 
Dissolution is when a friendship ends. Some friendships may go on for decades, with 
the assumption of indefinite existence, some end abruptly as a result of a disagreement 
or death, others simply wither away through 'benign neglect' (Adams & Blieszner, 
1 994). 
Thus, the types of interactions and the dynamics within friendships can change a great 
deal over time. Outside influences might impact on friendship processes, and the phase 
a given friendship is in might also influence the way a particular dyad might interact. 
Processes in friendships 
Friendship processes include both the covert cognitive and affective responses and overt 
behavioural events that occur when people interact (Adams & Blieszner, 1 994; 
Blieszner & Adams, 1 992). Cognitive processes reflect the internal thoughts of each 
partner (about herself, her partner, the friendship etc). These include interpretations of 
behaviour, assessments of the stability of the friendship, and evaluations of the 
attractiveness, value, character and similarity to oneself of the other person. Affective 
processes are the emotional reactions to friends; these may be positive (affection, trust, 
loyalty, commitment, joy) or negative (anger, hostility, jealousy). Behavioural 
processes are the action components of friendships and include 
communication/disclosure, displays of affection, resource exchange, co-operation and 
1 8  
Literature Review 
the sharing of activities. On a more negative level they might include concealment, 
manipulation, conflict and competition. 
The three processes interact with each other. Cognitive processes result in affective 
reactions that, in turn, influence future actions. Behaviours can affect thoughts and 
emotions and so on. People can express their thoughts, feelings and actions so that their 
friends are aware of them or they may keep them hidden (Adams & Blieszner, 1 994). 
Thus it is possible that a change in any one of these processes will impact on the other 
two. 
In relation to the current study, it is necessary to keep the structures, phases and 
processes of friendships in mind when interpreting findings. This is particularly 
important when investigating the way that friendships might influence people's 
experience of work" as the aspects of friendships described above may influence the 
way measured variables relate to one another. For example, a relationship between two 
individuals in a very interdependent work environment organisation may be more 
difficult to manage than one between peers who work together only occasionally, and 
will probably require different types of communication and behaviour processes. 
2.2.3 Gender Differences in Friendship 
While aspects of friendship relationships for men and women have been found to be 
similar in many respects (Wright, 1 988) some gender-based differences have emerged. 
Most research has pointed to the following broad differences between men and women. 
Women's friendships are described as 'communal' or ' face to face.' They tend to 
involve more self-disclosure, emotional supportiveness, extensiveness and complexity 
than those of men. In addition, women's friendships tend to stress reciprocity while 
men's  stress commonality. Men's friendships are described as ' instrumental' or 
' agentic' and ' side by side.' They tend to be organised around shared interests and 
activities and be action-oriented rather than person oriented (Markiewicz, Devine, & 
Kausilas, 2000; Winstead, 1 986; Wright, 1 988, 1 99 1 ). Literature with a focus on 
interpersonal relationships indicates men achieve and define closeness through the 
1 9  
Literature Review 
sharing of activities, while women define and achieve closeness through the sharing of 
feelings and emotions (Odden & Sias, 1 997; Wood & Inman, 1 993). 
These trends have also been found in a New Zealand context. Aukett, Ritchie and Mill 
( 1 988) examined the same-sex and opposite-sex friendship patterns of 66 men and 152 
women students at the University ofWaikato (N.Z.). Resembling findings in America 
by Winstead (1 995), Aukett et al. found that women in New Zealand also tend to have 
fewer, more intimate friends, which they place more importance on, than do men. In 
addition they also found that men tended to emphasise shared activities while women 
emphasised talking and emotional sharing. Both men and women were found to derive 
more emotional support and therapeutic value from their relationships with women. A 
possible implication of this, in a workplace context, is that friendships may have more 
salience in work environments that are predominantly female. 
In sum, research on friendships outside the workplace indicates that friendships 
involving at least one female are more satisfying for most people. Generally both males 
and females report greater satisfaction and more instrumental and emotional rewards 
from female friends than from male friends (Aukett et al., 1 988; Sapadin, 1 988;  
Veniegas & Peplau, 1 997), possibly as a result of women's greater comfort with 
intimacy and their emphasis on successful relationships as part of their self-concept 
(Markiewicz et al. ,  2000). 
It is worth noting here however, that generalising literature on gender difference in 
friendships to specific individuals should be done with caution. While findings of 
gender differences in friendship are very robust, exceptions are certainly found. 
Findings that on average relationships with women are more satisfying does not mean 
that every friendship with a woman is found to be satisfying. In addition there are 
certainly more similarities across genders than there are differences between genders in 
terms of friendships. Virtually all close friendships involve shared interests and 
activities, intimacy, emotional support, small talk and the exchange of tangible favours, 
regardless of the gender of the participants (Wright, 1 988). The often quoted finding 
that women self-disclose more than men in relationships is an example of the way 
20 
Literature Review 
gender differences have been somewhat exaggerated. Wright ( 1 988) states that the bulk 
of evidence indicates that most men, as well as most women, self-disclose with relative 
ease. It just so happens that, on average, women do so more easily than men. 
2.2.4 Why Do People Engage in Friendships? 
Two theories, which attempt to explain why people engage in friendships, are: (a) social 
exchange and (b) the intrinsic quality of friendship. Both are outlined below. 
Social exchange 
Social exchange theory proposes that whether we like somebody, or want to engage in a 
friendship with him or her, is determined by the cost-reward ratio .  People will evaluate 
the cost to themselves (e.g., time, money, effort) to get a positive reward (e.g. ,  
satisfaction, pleasure, support) from a particular person. Consequently social interaction 
entails both rewards and costs (Rook, 1 984). Social exchange may be thought of as a 
give-and-take relationship between people. According to this theory, " . . .  we choose our 
social ties on the basis of their capacity to provide rewards relative to costs and as a 
function of the alternatives available" (Rook, 1 984, p. 1 098). Social exchange theory 
holds that people enter into relationships because of the rewards (benefits, fulfilled 
needs and other "profits") that such relations are expected to bring. Although the notion 
that relationships are cemented according to "economic" considerations is somewhat 
distasteful, it is generally true that people' s behaviour in relationships is shaped by the 
expectation of rewards, even if the reward is no more than an expression of gratitude. 
When such an exchange occurs, positive feelings develop (Anderson & Hunsaker, 
1 985). 
The investment model (Rusbult, 1 980) also conceptualises relationships to involve costs 
and rewards; this model attempts to explain why people stay in relationships (i.e., an 
individual' s  commitment) by describing three processes (satisfaction, alternatives and 
investment) . The investment model is usually applied to romantic relationships but it is 
possible to conceptualise the three processes as being relevant to friendships also. 
Impett, Beals and Peplau (2001 )  state that individuals are generally satisfied when 
relationships provide high rewards and low costs. Rewards are things obtained from the 
2 1  
Literature Review 
relationship that an individual enjoys, such as social support; while costs are aspects of 
the relationship that are unpleasant, such as conflict or financial burdens. People will 
generally be more committed to a relationship that creates satisfaction (i.e., one that has 
low costs and high rewards) (Bui, Peplau, & Hill, 1 996; Rusbult, 1 980). 
A second predictor of commitment in the investment model is the quality of 
alternatives. Alternatives refer to an individual's assessment of the rewards and costs 
that could be obtained outside the current relationship. Third, commitment is affected 
by investments of resources such as time, effort or money that an individual has 
contributed to the relationship and would lose if the relationship were to end. Thus, 
according to the investment model, individuals who are highly satisfied, have invested a 
great deal, and perceive few appealing alternatives will be highly committed to their 
relationships. 
Most exchange theorists assume that relationships are more satisfying and stable when 
reciprocity is perceived and when the rewards for each partner are relatively equal 
(Buunk, Doosje, Liesbeth, Jans, & Hopstaken, 1 993; Buunk & Prins, 1 998). Equity 
theorists 4 argue that being both over benefited or under benefited in a relationship may 
cause negative feelings (Buunk & Prins, 1 998). For example, someone may feel 
indebted to a friend if they receive more rewards than they are willing or able to 
reciprocate, or when they receive a more favourable rate of outcomes than the friend. 
This is an aversive state because of feelings of obligation, guilt and shame (Buunk et aI. ,  
1 993). On the other hand, feeling that the costs of a relationship outweigh the rewards 
is also accompanied by negative emotions such as a feeling of unfairness and 
resentment towards the other person. Thus, according to exchange theorists, the best 
friendships are those that are perceived to be reciprocal and of equal benefit to both 
individuals. In relation to the current study, social exchange may be more salient in an 
organisation than in other contexts; salient rewards and benefits such as positive 
reviews, promotion and "perks" might be obtainable from workplace relationships and 
may be a reason people seek to form alliances and friendships at work. 
4 Equity theory is a special case of social exchange theory that defmes a relationship as equitable when 
the ratio of profit to contribution is perceived to be the same by each partner (Vaughan & Hogg, 1 998). 
22 
Literature Review 
The intrinsic quality of friendship 
While there is no doubt that friendships are fonned and maintained because they are 
rewarding (people are often able to clearly state what they "get from" particular 
friends), friendships that develop beyond superficial levels often have an intrinsic "end 
in themselves" quality. Each person perceives the other as being unique and 
irreplaceable. These friendships involve individuals whose participation transcends the 
importance of any easily specified set of rewards (Clark & Mills, 1 979; Wright, 1 984) . 
Thus friendships can become self-sustaining, inasmuch as they may be maintained even 
when the identifiable rewards that may have led to their development are no longer 
available. People do not stay in close friendships simply because they find that person 
or that friendship rewarding. Exchange theories are restricted to observations of 
rewards people obtain from friendships, this means-end thinking precludes regarding 
friendship as an intrinsic relationship (Clark & Mills, 1 979; Wright, 1 984). So, 
although social exchange does indeed take place in friendships, it does not fully explain 
them. 
Wright ( 1 984) presents a perspective that attempts to reconcile the intrinsic quality of 
friendship with the seemingly essential presence of some fonn of rewardingness. He 
states that it is possible to look at friendship as a communal relationship. A communal 
relationship is one where members are genuinely concerned about the welfare of each 
other. People are motivated towards benefiting the other because of "equality of affect." 
This is the idea that members in a dyad experience similar affect, if one feels good or 
bad, both do. Thus, it is in the best interests of both individuals to please the other 
(Clark & Mills, 1 979; Wright, 1 984). 
Conceptualising friendships as communal contrasts with the conceptualisation of 
friendships solely as exchange relationships, in which members assume that a benefit is 
given with the expectation of receiving benefit in return. In fact, communal 
relationships tend to be more rewarding if exchange is not in evidence (Wright, 1 984). 
Thus, although people often receive tangible rewards from their friends, the communal 
23 
Literature Review 
aspect of these relationships is one thing th'lt differenti'!tes "friends" from other "useful" 
people in a person's life. Many people probably make an effort, at least initially, to 
make friends at work because of the tangible benefits and perks that they may be able to 
obtain; but if geriuine", communal friendships form in a workplace, the rewards may be 
even greater. It is probably the more communal relationships that will affect 
employees' satisfaction with their jobs and commitment to their organisation. 
2.3 Relationships in a Workplace Context 
Because friendships have been found to commence and evolve in a similar manner 
across contexts, there is no reason to believe that literature on friendship in general 
cannot be applied to friendships at work. For example Fritz (1 997) conducted a study 
comparing men's and women's peer relationships in organisations, surveying 666 
organisational employees resident in a Midwestern (U.S.A.) metropolitan area. She 
found that women's  organisati<?nal relationships have the potential to be stronger and, 
among very close friends, are characterised by greater intimacy than those of men, 
which reflects findings in non-work contexts (e.g., Winstead, 1 986). In addition men's 
workplace relationships were found to be characterised by more mutual dependence and 
involved more activities than women's; also reflecting findings from non-work contexts 
(e.g., Aukett et aI., 1 988; Wright, 1 988). Fritz also found that women's relationships are 
similar to men's in terms of instrumentality, which supports a finding by Wright and 
Scanlon ( 1 99 1 )  that women's  relationships are instrumental as well as expressive. She 
states that; " . . .  the work setting is, for friendship purposes, similar to non-work 
contexts" (p.27) and that, "" . general characteristics of men's and women's 
relationships outside the organizational context may transfer to the organizational 
setting" (p. 44). In addition, the reasons people make friends (e.g., social exchange, 
intrinsic rewards) and the characteristics that are looked for in a potential friend (e.g., 
proximity, similarity, working towards a common goal) are as salient in an organisation 
as in any other setting (Fine, 1 986). 
24 
Literature Review 
2.3.1 Types of Organisational Peer Relationships 
This project will mainly focus on friendships in organisations, but to do this, friends 
must be distinguished from other organisational relationships. Kram and Isabella 
( 1 985) report three organisational peer types (information peer, collegial peer and 
special peer), differing in level of closeness. Research on these peer types has shown 
them to be both empirically discriminable and conceptually meaningful (Fritz, 1 997). 
In the current study, the three peer types identified by Kram and Isabella ( 1 985) were 
used to distinguish between categories of colleagues at work. The peer types were 
defined for the respondents, along with a fourth category termed "negative 
relationships" (refer appendix 3), and referred to in the questionnaires respondents 
completed. The four peer types, including negative relationships are defined as follows: 
The information peer: This relationship is characterised by low levels of self-disclosure 
and trust, little emotional support and little personal feedback. It is the most common of 
workplace relationships. The functions are to provide information about the job, task, 
organisation etc. It is not a friendship relationship, and relates most closely to Jehn's  
( 1997) definition of  acquaintance relationships (section 2 .2 . 1 ). 
The collegial peer: This relationship is characterised by moderate levels of trust and 
self-disclosure, more complex individual roles within the relationship and wider 
information boundaries. Information sharing occurs in addition to emotional support, 
feedback, and friendship. 
The special peer: This is the most intimate of peer relationships; formal workplace roles 
are ignored or downplayed, in favour of high levels of self-disclosure and self­
expression. It is characterised by intimacy (a measure of the strength of social ties), 
stability (a measure of how strongly established the relationship is) and continuity 
(whether the relationship is ongoing, without breaks). Special peer relationships take 
longer to develop than either collegial or information peer relationships and are 
relatively scarce in organisations, with many people having none, or only one, special 
peer at work. The special peer relationship relates most closely to the definitions of 
friendship given above (section 2.2. 1 ). 
25 
Literature Review 
The negative relationship: This is not a friendship relationship.  Interaction between 
individuals with a negative relationship is characterised by disrespect, disagreement, 
dislike, conflict and/or animosity. Individuals may interact with each other on a fairly 
regular basis but would definitely not continue the relationship if they did not work 
together. 
Most employees have the option of establishing relationships with their co-workers, but 
while some workplace relationships do develop into intimacy, many do not. "Friendly 
relations"S often develop out of formal role relationships, and may be the preliminary 
stage in the development of more intense friendships (Fine, 1 986). Fine states that this 
progression is especially characteristic within organisations, where individuals begin by 
interacting with each other because they have to, according to their organisational roles, 
and then, making a virtue of necessity, turn these exchanges into something more 
informal and pleasurable. 
Kram and Isabella (1 985) maintain that co-worker friendships are a valuable means of 
growth and support. They also argue that the characteristic of reciprocity sets them 
somewhat apart from other work relationships, such as mentoring or superior­
subordinate relationships. This ties in with the definitions given previously (section 
2.2. 1 ), which describe friendship relationships as voluntary, reciprocal and equal; while 
) 
the more formally defined organisational relationships (superior-s-ubordinate, mentor-
protege, etc.) will likely lack these characteristics. Buunk et al . (1 993) state that 
reciprocity is a particularly important issue in work relationships because the perception 
of receiving more help than one can return can result in negative feelings (such as guilt, 
indebtedness and inadequacy), as can the perception of giving more. When the 
relationship is between status-unequal friends (e.g., superior and subordinate), a degree 
of inequality might be . considered normal, because providing help and support is 
expected from the superior. In consequence, feeling over-benefited will be common in 
relationships with superiors. This situation will be less acceptable in relationships with 
SFriendly relations are characterised by social and personal interactions, but they lack the intimacy, sense 
of uniqueness, strength of affective bond and felt obligations associated with personal friendship (Bridge 
& Baxter, 1992). 
26 
Literature Review 
colleagues, who will probably aim at reciprocity in order to avoid either individual 
feeling indebted (Buunk et aI. ,  1 993). 
Tbarra ( 1 993) discusses characteristics of relationships at work in terms o f  "tie strength ."  
The strength of a tie is a function of the amount of time, the emotional intensity, 
intimacy and reciprocity in the relationship. A related concept is "multiplexity," which 
refers to the number of dimensions in the relationship, a multiplex relationship is one 
that involves the exchange of multiple resources; for example one where an individual 
might receive job related support, and a friendship, and practical career support. The 
more dimensions that link one individual to another, the stronger the relationship will 
be. 
Thus an individual will have a stronger tie to a special peer than a collegial peer, and the 
link is likely to be mUltiplex. Although strong ties may be useful for providing 
assistance in stressful situations at work and provide emotional and tangible support to 
individuals, Ibarra (1 993) states that weak ties can have significant instrumental 
benefits. The weaker ties tend to be the channel through which socially distant ideas 
and information reach an individual, as a weak tie is likely to come from a part of the 
organisation that is otherwise disconnected from the individual. Strong ties, on the 
other hand, bond similar people, and similar people tend to be interconnected. Thus 
information obtained through strong ties is more likely to be redundant (Ibarra, 1 993). 
In addition, different organisational peer types have been shown to vary in terms of the 
type' of communication they engage in. Myers, Knox, Pawlowski and Ropog ( 1 999) 
conducted a study in Southern U.S.A. in which 1 3 8  members of various organisations 
completed the Communication Openness Measure and the Communication Functions 
Questionnaire regarding either an information, collegial, or special peer. Results 
indicated that information peers are less open, and use less functional communication 
skills, than either collegial or special peers. Thus peers within organisations can differ 
in terms of their closeness, strength, multiplexity and communication. 
27 
Literature Review 
Collegial and special peer relationships are the primary focus of this thesis, as these are 
the relationships that people tend to think of as "friends." Because collegial and special 
peer relationships are more open and intimate than information peer relationships, and 
are associated with mQre tangible support, they are the relationships that are most likely 
to affect an employee' s  experience of work. 
Friendships compared to other organisational relationships 
Peer relationships at work are considered to be one of the primary means by which 
organisational socialisation takes place (Louis, Posner, & Powell, 1 983). Previous 
research has looked at romantic relationships between work-mates (e.g., Anderson & 
Hunsaker, 1 985; Quinn, 1 977; Schultz, 2003) but, as stated in section 2.2. 1 ,  sufficient 
research exists to suggest that friendship and romantic relationships are distinct 
relational types. 
Another organisational relationship that has received research attention is the mentor­
protege relationship. While there is little doubt that mentor relationships are valuable to 
organisational members, peer relationships are both more common and more long 
lasting. Kram ( 1 983) conducted an intensive biographical interview study of eighteen 
corporate relationships, investigating the phases of mentor-protege relationships and, as 
well as examining the utility and benefits to managers, discusses the limitations of these 
relationships. She claims that the mentor relationship is limited in value and duration 
and, in addition, is not available to all individuals in the early stage of their careers. 
Peer relationships, on the other hand, offer a valuable alternative to the mentor 
relationships offered by organisational superiors. Peer relationships can provide career 
and psychosocial functions; they offer the opportunity for greater mutuality and sense of 
equality, and they are more available in numbers (Kram, 1 983). Another difference 
between mentoring and peer relationships is that, in conventional, stereotypical 
mentoring relationships, there are significant differences in age and in hierarchical 
levels, while in peer relationships at least one of these characteristics is usually the same 
for both individuals. In addition, mentoring relationships involve a one-way helping 
dynamic while peer relationships are more likely to involve a two-way exchange (Kram 
& Isabella, 1 985). Thus it seems that peer relationships have long been undervalued 
28 
Literature Review 
and may be as (if not more) useful than traditional mentor relationships to individuals in 
organisations. 
There is an increasing' literature focusing on "peer mentoring" (e.g. , Allen, McManus, & 
Russell, 1 999; AlIen, Russell, & Maetzke, 1 997; Aston & Molassiotis, 2003 ;  Colling, 
Grabo, Rowe, & Straneva, 1 998; Ensher, Thomas, & Murphy, 200 1 ;  Jacelon, Zucker, 
Staccarini, & Henneman, 2003; McDougall & Beattie, 1 997) which builds on the ideas 
proposed by Kram ( 1 983) and Kram and Isabella ( 1 985). Peer mentoring is based on 
the principle of mutual involvement for learning and development between two peers 
(McDougall & Beattie, 1 997). Most authors writing about peer mentoring contend that 
these relationships provide a valuable source of learning for individuals in 
organisations. Although peer mentoring relationships are, perhaps, more likely to 
evolve into friendships than mentor-protege relationships between superiors arid 
subordinates, they are an aspect of the formal, rather than informal, structure ' of an 
organisation in that they are generally prescribed by the organisation (e.g., Allen et aI. ,  
1 997; Aston & Molassiotis, 2003) rather than being selected by individuals themselves. 
Thus, although peer mentoring relationships may be reciprocal and equal, they will 
likely lack the essential voluntary aspect of true friendships (refer section 2 .2 . 1 ). 
2.3.2 The Informal Structure of Organisations 
The friendships and negative relationships employees have at work will have an impact 
beyond each individual's  own sphere of experience at work. Taken together, the 
friendships, informal organisational relationships and unofficial power blocs in an 
organisation comprise the informal structure of that organisation. The type and quality 
of an individual' s  relationships at work will certainly alter his or her experience of 
work, while the informal structure of an organisation is likely to impact the functioning 
of the organisation itself. Ringer and Robinson ( 1 996) outline a layered systems model 
of organisational functioning, describing the organisation as having six different levels 
of functioning (the process level, formal structure, informal structure, interacting self, 
inner self and archetypal role). Friendships, informal organisational relationships and 
power blocs are an aspect of the informal structure. More precisely the informal 
structure involves: 
29 
• the unspoken unofficial power structure of the organisation; 
Literature Review 
• the networks of employees who subscribe to particular models or ideologies; 
• subgroups that fO ml unofficial power blocs; 
• semi-structured "secret societies" within an organisation; 
• relationships that unofficially carry influence in the organisation; 
• covert alliances, subgroups based on gender, culture, age etc. ;  
• informal use of databases and software; and 
• social groups and relationships (including friendships) (Ringer & Robinson, 
1 996). 
The informal structure is the network of relationships not described in job descriptions 
or role statements. It is the "unofficial organisation" that is implied by the saying "it is 
not what you know but who you know" and by the grapevine as a conduit of 
information flow in an organisation. If individuals bypass formal channels of command 
I or procedure, they will probably be utilising the informal structure. Ringer and 
Robinson ( 1 996) state that the informal structure is a complex, usually hidden, web of 
relationships and subgroups of people that interacts with, but is different from, the 
formal structure. In the informal structure there may be strong relationships that span 
horizontal and vertical boundaries. For example, informal structures are seen in action 
when one person tries to influence. another' s  work related behaviours in a non-work 
setting. The informal structure is not inherently negative or positive, in some 
organisations it may allow an organisation to function when the formal structure is not 
adequate, while in others the informal structure may undermine the formal. If the 
informal structures are analysed it may be possible to identify patterns and areas of 
influence that can be incorporated into the formal -structure. The presence of the 
informal structure of an organisation may be ignored, but is no doubt of great 
importance, both to an individual's experience of work and to the functiohing of the 
organisation itself. This further highlights the potential impact that workplace 
friendships may have in organisations. 
30 
Literature Review 
2.3.3 The Impact of Friendships in the Workplace 
The workplace context is quite unique, so generalising friendship literature from 
contexts other than work should be done with caution. Nonetheless, findings that 
friendships are initiated and develop in a similar manner across contexts (e.g., Fine, 
1986; Fritz, 1 997) provide some rationale for applying non-organisational literature on 
friendship to a work environment. It should be acknowledged, however that there are 
likely to be characteristics of organisations not present in other contexts (e.g., type of 
work, opportunity to interact, interdependent roles) that may influence the way 
relationships are formed and maintained, as well as influencing relationship 
characteristics such as intimacy and communication (Fritz, 1 997). 
The importance of studying friendships at work 
Although there are some authors who hold that friendships at work are to be avoided 
(e.g., Eisenberg, 1 994)6, the few empirical studies that exist, highlight the positive 
outcomes of these relationships (e.g., Richer et aI. ,  2002; Riordan & Griffeth, 1 995) .  
Informal social relations may offer significant and rewarding benefits to individuals at 
work. Workplace friendships can provide increased communication (Kramer, 1 996), 
support (Buunk et aI. ,  1 993), trust, respect, co-operation, growth, development, energy 
and security that, in turn, influence work related attitudes and behaviours (Foote, 1985 ;  
Krackhardt & Stem, 1988;  Riordan & Griffeth, 1 995). Thus, friendships developed 
within the workplace represent a key element in the informal structure of an 
organisation. Friendships are potentially powerful structural units that can either hinder 
or facilitate organisational effectiveness. Conversely, the formal structure of an 
organisation can contribute to the development of friendships in the work environment 
through the grouping of units and departments (inasmuch as close proximity between 
individuals leads to the opportunity for friendship) (Riordan & Griffeth; 1 995). Thus 
workplace friendships are likely to both affect, and be affected by, the workplace. 
6 Eisenberg ( 1994) states that, "Any number of serious problems, some equivalent to those in illegal 
sexual harassment, can arise in an organization when a relationship between co-workers takes on a greater 
intimacy than a nonnal working partnership. One problem has to do with morale among other employees 
who become aware of a relationship. A second concern is that fraternization affects employee 
productivity. Another problem associated with allowing personal relationships in an organization is the 
potential for legal liability. Many employers have recognized problems that can arise out of personal 
work-based relationships." (p.22) 
3 1  
Literature Review 
Friendships related to turnover 
Of the areas of organisational functioning that . can be affected by workplace 
relationships, turnove� is particularly important. As stated in the introductory chapter, 
an American study by Lozada ( 1 996) found that 90 percent of dismissals are the result 
of poor attitudes, inappropriate behaviour and difficulties with interpersonal 
relationships rather than deficient technical skills. The finding that people are so often 
dismissed for reasons other than being unable to do their jobs highlights the importance 
of informal interpersonal relationships at work; being good at your job is not sufficient 
if you cannot work with people. Thus, the informal relationships employees have at 
work seem to have a significant effect on several aspects of turnover; on whether 
employees choose to stay in their jobs (Richer et aI. ,  2002; Riordan & Griffeth, 1 995), 
on their subjective enjoyment of their jobs (Nielsen et aI ., 2000; Richer et aI., 2002), and 
on whether organisations want them to remain or decide to end their employment 
(Lozada, 1 996). 
Friendships related to performance 
Jehn and Shah ( 1 997) discuss the different processes occurring in a work group of 
friends as opposed to one of acquaintances, claiming that group processes needed for 
optimal performance may differ depending on the type of task being performed (off-task 
conversation may relieve boredom of a repetitive, manual task but distract from one 
requiring complex decision making). Thus the type of 'York an individual does may 
interact with the salience of friendships at work. Informal relationships may be  more or 
less important depending on whether they facilitate or interfere with the work to be 
done. Ross ( 1 997) states that workplace friendships significantly improve performance. 
There is, however, one condition: the friends must be strongly committed to the group's  
or the organisation's goals. Otherwise, managers may be better off keeping them apart. 
Although some research has indicated that friendships among group members may have 
a negative effect on the group' s  performance, because the focus may be on social 
interaction rather than 
'
on the task (Bramel & Friend, 1 987), other research assumes that 
improved interpersonal relationships (i.e., friendships) lead to improved task 
32  
Literature Review 
performance through a reduction in "process loss,,7. Jehn and Shah ( 1 997) found that 
friendship groups performed significantly better than acquaintance groups on both 
decision-making and motor tasks, because of a greater degree of group commitment and 
I 
co-operation. Jehn and Shah's study involved an experiment in which 26 groups of 
three friends and 27 groups of three acquaintances were asked to follow specific 
instructions for building models of Tinkertoys®. The friends built an average of 9 
models versus 2.45 for the acquaintances. The authors found that friendship groups 
communicate more and provide more positive encouragement than acquaintance groups. 
Friendship groups are more committed to the task at hand and co-operate more with 
others in their group, which leads to better performance. An individual in a group of 
friends has more incentive to perform better because poor performance reflects 
negatively on a group, and it is more difficult to disassociate oneself from a group of 
friends than from a group of acquaintances. 
These findings support those by Campion, Medsker and Higgs ( 1 993) who examined 
work-team characteristics and their relation to team effectiveness8 in a large financial 
organisation in Indiana (D.S.A.). Data were collected from 39 1  employees and 70 
managers. Campion et al. found that potency (the belief that the group can be 
effective)9, workload sharing (which enhances effectiveness by decreasing social 
loafing) and communication/co-operation within teams were significantly related to 
effectiveness (in terms of productivity and manager judgements). Thus, work teams 
were more effective when team members had team spirit, helped each other and had 
positive social interactions, all of which are likely to be positively influenced by 
friendship. 
Friendships related to organisational soCialisation 
On beginning work in a new organisation, a relatively quick adjustment or "settling in" 
period (termed organisational socialisation) is desirable from both an individual and an 
organisational perspective. For the individual, a quick adjustment period means a 
7 Process loss is the extent to which group processes are hindered by misunderstandings, 
miscommunication and dislike, resulting in wasted energy. 
8 Effectiveness criteria included productivity, employee satisfaction and manager judgments. 
9 Potency is similar to the lay term "team spirit" (Campion et aI., 1 993) and high expectancy (Vroom, 
1 964). 
33  
Literature Review 
reduction in anxiety resulting from a lack of situational identity. While for the 
organisation, a reduction in the anxiety of new employees will result in the individual 
being able to focus better on their job (Reichers, 1987). This socialisation period is 
when newcomers acquire information that turns them into productive employees 
(Corner, 1 99 1 ). 
Corner ( 1 99 1 )  discusses the fact that, because managers often have little time to spend 
on new employees and have many other responsibilities, peers' contributions towards 
newcomer socialisation are of great importance, stating that, in the organisational 
socialisation process, " . . .  the role of peers as primary socialization agents becomes 
critical" (p. 65).  Similarly, Louis, Posner and Powell ( 1 983) reported that peers are both 
more available and more helpful than superiors as socialisation agents. Results of 
Corner's study indicated that work interactions with peers facilitated newcomers' 
acquisition of information and subsequent socialisation into the organisation. Similarly, 
Reichers ( 1 987) claims that new employees who form peer relationships move through 
the initial, encounter stage of the occupational role socialisation process at a faster rate 
than those organisational members not involved in peer relationships, further 
highlighting the importance of workplace friendships. 
In addition, in a longitudinal study examining the impact of peer communication on 
employees moving from one location to another within organisations, Kramer ( 1 996) 
found that organisational members who engage in peer relationships adapt to the 
organisation more quickly and efficiently than those not involved in peer relationships. 
Similarly, Morrison ( 1 993) found that the frequency with which new staff accountants 
sought specific types of information during their first six months of employment was 
found to be related to how well they had mastered their job, defined their role, learned 
about their organisation's  culture, and became socially integrated. Thus, friendships, as 
well as having a potentially beneficial effect on performance, have been shown to 
facilitate the organisational socialisation process. 
34 
Literature Review 
Friendships related to job satisfaction 
Job satisfaction may be defined as a pleasurable, positive emotional state resulting from 
the appraisal of one's job or job experiences (Levy, 2003). There are several 
hyPothesised anteced�nts and consequences of job satisfaction. Aspects of an 
individual' s  job,  including the stress and workload are accepted antecedents of 
satisfaction, along with individual characteristics, social factors and opportunities for 
growth in the j ob. Conseq-qences of satisfaction include better performance, and a 
reduction in withdrawal and counterproductive behaviours (Levy, 2003). 
The antecedents of job satisfaction that are of primary interest in the current study are 
the social factors; the relationships that employees have with their supervisors and co­
workers seem to be significant determinates of whether these employees are satisfied 
with their jobs (Levy, 2003). There is a large body of research supporting the idea that 
j ob satisfaction is affected by employees' relationships with their supervisors (e.g., 
Amason & Allen, 1 997; Anderson & Martin, 1 995; Hayes, 1 982; Neubert, 1 999; Tao, 
Takagi, Ishida, & Masuda, 1 998). Recent examples include studies showing, among 
other things, that job satisfaction is affected by supervisory support (Gaertner, 1 999), by 
the level of tmst of a subordinate for a supervisor (Cunningham & MacGregor, 2000) by 
similarly between supervisors and subordinates (Ensher & Murphy, 1 997; Vecchio & 
Bullis, 2001 ; Wesolowski & Mossholder, 1 997), by the extent to which supervisors and 
subordinates like and respect each other (Murphy & Ensher, 1 999) and the way tasks are 
assigned by supervisors (Blau, 1 999). 
Although interesting, the impact of the supervisor/subordinate relationship on work 
related outcomes is not the primary focus of the current study; rather, the friends an 
employee has at work, and the impact these relationships have with satisfaction, 
commitment and turnover, are of interest (notwithstanding the possibility that a 
supervisor may also, on occasion, be a close friend). There is empirical evidence that 
suggests that co-worker relations are an antecedent of job satisfaction. The earliest of 
these studies was by Hackman and Lawler ( 1 97 1 )  who originally developed the job 
characteristic termed "friendship opportunities." Hackrnan and Lawler surveyed a 
sample of 208 employees of an Amelican telephone company who worked in 1 3  
35 
Literature Review 
different jobs, and found significant positive relationships (p < .05) between friendship 
opportunities and job satisfaction (however the authors found that friendship 
opportunities did not relate significantly to motivation or perfonnance). In a more 
recent study, Markiewicz et al. (2000) examined whether friendship quality would 
predict salary and job satisfaction, and if this would differ as a function of the sex of the 
employee or the friend. Markiewicz and her colleagues found that the quality of close 
friendships was, indeed, associated with both the career success and job satisfaction of 
employees. 
Another study examining the impact of friendship on satisfaction (among other things) 
was by Riordan and Griffeth ( 1 995), who hypothesised and tested a theoretical model of 
the relationship betWeen perceived friendship opportunities in the workplace and work­
related outcomes.-- Riofdari and Griffeth's study was based on the survey responses of 
1 74 employees in a small electric utility in the South-Eastern United States (refer 
appendix A for Riordan and Griffeth's  model of workplace relationships). Using 
structural equation modelling, Riordan and Griffeth found that friendship opportunities 
were associated with increases in job satisfaction, job involvement and organisational 
commitment, and with a significant decrease in intention to turnover. 
Riordan and Griffeth's findings are supported in a recent study by Nielsen et al. (2000) 
who conducted two studies to develop and provide evidence for the construct validity of 
the workplace friendship scale used in the current study. Nielsen et al. collected data 
from a total sample of 1 1 6 part-time graduate students who were employees of three 
American organisations. Support was provided for convergent, discriminant, and 
nomological validity of scale scores. In addition, Nielsen et al. reported a positive 
correlation between friendship opportunities, friendship prevalence and job satisfaction; 
furthermore they found that those experiencing friendship at work were less likely to 
want to leave their current jobs. These findings highlight the positive impact workplace 
friendships can have for employees within organisations, particularly in tenns of their 
satisfaction with their jobs. 
36 
Literature Review 
There is further support for Riordan and Griffeth' s  findings that friendships are 
associated with increased satisfaction. For example Winstead et al. ( 1 995) examined 
existing friendships of both general and academic staff in two universities in Virginia 
(U.S.A.). Using multiple regression, Winstead et al. found that the quality of an 
individual ' s  best friendship in the workplace was predictive of job satisfaction. 
Friendships at work were found to be something of a "double edged sword" however. 
, 
Results indicated that, although staff who reported higher quality friendships were also 
significantly more satisfied with their jobs, maintenance difficultylO was inversely 
related to satisfaction with work, indicating that, for many workers, the negative aspects 
of a close friendship have a significant negative impact on satisfaction with work. 
Interestingly, the effects of maintenance difficulty on satisfaction were only apparent for 
general staff (as opposed to faculty members). An explanation Winstead et al. give for 
this is that, for faculty, the intrinsic qualities of their job (teaching, research, supervising 
students) are more important to them than the non task-related aspects; while for general 
staff, whose work may be more routine or work roles more interdependent, the social, 
non-work aspects are more salient. In addition, for academic staff, social needs may be 
met in the course of their job; teaching, supervising and collaborative research all being 
very social activities. For general staff, on the other hand, rewarding social interactions 
may be more rare, and therefore more valued. In addition, faculty have significantly 
more autonomy at work, and more control over their personal space, usually having 
their own office. Consequently academic staff members are able to simply avoid 
negative interactions more easily than general staff, who may have to liase frequently 
with others, or work in an open plan office. Either way therefore, it seems likely that 
the type of work an individual does will interact with the impact of friendships at work. 
In general, friendships will probably be more salient for jobs that are relatively more 
interdependent. 
The most recent study examining the relationship between friendships at work and 
satisfaction is by Richer, Blanchard and Vallerand (2002) who used structural equation 
modelling to test their Motivational Model of Work Turnover (refer appendix B for a 
10 Maintenance difficulty reflects tension and strain being experienced by the individuals in a friendship 
dyad, as a result of aspects of the friendship. 
37  
Literature Review 
reproduction of this model). A relevant aspect of Richer et al. ' s  model is the path from 
"relatedness toward work colleagues" (a measure of relationship quality) to "work 
satisfaction." Although Richer et al. posit that " work motivation" mediates the 
relationship between "relatedness toward work colleagues" and satisfaction; their model 
too, provides support for their hypothesis that a positive relationship between 
friendships at work and job satisfaction would exist. Thus, in the current study, 
friendship opportunities are hypothesised to be antecedent to job satisfaction. 
Building a model of organisational relationships 
At this point, building the model of organisational relationships tested in the current 
study can commence. The degree to which friendship opportunities are antecedent to 
job satisfaction is a central focus of the current study, and is illustrated in Figure 1 with 
an arrow leading from friendship opportunities to job satisfactionl l . The "+" sign 
indicates that a positive relationship is hypothesised between these two variables. 
F r ie n d s h ip 
o p p o rtun ities 
+ 
, r 
Job satisfaction 
Figure 1 :  Showing the hypothesised relationship between friendship opportunities and 
job satisfaction 
1 1  The 'friendship prevalence' variable is not modelled at this stage but is included later (refer section 
2.3 .5) 
38  
Literature Review 
Job satisfaction related to organisational commitment 
Organisational commitment is distinguished from job satisfaction in that organisational 
commitment is a response to the whole organisation, while satisfaction is an affective 
response to specific aspects of the job (Williams & Hazer, 1 986). Organisational 
--
commitment and job satisfaction are generally found to be positively correlated with 
one another (Cohen, 1 993, 1 996; Fisher, 2002; Hackett & Lapierre, 200 1 ;  Kaldenberg, 
Becker, & Zvonkovic, 1 995 ; Meyer, Stanley, Herscovitch, & Topolnytsky, 2002; Porter, 
. Steers, Mowday, & Boulian, 1974; Rentsch & Steel, 1 992; Tett & Meyer, 1993). 
However, although the positive correlations between organisational commitment and 
job satisfaction are strong, Meyer and AlIen (2002) state that the correlations " . . .  are not 
of sufficient magnitude to suggest construct redundancy" (p.38). In other words, job 
satisfaction and organisational commitment are conceptually distinct. 
The concept of organisational commitment has been defined in various ways in the past, 
depending on the background of the scholars (Yousef, 2003). Lee ( 1 97 1 )  defines 
organisational commitment as identification with the organisation. Steers ( 1 977) and 
Levy (2003) define it as identification with and involvement in an organisation. 
Sheldon ( 1971)  states that organisational commitment is an attitude or an orientation 
towards the organisation which links or attaches the identity of the person to the 
organisation. The most frequently used definition of commitment, however, is that of 
Porter (1 974) who defined it  as a strong belief in and acceptance of the organisational 
goals and values, a willingness to exert effort on behalf of the organisation and a desire 
to remain in the organisation . . Allen and Meyer ( 1 990) identified three components of 
commitment. These were (1) Affective commitment which "concerns a person' s  
emotional attachment to their organization" (p. 1 ) ;  (2) Normative commitment or  "a 
moral dimension based on a person's felt obligation and responsibility to their 
employing organization" (p. 1 ) ;  and (3) Continuance commitment or "a person' s  
perception of  the costs and risks associated with leaving their present organization" (p. 
4). Continuance commitment is the intent to remain with the organisation, usually 
through an evaluation of the perceived costs of leaving the organisation, also 
39 
Literature Review 
conceptualised as "side bets" (Becker, 1 960) 12. Continuance commitment is sometimes 
referred to as "sunk-costs" commitment because it is a function of what an employee 
has sunk into the organisation (Levy, 2003). Normative commitment, on the other hand 
reflects a perceived obligation to remain in the organisation. As Meyer and AlIen 
( 199 1 )  put it, an individual with strong affective commitment stays with an organisation 
because he or she wants to do so, an individual with strong continuance commitment 
stays with an organisation because he or she needs to do so, while individuals with 
strong normative commitment stay with an organisation because they feel they ought to 
do so. The questionnaire used in the current study (the Organisational Commitment 
Questionnaire, refer appendix 6) taps into all three aspects of commitment. 
Antecedents and consequences of organisational commitment. Mathieu and Zajac 
(1 990) performed a large meta-analysis of the commitment literature and identified the 
antecedents, correlates and consequences of commitment. They identify personal 
characteristics of the employee as being antecedent to commitment; these include the 
attitudes beliefs and skills of individuals in the organisation. Other antecedents are job 
characteristics, group and leader relations, co-worker relations and organisational 
characteristics. A further antecedent of organisational commitment is the existence of 
organisational mechanisms such as reward systems, socialisation and newsletters; these 
mechanisms lead to affective commitment by communicating to employees that the 
organisation is both supportive of them and fair to them (Levy, 2003 ; Meyer & AlIen, 
1 997). Consequences of organisational commitment include a reduction in withdrawal 
behaviours such as absenteeism (Levy, 2003 ; Mathieu & Zajac, 1 990) and intention to 
leave. In addition Levy suggests that commitment will result in a reduction ill 
counterproductive behaviours such as theft and sabotage. Another consequence of 
commitment is improved performance, although, because of the complexity of 
performance, the relationship between performance and commitment is not strong 
(Levy, 2003). The findings of Mathieu and Zajac's ( 1 990) meta-analysis of the 
1 2  The term side-bet was used by Becker ( 1 960) to refer to the accumulation of investments valued by 
individuals that would be lost if they were to leave the organisation. Becker argued that organisational 
commitment is based on these investments. Working in an organisation increases an individual's 
investments, and the costs of leaving cause higher levels of organisational commitment. 
40 
Literature Review 
antecedents, correlates and consequences of commitment are summarised in a diagram 
in appendix C. 
Job satisfaction is hypothesised to be antecedent to commitment in the current study 
because attitude towards one's job is a more proximal evaluation than attitude towards 
one's organisation, i .e. , an employee is unlikely to form strong organisational 
commitment if they are not satisfied with their job. It is worth noting here that the 
commitment-to-satisfaction relationship may operate somewhat differently in Western, 
individualist cultures than in more collectivist societies 1 3 . Research in Western 
contexts, however, has indicated that commitment to the company develops from job 
satisfaction and mediates the effects of satisfaction on turnover and intention to leave 
(e.g., Porter et aI ., 1974 ; Riordan & Griffeth, 1 995 ; Williams & Hazer, 1 986) . This 
satisfaction-to-commitment model reflects Porter et al. ' s  ( 1 974) claim that commitment 
takes longer to develop and is more stable than satisfaction, stating, "We would expect 
that a relatively greater amount of time would be required for an employee to determine 
his level of commitment to the organization than would be the case with his level of his 
job satisfaction" (Porter et aI. ,  1 974, pg. 608) . This claim has received empirical 
support, for example, Williams and Hazer ( 1 986) used structural equation modelling to 
re-analyse data from two previous studies (Bluedorn, 1 982; Michaels & Spector, 1 982). 
They specifically examined the causal relationship between satisfaction and 
commitment, hypothesising that job satisfaction is a causal antecedent of commitment. 
Their . analysis supported the satisfaction-to-commitment relation (the model 
incorporating this relation having a significantly better fit than the model without), 
implying that satisfaction is antecedent to commitment in individuals. They state; 
13 It is possible that the satisfaction-to-commitment relationship will be less apparent in more collectivist 
cultures (i.e., Japan and Hong Kong). Empirical research on the relationship between Individualism­
Collectivism and commitment has been limited (Parkes, Bochner, & Schneider, 200 1 )  and a review of 
commitment across national cultures failed to fmd a link between collectivism and commitment (Randall, 
1 993). However, Parkes et al. (2001 )  state that theory and logic suggest such a link should exist. One of 
the defining elements of collectivist cultures is a more familial relationship between employee and 
employer, with protection and long-term job security provided in return for loyalty, trust, and 
organisational commitment (Kao & Sek-Hong, 1993). Because this implicit contract is between 
individuals and organisations, collectivism at both these levels should contribute to employees' greater 
commitment and longer stay in an organisation, regardless of job satisfaction. 
41  
Literature Review 
" . . .  through a process of the evaluation of costs and benefits, 
individual needs and desires are satisfied, and the resulting affective 
state becomes associated with the organization, which has provided 
the job and its associated characteristics and environment. 
Commitment results from this association" (p.230). 
In addition to findings suggesting that satisfaction is antecedent to commitment, 
Williams and Hazer (1 986) found satisfaction to be an intervening variable between 
workplace variables (e.g., leadership consideration, "routinization" and equity), and 
organisational commitment; workplace variables directly influencing only satisfaction, 
and influencing commitment indirectly through their impact on satisfaction (and the 
subsequent effect of satisfaction on commitment). Thus, there are reasons to believe 
that the relationship between commitment and variables such as friendship opportunities 
and cohesion may not be direct, but rather, will be mediated by satisfaction. 
Organisational commitment is the next variable to be added to the model (refer Figure 
2). The one-way arrow from satisfaction to commitment reflects the literature, 
described above, which suggests that job satisfaction is a causal antecedent of 
commitment. 
Friendship . 
opportunities 
+ 
Job + Organisational 
satisfaction commitment 
Figure 2: Showing the hypothesised relationship  between job satisfaction and 
organisational commitment 
42 
Literature Review 
Friendships related to organisational commitment 
As stated above, Riordan and Griffeth ( 1 995) found that friendship opportunities are 
associated with, among other things, increases in organisational commitment. Because 
Riordan and Griffeth (1 995) used a cross-sectional design, one limitation of the study is 
that the direction of causality is unclear, it cannot be determined with certainty if 
friendship opportunities caused the organisational outcomes or vice versa, or if some 
other variable affected them both. The qualitative data collected in the current study 
will hopefully go some way towards addressing the issue of causality. Another 
limitation of  the study by Riordan and Griffeth is that it only investigated opportunities 
for friendship rather than actual friendships. In the current study, the prevalence of  
friends is measured in addition to the opportunities for friendship in the workplace. 
With respect to the relationship between friendship and commitment, in his early work 
in this area, Becker (1 960) suggests that workplace associates helped produce 
commitment to an individual' s  job, mentioning the " . . .  loss of connections in his present 
firm . . .  " (p. 3 8) if the employee was to move. Related to this Cherniss ( 199 1 )  found a 
link between occupational commitment and having a supportive organisational climate, 
he conducted a study using questionnaires and in-depth biographical interviews of 
twenty-five human service professionals (teachers, lawyers, nurses) in America over a 
twelve-year period. Cherniss found that those with high commitment had worked in 
especially supportive settings during the much of the first decade of their careers, while 
those with low commitment had worked in negative climates during the same period 
(occupational commitment is thought to be a correlate of organisational commitment) 
(Meyer et aI. ,  2002). 
In a study based in New Zealand, Shouksmith (1 994) administered the 
Multidimensional Job Satisfaction Scale to a sample of 1 1 2 1  health professionals to 
ascertain which organisational factors were associated with commitment. The existence 
of coworkers with positive attributes was found to strongly affect normative 
commitment (p < .001) .  This implies that the aspect of commitment concerning 
individuals' felt obligation and responsibility to their organisation will be significantly 
43 
Literature Review 
affected by whether or not they view their coworkers in a positive light (i.e., have 
friendly relationships with them). 
A recent Gallup study� which found that having a best friend at work is related to how 
engaged and committed an individual is to his or her job, further supports the findings 
that friendships and socially supportive envirorunents at work are related to commitment 
(Ellingwood, 2001 ). In the Gallup study, a random sample of American workers over 
the age of 1 8  were given the Q 1 2  workplace evaluations; "Do I have a best friend at 
work? " is one if the 1 2  questions in this survey. It was found that 5 1  % of respondents 
who agreed with this statement were engaged in their jobs, compared to only 1 0% who 
disagreed. Ellingwood (200 1)  states that in the thousands of employee interviews and 
hundreds of focus groups conducted by Gallup, "friendship trumped such seemingly 
obvious employee motivators as pay and benefits." In addition, having a best friend at 
work was strongly related to intention to leave. Seventy-five percent of respondents 
who had a best friend at work planned to be with the company for at least another year, 
as opposed to only 5 1  % who did not have a best friend at work. 
In one of the few early studies exammmg the relationship between organisational 
commitment and informal organisational relationships, Sheldon ( 1 97 1 )  investigated the 
link between social involvement and the commitment of employees to their 
organisation. The sample in this study was entirely male scientists and engineers with 
PhD's,  who were understood to have high career commitment. It was assumed that 
these individuals would be more socialised into their profession than their less qualified 
colleagues, and "would be more likely to expect and to have professional prerogatives 
and less likely to desire organisational rewards" (p. 144). Because these individuals are 
less likely to seek organisational rewards, it is not unreasonable to expect relatively low 
organisational commitment and indeed, Sheldon found that those who were not socially 
involved with peers in the organisation were also not committed to the organisation, 
compared to those who were socially involved (i.e., had friends). This finding sugs.ests 
that workplace relationships are one way that commitment to an organisation is shaped. 
44 
Literature Review 
Other research investigating the link between commitment, and informal 
communication with co-workers was carried out by Anderson and Martin ( 1 995) who 
\ 
surveyed 202 full time employees from the Midwestern United Sates. Respondents 
completed a questionnaire, which asked questions about their motives for 
communication at work, and measured their organisational commitment and job 
satisfaction. Results supported the hypothesis that when people communicate with their 
co-workers and superiors regularly for inclusion, affection, relaxation (non-task related 
communication) or duty (task related communication) they will have increased 
commitment to, and satisfaction with, their job. If they tended to communicate for 
control and escape (i.e., biding time, avoiding work), commitment and job satisfaction 
would be reduced. The finding that commitment is increased when employees "chat" at 
work provides evidence that some non-task oriented communication (for example, 
communicating for affection or inclusion) serves a valuable function for individuals in 
organisations. People spending 40 hours a week in a particular environment need to 
know that others around them care about them and like them. Work friends may 
provide understanding that eases frustration and job-related anxiety and stress 
(Anderson & Martin, 1 995). 
In sum, the research outlined above suggests a positive relationship between a friendly 
work environment and organisational commitment. The friendship/commitment 
relationship may be either direct or mediated by satisfaction, which is thought to be a 
possible precursor to commitment (as discussed above). 
The relationship between friendship opportunities and organisational commitment is the 
next aspect of the model to be added. Figure 3 shows the hypothesised positive 
relationship between these two variables. 
45 
Friendship 
opportunities 
Job satisfaction 
+ Organisational 
commitment 
Literature Review 
Figure 3: Showing the hypothesised relationship between friendship opportunities, job 
satisfaction and organisational commitment 
Intention to leave 
Over the last two decades the development of predictive models of voluntary turnover 
has been an aim of many researchers in this area. Job satisfaction, organisational 
commitment, and intention to leave are among the most commonly proposed 
antecedents to turnover (Tett & Meyer, 1 993). Satisfaction and commitment are 
invariably reported to be negatively related to turnover and intention to leave (e.g., Blau 
& Boal, 1 989; Cohen, 1 993; Cohen & Hudecek, 1 993; Hackett & Lapierre, 200 1 ;  Irvine 
& Evans, 1 995; Irving & Meyer, 1 994; Kaldenberg et al., 1 995; Lee, Carswell, & Allen, 
2000; Lu, Lin, Wu, Hsieh, & Chang, 2002; McEvoy & Cascio, 1 987; Meyer et al. ,  
2002; Mobley, 1 977; Porter et al. ,  1 974; Riordan & Griffeth, 1 995; Steers, 1 977; Tett & 
Meyer, 1 993). The finding that turnover intention is the strongest predictor of turnover 
is similarly consistent (e.g., Irvine & Evans, 1 995; Meyer et al. ,  2002; Mitchel, 1 98 1 ;  
Mobley, 1 977; Tett & Meyer, 1 993). 
The recent Motivational Model of Work Turnover, proposed by Richer et al. (2002) 
(refer appendix B for a reproduction of Richer et al . '  s model) supports the negative 
relationship between satisfaction and turnover. Richer et al . ' s  model shows satisfaction 
46 
Literature Review 
to be antecedent to turnover intention. Similarly, Irvine and Evans ( 1 995) published a 
meta-analytic study investigating the causal relationship between job satisfaction, 
behavioural intentions and turnover behaviour in nurses. The model proposed by Irvin� 
and Evans (1 995) also. shows satisfaction to be negatively related to turnover intentions 
(and, in turn, turnover) (refer appendix D for a reproduction of Irvine and Evans' 
model). 
In addition to job satisfaction, organisational commitment has been shown to affect 
turnover. Much of the research in this area, while it does not dispute that commitment 
develops from satisfaction (as described above), indicates that both satisfaction and 
commitment contribute independently to the turnover process. This independent-effects 
model follows Porter et al. ' s  ( 1 974) suggestion, based on their research, that job 
satisfaction and organisational commitment, though related, are distinct constructs. 
They state, " . . .  although we would expect commitment and satisfaction to be related, 
each construct appears to contribute unique information about the individual 's  
relationship to the organisation" (p. 608). 
Further support for the notion that both satisfaction and commitment contribute to 
turnover comes from a study by Tett and Meyer ( 1993) who conducted a large study 
investigating job satisfaction, organisational commitment, turnover intentions, and 
turnover; applying path analysis to data comprised of aggregations involving a total of 
1 78 independent samples from 1 55 studies. Tett and Meyer's meta-analysis of the 
attitude and turnover intentions literature concluded that the two attitudes of job 
satisfaction and organisational commitment are independent contributors to turnover 
intention, supporting the independent-effects perspective. The independent-effects 
perspective is also supported by Riordan and Griffeth ( 1 995) who proposed a model of 
work related outcomes of friendship opportunities. As discussed above, Riordan and 
Griffeth hypothesised that friendship opportunities would have a significant direct effect 
on an individual's  satisfaction with his or her job, which would, in turn, have a 
significant positive relationship with organisational commitment. Because prior 
research has suggested that both job satisfaction and organisational commitment reduce 
intention to turnover (Angle & Perry, 1 98 1 ;  Stumpf & Hartman, 1 984; Tett & Meyer, 
47 
Literature Review 
1 993) their model (like the model for the current study; refer Figure 4) included direct 
negative relationships between (a) job satisfaction and intention to leave, and (b) 
organisational commitment and intention to leave (refer appendix A for a reproduction 
of Riordan and Griffeth's model). 
The hypothesised relationships between commitment, satisfaction and turnover are 
shown in Figure 4 by arrows leading from satisfaction to intention to leave, and from 
commitment to intention to leave. The "-" signs indicate that the relationships between 
satisfaction/commitment and turnover are hypothesised to be negative. 
Friendship 
opportunities 
Job satisfaction 
+ 
Intention to leave 
Organisational 
commitment 
Figure 4: Showing the hypothesised relationship between job satisfaction, commitment 
and intention to leave 
The current study investigates intention to leave (as opposed to actual turnover) but 
although it is not measured in the current study, a link between behavioural intentions 
and behaviour has been documented in the social psychological literature (Ajzen & 
48 
Literature Review 
Fishbein, 1 980), as has a link between turnover intentions and turnover (Irvine & Evans, 
1 995; Richer et al. ,  2002; Tett & Meyer, 1993). For example, Irvine and Evans ( 1 995) 
published a meta-analytic study investigating the causal relationship between, among 
other things, behaviollral intentions and turnover behaviour. Irvine and Evans found a 
positive relationship between behavioural intentions (intention to leave) and turnover14. 
Research has shown that most employees who intend to leave their jobs, and who feel 
they have the choice to do so, will most likely quit (Richer et al. , 2002). 
Conclusions 
To summarise the review of literature focusing on organisational peer relationships 
outlined so far in this chapter, there are several conclusions that may be drawn: 
• Peer relationships are more intimate, have a longer duration, and are 
characterised by higher levels of trust and self-disclosure than sup eri or­
subordinate or mentor-protege relationships. 
• Organisational outcomes such as job productivity, job satisfaction and 
commitment are linked closely with peer relationships. 
• Organisational members who engage in peer relationships adapt to the 
organisation more quickly and efficiently than those not involved in peer 
relationships. 
• Organisational members are likely to interact more with peers than with 
superiors. Because of this, the development of satisfying peer relationships is 
likely to be of great importance for employees, and may impact on turnover 
intentions. 
It is worth noting that, other than Shouksmith (1 994) none of the studies described 
above have been conducted in New Zealand; in fact little of the research discussed was 
14 The mean correlation between turnover behaviour and behavioral intentions, corrected for attenuation 
was .34. The positive relationship indicates that as employees develop their intention to leave, they are 
inclined to follow through with turnover behaviour (Irvine & Evans, 1995). 
49 
Literature Review 
conducted outside the United States. This highlights the importance of examining the 
generalis ability of these findings, looking at the impact of friendships at work in 
environments and countries other than America. This is a major contribution the current 
study will make. Th� respondents to the Hospital Study (Chapter Three) being solely 
from NZ, and the Internet Study (Chapter Four) drawing on responses from the world 
over. 
Friendships may be more important in some types of jobs. For example where the job 
involves long hours or high levels of stress, strong friendships may increase retention. 
Conversely negative relationships could increase intention to turnover, or at least 
produce stress at work which, in turn, can cause other problems (Riordan & Griffeth, 
1 995). Most of the literature focusing on organisational relationships suggests that 
friendships have a significant impact within organisations, both at the individual and 
organisational level of functioning. Similarly, the organisation is likely to impact on the 
friendship dyads within it. The impact of the organisation on friendships is addressed in 
the following section. 
2.3.4 The Impact of the Organisation on Friendships 
Although types of work differ widely among occupations, m general, workers are 
expected to remain together throughout their workday. Some workers, such as nurses, 
work in close proximity, whereas others, such as academics, can retreat to their offices. 
Others may work at separate desks but congregate regularly for breaks. Whatever the 
structure of the workplace, common areas are typically found where personal interaction 
can occur, for example, staff rooms, lifts, bathrooms, and foyers. People do not want to 
see the same strangers week after week; therefore co-workers will be incorporated into 
people's acquaintance-networks (Fine, 1 986). 
Aspects of the workplace which facilitate friendships 
Fine (1 986) states that workplaces are significant in the development of friendships 
beyond the pragmatic reality that they force individuals together. The culture of work 
allows and, in some cases, requires friendly behaviour. Similarly, Sias (1 998) found in 
a recent study that the organisation is not merely a "container" for relationships, but is 
50 
Literature Review 
an integral part of the relationship itself, playing a key part in its developmental process. 
That is to say that, in addition to providing a context within which people become 
friends, the workplace itself not only creates opportunities for friendship but also 
impacts on the way friendships develop. 
Literature on friendship formation from social psychology holds that people are more 
likely to become friends if they are in close proximity. Workplaces, and particularly 
work-groups require individuals to work together on a daily basis, so it is hardly 
surprising that friendships develop among work-mates. In addition, frequent exposure 
to another person is often enough to lead to more positive evaluations of that person, 
increasing the likelihood of friendship (Anderson & Hunsaker, 1 985 ;  Sias & Cahill, 
1 998; Zajonc, 200 1 ). 
Similarity also facilitates friendship formation. Individuals will be attracted to people 
they perceive to be similar to themselves, particularly with respect to attitudes, values 
and interests (Anderson & Hunsaker, 1 985 ;  Duck & Perlman, 1 985). By definition 
organisational peers are similar with respect to status level. In addition they will 
perform similar work, in similar occupations, probably with a similar educational 
background (Sias & Cahill, 1 998). 
Working towards a common goal is a third aspect of the workplace that may facilitate 
friendships. People who share common goals (such as members of a work group) are 
more likely to achieve these goals if they co-operate with one another, this shared goal, 
along with co-operative behaviour also facilitates friendship. Moreover, people 
working toward the same goal see themselves as a "unit" - a view that can increase 
positive feelings about others working in the same team (Anderson & Hunsaker, 1 985). 
In addition to the three factors described above, an aspect of the workplace that may not 
exist in other contexts, which is likely to facilitate friendships is that when people are 
committed to workin� together there is a desire to like the other person if only because a 
pleasant work environment is more rewarding than an unpleasant one. This is a 
5 1  
Literature Review 
situation which will further increase the likelihood of friendships forming (Anderson & 
Hunsaker, 1 985). 
Hierarchy 
Another aspect of the workplace that may affect the formation of friendships is the 
organisational hierarchy. As well as peer relationships forming in organisations, 
friendships can and do form across hierarchical levels. These relationships come with 
their own complications but have been shown to offer a number of benefits such as 
improved communications and a less stressful work environment. In spite of this, 
companies have a long-standing position against friendship between managers and their 
subordinates, largely to avoid management related problems such as biased performance 
appraisals (Marelich, 1 996). 
Autonomy 
The amount of autonomy in a job may also affect social connections at work. If 
individuals have autonomous roles (such as academics), informal relationships among 
peers are not essential for completing the work. Friendly relationships may still occur 
however because of shared values, norms and behaviours (Fine, 1986). People in 
autonomous occupations have a great deal of control over the content of their workplace 
relationships so they can, perhaps, be more selective about who they maintain 
relationships with. Fine (1 986) �laims that those in occupations with relatively less 
autonomy, however, have a proportionately greater need for solidarity with their peers. 
Thus it is possible that autonomy will be negatively related to friendships forming in the 
workplace. 
Cohesion 
Cohesion is one of the measured variables in the current study; the cohesiveness of their 
work group or team is another aspect of an employee's experience of their workplace. 
As previously stated (section 2.3 . 1 ), Kram and Isabella (1 985) maintain that the 
characteristic of reciprocity sets co-worker friendships apart from other work 
relationships, such as mentoring or superior-subordinate relationships. Buunk et al. 
( 1 993) identify perceived reciprocity (mutual flow of support and help) as an important 
52 
Literature Review 
characteristic of relationships with colleagues. Results from Buunk et al . '  s study 
indicate that when co-workers do not reciprocate favours, individuals expenence 
negative emotions such as irritation, depression, and confusion. If an individual 
perceives their working climate to be low in cohesion, they may believe that their peers 
will not reciprocate emotional and instrumental support, which will likely hinder the 
formation of friendships. Alternatively, if someone perceives that their efforts will be 
reciprocated (e.g. ,  in a climate perceived as high in cohesion) they may be more likely 
to develop collegial and special peer relationships. Supporting this, Odden and Sias 
( 1 997) conducted a study using teachers, which examined the association between 
psychological climate and the types of communication relationships employees form 
with their peers. Odden and Sias found that climates perceived as high in cohesion were 
related to larger proportions of collegial and special peer relationships. Moreover, the 
cohesion dimension reflects a general liking of one's co-workers, as well as perceptions 
that an employee shares a great deal of "common ground" with hislher co-workers 
(Odden & Sias, .1 997). Existing literature regarding friendship development identifies 
liking and perceived similarity as factors that enhance the formation of friendships. 
Therefore, it is not unreasonable to expect a positive relationship between perceived 
cohesion and friendship opportunities. 
Thus, a cohesion variable can be added to the model. In the current study, opportunities 
for friendship are hypothesised to impact positively on the cohesiveness of the 
workgroup or team in that, if there is a generally friendlier environment, this will likely 
increase the cohesiveness of the group. Cohesiveness is likely to impact on the 
perceived opportunities for friendship, (a more highly cohesive team will probably 
result in more opportunities for friendship). The relationship between cohesion and 
friendship opportunities is shown in Figure 5 with a two-way arrow at the top of the 
diagram, the "+" indicating a positive correlation. 
53  
Friendship 
opportunities 
Job satisfaction 
+ 
+ 
Intention to leave 
Cohesion 
Organisational 
commitment 
Literature Review 
Figure 5: Showing the hypothesised relationship between friendship opportunities and 
cohesion 
Cohesion related to intention to leave 
Sheridan (1 992) investigated employee retention; he compared the voluntary turnover 
rates of new employees of accounting firms with varying organisational cultures. 
Sheridan examined the effects of interpersonal relationships; measured by a "team 
orientation" dimension and a "respect for people" dimension (measures closely related 
to the cohesion of a workplace) on retention. Sheridan found that hazard rates1 5 were 
significantly lower for employees working in a culture emphasising interpersonal 
relationship values than they were in a culture emphasising work task values. Findings 
indicated that new employees voluntarily stayed fourteen months longer in the culture 
emphasising interpersonal relationship values (i.e., one which was highly cohesive) than 
15 Hazard rates are an estimate of the probability of an employee leaving during a particular month of 
seniority given that they have survived to the beginning of the month. 
54 
Literature Review 
in the culture emphasising work task values (the median survival time was 45 months 
compared to 3 1  months). Sheridan states; 
" the most parsimonious explanation of employee retention may 
simply be that an organisational culture emphasising interpersonal 
relationship values is uniformly more attractive to professionals than a 
culture emphasising work task values" (p. 1 052). 
The possibility that an organisational culture which emphasises work task values, as 
opposed to interpersonal relationship values, will not be conducive to friendship 
formation is also discussed by Odden and Sias ( 1 997). They claim that because "special 
peer" relationships require trust and self-disclosure, as well as high levels of emotional 
support and personal feedback, these relationships are likely to require a great deal of 
time and energy to sustain; far more than is required for "information peer" 
relationships. Employees in high-pressure, task oriented work environments will 
probably devote more of their limited time to task completion than to developing their 
relationships. This will be likely to result in a smaller proportion of special peer 
I ·  h' 1 6  re atlOns IpS . In addition, Eisenberg and Goodall ( 1 997) note that one of the 
consequences of a high-pressure work environment is that individuals develop negative 
attitudes toward others, thus it also follows that a work environment that is task, as 
opposed to relationship, oriented �il1 provide fewer opportunities for friendship, be less 
cohesive and, extrapolating from Sheridan's ( 1 992) findings, result in higher turnover. 
Thus, Figure 6 shows an arrow added to the model, leading from cohesion to intention 
to leave, indicating that employees' perceptions of the cohesiveness of their work 
environment will have negative correlations on their leaving intentions. The 
relationship between cohesion and intention to leave may be a direct relationship or one 
mediated by job satisfaction (i.e. , having a cohesive work environment will have an 
impact on leaving intentions only if an individual is satisfied with their job). 
16 It is worth noting that the opposite may also be true, as it is often in times of adversity (i.e., poor 
working conditions) that strong friendships will form (Carr, 2003 ; Odden & Sias, 1997), refer section 
2.3.5 for more detail. However, a "task oriented" work environment should probably be differentiated 
from "poor conditions" at work, this differentiation goes some way towards explaining the seemingly 
contradictory findings. 
55  
Friendship 
opportunities 
Job satisfaction 
+ 
+ 
tntention to leave 
Cohesion 
Organisational 
commitment 
Literature Review 
Figure 6: Showing the hypothesised relationship between cohesion and intention to 
leave 
Cohesion, organisational commitment and job satisfaction 
As discussed above, an organisational culture emphasising interpersonal relationship 
values will provide increased opportunities for friendship and, at a group level, 
cohesion. The "opportunities for friendship" variable has received only a small amount 
of research attention to date, but has been empirically related to cohesion (e.g., Odden & 
Sias, 1997), satisfaction (e.g., Riordan & Griffeth, 1 995) and commitment (e.g., 
Kaldenberg et al . ,  1 995). The current model hypothesises that increased opportunities 
for friendship will lead to improved job satisfaction and organisational commitment. 
Hence there are arrows leading from the friendship opportunities variable to both job 
satisfaction and organisational commitment. Likewise, the cohesion variable is also 
hypothesised to result in improved job satisfaction and organisational commitment. 
AlIen and Meyer (1 990) specifically linked organisational commitment to cohesion, the 
56 
Literature Review 
results of their study iridicated that cohesion IS antecedent to commitment. These 
relationships are indicated in Figure 7. 
Friendship 
opportunities 
+ 
Job satisfaction 
+ 
+ 
Intention to leave 
Cohesion 
Organisational 
commitment 
/ 
Figure 7: Showing the hypothesised relationship between cohesion and organisational 
commitment and job satisfaction 
2.3.5 Friendship Prevalence 
A key variable measured in the current study is the actual prevalence of workplace 
friendship. Friendship prevalence is hypothesised to be a consequence of  both 
friendship opportunities and cohesion. Logically, if there are increased opportunities for 
friendship, the prevalence of friendships should also increase, or alternatively 
opportunities for friendship may be a necessary condition for actual friendships to form. 
In addition, as stated above, Odden and Sias ( 1 997) found that climates perceived to be 
high in cohesion were related to more special peer relationships. The cohesion 
dimension reflects a general liking of one's co-workers, as well as perceptions that an 
employee shares a great deal of "common ground" with them; both factors likely to 
57 
Literature Review 
contribute to the formation of friendships. Thus Figure 8 shows arrows leading to 
friendship prevalence from both friendship opportunities and cohesion, again the "+" 
indicates the positive nature of the hypothesised relationship. 
Friendship 
prevalence 
Friendship 
opportunities 
+ 
+ 
+ 
Job satisfaction I------+-l 
Cohesion 
+ 
Organisational 
commitment 
Figure 8: Showing friendship prevalence as the result of i ncreased friendship 
opportunities and cohesion 
From Figure 8, the friendship prevalence variable is not hypothesised to impact directly 
on job satisfaction or commitment. Initially this may seem counterintuitive, given the 
evidence that friendship opportunities and cohesion have positive correlations with 
these outcome measures. However, the actual prevalence of friendship may not be 
related to satisfaction and commitment (as opportunities for friendship are) because the 
effects of need fulfilment (i.e., fulfilling the need for rewarding relationships) by the 
\ 
organisation may be conditional. It is probably necessary that the employee see the 
organisation as, in some way, responsible for the friendship. That is to say, employees 
58 
' .  
Literature Review 
may need to believe that their friendships are the consequence of the way the company 
is managed, and perhaps not easily achieved in another organisation (Meyer & Allen, 
1 99 1). In relation to the current study, this implies that having friendships at work 
andlor feeling that they work in a friendly environment will only improve commitment 
if the employee feels that the organisation is in some way responsible for the positive 
environment. A generally friendly environment is likely to be seen as something that 
the organisation has made available, while the actual friendships an individual has 
formed and maintained are probably more likely to be seen as something which the 
organisation is not directly responsible for. 
Amason and AlIen ( 1997) also discuss the concept of employees not crediting the 
organisation for their own relationships. Amason and Allen state that women, 
partiCUlarly, may see their identities as separate from their work identities, and their 
relationships not defined by organisational roles. While these women will find their 
friendships at work pleasurable, the friendships may be viewed as separate from their 
relationship with the organisation as a whole (Amason & AlIen, 1 997) . Thus, these 
women will not 'credit' the organisation for any positive outcomes resulting from their 
friendships. 
In addition, it is often in times of adversity (when morale and satisfaction may be low) 
that strong friendships will form (Carr, 2003). Similarly, Sias and Jablin (1 995) found 
that, when a supervisor or group leader was perceived to treat group members unfairly, 
group members b�came more cohesive; they interacted more and their communication 
relationships became more intimate, suggesting that perceptions of low quality 
supervision and feelings of dissatisfaction may lead co-workers to maintain more close 
friendships as a form of alliance against the organisation. Sias and Jablin' s  finding was 
supported by Odden and Sias (1997), who found that perceptions of inconsiderate 
supervision were related an increase in special peer relationships. Thus, while Kram 
and Isabella (1 985) suggest that special peers provide a sounding board, or social 
support, Sias and Jablin ( 1995) and Odden and Sias ( 1 997) suggest that special peers 
may act as confidantes with whom to discuss bad experiences at work and unpleasant 
experiences with supervisors. 
59 
Literature Review 
Informal relationships should still have an effect on intention to leave however, both 
because the presence of negative relationships is a reason people seek to leave their 
organisation, and because strong relationships make it harder for people to leave. Social 
involvement may be characterised as interactions and relationships with other members 
of the organisation. This involvement is intrinsically rewarding, so that the decision to 
depart from the organisation (and leave these significant others) is made more difficult. 
In addition Ellingwood (2001 )  states that another finding in the Gallup research (refer 
section 2.3 .3) was that having a best friend at work was strongly related to intention to 
leave. Seventy-five percent of respondents in the Gallup research who had a best friend 
at work planned to be with the company for at least another year, as opposed to only 
5 1  % who did not have a best friend at work. Thus, Figure 9 illustrates the hypothesis 
that friendship prevalence will be antecedent to leaving intention, a "-" sign indicates 
the negative relationship between the two variables. 
Figure 9: Showing the hypothesised relationship between friendship prevalence and 
intention to leave 
60 
/ 
Literature Review 
2.3.6 Gender Differences in Organisational Relationships 
As discussed at the beginning of section 2.3 ,  Fritz ( 1 997) conducted a large study 
looking at differences in men and women's  organisational peer relationships. Findings 
of gender differences in friendships were largely consistent with studies conducted in a 
non-work setting. Fritz claimed that the work setting is, for friendship purposes, similar 
to non-work contexts. Sapadin ( 1 988) too, conducted an investigation of friendship in 
the workplace; she investigated same-sex and cross-sex friendships of 1 56 professional 
men and women 1 7 using a self-report questionnaire consisting of a. r�ting scale and 
open-ended questions. Sapadin's results were also largely consistent with research from 
non-work contexts, finding for example, that women's same-sex friendships were rated 
higher for overall quality, intimacy, enjoyment and nurturance. Men, on the other hand, 
rated their cross-sex friendships higher in these areas. This research can be interpreted 
as suggesting that women's participation in professional roles has not resulted in 
noticeable changes to friendship patterns. Gender differences in friendship remain 
evident despite new career roles for women. As in the non-work context, findings 
regarding friendships in the workplace generally indicate that friendships with women 
are rated as more enjoyable, nurturing and of an overall higher quality (by both women 
and men) (Sapadin, 1 988). 
In contrast to this however, Markiewicz, Devine, and Kausilas (2000) state that some 
findings related to gender differences in friendships are not carried over into a 
workplace context. While research has shown that friendships involving women outside 
the work environment are consistently evaluated as more satisfying, Markiewicz et al. 
found this not to be the case .in an organisational setting, instead they found that 
friendships involving women provided less rewards and were less satisfying. 
Thus, it is possible that the work context may influence the nature of relationships 
between men and women. Factors that might account for this include women's 
relatively lower status in many professions, making them less desirable as friends, and 
sex-role stereotypes leading to unfavourable attributions about work-related 
competencies. In addition, the relative proportion of women in a given occupation may 
17 Respondents were from New York, Los Angeles and Boston (U.S .A.). 
61  
Literature Review 
mean their token status might make women less able to provide salient rewards, as they 
tend to have less power (Ibarra, 1 993). One explanation for the conflicting findings is 
the notion that gender is socially constructed. Organisations may differ in the meanings 
ascribed to being 'male' or 'female' such that being female is linked to different 
expectations and behaviours in one organisation versus another, or in an organisation 
versus other aspects of society. 
As in other contexts, women in organisations have been found to place more importance 
on their friendships than do men. Further, it is possible that the organisational 
commitment of women may be more affected by affiliation opportunities than that of 
men. Evidence that friendships have more salience for women can be seen in a study 
examining work attitudes of dentists which found that affiliation opportunities such as a 
highly cohesive work environment was related positively to commitment among female 
(but not male) dentists (Kaldenberg et aI ., 1 995). The authors state " . . .  affective 
commitment was related to proportion of friends in the workplace for female dentists" 
(p. 1 37 1 ) . 
This may point to a gender difference in the commitment process, which relates to 
generalisations that males seek independence while women seek intimacy (Kaldenberg 
et aI. ,  1 995). Similarly, in an investigation using students in Miami, Ashton and Fuehrer 
(1 993) found that males reported a significantly lower likelihood of seeking emotional 
support than instrumental support. Instrumental support is the basis of the information 
peer relationship, while the definition of a collegial peer or special peer includes both 
instrumental and emotional support (Kram & Isabella, 1 985). Thus, men may be more 
likely to form larger proportions of the less intimate, information peer relationships, 
while women may develop larger proportions of collegial and special peer relationships. 
If men and women differ in their quests for intimacy, emotional support and 
independence it make sense that these differences would be apparent in how each 
gender grounds its commitment. A possible gender difference in the importance or 
salience of organisational relationships is a question that will be addressed in the current 
research (refer Chapter Four). 
62 
Literature Review 
2.3.7 Negative Relationships 
Friendships can degrade and turn sour. When this happens in a workplace the 
individuals co�cerned often have to continue to interact. Although no standard 
definition of such relationships yet exists, they can be characterised by conflict and 
disagreement, with communication ranging from " . . .  passive to active dislike, 
animosity, disrespect, or destructive mutual interaction" (Dillard & Fritz, 1 995, p. 1 2) .  
Such relationships have been shown to affect both individuals (Rook, 1 984) and 
organisations (Dillard & Fritz, 1 995) adversely, causing stress and 'turnover. Although 
there is little research to date looking at the effects of negative relationships on 
productivity it seems likely that it would interfere with co-operation and communication 
in work groups, and direct attention and energy away from the task at hand. 
The marked lack of attention to this topic in the current literature may be due to 
researchers' focus on the identification of factors that reduce the stress and anxiety of 
employees. In addition, in organisational literature, there is a lot of attention given to 
how negative relationships might affect, or be dealt with, by managers (Bramson, 1 98 1 ;  
Grove, 1 993 ; Monroe, Borzi, & DiSalvo, 1 992). Research attention given to managers' 
relationships might explain the lack of focus on co-worker relationships. The current 
study will focus on co-worker's negative relationships and how they might impact on 
individuals'  experience of work. 
What are the causes of negative relationships? 
The personality characteristics of individuals are certainly likely to contribute to 
negative relationships in the workplace. For example, 'Type A' behaviour patterns 1 8 are 
related to more difficulty dealing with workplace conflict (Baron, 1 989). Dillard and 
Fritz (1 995) also claim that people who exhibit verbal aggressiveness or have a 
generalised negative outlook (i.e., negative affectivity) are more likely to be involved in 
negative relationships at work. 
18 The core elements of the Type A behaviour pattern are extremes of aggressiveness, easily aroused "free 
floating" hostility, a sense of time urgency and competitively striving for achievement (Mathews, 1 982). 
63 
Literature Review 
The organisation itself may provide the opportunity for negative relationships to form, 
people are rarely in a position to choose who they work with. If an individual 
continually has to interact and work with a person with whom they do not get along, the 
potential for deepening hostility exists (Dillard & Fritz, 1 995). The organisational 
environment may provide other elements conducive to the formation of negative 
relationships. People may have to compete for resources or individuals might have 
incompatible goals. Combined with other factors, such as personality or an unhealthy 
organisational climate, a previously benign relationship could escalate into a hostile one. 
In addition, workplace envy is a factor that is likely to create negative emotion in the 
workplace. Envy is common in businesses and organisations. The way that limited 
resources (such as office space, company cars, promotions, secretarial support and so 
on) are distributed creates an environment where envy is not only possible but almost 
inevitable. Envy may be defined as an emotion occurring when a person begrudges 
another for having something that he or she does not have, or seeing another individual 
gain advantage and viewing it with displeasure (Bedeian, 1 995). Envy implies hostility; 
it is generally viewed as a dangerous emotion, and not one that people will easily 
acknowledge, (even to themselves, often attributing and rationalising behaviour in other 
ways where they can). 
One reason people may be unwilling to admit to envy is that it usually reqUIres 
admitting the other person is in some way superior to them, difficult to do without 
damaging the ego. The discrepancy between the achievements of one colleague and the 
success of another can serve to highlight shortcomings in the first individual; this can 
have psychological consequences such as a drop in productivity and an decrease in job 
satisfaction. Envy can have unpleasant consequences, both for an individual' s  
interactions with others, and for their own well-being. Because they are continually 
comparing themselves to others, enviers will experience unfavourable comparisons 
more often than their less envious colleagues. In an effort to reduce this, these 
individual may sabotage other's work efforts (professional ambush), demean the 
achievements of their colleagues and positively distort their own efforts (Bedeian, 
1 995). 
64 
Literature Review 
Bedeian (1 995) suggests that one way to reduce workplace envy is to become a "strong 
team player" and to adhere to the informal, unofficial workplace "rules." Perhaps 
forming and maintaining friendships in one's work environment is a way to avoid 
becoming a target of hostile envy. Bedeian also recommends not revealing too much 
information about oneself to others in the workplace. However, because disclosing 
personal information IS an important relationship-building technique, withholding 
personal information might interfere with useful workplace friendships !  Being 
somewhat selective about the information employees give each other is probably wise 
however. Becoming close friends with a person with whom you work is a greater risk 
than that taken when embarking on friendships in other contexts. 
Negative relationships will also differ from workplace friendships in that a friendship 
can exist in the workplace between two people from the same organisation who never 
interact professionally or formally :(perhaps they met at the staff Christmas function, 
have developed a friendship, but do not work together). In a case like this role tension 
(which may exist for those who are attempting to maintain a dual relationship at work) 
will probably not exist. A negative relationship, on the other hand is very unlikely to 
occur at work between two people who are not required to work with each other. 
People will seldom be motivated to maintain a negative relationship voluntarily. Thus 
negative relationships will almost always be dual relationships (work-place and 
negative). 
Negative relationships can be added to the model at this point (refer Figure 1 0) .  There 
are no measured antecedents to negative relationships but, as stated above, the presence 
of negative relationships is likely to be positively associated with intention to leave (i.e. ,  
if people experience negative relationships at work they are hypothesised to be more 
inclined to leave their job). These relationships are also likely to adversely effect an 
individuals' experience of work so are hypothesised to be negatively related to job 
satisfaction and organisational commitment. 
65 
Friendship 
prevalence 
Friendship 
opportunities 
+ 
Cohesion 
Literature Review 
Negative 
relationship 
Figure 1 0: Showing negative relationships added to the model 
2.3.8 Blended Relationships: Friends as Work Associates 
It is well established that friendships within the workplace are in existence (along with 
the benefits and other effects of these relationships). Yet despite the frequency of close 
friends who are also work associates, we know very little about how this dual 
personal/role relationship functions. This thesis examines the positive and negative 
impact of these relationships in organisations. In the organisational literature these 
relationships are known as "blended." 
Bridge and Baxter (1 992) conducted a study examining the extent to which employees 
experience dilemmas or contradictions posed by the friendship and work-association 
components of their relationships at work. In this context friendship is conceptualised 
as a role. Though this conceptualisation is somewhat unusual (given that friendships are 
not formally prescribed) friendship relationships do nonetheless entail felt obligations 
and responsibilities for the parties. Potentially incompatible demands associated with 
the role of "friend" and the role of "work associate" are outlined below. Related to this 
66 
Literature Review 
apparent dual role conflict, a recent met a analysis of antecedents, correlates, and 
consequences of commitment to the organisation Meyer and AlIen (2002) found that 
role conflict is strongly negatively correlated with commitment to the organisation. 
Ways in which the expectations of close friendship may contradict with the role-based 
expectations of work-association 
Instrumentality and affection: Friendship can enhance feelings of trust, believability and 
acceptance of feedback. On the other hand, utilitarian support may create feelings of 
indebtedness exploitation or suspicion of another's motives thereby undermining the 
friendship. Reciprocity has been shown to be an important aspect of friendship, as a 
lack of reciprocity is associated with negative affect (Buunk et aI . ,  1 993). A situation 
where one member of the dyad receives more benefits (as in a relationship with an 
organisational superior) will cause a lack of reciprocity and may cause tension for the 
dyad. 
Impartiality and favouritism: Organisational practices usually aim to provide equitable 
treatment for everyone with no personal bias. However people usually expect their 
friends to display special treatment and favouritism, thereby indicating that they regard 
each other as unique and special. 
Openness versus closedness: Friends are expected to be fully open and honest with one 
another, trusting and displaying trustworthiness. However confidentially practices may 
mean that close friends refrain from full disclosure. Disclosing other information for 
work role may violate a friend's  expectation of confidentiality. 
Judgement and acceptance: Friendship is built on an expectation of mutual affirmation 
and acceptance (work-mates may be ideal to provide empathy regarding work-related 
angst) . However, work associates may find themselves in conflict because of 
competing interests associated with their work roles or because of a performance 
evaluation process. 
67 
Literature Review 
Autonomy versus connection: The sheer proximity afforded by the workplace facilitates 
interpersonal attraction between persons. · Hiring practices (normally similar people 
would be hired for similar jobs) mean that they are. likely to perceive themselves as 
similar to their work-mates, which in turn facilitates friendship development and 
maintenance. However, daily contact with the other person may provide too little 
autonomy or separation, thereby jeopardising the friendship (Bridge & Baxter, 1 992). 
Thus, although close friendships at work are hypothesised to have predominantly 
positive consequences, the possibility that friendships at work will result in stress and 
conflict is also present. Dual role tension may be a possible moderator between 
. .  ' -� .friendship prevalence and intention to leave. This is because it is not unreasonable to 
assume that if an employee has experienced increased tension, stress and conflict as a 
result of trying to maintain their workplace friendship, the positive effects of these 
relationships on leaving intention may be reduced. Thus the final variable can be added 
to the model, dual role tension is hypothesised to moderate the relationship between 
friendship prevalence and intention to leave, this is indicated in Figure 1 1  by the arrow 
leading from dual role tension to the arrow linking friendship prevalence and intention 
to leave. Hereafter the completed model below will be referred to as 'Model 1 . ' 
68 
Friendship 
opportunities 
Friendship 
prevalence + 
Dual role 
tension 
(moderalor 
variable) 
+ 
Cohesion 
Literature Review 
Negative 
relationships 
Figure 1 1 :  Showing the hypothesised moderating effect of dual role tension (Model 1 )  
69 
Hospital Study 
Chapter 3 :  Testing the Model in a New Zea.Iand Hospital 
3.1  Introduction 
The previous chapter derived a model of workplace relationships and organisational 
outcomes. The model contains a number of variables from friendship opportunities to 
intention to leave that are linked together, hypothetically, in several ways including 
correlations, direct relationships and mediated relationships. This chapter reports on a 
first attempt to test this model, as a whole, in a hospital setting within New Zealand. 
3.1 .1 A New Zealand Hospital as a Setting for Organisational Research 
Previous organisational research in hospitals 
Hospitals are accepted settings for organisational research worldwide. Studies 
conducted within hospitals, which have focused on topics related to the current study, 
include research into organisational commitment in Canada (Cohen, 1 996, 1 999) the 
United States (Meyer, Allen, & Smith, 1993) and New Zealand (Shouksmith, 1 994), a 
Belgian study focusing on the relatIonship between person organisation fit and turnover 
(Vandenberghe, 1 999), and an investigation of organisational culture and behavioural 
outcomes that was conducted in both Canadian and Tennessee hospitals (Gilbert & 
Sneed, 1 992). There has been English research into nurses' job satisfaction (Tovey & 
Adams, 1 999) and morale (Cochrane & Jowett, 1 994) and research into nurse's job 
satisfaction in the United States (Blegen & Mueller, 1 987); a study investigating the 
relationship between job satisfaction and turnover in Taiwan (Lu et aI., 2002) and one 
investigating the relationship between job satisfaction and the work environment in the 
Middle East (Tumulty, Jernigan, & Kohut, 1 995). Supervisory support and turnover has 
been studied in a New Zealand hospital (Kalliath & Beck, 2001)  and there has been both 
an English study focusing on peer support (Aston & Molassiotis, 2003), and a Finnish 
70 
/ 
Hospital Study 
study focusing on peer evaluation (Vuorinen, Tarkka, & Meretoja, 2000). In addition, 
the work/life balance of hospital consultants was investigated in England (Dumelow, 
Littlejohns, & Griffiths, 2000). This is by no means an exhaustive list of organisational 
research set in hospitals but it is worth noting that, of the articles published, only the 
studies by Shouksmith ( 1 994) and Kalliath & Beck (200 1 )  were based in NZ hospitals. 
No research has specifically examined at the impact of friendships in this setting, 
although Froebe, Deetz and Knox ( 1 983) did ask American nurses to rank the 
importance of factors important to their remaining employed in a hospital and found 
"relations with co-workers" to be ranked as the third most important with 97% of 
administrator and staff nurses rating relationships with co-workers as "very important" 
or "important." 
Wards and teams 
Team dynamics, in wards and elsewhere, are of great interest. People in hospitals often 
work closely together in what can be a highly stressful, charged environm�nt or, at other 
times (for example the night shift), one which might involve a lot of 'down time' with 
just one other person. Both of these environments are conducive to the formation of 
close relationships, first because people tend to 'bond' in times of adversity (Aronson & 
Cope, 1968; Carr, 2003 ; Sias & Jablin, 1 995), and second because propinquity is an 
important antecedent to relationships forming (Adams & Blieszner, 1 994). On the other 
hand, close proximity may be as likely to create strongly negative relationships (e.g., 
through forced exposure to an individual one might not choose to spend time with) as 
strong and enduring friendships. 
New Zealand context 
Of the research examining friendships in the workplace, none has been placed within a 
New Zealand context. The effects of interpersonal relationships, and the relationships 
themselves, may vary across cultural boundaries. Until there is New Zealand-based 
data and studies pertaining specifically to New Zealand organisations, it is impossible to 
know with any certainty whether findings from overseas are applicable here. 
7 1  
Hospital Study 
3.1 .2 Research Questions 
The model itself poses three general research questions, which address the key enquiry 
of this thesis. First, how do friendships at work relate to organisational outcomes (such 
as satisfaction, commitment and turnover)? Second, how does the perceived level of 
cohesion relate to friendships at work and organisational outcomes? And third, what are 
the effects of negative relationships in the workplace? This study aims· to answer each 
of these questions empirically. The research questions and hypotheses are described 
below. 
Research Question 1 
Do the opportunities for, and prevalence of, friendships within a hospital have an impact 
(either mediated or direct) on job satisfaction, organisational commitment, and intention 
to leave? The discussion on the effects of relationships in the workplace (Chapter Two) 
is the focus of the first set of hypotheses. 
Hypothesis 1 a & 1 b: That having more opportunities for friendship will have a positive 
relationship with job satisfaction and organisational commitment. 
Hypothesis 1 c: That having more opportunities for friendship within the workplace will 
have a negative relationship with intention to leave. 
Hypothesis 1 d & 1 e: That the prevalence of friendships at work will not be related to 
job satisfaction or organisational commitment. 
Hypothesis 1f That the prevalence of friendships at work will have a negative 
relationship with intention to leave. 
Research Question 2 
Is the perceived level of cohesion related to opportunities for friendship, prevalence of 
friendship, job satisfaction, organisational commitment and intention to leave? The 
discussion on the relationship between cohesion and these variables (Chapter Two) is 
the focus of the second set of hypotheses. 
Hypothesis 2a: That cohesion will be positively related to opportunities for, and 
prevalence of, friendship in the workplace. 
72 
Hospital Study 
Hypothesis 2b & 2c: That cohesion will have a positive relationship with job 
satisfaction and organisational commitment. 
Hypothesis 2d: That cohesion will have a negative relationship with intention to leave. 
Research Question 3 
Do negative relationships in the workplace worsen organisational commitment, job 
satisfaction and intention to leave? The discussion on the relationship between negative 
relationships and these variables (Chapter Two) is the focus of the third set of 
hypotheses. 
Hypothesis 3a: That the prevalence of negative relationships within the workplace will 
be negatively related to job satisfaction. 
Hypothesis 3b: That the prevalence of negative relationships within the workplace will 
have a negative relationship with organisational commitment. 
Hypothesis 3e: That the prevalence of negative relationships within the workplace will 
have a positive relationship with intention to leave. 
3.1 .3 Research Methodology 
Experimental designs tend to have high internal validity; often generating definitive 
findings and establishing causality, i .e. , researchers using the experimental method are 
often able to establish whether or not the independent variable really does affect the 
dependent variable. However, experimental research and laboratory studies in 
particular, may not have external validity. That is to say, it is questionable how 
generalisable findings might be outside the laboratory environment (Bryman, 1 989). 
Survey research, on the other hand, enables researchers to collect data relating to a 
number of different concepts from a large number of individuals. While obtaining a 
representative sample may be a problem (due to issues such as non-probability sampling 
and non-response bias), surveys do allow researchers to examine data with respect to a 
number of variables to discern patterns of association. However, because data from 
surveys are usually collected at a single juncture, researchers can seldom establish 
causality with certainty (Bryman, 1 989). 
73 
Hospital Study 
Scholars studying friendship have acknowledged the need to employ, in conjunction 
with quantitative techniques, qualitative methods that permit detailed open-ended 
questioning about friend interactions and the effects and meanings people ascribe to 
them (Adams & Blieszner, 1 994; Blieszner & Adams, 1 992; Rubin, Perse, & Barbato, 
1 988). Both qualitative and quantitative methods are incorporated in the current study. 
These include pen and paper tools such as the Organisational Commitment 
Questionnaire, and a scale measuring the presence and importance of workplace 
relationships to respondents (Nielsen et al. ,  2000). In addition, a qualitative technique, 
(open-ended questions) will be utilised. 
3.2.1 Participants 
3.2 Method 
Several large New Zealand organisations were approached to assess their interest in 
being involved in the first study. Waitemata District Health Board expressed interest 
and was selected. 
The population sampled for the first study consisted of employees of the Waitemata 
District Health Board, which is spread over three sites in Auckland (New Zealand). All 
employees of the Waitakere Hospital site, except those working in mental health, were 
included in the population, as were all employees of the Community and Disability 
Service (a service which has employees at all sites) 1 9. The employees in the sample 
potentially included all hospital staff; doctors, nurses, therapists, hospital aids and 
caregivers and clerical staff. Of these employees, every 5th person on payroll (listed 
alphabetically) received a questionnaire through the internal mail. 
Of the 408 questionnaires sent out, 1 24 were returned (a 30% response rate). The 
demographics of the sample are shown in Table 1 (refer appendix 1 0  for the sheet used 
to gather demographic information and pie charts showing the results). 
19 The total N of the sampled population was 2040. 
74 
Hospital Study 
Table 1 :  Oemographics of Respondents of the Hospital Study 
Variable Frequency (n) Valid percent 
Sex (1 missing) 
Males 6 5 
Females 1 1 8  95 
Age (3 missing) 
20-29 years 16  13 .2  
30-39 years 16  1 3 .2  
40-49 years 53 43.9 
50-59 years 24 19 .8  
Over 60 years 12  9.9 
Ethnicity (4 missing) 
NZ European or New Zealander 107 89.2 
Maori 6 5 .0 
Pacific Island 4 3.3 
Chinese 1 0.8 
Canadian 1 0.8 
Filipino 0.8 
Role (1 missing) 
Clerical 8 6.5 
Technical 4 3 .3 
Clinical 97 78.9 
Management 1 1  8 .9 
Other 3 2.4 
Tenure (0 missing) 
> 1 year 14 1 1 .6 
1-5 years 50 40. 1  
5- 1 0  years 17  14 
10- 1 5  years 14  1 1 .6 
1 5-20 years 17  13 .4  
20+ years 1 0  8 . 1 
Note: not all people answered all questions; the number of respondents declining to answer each question 
is indicated in Table 1 .  
Because they worked in  a hospital, many of  the respondents were from predominantly 
traditionally female occupations; nurses and clerical staff made up 84.7% of the sample. 
As a result, the sample was almost entirely female (95%).  In addition, most were over 
, the age of 40 (75 .8%), the mean age was 44, most self-reported as being of "European 
descent" (89.2%). Almost halfhad been at Waitemata DHB for 5 years or longer. 
3.2.2 Materials 
Data were gathered using a self-administered questionnaire (refer appendices 1 - 1 0), 
which consisted of an information sheet, several instruments and survey questions. The 
instruments and survey questions are as follows. 
75 
Hospital Study 
Workplace Friendship Scale 
Prior to investigating the possible effects of friendships in the workplace, it is first 
necessary to establish if, in fact, friendship relationships are present in the workplace. 
To do this, the Workplace Friendship Scale (Nielsen et aI. ,  2000) was incorporated into 
the final questionnaire. The scale measures two aspects of work place friendship: (a) the 
opportunity for friendship (e.g., I have the opportunity to get to know my co-workers), 
and (b) the prevalence of friendship (e.g., I have formed strong friendships at work). 
There are twelve items, rated on a 5-point likert scale from strongly disagree to strongly 
agree (refer appendix 2 for the items in the workplace friendship scale). 
The items in the workplace friendship scale were derived by Nielsen et at (2000) from 
previous workplace friendship measures (Riordan & Griffeth, 1 995;  Winstead et aI. ,  
1 995;  Wright, 1 969, 1 974) and a literature review of previous social psychological 
research. Nielsen et al. conducted item analysis to assess how each item related to the 
overall scale. The items in the scale are the 1 2  (of an original 3 5) that correlated most 
highly with the total sub-scale score. 
Nielsen et al. (2000) assessed the internal consistency reliability of the scores of the two 
six-item sub-scales of the Workplace Friendship Scale (i.e., "prevalence of friendship" 
and "opportunities for friendship"). The internal consistency reliability estimates 
(Cronbach's alpha) for the scores were .84 and .89 for the 'opportunities for friendship' 
and 'prevalence of friendship' sub-scales respectively. In addition, Nielsen et al. 's 
factor analysis supported a two-dimensional scale structure. 
After determining the internal consistency reliability and dimensionality of the scale 
scores, Nielsen et al. (2000) conducted a second study to provide evidence of construct 
validity. Construct validity was assessed through convergent validity evidence, 
discriminant validity evidence, and the predictions that comprised the proposed 
nomological network (Cronbach & Meehl, 1 955). Nielsen et al. (2000) believe 
adequate evidence was provided for the construct validity of the scale. Based on their 
results, it appears that the newly developed scale yields internally reliable scores, and 
effectively measures both friendship opportunities and friendship prevalence. 
76 
Hospital Study 
Relationships in the workplace questionnaire 
As part of the questionnaire for the current study, respondents were also given the 
definitions of the three peer types identified by Kram and Isabella ( 1 985) (i .e. ,  special 
peer, collegial peer, and information peer) and asked to indicate how many peers of 
each type they had at work. Negative relationships were also defined, and respondents 
indicated how many negative relationships (if any) they had (refer appendix 3 for the 
definitions given to respondents). 
Dual Role Tension Questionnaire 
As discussed in section 2.3 .8 ,  Bridge and Baxter conducted a study examining the 
extent to which employees experience dilemmas or contradictions posed by the 
friendship and work-association components of their relationships at work. An 
adaptation of the Dual Role Tension Questionnaire, designed by Bridge and Baxter 
( 1 992), was included to measure the dual role tension experienced by individuals who 
have blended relationships, i .e. ,  relationships which occur when people have a work­
associate who is also a close friend (e.g., Our relationship would be a lot easier if we 
were only friends or only work associates instead of being both; refer appendix 4 for the 
items in the Dual Role Tension Questionnaire). The six-items used in the current study 
are those that were found to tap into overall dual role tension, Bridge and Baxter state 
that these items load together as a single factor and found the six-item index to have an 
alpha reliability of .85 .  
Workgroup Cohesion Scale 
Cohesion was measured using a nine-item workgroup cohesion scale, rated on a 5 -point 
Likert type scale from strongly disagree to strongly agree (e.g., Members of my team are 
very willing to share information with other team members about our work). Items 
measuring this variable were selected from a 54 item Work Group Characteristics 
Measure developed by Campion (1 993). Only those items from the Work Group 
Characteristics Measure relating to cohesion were used in the current study. The items 
are those relating to ( 1 )  Social Support, (2) Workload Sharing and (3) .,' 
Communication/Co-operation within the work group (refer appendix 5 for the items in 
the workgroup cohesion scale). Campion et al. provided evidence that a composite of 
77 
Hospital Study 
these items reliably predicted effectiveness criteria (productivity and manager 
judgements of effectiveness, (p < .05)). In addition the sub scales had adequate internal 
consistency reliability ( a  = .78, .84 and . 8 1  respectively). 
Organisational Commitment Questionnaire 
To assess the organisational commitment of respondents the Organisational 
Commitment Questionnaire was included in the survey (refer appendix 6 for the 1 5  
items in the Organisational Commitment Questionnaire). This is the most commonly 
used measure of employee's organisational attachment to an organisation (Meyer & 
AlIen, 1 99 1 ). The Organisational Commitment Questionnaire IS a 1 5-item scale, 
designed to assess acceptance of organisational values, desire to remain with the 
organisation and willingness to exert effort (e.g., I am proud the tell others I am part of 
this organisation). Items are rated on a 7-point Likert scale from strongly disagree to 
strongly agree. Although it  was validated with a western population in mind, recent 
research using the Organisational Commitment Questionnaire has provided evidence for 
its generalis ability and validity in the measurement of organisational commitment in 
non-western populations such as China (Siu, 2002), the United Arab Emirates (Yousef, 
2003) and South Korea (Lee, AlIen, & Meyer, 200 1 ). In addition Mowday and 
colleagues have provided evidence for the test-re-test reliability, convergent and 
discriminant validity, internal consistency, and predictive validity of the Organisational 
Commitment Questionnaire (Mowday, Steers, & Porter, 1 979). 
For instance Mowday et al. found the overall measure of organisational commitment to 
be relatively stable over time (r = .53, .63 and .75 over 2-, 3- and 4-month periods), 
demonstrating good test-re-test reliability. Significant correlations were found between 
Organisational Commitment Questionnaire scores and "intention to remain with the 
organisation", illustrating convergent validity. In addition, the Organisational 
Commitment Questionnaire was found to correlate significantly with scores from the 
Organisational Attachment Questionnaire (convergent validities across six diverse 
samples ranged from .63 to .74) (Mowday et al. ,  1 979). Meyer and Allen ( 199 1 )  also 
examined convergent validity, finding that the Organisational Commitment 
Questionnaire correlated significantly with their measure of normative commitment (r = 
78 
Hospital Study 
. 5 1 ). Evidence was reported for a consistent inverse commitment-turnover relationship 
with correlations, ranging from - . 1 7  to -.43, demonstrating the predictive validity of the 
measure. Mowday and colleagues calculated internal con"sistency using coefficient a ,  
item analysis and factor analysis, finding coefficient a to be consistently high, ranging 
from .82 to .93 with a median 'of .90. Item analysis20 indicated that each item had a 
positive correlation with the total score for the Organisational Commitment 
Questionnaire, with the range being from .32 to .72. Mowday et al. ' s  ( 1 979) original 
factor analysis (rotated to Kaiser varimax solution) resulted in a single factor solution 
and suggested the 1 5  items of the Organisational Commitment Questionnaire are 
relatively homogeneous with respect to the underlying attitude construct they measure. 
Since Mowday et al. ' s  study, several researchers have supported the notion that the 
Organisational Commitment Questionnaire measures a single common underlying 
construct. For example Ferris and Aranya ( 1 983) employed factor analysis (varimax 
rotation with Kaiser normalisation) to analyse data collected from 1 , 1 05 Canadian 
accountants. Ferris and Aranya found the I S-item scale yielded a single factor. 
Morrow and McElroy ( 1 986) used data collected from 536 supervisors in a public 
agency in the mid western United States ;  they also found that items on the 
organisational commitment scale used in the current study loaded together as a single 
factor (varimax rotation). 
There have been other researchers, however, who have found support for the notion that 
the Organisational Commitment Questionnaire is a multidimensional measure of 
organisational commitment. For example Barhayim and Berman ( 1 992) examined the 
dimensions of organisational commitment using data collected from 1 ,299 industrial 
workers in Israel. Barhayim and Berman's results revealed organisational commitment 
to have two dimensions; they made a distinction is between ' loyalty to the organisation' 
and 'identification with and readiness to exert effort on its behalf. ' Barhayim and 
Berrnan state that this is a distinction between passive and active organisational 
commitment. Angle .and Perry ( 1 98 1 )  studied a sample of 1 244 bus-service employees 
and 96 transit managers in twenty-four organisations in the Western United Sates. 
20 Correlation between each item of the commitment scale and the total score, less the item 
79 
Hospital Study 
Utilising factor analysis, Angle and Perry also found a two-dimensional 
conceptualisation of organisational commitment, which they labelled 'value 
commitment' and 'commitment to stay. ' In a more recent study using data collected 
from employees from- the United Arab Emirates (two independent samples, one of 430 
and another of 567 employees), Yousef (2003) also found that the Organisational 
Commitment Questionnaire to be a multidimensional concept. 
For the purposes of the questionnaire used in the current study, organisational 
commitment was defined as " . . .  the relative strength of an individual ' s  identification 
with, and involvement with, a particular organisation" (Mowday et aI., 1 979, p.226). As 
discussed in Chapter Two, organisational commitment can be characterised by a strong 
belief in and acceptance of the organisation's goals and values, a willingness to exert 
effort on behalf of the organisation and a strong desire to maintain membership in the 
organisation. Exploratory factor analysis in the current study indicated that the items on 
the Organisational Commitment Questionnaire generally loaded together as one factor, 
the coefficient a for the items remaining after factor analysis for the current study was 
.87, indicating that, for this sample, the items used to measure organisational 
commitment can be used as a single factor. 
Measure of job satisfaction 
Job satisfaction is the degree to which a person reports satisfaction with intrinsic and 
extrinsic features of their job; it is also defined as a positive emotional state resulting 
from the appraisal of aspects of one' s job (Levy, 2003) .  The job satisfaction scale used 
in the current study was one part of a larger battery of eight scales devised by Warr, 
Cook and Wall ( 1 979). The whole battery included measures of work involvement, 
intrinsic job motivations, higher order need strength, perceived intrinsic job 
characteristics, job satisfaction, life satisfaction, happiness and self-rated anxiety. Only 
the 16-item scale relating to job satisfaction was used for this study. Respondents 
indicate how satisfied or dissatisfied they feel with each of 1 5  aspects of their job (e.g. 
The recognition you get for good work). Items are rated on a 7-point Likert type scale 
from very dissatisfied to very satisfied. Total job satisfaction is the sum of all the 
80 
.-
Hospital Study 
separate items. There is also a 16th item, which is a measure of satisfaction with the job 
as a whole (refer appendix 7 for the 1 6  items in the Job Satisfaction Questionnaire) . 
Warr, Cook and Wall ( 1 979) reported the test re-test reliability of their job satisfaction 
measure (respondents in the study were tested six months after their initial interview). 
The test-retest correlation co-efficient was shown to be .63 for total job satisfaction, but 
only .27 for overall job satisfaction, the single item measure. For this reason total job 
satisfaction will be used in the current study. Warr, et aI . (1 979) also note that, although 
the scale can be factor analysed into several job satisfaction clusters, (for example 
'extrinsic satisfaction' and 'intrinsic satisfaction' items were identified by Warr, et al. )  
they are closely and positively related to one another, suggesting that separate subscale 
scores will not always warrant calculation. 
Although Warr et all found, using cluster analysis, that items clustered together into 
intrinsic and extrinsic satisfaction subscales; exploratory factor analysis in the current 
study indicated that the items on the Job Satisfaction Scale generally loaded together as 
one factor, the coefficient a for the items remaining after factor analysis for the current 
study was .83,  indicating that, for this sample, the items used are sufficiently 
homogeneous with respect to the construct they measure, to use as a single factor. 
Measure of intention to leave 
Quitting a job is often portrayed as a carefully considered multi-step decision 
culminating in an intention to quit (Fisher, 2002; Mobley, 1 977; Sager, Griffeth, & 
Hom, 1998). Intention to leave has been described as the last in a sequence of  
withdrawal cognitions, along with thinking of quitting and intent to search for 
alternative employment (Mitchel, 1 98 1 ;  Mobley, 1 977; Mobley, Homer, & 
Hollingsworth, 1 978; Tett & Meyer, 1 993). It is often measured with reference to a 
specific interval such as "within the next 6 months" or "within the next year." In this 
study intention to leave was measured with three items theorised to be important 
precursors to turnover; thinking of quitting, intention to search for alternative 
employment, and intention to quit (Chang, 1999; Mobley, 1 977; Mobley et aI . ,  1 978) 
(e.g. I will probably quit my job in the next year). Some or all of these three items have 
8 1  
Hospital Study 
been used by researchers in some form for the past few decades (e.g., Chang, 1 999; Lu 
et aI . ,  2002; Rambur, Palumbo, McIntosh, & Mongeon, 2003 ; Sager et aI. ,  1 998). 
Answers to each item were recorded on a seven-point likert scale (refer appendix 8). In 
the current study the internal reliability of the three items was good, (coefficient a. = 
.82). 
Open ended questions 
Six open-ended critical incident questions were used to give respondents an opportunity 
to comment on the impact workplace friendships and negative relationships had on their 
experience of work, with the aim of providing the researcher with a source of qualitative 
data to provide further insight to quantitative findings. In these, respondents outlined 
how their informal interpersonal relationships either improved or negatively affected 
their work environment and, if appropriate, how the workplace might have affected their 
relationships (refer appendix 9 for the actual questions asked of respondents). 
3.2.3 Procedure 
Ethics approval was applied for and obtained from the Massey University Human Ethics 
Committee (MUHEC) and from Waitemata District Health Board. Every fifth person 
(listed alphabetically) on the payroll of the Waitakere Hospital site (except those 
working in mental health) and all employees of the Community and Disability Service 
were selected to receive an information sheet and questionnaire in the organisation's 
internal mail system (refer appendices 1 - 10). A full information sheet was delivered 
along with the questionnaire, so submission of the questionnaire was taken as informed 
consent. In order to maintain confidentiality there was no way to identify individual 
\ 
respondents. A stamped envelope, addressed to the researcher was included with each 
questionnaire for ease of return. In total, 419  questionnaires were sent out through the 
internal mail system of the Waitemata District Health Board. Of the 419  sent out, 1 1  
were not delivered as the staff members had left the organisation. Thus the total number 
of questionnaires delivered to employees at Waiterp.ata DHB was 408. 
One month later, a follow up letter was sent to the original list of 419  employees (refer 
appendix 1 1 ) .  This letter thanked those who had returned their questionnaires and 
82 
Hospital Study 
reminded the others that they could still submit theirs if they wished to participate. 
Fourteen participants responded to the second call, bringing the final return rate up to 
the figure of 30% reported in section 3 .2 . 1 .  
3.3 Analysis and Results 
3.3.1 Imputing Missing Values 
Respondents 1 6  and 1 0 1  failed to complete the cohesion scale so these two respondents 
were removed from all further analysis (leaving the number of respondents at 1 22). 
Prior to computing the scales, missing values (items which individuals failed to respond 
to) were imputed. First the responses to the five scales (cohesion, friendship, 
commitment, satisfaction and intention to leave) from all remaining respondents were 
placed into separate SPSS files. Following this, the inversely worded items from the 
various scales were reversed2 1 . Finally, missing values for each scale were imputed 
using the 'missing value analysis' option in the analyse menu of SPSS, which fills in 
(imputes) missing values with estimated values using regression22. 
3.3.2 Factor Analysis 
There is the possibility of some overlap in the constructs measured by the questionnaire 
in the current study. If this were the case it would "force" a relationship between 
variables. Exploratory factor analysis was carried out to verify that the measures were 
both internally coherent and conceptually and empirically distinct from each other. 
Because overlap in constructs is only likely for conceptually similar variables, the three 
measures of relationship quality (workplace friendships, measure of work group 
cohesion and the dual role tension questionnaire) were factor analysed together, as were 
the three outcome variables (organisational commitment, job satisfaction and intention 
to leave). 
21 Inversely worded items �ere; item 1 2  of the workplace friendship scale, item 4 of the dual role tension 
questionnaire, and items 3, 7, 9, 1 1 , 12 and 15 of the organisational commitment questionnaire. 
22 There were few missing values in these scales. Eighteen values in total were missing; three from the 
cohesion scale, two from the friendship scale, four from the commitment scale, nine from the satisfaction 
scale and none from the intention to le.ave scale. 
83 
Hospital Study 
Relationship-quality measures (friendship, cohesion and dual role tension measures) 
The friendship, cohesion and dual role tension measures from the questionnaire were 
factor analysed together. As stated above, it is possible that the scales tap into similar 
constructs, so the analysis is carried out to ensure that the scales are internally coherent 
and also distinct from the other relationship-quality measures. The 27 items (i.e., a 1 2  
item Friendships scale, a 6 item Dual Role Tension (Dual Role Tension) scale and a 9 
item Cohesion scale) were factor analysed with a maximum likelihood extraction23 and 
direct oblimin rotation24• The solution was set for 4 factors to represent the four 
components (i.e. , cohesion, dual role tension, friendship opportunities and friendship 
prevalence). This four-factor construct was suggested by the scree plot (refer Figure 
1 2. 1 ;  appendix 1 2), which shows four clear factors. 
The Kaiser-Meyer-Olkin Measure of Sampling Adequac!S had a value of .83 (higher 
than the criterion of .7). The resulting matrix accounted for 52.9% of the variance. 
Most of the items loaded on the hypothesised factor (refer Table 2). For example, all 
the 'dual role tension' items loaded on one factor, all the items measuring friendship 
prevalence loaded on one factor and all but one of the items measuring friendship 
opportunities loaded on one factor. All but two of the cohesion scale items loaded on 
one factor; items 2 and 3 of the workgroup cohesion scale "My team increases my 
opportunities for positive social interaction " and "Members of my team help each other 
out at work when needed" cross-loaded on the friendship prevalence factor. The two 
cross loading cohesion items measure social support, and presumably tap into much the 
same construct as the friendship scale does. 
23 Maximum likelihood factor analysis, originally introduced by Lawley ( 1 940), was chosen because it is 
based on a flnn mathematical foundation that allows hypothesis testing when nonnality is assumed (Chen, 
2003). Statistical tests using maximum-likelihood estimation, assume that the data are multivariate 
nonnal (Reise, Wailer, & Cornrey, 2000). Because the data obtained from the likert-type scales in the 
current study are multivariate normal (in contrast to dichotomously scored items, for example) maximum 
likelihood factor analysis was appropriate. In addition, Cudeck and O'Dell ( 1 994) state that the properties 
of the maximum likelihood estimator, especially with regard to estimated standard errors for rotated 
loadings, have been most completely developed, calling it a "good statistical method." 
24 Because the factors may be correlated with one another, direct oblimin was chosen as the rotation 
method. Direct oblimin is an oblique rotation and was selected over orthogonal rotation, as orthogonal 
rotation is based on the assumption that the factors are uncorrelated. 
25 The Kaiser-Meyer-Olkin measure of sampling adequacy tests whether the partial correlations among 
variables are small. The measure needs to be higher than the criterion of . 7 to indicate sampling adequacy. 
84 
Hospital Study, 
Items 1 ,  2 and 3 of the cohesion scale measure social support (Campion et aI . ,  1 993), a 
construct that is also likely to be measured by the friendship scale. The relationship 
between the cohesion and friendship constructs will be measured, and leaving these 
items in may force a relationship between the constructs. In order to make the eventual 
data analysis as conservative as possible, all three were removed. In addition, items 1 
and 5 of the workplace friendship scale "I have the opportunity to get to know my co­
workers 11 and "I have the opportunity to develop close friendships at my workplace 1 1  
were cross loading between the two friendship factors (prevalence and opportunities) so 
were also removed. Item 4 of the Dual Role Tension scale had an absolute value less 
than .3 so was also dropped. The "dropped" items, those that did not load on their 
hypothesised factor, cross-loaded with other factors or had a factor loading with an 
absolute value less than .3 are shaded in grey for ease of identification. 
85 
Hospital Study 
Table 2: First Pattern Matrix of the Factor Analysis of the Friendship, Cohesion and Dual 
Role Tension Measures items to be removed are shaded in 
Dual role tension 2 
Dual role tension 1 
Dual role tension 3 
Dual role tension 6 
Dual role tension 5 
Workplace Friendship Scale 
Workplace Friendship Scale 
Workplace Friendship Scale 
Workplace Friendship Scale 
Our relationship would be a lot easier if we were only 
friends or only work associates instead of being both 
Overall, the friendship half and the work half of our 
relationship interfere with each other, creating problems 
for us 
Our work relationship and our friendship are often in 872 
conflict with one another 
My friend and I have lowered our expectations about 736 
what we expect as both friend and co-worker in order to 
maintain our relationship 
. 
It requires extra effort to maintain both the friendship -,599 
side and the work side of our H:;l':1Ll�I11"IWIl 
In my organisation 
informally and visit with others 
Communication among employees is encouraged by my 
organisation 
I am able to work with my co-workers to collectively 
solve problems 
Informal talk is tolerated by my organisation as long as 
the work is completed 
�grMrRt[8]tf.li;�!iitC§;� ]�}l1 'f,Bl�;i;'ii'/F"" ; ' '''RMll''�;f&�'>' i1;t']�iiliW,/'<' ''''f(;"b��'�k';;;-� �!; &;�,�  .. 9,HP&=,�.Jy..� .,g� .. AL� _!'tm�'��:l.d�L�i! 
W orkgroup cohesion 9 Members of my team cooperate to get the work done 
W orkgroup cohesion 7 Members of my team are very willing to share 
W orkgroup cohesion 5 
W orkgroup cohesion 6 
W orkgroup cohesion 8 
Workplace Friendship Scale 
Workplace Friendship Scale 
Workplace Friendship Scale 
Workplace Friendship S 
1 0  
Workplace Friendship 
1 2  
W orkplace Friendship 
1 1  
information with other team members about our work 
No one on my team depends on other team members to 
do the work for them 
Nearly all the members of my team contribute equally to 
the work 
Everyone on my team does their fair share of the work 
"�I�ritlThf�W)";�;:St(r��··�If�fl.�eiicrW?lih��N��i,)*76,fiGI"!�;t:�� ; ;l'#."'�i «",;>,_�'.. ,¥k. �iID!1(''''i","Jli&'>'L_ ''m!ili.U,''''��-''W!EM 
bS12!l.9. 
Teams enhance the communication among people 
I have formed strong friendships at work 
I can confide in people at work 
I socialise with co-workers outside the workplace 
I feel I can trust many co-workers a great deal 
I do not feel that anyone I work with is a true friend 
,932 
.699 
,6 10  
.427 
"�� ;l..k . 
� 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 1 1  iterations. Loadings less than .3 are not shown. 
86 
� 
.710 
,710 
.666 
.646 
.640 
� t . . m 
.42 1 
m ," , 
. 8 1 1  
.682 
.677 
.606 
.585 
.573 
III 
m! ""'M 
Hospital Study 
With the 6 "dropped" items not included in the analysis, the solution accounted for 
56.4% of the variance, and the items in each scale loaded together as single factors. The 
Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a value of .79. One final item 
(item 8 of the workgroup cohesion scale) "Teams enhance the communication among 
people working on the same product/toward the same goals " was cross loading with the 
workplace friendship scale, so it was also removed. 
Table 3 shows the final factor analysis, with the solution still set to four factors to 
represent the four components (friendship prevalence, friendship opportunities, dual role 
tension and workgroup cohesion). The solution accounted for 57 .464% of the variance, 
and revealed four distinguishable factors representing the four scales. The scree plot for 
the final factor analysis is shown in Figure 12 .2, appendix 1 2. The Kaiser-Meyer-Olkin 
Measure of Sampling Adequacy had a value of .80. 
87 
Hospital Study 
Table 3: Final Pattern Matrix of the Factor Analysis of the Friendship, Cohesion and Dual 
Role Tension Measures 
Scale Item 
Dual role tension 2 
Dual role tension I 
Dual role tension 3 
Dual role tension 6 
Dual role tension 5 
W orkgroup cohesion 5 
Workgroup cohesion 4 
Workgroup cohesion 7 
Workgroup cohesion 6 
Workgroup cohesion 9 
Workplace Friendship Scale 7 
Workplace Friendship Scale 9 
Workplace Friendship Scale 8 
Workplace Friendship Scale 1 0  
Workplace Friendship Scale 1 1  
Workplace Friendship Scale 1 2  
Wordinl! Factor 
Our relationship would be a lot easier if we were only 
friends or only work associates instead of being both 
Overall, the friendship half and the work half of our 
relationship interfere with each other, creating 
problems for us 
Our work relationship and our friendship are often in 
conflict with one another 
My friend and I have lowered our expectations about 
what we expect as both friend and co-worker in order 
to maintain our relationship 
It requires extra effort to maintain both the friendship 
side and the work side of our relationship 
No one on my team depends on other team members 
to do the work for them 
Everyone on my team does their fair share of the 
work 
Members of my team are very willing to share 
information with other team members about our work 
Nearly all the members of my team contribute equally 
to the work 
Members of my team cooperate to get the work done 
I have formed strong friendships at work 
I can confide in people at work 
I socialise with co-workers outside the workplace 
I feel I can trust many co-workers a great deal 
Being able to see my co-workers is one reason I look 
forward to my job 
I do not feel that anyone I work with is a true friend 
1 
-.936 
-.929 
-.870 
-.725 
-.592 
2 
.724 
.679 
.66 1 
.659 
.592 
3 
.760 
.7 1 1  
.700 
.684 
.636 
.579 
4 
Workplace Friendship Scale 3 In my organisation I have the opportunity to talk -.885 
informally and visit with others 
Workplace Friendship Scale 4 Conununication among employees is encouraged by -.7 1 8  
my organisation 
Workplace Friendship Scale 2 I am able to work with my co-workers to collectively - .614 
solve problems 
Workplace Friendship Scale 6 Informal talk is tolerated by my organisation as long -.43 1 
as the work is completed 
Extraction Method: Maximum Likelihood. Rotation Method: Oblirnin with Kaiser Normalization. 
Rotation converged in 9 iterations. Loadings less than .3 are not shown. 
Table 3 clearly shows that all of the remaining items loaded on the expected factor, 
indicating that the scales are both internally coherent, and distinct from each other. The 
Cronbach's alpha reliability estimates of the friendship opportunities, friendship 
prevalence, dual role tension and cohesion subscales are all acceptable, (a = .76, .86, 
.91 and .82 respectively). The resulting scales were used in all subsequent analyses by 
obtaining a mean score on each scale for all respondents. 
88 
Hospital Study 
Outcome variables (organisational commitment, job satisfaction and leaving intention) 
An exploratory factor analysis on the three outcome measures (organisational 
commitment, job satisfaction and leaving intention) was performed. As with the 
relationship-quality measures (above), it is possible that the three scales measuring 
organisational outcome variables would tap into similar constructs, so the analysis was 
carried out to ensure that the scales are internally coherent and also distinct from each 
other. The 33 items (i.e., a 1 5  item Organisational Commitment Questionnaire, a 1 5  
item Job Satisfaction Scale and a 3 item Intention to Leave Scale) were factor analysed 
with a maximum likelihood extraction and direct oblimin rotation. Because the scree 
plot indicated three factors (refer Figure 1 2. 3 ;  appendix 1 2), the solution was set to three 
factors reflecting the three components (satisfaction, commitment and leaving 
intention). 
The Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a value of .89.  The 
resulting matrix accounted for 42.59% of the variance. Most of the items .loaded on the 
hypothesised factor, thus yielding a reasonably interpretable simple structure (refer 
Table 4). For example, all the Organisational Commitment items loaded on one factor, 
all the Intention to Leave items loaded on one factor and all but four Job Satisfaction 
items loaded on one factor. 
Item 3 of the Job Satisfaction Scale measures satisfaction with relationships in the 
workplace (rate your satisfaction 'with "your fellow workers"). Leaving this item in 
may partially force a relationship between the ' Job Satisfaction' and 'Friendships in the 
Workplace' constructs. In order to make the eventual data analysis as conservative as 
possible, item three was removed. 
Items (4, 9 and 1 1 ) from the Organisational Commitment Questionnaire measure the 
respondents' intention to remain in their current jobs, and tap into the same construct 
measured by the Intention to Leave scale. Organisational Commitment may be one 
predictor of intention to leave; if intention to leave is measured within the organisational 
commitment scale it will artificially increase the link between them, i .e . ,  will force a 
relationship as above. Thus, in order to measure Organisational Commitment and 
89 
Hospital Study 
Intention to Leave cleanly and separately, items 4, 9 and 1 1  of the Organisational 
Commitment Questionnaire were removed (refer Table 4) . 
An examination of th� individual items revealed that items 2, 1 0  and 1 2  of the Job 
Satisfaction Scale cross-loaded onto the Organisational Commitment factor. In 
addition, item 3 of the Organisational Commitment Questionnaire and items 1 6, 1 and 
1 4  had absolute values less than .3 . These seven were dropped. Again, the "dropped" 
items (i.e., those that did not load on their hypothesised factor, cross-loaded with other 
factors or had a factor loading with an absolute value less than .3)  are shaded in grey for 
ease of identification. 
90 
Hospital Study 
Table 4: First Pattern Matrix of the Factor Analysis of the Satisfaction, Commitment and 
Intention Measures items to be removed are shaded in 
Organisational Commitment 1 I describe this organisation to my friends as a great 
organisation to work for 
Organisational Commitment 6 I am proud to tell others that I am part of this .849 
organisation 
Organisational Commitment 1 4  For me this is the best o f  all possible organisations for .79 1 
Organisational Commitment 10  
Organisational Commitment 5 
Organisational Commitment 8 
Organisational Commitment 2 
Organisational Commitment 1 5  
Organisational Commitment 7 
Organisational Commitment 1 2  
Intention to leave 3 
Intention to leave 1 
Intention to leave 2 
Job Satisfaction Scale 9 
Job Satisfaction Scale 4 
Job Satisfaction Scale 6 
Job Satisfaction Scale 5 
Job Satisfaction Scale 1 1  
Job Satisfaction Scale 1 3  
Job Satisfaction Scale 7 
�� 
Job Satisfaction Scale 1 5  
which to work 
I am extremely glad that I chose this organisation to 
work for over others I was considering at the time I 
joined 
I fInd that my values and the organisation's values are 
very similar 
This organisation really inspires the very best in me in 
the way of job performance 
I am willing to put in a great deal of effort beyond that 
normally expected in order to help this organisation be 
successful 
Deciding to work for this organisation was a defInite 
mistake on my part . , . . 
I could just as well be working for a different 
organisation as long as the type of work was similar 
Often, I find it difficult to agree with this organisation's 
policies on important matters relating to its employees 
I really care about the fate of this organisation 
��lt��a.9mtmf��'I _� 
�(fef�E"'�.4�;l;:Fil'�1i1i��!l!1;Q;�' .' � �,,�tllJ'ii  
I will probably look for a new job in the next year 
How likely is it that you will actively look for a new 
job in the next year 
I often think about quitting 
Your opportunity to use your abilities 
The recognition you get for good work 
The amount of responsibility you are given 
Your immediate boss 
Your chances of promotion 
The attention paid to suggestions you make 
Your rate of pay 
�l-'illJ 
The amount of variety in your job 
.735 
.684 
.553 
.508 
.502 
.42 1 
.337 
.32 1 
�'!'J � 
.863 
.848 
.361  
Extraction Method: Maximum Likelihood. Rotation Method: Ob1imin with Kaiser Normalization. 
Rotation converged in 1 6  iterations. Loadings less than .3 are not shown. 
9 1  
.376 
. 802 
.784 
. 683 
.632 
.580 
.483 
.469 
ID 
. 378 
Hospital Study 
A second factor analysis (with the above 1 1  items dropped) accounted for 5 1 .533% of 
the variance, and revealed three distinguishable factors representing the three scales. 
The scree plot for the final factor analysis is shown in Figure 12.4, appendix 1 2. The 
Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a value of .89.  
Only item 13 of the job satisfaction scale and item 8 of the Organisational Commitment 
Questionnaire were cross loading with other factors. They were also dropped. Table 5 
shows the final solution, which accounted for 50.983% of the variance, and revealed 
three distinguishable factors representing the three scales. The Kaiser-Meyer-Olkin 
Measure of Sampling Adequacy had a value of .87.  
92 
Hospital Study 
Table 5: Final Pattern Matrix of the Factor Analysis of the Satisfaction, Commitment and 
Leaving Intention Measures 
Scale Item 
Intention to leave 3 
Intention to leave 1 
Intention to leave 2 
Organisational Commitment 1 
Organisational Commitment 6 
Organisational Commitment 14  
Organisational Commitment 5 
Organisational Commitment 1 0  
Organisational Commitment 1 5  
Organisational Commitment 2 
Organisational Commitment 7 
Organisational Commitment 13  
Organisational Commitment 1 2  
Wording Factor 
I will probably look for a new job in the next year 
How likely is it that you will actively look for a new 
. job in the next year 
I often think about quitting 
I describe this organisation to my friends as a great 
organisation to work for 
I am proud to tell others that I am part of this 
organisation 
For me this is the best of all possible organisations 
for which to work 
I fmd that my values and the organisation's values are 
very similar 
I am extremely glad that I chose this organisation to 
work for over others I was considering at the time I 
joined 
Deciding to work for this organisation was a definite 
mistake on my part 
I am willing to put in a great deal of effort beyond 
that normally expected in order to help this 
organisation be successful 
I could just as well be working for a different 
organisation as long as the type of work was similar 
I really care about the fate of this organisation 
Often, I find it difficult to agree with this 
organisation's policies on important matters relating 
to its employees 
1 2 3 
-.966 
- .85 1 
-.384 
.867 
.855 
.792 
.754 
.735 
.550 
.5 17  
.358 
.33 1 
.309 
Job Satisfaction Scale 9 Your opportunity to use your abilities .862 
Job Satisfaction Scale 6 The amount of responsibility you are given .738 
Job Satisfaction Scale 4 The recognition you get for good work .634 
Job Satisfaction Scale 1 1  Your chances of promotion .566 
Job Satisfaction Scale 5 Your immediate boss .454 
Job Satisfaction Scale 7 Your rate of pay .4 14 
Job Satisfaction Scale 1 5  The amount of variety in your job .377 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 8 iterations. Loadings less than .3 are not shown. 
The Cronbach's alpha reliability estimates of the organisational commitment, job 
satisfaction and intention to leave subscales were acceptable, a.'s = .87, .83 and .82 
respectively, indicating the items used are sufficiently homogeneous with respect to the 
construct they measure. The finding that the Organisational Commitment Questionnaire 
items loaded together as one factor is consistent with Mowday et al. (1 979) who also 
found that factor analysis of the Organisational Commitment Questionnaire items 
resulted in a single factor solution. With respect to the Job Satisfaction Scale, although 
WaIT et al. ( 1 979) found, using cluster analysis, that items clustered together into 
93 
Hospital Study 
intrinsic and extrinsic satisfaction subscales; factor analysis in the current study 
indicated that the items on the Job Satisfaction Scale generally loaded together as one 
factor. The resulting scales were used in all subsequent analyses. 
The items that loaded together as factors (as indicated by the factor analyses) were 
combined into new scales. Following this, the new files, and scale scores (the mean for 
each scale) were calculated for each respondent. This was the data used to test the 
model using path analysis. To further test the validity of the measures and to illustrate 
the relationships between them, Table 6 shows the correlations between the scale scores 
of the remaining composite items in each measure, all hypothesised correlations are 
significant and in the expected direction. 
Table 6: Correlations between Measured Variables : Hospital Study 
rganisational Friendship Friendship Cohesion Job Satisfaction 
ommitment prevalence opportunities 
Friendship prevalence . 1 6  
Friendship opportunities .36** .37** 
Cohesion .23* .32** 
Job Satisfaction .63** . 1 2  
Intention to leave .45** -.23* 
* *  Correlation is significant at <the .01  level (2-tailed) 
* Correlation is significant at the .05 level (2-tailed) 
.3 1 ** 
.36** 
- . 19*  
.37** 
-.22* - .47** 
To further verify the factor structure of the whole questionnaire, a third factor analysis 
was performed for all the 54 items from the six scales in questionnaire (i.e., the 1 2  item 
Friendships scale, 9 item Cohesion scale, 1 5  item Organisational Commitment scale, 1 5  
item Job Satisfaction scale and 3 item Intention to Leave scale). The resulting factor 
structure supported the findings described above. The items in each factor after the 
analysis of all the component scales together are almost identical to those remaining 
after the two separate factor analyses described above. See appendix 1 3  for a full 
description of the third factor analysis. 
94 
Hospital Study 
3.3 .3 Path Analysis 
The factor analysis described above (and in appendix 1 3), verified that the scales used in 
the questionnaire are not only internally coherent, but also measure conceptually and 
empirically distinct variables. Because of this, it is possible to statistically test the 
relationship between the variables. Due to the relatively small sample size ( 1 22) and the 
fact that the variables examined in this study are measured with scales shown to have 
adequate validity and reliability (refer section 3 .2.2.); an "observed variable" model was 
tested. A structural equation model using observed (as opposed to latent) variables is 
termed path analysis (Pedhazur, 1 982). In path analysis, the relationships between the 
observed variables are measured directly and the significance (or lack thereof) of the 
relations between the variables can be ascertained. Pedhazur states that path analysis 
was developed as a method for studying the direct and indirect effects of variables 
hypothesised as causes on variables hypothesised as effects. Path analysis is used to 
shed light on the plausibility of the causal models, which are based on knowledge and 
theoretical considerations. 
Having decided on path analysis, the AMOS computer application (Arbuckle, 1 999) 
was used to complete the analysis, testing the model presented in Chapter Two. Table 7 
shows the statistical significance of the various regression paths tested (the AMOS 
output is provided in appendix 14). 
95 
Hospital Study 
Table 7: Showing Critical Ratio (C.R.) Values (parameter estimate divided by standard 
error) of the Regression Paths in Model 1 
Regression Path CR. Estimate S.E. Stand p 
Reg 
wei ht 
Friendship -7 Job Satisfaction 3 . 127 .854 .273 .271 .0 1 1  
Opportunities 
Cohesion -7 Job Satisfaction 3.32 1 .676 .204 .287 .0 1 3  
Friendship -7 Friendship 3.499 .6 10  . 174 .307 .046 
Opportunities prevalence 
Cohesion -7 Friendship 2.628 .342 . 1 30 .230 .025 
prevalence 
Cohesion -7 Org Commitment -0.5 1 1  -. 1 26 .247 -.039 .798 
Job Satisfaction -7 Org Commitment 7.494 .800 . 1 07 .586 .0 1 9  
Friendship -7 Org Commitment 2 . 149 .7 10  .330 . 165 .056 
Opportunities 
Friendship prevalence -7 Intention to leave - 1 .846 -. 1 59 .086 -. 1 54 .046 
Job Satisfaction -7 Intention to leave -2.809 -. 194 .069 -.297 .004 
Org Commitment -7 Intention to leave -2.435 - . 1 1 6 .048 -.242 .034 
Cohesion -7 Intention to leave -0.0 1 3  -.002 . 1 32 -.00 1 .928 
C. R. values >±1 .96 are statistically significant. Parameter estimates and standard errors are also shown, 
along with the standardised regression weights and the probability level (p)' indicated in Figure 12 .  
The test statistics of interest in Table 7 are the critical ratio (C.R.) and the probability 
level. The critical ratio operates as a z-statistic in testing that the estimate is statistically 
different from zero. Based on a probability level of .05, the test statistic needs to be 
>±1 .96 to achieve statistical significance (Byrne, 200 1 ). The probability level in the 
final column indicates the likelihood that the tested relationship would have occurred by 
chance. Thus, values less than .05 indicate a significant relationship. From Table 7, the 
paths from 'cohesion -7 organisational commitment' and ' cohesion -7 intention to 
leave' are shown to be non significant, both because the critical ratio for these 
relationships is less than ±1 .96, and because the probability level is greater than .05 .  
Although the critical ratio for the relationship 'friendship prevalence -7 intention to 
leave' is also slightly smaller than ± 1 .96, the probability level (p < .05) suggests that the 
relationship between the two variables is significant. In addition, when Model 1 is 
respecified with the non-significant regression paths removed, the critical ratio for the 
'friendship prevalence -7 intention to leave' path is more than ±1 .96 (refer Table 8; 
showing the results of the analysis for Model 1 ) . Thus the 'friendship prevalence -7 
intention to leave' path is reported as being significant and is included in the graphical 
representation of the path analysis (Figure 12). 
96 
Hospital Study 
Table 8: Showing Critical Ratio Values (parameter estimate divided by standard error) of 
the Regression Paths in Model 1 When Specified with Non-Significant Regression Paths 
Removed 
Regression Path CR. Estimate S.E. StandReg p 
wei ht 
Friendship -7 Job Satisfaction 3 .222 .934 .290 .287 .006 
Opportunities 
Cohesion -7 Job Satisfaction 3 . 1 99 .656 .205 .278 .028 
Friendship -7 Friendship 3 .036 .556 . 1 83 .274 . 1 20 
Opportunities prevalence 
Cohesion -7 Friendship 2.723 .355 . 1 3 1  .24 1 .029 
prevalence 
Job Satisfaction -7 Org Commitment 8 .9 14 .863 .097 .646 .032 
Job Satisfaction -7 Intention to leave -2.560 - . 1 77 .069 -.27 1 .030 
Org Commitment -7 Intention to leave -2.656 -. 134 .050 -.273 .026 
Friendship prevalence -7 Intention to leave - 1 . 963 -. 1 66 .085 -. 1 59  .044 
C.R values >± 1 .96 are statistically significant. Parameter estimates and standard errors are also shown, 
along with the standardised regression weights and probability level (p). 
Figure 1 2  shows the standardised regression weights and the level of significance of the 
relationships26. Dashed lines indicate that the regression between two variables is not 
significant at the .05 level of significance. 
26 The potentially moderating effect of dual role tension cannot be tested in this analysis, as the number of 
respondents for whom this was relevant was too small. The 'negative relationship' variable was not 
included in this path analysis because 'peer type' data were dichotomous. The analysis of the negative 
relationship data is described in section 3 .3 .4 .  
97 
v1 
Friendship 
opportunities 
0.230 l...-_...-_-' 
0.314 
(p<O.OS) 
v3 
(p<O.OS) 
Friendship 
Prevalence 
-0.1 S4 
(p<O.S) 
v4 
0.1 6S '(n5) �<0.1 ) 
0.271 
(p<O.OS) 
" 
" 
" 
Job satisfaction 
0.S86 
(p<O.OS) 
Intention to leave 
v6 
Hospital Study 
v2 
Cohesion 
'--�--' \ 
\ 
-0.001 \ (n5) 
" 
" 
1 -0.039 (n5) 
I 
I 
vS / 
Organisational 
commitment / 
/" 
/ 
/ 
\ 
I 
I 
/ 
Figure 12 :  Path analysis of Model 1 
Explanation of the relationships between variables in the proposed model 
The diagrams that follow are sections of the model shown in Figure 1 2  and warrant 
some explanation. Several of the hypothesised relationships between variables have 
been shown not to be direct; instead they are mediated by other variables. Each is 
explained below. 
The path analysis of the proposed model (refer Figure 1 2) indicates that there is no 
direct relationship from either friendship opportunities or cohesion to organisational 
commitment. Instead these relationships are mediated by job satisfaction. The absence 
of a direct path from friendship opportunities or cohesion, and organisational 
commitment is indicated by the non-significant regression weights (shown in the model 
as dashed lines) between these variables. The path leading from friendship 
98 
Hospital Study 
opportunities to job satisfaction to organisational commitment is significant, however, 
as is the one from cohesion to job satisfaction to organisational commitment. 
Figure 1 3, below, shows the portion of Model 1 described in the previous paragraph. 
Cohesion and friendship opportunities are significantly positively correlated with each 
other. Both cohesion and friendship opportunities are shown to impact on job 
satisfaction, which in turn, impacts on organisational commitment. 
Friendship 
opportunities 
0. 1 65 ,(ns) �<0. 1 )  
0.271 
(p<CJ.05) 
Job satisfaction 
"-
0.314 
(p<0.05) 
" 
" 
0.586 
(p<0.05) 
"-
"-
Cohesion 
1 -0.039 (ns) 
I 
I 
Organisational 
commitment 
Figure 1 3: Path from friendship opportunities and cohesion to organisational 
commitment 
In addition, the analysis of Model 1 indicates that satisfaction and commitment each 
contribute to intention to leave. The relationship between these three variables is shown 
in Figure 1 4. The analysis of Model 1 is consistent with the notion that job satisfaction 
will lead to organisational commitment, which will, in turn, impact on leaving intention. 
However, in this sample, job satisfaction is also shown to directly affect intention to 
leave (i.e., the satisfaction/intention to leave relationship is not mediated by 
organisational commitment). 
99 
Job satisfaction 
0.586 
(p<0.05) 
Intention to leave 
Organisational 
commitment 
Hospital Study 
Figure 14: Path from job satisfaction and organisational commitment to i ntention to 
leave 
Further, the analysis of Model 1 indicates that cohesion and friendship opportunities 
each contribute individually to friendship prevalence. These relationships are the focus 
of Figure 1 5, which shows significant regression weights from both cohesion and 
friendship opportunities to friendship prevalence. Thus it seems that perceptions of a 
cohesive work environment will be correlated with increased opportunities for 
friendship, and both, in turn, are related to more friendships at work (friendship 
prevalence). 
Friendship 
Prevalence 
Friendship 
opportunities 0.314 
(p<0.05) 
Cohesion 
Figure 1 5: Path from friendship opportunities and cohesion to friendship prevalence 
The analysis of Model 1 (Figure 1 2) suggests that the relationship between cohesion and 
intention to leave is mediated by both job satisfaction and friendship prevalence; this is  
supported both by the non-significant regression weight between cohesion and intention 
to leave (indicated by the dashed line between these two variables) and by the 
1 00 
Hospital Study 
significant weights from cohesion to satisfaction to intention to leave, and from 
cohesion to friendship prevalence to intention to leave, refer to Figure 1 6  for a focus on 
this path. 
Friendshrp 
Prevalence 
'{).154 
(p<O.5) 
Cohesion 
�---' \ 
/ 
/ 
\ 
\ 
\ 
.{).001 I 
(ns) 
I 
/ 
/ 
/ 
Figure 1 6: Path from cohesion to intention to leave 
Figure 1 7  focuses on the relationship between friendship opportunities and intention to 
leave. As is the case for cohesion, the impact of friendship opportunities on intention to 
leave appears to be mediated by both friendship prevalence and job satisfaction, 
indicated by a non-significant regression weight between friendship opportunities and 
intention to leave27. The significant relationship between friendship prevalence and 
intention to leave, shown in Figures 1 6  and 1 7, is worth noting here also. A negative 
friendship prevalence -7 intention to leave relationship is in line both with the existing 
literature and with the qualitative data from respondents in this study and will be 
discussed more fully in the discussion section of this chapter. 
27 When this relation was included (in a separate path analysis) the standardised regression weight 
between friendship opportunities and intention to leave was .088 (ns) 
1 01 
,. 
friendship 
prevalence 
Friendship 
opportunities 
0 . 1 65 ' ( n s )  (�<0 . 1 )  
" 
0 . 2 7 1  '-(p<0 . 05) , 
" 
"-
" 
Job satisfactiOn !--�0�. 5�8 6�--t� ( p < 0 . 0 5 )  
Organisational 
commitment 
Hospital Study 
Figure 17 :  Path from friendship opportunities to i ntention to leave 
Taking the findings of the first path analysis into consideration, Model 1 (refer to Figure 
1 1 , at the end of Chapter Two) may be redrawn without the non-significant regression 
weights; this is shown in Figure 1 8 .  All the relationships between variables proposed in 
the discussion of Model 1 do exist; however some are mediated by other variables rather 
that having direct relations. For example Model 1 suggested that friendship 
opportunities might impact on organisational commitment, the results of the path 
analysis of Model 1 suggests that, while opportunities for friendship will affect 
organisational commitment, the effect will be mediated by job satisfaction. 
1 02 
v3 
Friendship 
prevalence 
-0 . 1 54 
(p<0. 1 )  
v1 
Friendship 
opportunities 
v4 
0.271 
(p<0.05) 
Jab satisfaction 
0.31 4 
(p<0.05) 
0.586 
(p<0.05) 
Intention to leave 
v6 
v2 
Cohesion 
v5 
Organisational 
commitment 
Hospital Study 
Figure 18 :  Model 1 redrawn showing only the significant regression weights (Hospital 
Study) 
3.3.4 Peer Types 
It is possible that the presence of even one special peer, or the existence of a single 
negative relationship at work will impact on an employee's experience of work. To 
investigate this, two-tailed, independent samples t-tests28 were performed to see if there 
were significant differences in respondents' scores on the workplace friendship scale, 
the job satisfaction scale, the organisational commitment questionnaire and the cohesion 
scale, between those who reported having close friends (special peers) at work and those 
who did not, and between those who reported having negative relationships and those 
who did not. It was found that people who reported having at least one special peer had 
28 The reason that the presence negative relationships and special peers were not tested in the path 
analysis is that the data gathered, relevant to negative relationships and special peers, was dichotomous; 
respondents either did or did not have negative relationships or special peers in the workplace. Thus, the 
data did not lend itself to use in path analysis. 
1 03 
Hospital Study 
a significantly higher score on the 'prevalence of friendships at work' scale t( 1 22) = 
4.05, P < .01 . Thus, individuals who had special peer relationships also tended to report 
having more friends at work. On the other hand, people who reported having at least 
one negative peer had a significantly lower score on the ' prevalence of friendships at 
work' scale, t( 1 22) = -2. 1 3 ,  p < .05. This indicates that individuals who have negative 
relationships will also tend to report having fewer friends at work overall . In addition, 
people who reported having at least one negative peer had a significantly lower score on 
both cohesion sub-scales; social support / communication t( 1 20) = -3 .476, p < .0 1  and 
workload sharing, t( 1 20) ::::: -4.653, p < .0 1 , indicating that individuals who have 
negative relationships will also tend to report being part of a less cohesive and 
supportive team, with reduced communication between team members and less 
workload sharing. 
3.4 Discussion 
Overall there is good support for the proposed model. While some of the hypothesised 
relationships are not direct, the results do support a mediated relations model. A review 
of the literature discussed in Chapter Two, presents previous support for the proposed 
relationships between variables. The model, and associated research questions and 
hypotheses, are discussed below in relation to both previous literature, and the analysis  
of the data collected for this study. 
3.4.1 Research Question 1 
Research Question 1 focused on the relationship between friendships and the outcome 
variables measured. It was hypothesised that opportunities for friendships and informal 
relationships would have a significant effect on how satisfied people are with their jobs, 
that having more opportunities for friendship would increase organisational 
commitment, and that both opportunities for friendship and the prevalence of friends at 
work would reduce turnover. Researc� Question 1 and the associated hypotheses were 
outlined previously in section 3 . 1 .2 .  
1 04 
Hospital Study 
The relationships between variables related to Research Question 1 are illustrated in 
Figure 1 8  with the 'friendship opportunities' variable drawn as being antecedent to both 
job satisfaction and organisational commitment, both of which, in turn, impact on 
intention to leave. The model also shows friendship prevalence as antecedent to 
intention to leave, but to have no direct relationship with job satisfaction or 
organisational commitment. 
Hypothesis la: Having more opportunities for friendships within the workplace will 
have a positive relationship to job satisfaction 
As discussed in Chapter Two, although there have been few published studies directly 
examining the relationship between friendship and satisfaction, those that have 
investigated the friendship 7 satisfaction link have found positive relationships 
between the two variables (e.g., Nielsen et aI . ,  2000; Richer et aI . ,  2002; Riordan & 
Griffeth, 1 995;  Winstead et aI . ,  1 995). Similarly, Markiewicz et al. (2000) found that 
the quality of close male friendships was associated with both career success and job 
satisfaction. The data from the current study support this previous literature. Figure 1 2  
showed the significant regression weight from 'friendship opportunities' to 'job 
satisfaction,' which supports the hypothesis that friendship opportunities have a positive 
impact on job satisfaction. The positive relationship between these two variables is also 
supported by qualitative data from the current study?9, for example: 
# 103 Our friendly work environment has improved my job satisfaction. 
#98 [Having friends at work] generally makes work more pleasant. 
Hypothesis 1 b: Opportunities for friendships within the workplace will have a positive 
relationship with organisational commitment 
The ' opportunities for friendship' variable has been empirically related to commitment 
by previous researchers. For example, and as previously discussed in Chapter Two, 
Kaldenberg et al. ( 1 995) found that affiliation opportunities were positively related to 
commitment among female dentists; and Sheldon ( 1 97 1 )  found a link between social 
29 For a summary of responses to the open ended questions in this study see Appendix 1 5. 
1 05 
Hospital Study 
involvement with fellow employees and the commitment of employees to their 
organisation. The relationship between 'friendship opportunities' and ' organisational 
commitment' is also present in the data obtained in the current study, but is mediated by 
job satisfaction (refer: Figure 1 2) . The mediated relationship between ' friendship 
opportunities' and 'organisational commitment' supports hypothesis Ib, and i s  in line 
with findings by Williams and Hazer ( 1 986), who used structural equation modelling to 
re-analyse data from two previous studies (Bluedom, 1 982;  Michaels & Spector, 1 982) . 
Williams and Hazer specifically examined the causal relationship between satisfaction 
and commitment, hypothesising that job satisfaction is a causal antecedent of 
commitment. In addition they proposed that variables associated with the work 
environment (such as social opportunities) influence commitment only indirectly via 
satisfaction. Both of Williams and Hazer' s hypotheses were supported in the current 
study. As in the current study, Williams and Hazer ( 1 986) found satisfaction to be an 
intervening variable between workplace variables30 and organisational commitment; 
workplace variables directLy influencing only satisfaction, and influencing commitment 
indirectly through their impact on satisfaction (and the subsequent effect of satisfaction 
on commitment). 
Hypothesis 1 c: Opportunities for friendships at work will be negatively related to 
intention to leave 
The relationship between friendship opportunities and intention to leave is  also present, 
but is mediated by both friendship prevalence and job satisfaction. Thus, it is unlikely 
that opportunities for friendship will directly affect respondents' leaving intentions 
unless actual friendships are formed as a result of the friendly workplace. Similarly, 
opportunities for friendship will  be unlikely to affect respondents' leaving intentions 
unless having a friendly workplace is first associated with an increase in job 
satisfaction. 
Thus, friendship opportunities were found to relate directly to job satisfaction, which in 
turn, has a direct relationship with organisational commitment. Both job satisfaction 
30 Examples of workplace variables include: friendship opportunities, leadership consideration, 
"routinization" and equity. 
1 06 
Hospital Study 
and commitment were directly related to intention to leave. The three linkages relate to 
hypothesis 1 a, 1 b and 1 c, and direct! y support findings by Riordan and Griffeth ( 1 995) 
who hypothesised and tested a model of the relationship between perceived friendship 
opportunities in the workplace and work-related outcomes. Riordan and Griffeth found 
that employees' perceptions of friendship opportunities in the workplace had direct 
effects on job satisfaction, but (as in the current study) only indirect effects on 
organisational commitment and intention to leave. 
Hypothesis 1 d and 1 e: Prevalence of friendships at work will not impact directly on job 
satisfaction or organisational commitment 
The "friendship prevalence" variable is not hypothesised to impact directly on job 
satisfaction or commitment. Initially this may seem counterintuitive, given the evidence 
that friendship opportunities and cohesion have positive correlations with these outcome 
measures. As previously stated, it is probable that prevalence of friends will be likely to 
stop people from leaving their jobs but, as the formation of friendships is not something 
that the organisation is responsible for, it may not affect satisfaction or commitment. 
That is to say, employees may need to believe their friendships are a consequence of the 
way the company is managed, and perhaps not easily achieved in another organisation 
in order for their friendships to affect organisational variables such as satisfaction and 
commitment (Meyer & AlIen, 1 99 1 ). It is probably necessary that the employee see the 
organisation as, in some way, responsible for the friendship in order for job satisfaction 
to result. 
In addition, it is often in times of adversity (when morale and satisfaction may be low) 
that strong friendships will form. Qualitative data from the current study indicate that 
this is one of the functions that strong friendships had for respondents, and one of the 
ways that the organisation actually facilitated the friendships, for example: 
#85 Poor work conditions mean we rely on each other more. 
#90 We have been through a lot of changes in the workplace in the last 1 0  years, and we 
don' t  feel that management has listened to our voices, we have only had each other to 
turn to and that has probably strengthened friendships and contacts with each other. We 
1 07 
Hospital Study 
work with poor equipment and an ever-increasing workload and still we are dedicated to 
what we actually trained for. That's what we have in common. The friendships have 
probably been one of the most positive outcomes of a poor employer. 
These comments indicate that the prevalence of friendships may not necessarily be 
strongly related to high organisational commitment or job satisfaction. Along these 
lines Sias and Jablin ( 1 995) found that when a supervisor or group leader was perceived 
to treat group members unfairly, group members became more cohesive and their 
communication relationships more intimate. This suggests that perceptions of low 
quality supervision and feelings of dissatisfaction may lead co-workers to form more 
close friendships, i.e., a case of adversity-attraction (Carr, 2003). 
Hypothesis 1f The prevalence of friendships within the workplace will be negatively 
related to intention to turnover 
The 'friendship prevalence' variable is hypothesised to impact directly on 'intention to 
leave,' because having friendships gives employees an added incentive to remain in 
their jobs. Some respondents specifically mentioned friends at work as a reason they 
worked where they did, for example: 
#24 I personally went back to work for this organisation because she [a special peer] still 
works here. 
Social involvement is intrinsically rewarding (i. e. ,  having friends is pleasant and fun), 
so that the decision to depart from the organisation, and leave these significant others, is 
made more difficult. The data from the current study support this hypothesis, as there is 
a significant relationship between friendship prevalence and intention to leave (refer 
Figure 1 7). The negative relationship between friendships at work and intention to 
leave is in line with both the existing literature and the qualitative data from respondents 
in this study in which respondents wrote frequently of how friendships made their 
experience of work more pleasurable, for example: 
#39 Friendships make it [work] a happier place with friendly environment and I enjoy 
catching up with them each day. / 
1 08 
#126 I feel the friendships at work make the shifts more enjoyable. 
Hospital Study 
#125 Friendships at �ork make the experience of working in that environment all the more 
positive. I look forward to coming to work because I enjoy the camaraderie of the 
workplace. ( 
#1 1 8  Makes work more pleasant, knowing there are people who are pleased to see you each 
day. 
3.4.2 Research Question 2 
Research Question 2 asked: Is the perceived level of cohesion related to opportunities 
for friendship, prevalence of friendship, job satisfaction, organisational commitment and 
intention to leave? It was hypothesised that cohesion would be positively related to 
opportunities for, and prevalence of, friendship in the workplace and would be 
associated with an increase in job satisfaction and commitment and a decrease in 
intention to leave. These hypotheses were presented previously in section 3 . 1 .2 along 
with the relevant sub-hypotheses. 
Hypothesis 2a: That cohesion will be positively related to opportunities for, and 
prevalence of, friendship in the workplace 
The relationships between friendship and cohesion are illustrated in Figure 1 8  where the 
cohesion variable is connected to both friendship opportunities and friendship 
prevalence. 
Hypothesis 2a was supported by the data analysis in the current study. A significant 
positive correlation was found between the cohesion and friendship opportunities 
variables. In addition, the regression weights to friendship prevalence, from both 
friendship opportunities and cohesion, are significant. This finding is in line with 
previous research in this area. As previously stated (section 2.3 .4), Buunk et al. ( 1 993) 
found that if an individual perceives their working climate to be low in cohesion it  
would be likely to hinder the formation of friendships. Odden and Sias ( 1 997) found 
that climates perceived to be high in cohesion were related to more collegial and special 
1 09 
Hospital Study 
peer relationships. Existing literature regarding friendship development identifies liking 
and perceived similarity as conducive to the formation of friendships, so it is not 
unreasonable to expect a positive relationship between the perceived cohesiveness of a 
workgroup or team &nd friendship prevalence. In addition, in the answers to the 
qualitative questions, respondents often linked co-cooperativeness, communication and 
support (measures of cohesion) with friendships at work (italics added), for example: 
#1 25 Friendships help us to work cooperatively with each other and support each other at 
times of stress. 
# 124 I feel open to suggestions and so is she [the friend] . There is an open pathway for 
communication between us and we feel comfortable working with each other and 
respect each other's  decisions. 
#97 By all of us working as a team supporting one another during our very stressful times. 
Having a laugh and keeping our sense of humour. Being able to count on one another. 
Working towards a common goal, too was often mentioned as a way a cohesive 
environment facilitated friendships, for example: 
#28 Constantly working with the same people, encountering the same frustrations and 
amusing incidents, and working together to solve problems creates a bond and insight 
into each others character, which is respected and enjoyed. 
#68 Working together on difficult cases where you achieve a good result despite the 
circumstances is a real buzz, and can make you feel closer to the people you work with. 
Hypothesis 2b: That cohesion will have a positive relationship with job satisfaction. 
The relationship between cohesion and job satisfaction is illustrated in Figure 1 8  with an 
arrow leading from 'cohesion' to 'job satisfaction.' It was hypothesised that increased 
cohesion would lead to improvements in job satisfaction and this hypothesis was 
supported by the analysis of the model, with the regression weight for the relationship 
between cohesion and job satisfaction being significant. Thus, it seems that increased 
cohesion in the workplace improves satisfaction in this hospital setting. 
1 1 0 
HOspItal Study 
Hypothesis 2c: That cohesion will have a positive relationship with organisational 
commitment. 
The relationship between cohesion and organisational commitment is illustrated in 
Figure 1 8 . The hypothesis was not fully supported by the analysis of the model, with 
the regression weight for the relations�ip between cohesion and organisational 
commitment being non-significant. The cohesion and organisational commitment 
variables were correlated at the simple bivariate level however, which suggests that the 
relationship between cohesion and organisational commitment is mediated by job 
satisfaction Gust as the relationship between friendship opportunities and organisational 
commitment is mediated by job satisfaction). This is shown in Figure 1 2  and suggests 
that having a cohesive work environment will have an impact on organisational 
commitment only if an individual is satisfied with their job .  
Hypothesis 2d: That cohesion will have a negative relationship with intention to leave. 
Figure 1 6  illustrates the relationship between 'cohesion' and ' intention to leave' with 
an arrow indicating that employees' perceptions of the cohesiveness of their work 
environment may have a negative relationship with leaving intentions (i.e. , as cohesion 
increases, intention to leave decreases) .  Justification for the hypothesised relationship 
between cohesion and intention to leave comes from the study by Sheridan ( 1 992), 
which is described more fully in Chapter Two. Sheridan compared the voluntary 
turnover rates of new employees of accounting firms with varying organisational 
cultures, finding that the probability of an employee leaving was significantly lower for 
employees working in a culture emphasising interpersonal relationship values (a 
cohesion measure) than it was in a culture emphasising work task values. 
In terms of the results of the current study it was found that, while the relationship 
between cohesion and intention to leave was not direct (illustrated by the non­
significant regression weight between cohesion and intention to leave, refer Figure 1 6) ,  
there was a relationship, which was mediated by  job satisfaction and friendship 
prevalence. This suggests that those in a work environment perceived to be highly 
1 1 1  
Hospital Study 
cohesive will be less likely to leave their jobs only if the high levels of cohesion 
increase satisfaction or improve friendships. 
3.4.3 Research Question 3 
Research Question 3 asked: "Do negative relationships in the workplace worsen 
organisational commitment, job satisfaction and intention to leave?" It was 
hypothesised that individuals who experienced negative relationships at work would 
report decreased organisational commitment and job satisfaction and increased intention 
to turnover (refer section 3 . 1 .2.). 
Although negative relationships were not themselves analysed during the path analysis 
it was found that people who reported having at least one negative peer had a 
significantly lower score on the 'prevalence of friendships at work' scale. In addition, 
people who reported having at least one negative peer have a significantly lower score 
on both cohesion sub-scales (social support / communication and workload sharing). 
Hypothesis 3a and 3b: That the prevalence of negative relationships within the 
workplace will have a negative relationship with job satisfaction and organisational 
commitment. 
Hypothesis 3a  focuses on the relationship between negative relationships and job 
satisfaction, and 3b focuses on the relationship between negative relationships and 
organisational commitment. Although there were no significant differences in these 
measures between those who did and those who did not report having at least one 
negative relationship in terms of their satisfaction and commitment (when performing a 
t-test between the two groups), the findings outlined above suggest that individuals who 
report negative relationships at work will report being part of a comparatively less 
cohesive team. Cohesion was correlated significantly with job satisfaction, which 
mediates the relationship to organisational commitment, thus it seems likely that the 
presence of negative relationships will be related to satisfaction and commitment. 
Further support for hypotheses 3 a  (that negative relationships within the workplace 
affect job satisfaction) comes from the qualitative data gathered from respondents in this 
1 1 2 
Hospital Study 
study, which suggests that having a negative relationship affects employee' s  enjoyment 
of work, for examJXe:  
#85 [the negative relationship] has generally made the prospect of being at work less than 
palatable 
#86 Negativity makes for an unpleasant working environment. 
As stated above, it was found that individuals who have negative relationships tended to 
report being part of a less cohesive and supportive team. The question of whether this is 
because less cohesive teams tend to facilitate more negative relationships, or whether an 
existing negative relationship will destroy team cohesion is an interesting one. Several 
individuals touched on this in their qualitative answers, for example: 
#66 If a friend in the team is not getting along with someone else in the team it can divide 
the team. 
#68 Sometimes it is hard to believe that people can bring their personal, negative feelings 
about an individual into the workplace. This is not appropriate, especially when that 
person is doing a good job. These negative feelings amongst other people in the team 
disrupt the team as a whole. 
#83 [negative relationships] interfere with true team work. 
#87 This person demotivates other team members. 
From these statements it seems the negative relationship is the antecedent to the poor 
team cohesion. 
Hypothesis 3c: That the prevalence of negative relationships within the workplace will 
have a positive relationship with intention to leave. 
With respect to the hypothesis that the prevalence of negative relationships within the 
workplace will increase intention to turnover, it was found that people who reported 
having at least one negative peer had a significantly lower score on the prevalence of 
1 1 3  
Hospital Study 
friendships at work, and because prevalence of friendships at work is significantly 
related to intention to leave it seems likely that negative relationships will be somewhat 
related to intention to leave. Although it is not possible to draw firm conclusions from 
this data, this hypothe�is, too, is indirectly supported. Further support for hypothesis 3 c  
comes from the qualitative data gathered from respondents i n  this study, for example: 
#26 [A negative relationship] can be enough to make one leave. 
# 108 It [the negative relationship] has caused frustration, tips the balance to wanting to leave 
the job if already feeling stressed. Their negativity/resistance can affect the other team 
members. 
# 124 [because of the negative relationship] I don't feel like coming to work. I am on 
tenterhooks waiting for the next uncomfortable experience. I go home unhappy 
3.4.4 Job Satisfaction and Commitment 
Organisational commitment is generally characterised by factors relating to belief in the 
organisation's goals and values, willingness to exert effort for the organisation and a 
desire to remain in the organisation (Meyer & AlIen, 1 99 1 ;  Meyer et aI. ,  1 993 ; Meyer et 
aI . ,  2002; Porter et aI . ,  1 974; Steers, 1 977). Organisational commitment is distinguished 
from job satisfaction in that it is an affective response to the whole organisation, while 
satisfaction is an affective response to specific aspects of the job (Williams & Hazer, 
1 986). 
Although there were no specific research questions or hypotheses focusing on the 
relationship between job satisfaction and organisational commitment, the model 
hypothesised job satisfaction to be antecedent to organisational commitment. The data 
supported this relationship, as there was a significant regression weight for the path 
from satisfaction to commitment, the largest regression weight in the model. The 
finding that satisfaction appears to be antecedent to commitment is consistent with some 
previous literature, which has suggested that commitment takes longer to develop and is 
more stable than satisfaction, and that an employee is unlikely to fprm strong 
organisational commitment if they are not first satisfied with their job (Porter et aI. ,  
1 1 4 
/ 
Hospital Study 
/ 
1 974). The finding also supports the satisfaction-to-commitment relation proposed by 
Williams and Hazer ( 1 986) who used structural equation modelling to re-analyse data 
from two preyious studies (Bluedorn, 1 982; Michaels & Spector, 1 982). Williams and 
Hazer specifically examined the causal relationship between satisfaction and 
commitment. Their analysis supported the satisfaction-to-commitment relation, 
implying that satisfaction is antecedent to commitment in employees 
3.4.5 Satisfaction/Commitment and Turnover 
The analysis of the data in the current study also supports the relationships hypothesised 
in the theoretical model, between job satisfaction, organisational commitment and 
turnover. That is, that satisfaction and commitment each contribute individually to 
intention to leave. This is in line with the independent-effects model. The independent­
effects model follows Porter et al . ' s  ( 1 974) findings that job satisfaction and 
organisational commitment, though related, are distinct constructs. Further evidence 
that both satisfaction and commitment contribute to intention to leave comes from a 
study by Tett and Meyer ( 1 993) who used path analysis to analyse data from 1 55 
different studies. Tett and Meyer concluded from their meta-analysis, that job 
satisfaction and organisational commitment are independent contributors to turnover 
intention. The independent-effects perspective is also supported by Riordan and 
Griffeth ( 1 995) who hypothesised direct negative relationships between intention to 
leave and both job satisfaction and organisational commitment. 
3.5 Limitations of the Study and Directions for Future Research 
Homogeneity of sample 
It is worth noting that the overwhelming majority (95%) of respondents were women. 
As a gender, women tend to both place more importance on their personal relationships 
and to be somewhat "better" at both initiating and maintaining them (Andrew & 
Montague, 1 998;  Aukett et aI. ,  1 988; Winstead, 1 986), Coupled with the fact that 
nurses as a group have a higher than average need for affiliation, and nursing and 
related professions (midwifery, occupational therapy, social work etc.) are 
characterised, in part, by affiliation (social) opportunities (Medcof & Hausdorf, 1 995), it 
1 1 5 
Hospital Study 
is not surprising that friendships have as much salience for the respondents in the 
current study. 
Perhaps for women working in a hospital setting, the presence or absence of friendships 
is a particularly salient aspect of their workday, enough perhaps, to tip the balance 
between staying and leaving their organisation. It is possible that individuals from other 
professions would have responded differently to the questionnaire. 
Problems with the dual role questionnaire 
Respondents were asked to identify a "special peer" rather than just "think of a friend at 
work." This made respondents unlikely to agree with negative statements about one's 
best friend at work (i.e., say bad things about the relationship). As a result, almost all 
respondents who answered this questionnaire "strongly disagreed" with items outlining 
aspects of dual role tension. It also meant that only about half of the respondents (those 
who considered themselves to have a best friend at work) answered this questionnaire. 
Thus, the data were not useable. The Dual Role Tension Questionnaire seems not to 
work in the context of the set of questionnaires used here. 
For the next study, respondents will be asked to think of a close friend (any friend) they 
have had at work and answer the questions. In this way there may be a higher response 
rate to the dual role tension questionnaire and a wider range of responses. If not, 
analysis of responses to the Dual Role Questionnaire it will be excluded from 
subsequent studies3 I . 
Low rate of return / small sample size 
There was a 30% return rate, which meant the sample size was relatively small (n = 
1 24) possibly due to the length of the survey and the workload of the respondents. 
Workload was an often mentioned complaint in the qualitative answers given by 
respondents, in fact the most frequent answers given to question 5 ("Please briefly 
outline how your work environment has strained a workplace friendship") were 11 heavy 
3 1  Note: Respondents to the Internet Study (Chapter Four) mainly "strongly disagreed" with all the items 
in the Dual Role Tension Questionnaire. Again, data obtained from the Dual Role Tension Questionnaire 
were not useful and were excluded from subsequent analyses. 
1 1 6 
Hospital Study 
workload / unable to find time for breaks / unrealistic workload expectations of 
management. " 
3.6 Conclusions and Implications 
Overall there is good support for the proposed model. The results suggest · that 
friendships at work have a significant impact on respondent' s  satisfaction with their 
jobs, commitment to their organisation and intention to leave. In addition, the 
cohesiveness of employees' workgroups is significantly related to both the opportunities 
for, and prevalence of, friends at work. While some of the hypothesised relationships 
are not direct, the results support a mediated relations model. For example, 
opportunities for friendship at work appear to be related indirectly to organisational 
commitment, through enhanced job satisfaction. 
Another interesting aspect of the model is that the 'prevalence of friendships'  variable is 
not significantly related to outcome measures other than intention to leave. As 
previously stated, it is probable that prevalence of friends will be likely to stop people 
from leaving their jobs but, as the formation of friendships is something that people feel 
that the organisation cannot be responsible for, it will not affect satisfaction or 
commitment (Meyer & AlIen, 1 99 1 ). Opportunities for friendship, on the other hand, 
are more likely to be viewed as something that an organisation has provided; so these 
perceived opportunities for friendships will be likely to be positively related to job 
satisfaction and organisational commitment. 
It was found that individuals who report having at least one negative relationship, also 
tended to report being part of a less cohesive and supportive team. It could be that less 
cohesive teams tend to facilitate more negative relationships; on the other hand, it  is 
possible that an existing negative relationship will damage team cohesion. Several 
individuals touched on this issue in their qualitative answers, stating that a negative 
relationship would: "Demotivate other team members, " "interfere with team work, " 
and/or "weaken the sense of team. " These statements imply that negative relationships 
are antecedent to poor team cohesion. To improve cohesion in a team it may therefore 
be useful to first identify, and attempt to deal with, any existing negative relationships.  
1 1 7 
Hospital Study 
The qualitative data gathered revealed that, in the main, the informal relationships and 
friendships are of importance to employees at Waitemata D.H.B.  Respondents 
described time and again ways that these relationships helped them at work, both in 
doing their actual job (professional support I advice etc) and in making their work day 
more pleasant and enjoyable. These findings are supported by the quantitative data, 
which show significant relationships between having opportunities for friendship, job 
satisfaction and commitment to the organisation. 
The most often mentioned benefits of having close friends at work were: "Making work 
more enjoyable JJ (the most common response, with 3 0% of respondents who answered 
this question mentioning this as a benefit), "social support, JJ  "learning from 
colleagues JJ and "receiving consideration or understanding. JJ  These benefits are also 
mentioned by Campion ( 1 993) as being positive outcomes of cohesive 
workgroups/teams, leading to increased productivity. 
Informal connections and friendships among employees are vital in building social 
capital (Inkson & Arthur, 200 1 ). Even a small number of people who share common 
experiences can create larger networks within an organisation to more effectively solve 
work problems, and share knowledge. Acknowledging the value of these links, 
networks and information channels is therefore important. 
The themes in the qualitative data also highlight the importance for management to 
acknowledge and nurture the relationship structures that exist. In response to the 
question about how the workplace might facilitate informal relationships for employees, 
respondents mentioned, "working as a team JJ and "shared work experiences, which give 
common ground. JJ In addition, some respondents mentioned "open plan / shared 
offices JJ  as being conducive to friendship, allowing lots of contact with colleagues, 
while other respondents stressed the importance of "being able to take tea breaks 
together, giving opportunities to catch up and debrief " 
1 1 8 
Hospital Study 
In sum, most of these findings suggest that, not only are mendship and informal 
relationships of importance to the employees of Waitemata D.H.B. ,  but also that (for 
most respondents) this is an aspect of their work environment with which they are 
satisfied. Most respondents completing the open ended questions seemed to feel that 
their needs for social support were, at least partially, met in their jobs. 
Acknowledgements 
This study has been published in the New Zealand Journal of Psychology: 
Morrison, R. (2004). Informal relationships in the workplace: Associations with j ob 
satisfaction, organisational commitment and turnover intentions. New Zealand Journal 
of Psychology, 33(3), 1 1 4- 128 .  
1 1 9 
Chapter 4:  Internet Study 
4.1  Introduction 
Internet Study 
In order to establish the robustness of any theoretical model it is important to test it 
across different settings. The data obtained in the Hospital Study (Chapter Three) 
supported the proposed model and associated hypotheses; however limitations, 
particularly the small and homogeneous sample, meant that it was not possible to 
evaluate the generalisability of the findings. In essence, the model was tested in a single 
human service organisation. In order to assess how robust the proposed model is across 
different contexts, and to deal with some of the limitations of the first study, a second 
study was conducted. For this, second, study the Internet was utilised; an on-line 
version of the questionnaire was created in which respondents could link, via email, to a 
secure site to submit their responses. This meant access to a broad diversity of 
respondents from a wider variety of organisations, improving the generalisability of the 
findings. In addition, the relative ease of data collection made it possible to collect data 
from a larger number of respondents. 
Because it was not possible to make assumptions about the relative needs of participants 
in this wider, internet-based study, as it was in the hospital research setting, the Needs 
Assessment Questionnaire (Heckert et aI ., 2000) was included in the set of questions 
answered by respondents, and is included in the factor analysis described in section 
4.3 . 1 .  In addition, the level of interdependence in respondents' jobs was assessed with 
two additional questions. The Needs Assessment Questionnaire and the 
interdependence questions were not used in the analysis of the data in the study 
described in this chapter, they are, however, relevant to the research questions addressed 
in Chapter Five, and will be discussed fully there. 
1 20 
Internet Study 
4.1.1  Gender Differences in Friendships 
There have been consistent findings in both the social psychology and organisational 
psychology literature of gender differences in friendships (refer Chapter Two; sections 
2.2.3 and 2.3 .6). In sum women' s  friendships tend to involve more self-disclosure, 
supportiveness, and complexity than friendships between men (Markiewicz et aI . ,  2000; 
Winstead, 1 986; Wright, 1 988, 1 99 1 ) .  Findings also generally indicate that friendships 
with women are rated as more enjoyable, nurturing and of an overall higher quality (by 
both women and men) (Sapadin, 1 988). In addition, males have been found to be less 
likely than females to seek instrumental or emotional support (Ashton & Fuer-hrer, 
1 993). Interpersonal relationship literature indicates men achieve and define closeness 
through the sharing of activities, while women define and achieve closeness through the 
sharing of feelings and emotions (Odden & Sias, 1 997; Wood & Inman, 1 993). Thus, it 
is likely that the presence or absence of friendships might impact men and women quite 
differently. Because of consistent findings that women tend to place more importance 
and value on their friends, and to devote more time and energy to maintaining their 
friendships (Andrew & Montague, 1 998; Markiewicz et aI . ,  2000; Winstead, 1 986;  
Wright, 1 988, 1 99 1 ), i t  is  not unreasonable to propose that the presence or absence of 
friendships will have more salience for women than men, and that there will be 
differences in the way men and women initiate and maintain relationships. Thus, in 
addition to the three research questions addressed in the Hospital Study, an additional 
research question, focusing on possible gender differences, is addressed in the current 
study. 
Research Question 4 
Are friendship variables differently correlated with other organisational variables for 
men and women? The discussion on gender difference in organisational relationships 
(Section 2.3 .6) is focused in the hypotheses below. 
Hypothesis 4a: Friendships at work will be more strongly related to organisational 
outcomes for women than for men. 
Hypothesis 4b: Organisational antecedents to friendship (specifically sharing activities) 
will be differently correlated to workplace friendships for men and women. 
1 2 1  
4.2.1 Participants 
4.2 Method 
Internet Study 
The demographic dat<� indicated that the respondents were diverse, there was a wide 
range of ages and industries and, although there was not an equivalent number of men 
and women, of the 406 respondents who indicated their gender, 279 were female 
(67.7%) and 1 27 were male (30.9%). Respondents ranged in age from 1 9  years to 64 
years, with a mean age of 35 and most respondents were from NZ (68%). The reason 
that the majority of the respondents were from NZ is because snowballing was used as 
the sampling technique. In the snowballing procedure, people who are initially selected 
for the sample, on the basis of convenience or purposive sampling, nominate 
acquaintances that they think might be willing to participate in the research. The 
nominees who agree to participate are then asked to nominate other potential 
participants (Whitley, 2002). The demographic data collected are presented below, in 
Table 9. 
Table 9: Demographic Data for the Internet Study 
Variable 
Sex 
Males 
Females 
Age (mean 35 .23, s.d. 1 1 .07) 
>20 years 
20-29 years 
30-39 years 
40-49 years 
50-59 years 
Over 60 years 
Country of origin 
New Zealand 
U.S .A. 
United Kingdom 
Australia 
Canada 
Other 
(6 missing) 
(6 missing) 
(5 missing) 
Frequency (n) 
127 
279 
8 
150 
1 1 6 
70 
57 
5 
277 
52 
33 
20 
5 
20 
Valid percent 
3 1 .3 
67.7 
2.0 
37.0 
28.6 
1 7.2 
14.0 
1 .2 
68 . 1  
12 .8  
8 . 1  
4.9 
1 .2 
4.9 
Note: Values are presented in percentages excluding respondents who declined to answer 
1 22 
Internet Study 
Industry 
There was variety in the kinds of industries/sectors in which the respondents reported 
working. This infonnation is summarised in Table 1 0. The largest reported sector was 
tertiary education (uniyersities and polytechnics, n = 92), the reason that a large number 
of respondents were from the tertiary sector is that many of the people who were 
initially sent the email invitation to participate in the research were from universities, 
and they would have passed it on to friends and colleagues also likely to work in tertiary 
institutions. In addition, academics are perhaps more likely to be willing to participate 
in research, most having had to recruit respondents themselves at some stage. The 
second largest reported industry was health care (including psychology, psychiatry and 
physiotherapy, n = 53) .  Again many of the people in the initial mail-out were from the 
health sector. The IT industry also provided a large number of respondents (n = 34). It 
is possible that the large response from those in the IT sector reflects the mode of data 
collection. Linking to an Internet site and completing an online questionnaire would, 
perhaps, be something people who spend much of their day online would be relatively 
more willing to do. There were 78 individuals who either did not respond to this 
question, or who reported an industry that did not fit into any of the categories. 
Table 10 :  Industry of Respondents 
lndust 
Tertiary education 
IT rr elecommunications 
Health 
F inancel Accounting 
Government 
Legal 
HRlHR consultancy 
Research 
Manufacturing 
'Other education 
Retail 
Marketing 
Publishing 
Insurance 
Real Estate 
MilitarylDefense 
OtherlMissing 
123 
n) Percent 
22.3 
8.3 
1 2.9 
7.3 
5 . 1  
3 .6 
3.4 
3 .4 
2.7 
2.7 
2.2 
2.2 
1 .9 
1 .7 
.7 
.7 
8 1 . 1  
1 8 .9 
1 00.0 
Valid Percent 
27.5 
1 0.2 
1 5 .9 
9.0 
6.3 
4.5 
4.2 
4.2 
3.3 
3.3 
2.7 
2.7 
2.4 
2 . 1 
.9 
.9 
1 00.0 
Internet Study 
Job type 
As there were no exclusion criteria (other than having a job) the variety in responses to 
the question asking what job type individuals has was almost as wide as the number of 
respondents. A complete list of answers is provided in appendix 1 7 .  Respondents were 
from almost every type of profession, from medical doctors, to secretaries, to 
academics, to police. 
4.2.2 Materials 
In addition to the tools used for the Hospital Study (described fully in Chapter Three), 
there were two additions to the questionnaire: The Needs Assessment Questionnaire 
(Heckert et aI . ,  2000) (refer appendix 1 6  for the items in the Needs Assessment 
Questionnaire), and two questions assessing the interdependence of respondents' jobs. 
The factor analysis of the Needs Assessment Questionnaire is described in this chapter, 
along with the factor analyses of the other scales (refer section 4.3 . 1 ), but because it is 
not used in the analysis of the data in the current study, and is relevant to the research 
questions in the next study, the Needs Assessment Questionnaire is described along with 
the interdependence questions in Chapter Five (refer section 5 .2.2). 
4.2.3 Procedure 
Initially, 68 individuals (friends and acquaintances) were sent an email inviting them to 
complete an online questionnaire, which included a link to the data collection site 
(www.studentresearcher.com). They were encouraged to pass it on to friends and 
colleagues. In addition, two emai1 lists, Emonet (a list of academics and practitioners in 
the field of emotions in organisations) and IOnet (a list of Industrial Organisational 
psychologists in New Zealand) were sent the email. Once at least 400 people had 
submitted responses to the database through the Internet site, the data were downloaded 
and used (the final number of respondents used for this study was 4 1 2) .  
124 
4.3.1 Factor Analysis 
4.3 Analysis and Results 
Internet Study 
Prior to beginning the factor analysis and subsequent structural equation modelling 
(SEM) the data had to be "cleaned." Initially the word answers given by respondents 
(e.g., "strongly disagree" to "strongly agree") were converted to numbers (y.g., the 1 to 
5 likert scale) and transferred into the statistical programme SPSS. Following this, the 
inversely worded items from the various scales were reversed32 . The scales were then 
saved as separate files in SPSS and missing items were imputed, using the 'missing 
value analysis' feature of the programme33 . Finally the scales were recombined into a 
master document and, using the data from the newly formed master document (n = 
4 12), each of the scales was factor analysed. 
Exploratory factor analysis (EF A) was utilised in order that the psychometric properties 
of the set of scales used in the current study could be ascertained. Although the scales 
used were previously validated (Campion et al. ,  1 993 ;  Mobley, 1 977;  Mowday et al. ,  
1 979; Nielsen et  al., 2000; Warr et al . ,  1 979), the samples used by the original authors 
are likely to be somewhat different from the group of individuals who responded in the 
current study. Thus, it is necessary to validate and, if necessary, adapt these original 
scales for use with this new sample. After the EF A, confirmatory factor analysis (CF A) 
was carried out in order to confirm the factor structure of the measurement models used. 
Froman (2001 )  states that CFA is best accomplished on a data set independent of the 
initial EF A, and suggests dividing the data pool into two, with responses from half the 
subjects being used for the EF A, and the other half for the CF A. The data were divided 
into two groups prior to the factor analysis, using the ' select random cases' feature of 
SPSS and resulted in two new data sets : Random Group 1 (n = 2 1 5) and Random Group 
2 (n = 1 97). 
32 Inversely worded items were; item 12 of the workplace friendship scale, item 4 of the dual role tension 
questionnaire, items 3, 7, 9, 1 1 , 1 2  and 1 5  of the organisational commitment questionnaire, items 8, 1 4  
and 1 8  of  the needs assessment questionnaire and the second interdependence question. 
33 The percentage of missing values from each scale are as follows: Cohesion Scale (4.4%), 
Interdependence questions ( 1 .4%), Intention to Leave questions ( 1 .2%), Needs Scale ( 1 .3%), 
Organisational Commitment Questionnaire ( 1 .3%), Job Satisfaction Scale ( 1 .4%), Workplace Friendship 
Scale (0.7%). 
1 25 
Internet Study 
The general procedure used in the analysis for each scale was to perform EF A, using 
Random Group 1 ,  in SPSS to ascertain the number of factors and the loadings of the 
items to the various factors for each scale. Items that were cross loading or had very 
small factor loadings were removed. Following this exploratory factor analysis, the 
factors within the scales were confinned in AMOS, using Random Group 2, by creating 
measurement models of the scales. 
Assessment of model fit was based on multiple criteria, reflecting statistical, theoretical 
and practical considerations (Byrne, 200 1 ). Pedhazur ( 1 982) points out that there have 
been numerous articles, both criticising existing indices and proposing new ones. 
Although there is little agreement about the value of various fit indices, Pedhazur states 
that there does seem to be unanimity that no single fit index should be relied upon; and 
that just as model formulation should be theory driven, so must theory and substantive 
considerations play a part in model evaluation; " . . .  the test statistics and fit indices are 
very beneficial, but they are no replacement for sound judgement and substantive 
expertise" (Pedhazur, 1 982, p.832). The indices used in the current study were (a) the 
X2 likelihood ratio statistic, (b) the Comparative Fit Index (CFI: Bentler, 1 990), the 
Parsimonious Comparative Fit Index (PCFI :  Mulaik et aI. ,  1 989), and (c) the Root Mean 
Square Error of Approximation (RMSEA: Browne & Cud eck, 1 993). 
The CFI is a revised version of the Bentler-Bonnet (Bentler & Bonett, 1 980) normed fit 
index that adjusts for degrees of  freedom. It ranges from zero to 1 .00 and provides a 
measure of complete covariation in the data; a value >.90 indicates a good fit to the data 
(B yrne , 1 994, 2001 ) .  The PCFI is calibrated from the CFI; it weighs the parsimony of 
the model against its use of the data in achieving goodness of fit. Mulaik et al. caution 
that PCFI values are often lower than what is generally considered acceptable on the 
basis of normed indices of fit; goodness of fit indices in the .90s accompanied by PCFI 
indices in the 50s are not unexpected. Byrne (200 1 )  maintains that the RMSEA is one 
of the most informative indices in Structural Equation Modelling, stating; "the RMSEA . --.. 
takes into account the error of  approximation and asks the question, 'How well would 
the model, with unknown but optimaUy chosen parameter values, fit the population 
1 26 
Internet Study 
covariance matrix if it were available? ' "  (p. 84). The RMSEA is sensitive to the 
complexity of the model; values less than .05 indicate excellent fit, and values less than 
.08 represent an adequate fit. 
Where the fit indices did not indicate a good fit to the model, the modification indices34 
and expected change statistics related to the covariances for each model were inspected 
for evidence of misspecification associated with the pairings of items. Large 
modification indices represent misspecified error covariances, indicating systematic, 
rather than random measurement error in item responses. Systematic error can derive 
from characteristics specific to the items or the respondents. Alternatively, a high 
degree of overlap in item content can trigger correlated errors. Correlated errors occur 
when two items, although worded differently, ask the same question (Byme, 200 1 ). 
Thus, if there was evidence that the model was misspecified, the "problem" items (i.e., 
those which had overlapping content with other items) were first examined to ascertain 
if there was a substantive justification for removal and, if there was, the items were 
removed in a post hoc analysis and the model respecified without the items. Finally the 
measurement models were run again using the whole data set and overall fit indices 
were ascertained. 
Workplace Friendship Scale 
EFA in SPSS: The workplace friendship scale was factor analysed in SPSS with the 
solution set to two factors to refleCt the workplace friendship opportunities and the 
workplace friendship prevalence subscales. The initial pattern matrix is shown in Table 
1 1 . 
34 Modification indices are a measure of model misspecification; a large MI would argue for the presence 
of factor cross-Ioadings. 
1 27 
Internet Study 
Table 1 1 :  Initial Pattern Matrix for the Workplace Friendship Scale (items to be removed 
are shaded in  
Scale Item Wording Factor Friendship Friendship 
Workplace Friendship Scale 7 
Workplace Friendship Scale 8 
Workplace Friendship Scale 1 2  . 
Workplace Friendship Scale 1 1  
:Wori%lafd'Ffiefiasm""Sc�16� ;..,. ... _ ...  _� ... : __  . _ ...... _�� _� ....... _p:>-... _t<.) .....  � . • �, 
Workplace Friendship Scale 1 
Workplace Friendship Scale 2 
Workplace Friendship Scale 4 
Workplace Friendship Scale 3 
Workplace Friendship Scale 5 
I have formed strong friendships at work 
I socialise with co-workers outside the workplace 
I do not feel that anyone I work with is a true friend 
Being able to see my co-workers is one reason I look 
forward to my job 
�ii(b]iill�)!i;�§Pl��f@2iM 
I have the opportunity to get to know my co-workers 
I am able to work with my co-workers to collectively 
solve problems 
Communication among employees is encouraged by 
my organisation 
In my organisation I have the opportunity to talk 
informally and visit with others 
I have the opportunity to develop close friendships at 
Extraction Method: Maximum Likelihood. 
Rotation convt:rged in 8 iterations. Loadings less than .3 are not shown. 
Prevalence 
.856 
.807 
.708 
.6 1 1  
. 54 1  
.525 
. 5 14  
Items that were cross-loading (item 9 and 1 0) were removed, along with item 6 which 
had a small factor loading (less than .3). The "dropped" items, those that did not load 
on their hypothesised factor, cross-loaded with other factors or had an absolute value 
less than .3 are shaded in grey for ease of identification. The final pattern matrix for the 
workplace friendship scale is shown below in Table 12 .  
Scale Item 
Table 1 2: Final Pattern Matrix for the Workplace Friendsh ip  Scale 
Wording Factor Friendship Friendship 
Prevalence Opportunity 
Workplace Friendship Scale 7 I have formed strong friendships at work . 877 
Workplace Friendship Scale 8 I socialise with co-workers outside the workplace .787 
Workplace Friendship Scale 1 2  I do not feel that anyone I work with is a true friend .698 
Workplace Friendship Scale 1 1  Being able to see my co-workers is one reason I .59 1 
look forward to my job 
Workplace Friendship Scale 1 I have the opportunity to get to know my co- . 846 
workers 
Workplace Friendship Scale 2 I am able to work with my co-workers to .763 
collectively solve problems 
Workplace Friendship Scale 5 I have the opportunity to develop close friendships . 520 
at my workplace 
Workplace Friendship Scale 4 Communication among employees is encouraged .476 
by my organisation 
Workplace Friendship Scale 3 In my organisation I have the opportunity to talk .469 
informally and visit with others 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 5 iterations. Loadings less than .3 are not shown. 
1 28 
/ 
Internet Study 
CFA in AMOS: When the two-factor structure, with the nine remaining items was 
modelled in AMOS, with friendship opportunities specified as antecedent to friendship 
prevalence (as friend�hip opportunities are likely to be a necessary condition for 
friendships to form), the output indicated a poor fit of the model to the data, (CFI = 
.870, PCFI = .628,  RMSEA = . 1 38 ;  see appendix 1 8a for complete table of fit indices). 
On inspection of the modification indices and expected change statistics related to the 
covariances for this model, there was evidence of misspecification associated with the 
pairings of item 3 with item 4, as this was the largest modification index value and, as 
such, represents measurement error covariance. As stated above, measurement error 
covariance can derive from a high degree of overlap in item content, which occurs when 
two items, although worded differently, ask the same question (Byme, 200 1 ) . 
On inspection of the scale items35, it is clear that they tap into the same notion, i .e. ,  that 
the organisation allows / encourages talk between workmates. Based on this logical 
rationale, the model was respecified with item 3 excluded36• In addition, item 5 (a 
friendship opportunities factor) was cross-loading on to the friendship prevalence 
variable so it too was removed. 
Post Hoc Analysis: On removal of items 3 and 5 ,  the fit indices were improved. The 
AMOS output for the 7-item friendship model is shown in appendix 1 8b, the CFI = 
.960, the PCFI = .594 and RMSEA = .088, indicating an adequate fit of the model to 
the data. When the friendship measurement model was confirmed with the full data set 
the there was an excellent fit of the model to the data, with the CFI = .974, the PCFI = 
.603 and RMSEA = .070 (refer appendix 1 8c for the complete table of fit indices). The 
AMOS diagram for the friendship scale is shown in Figure 1 9. 
35 Item 3: In my organisation I have the opportunity to talk infonnally and visit with other 
Item 4: Communication among employees is encouraged by my organisation 
36 Item 3 was chosen because, in addition to having measurement error covariance with item 4, there was 
also evidence of misspecification associated with the pairings of item 3 with items 2, 1 1  and 7 
1 29 
Internet Study 
Figure 1 9: AMOS diagram for friendship scale 
The items in the two factors are: 
Factor 1 :  friendship opportunities 
4. Communication among employees is encouraged by my organisation 
1 .  I have the opportunity to get to know my co-workers 
2. I am able to work with my co-workers to collectively solve problems 
Factor 2 :  friendship prevalence 
7 .  I have formed strong friendships at work 
8. I socialise with coiworkers outside the workplace 
1 2. I do not feel that anyone I work with is a true friend (R) 
1 1 . Being able to see my co-workers is one reason I look forward to my job 
Satisfaction Scale . 
EFA in SPSS: The Job Satisfaction Scale was factor analysed in SPSS using half the 
data pool, as discussed above. Table 1 3  below, shows the initial pattern matrix for the 
job satisfaction scale. 
1 30  
Internet Study 
Table 13: Initial Pattern Matrix for the Job Satisfaction Scale ( items to be removed are 
shaded in grey) 
Scale item 
Job Satisfaction Scale 8 
Job Satisfaction Scale 6 
Job Satisfaction Scale 14  
Job Satisfaction Scale 10  
i9}l�.��.ii§{�c@n::Scaf�J� 
12!i,�!�i�f�.t@E�s��s.�i� 
Job Satisfaction Scale 5 
Job Satisfaction Scale 3 
iJ oRS�M fu�ti61f .s·<!ale7� ,._ ....... _, L . . ... ..r. -l'" .04 .. '. �·. "',1._ ... a,..· � _� c '-_.j 
Job Satisfaction Scale 1 
�2�-;[i(@!�91tgK��i¥Z[� 
J ob Satisfaction Scale 2 
Job Satisfaction Scale 1 3  
Job Satisfaction Scale 7 
Job Satisfaction Scale 9 
Wordinf! Factor 
Your opportunity to use your abilities 
The amount of responsibility you are given 
rhe amount of variety in your job 
��u�C"c�����,��X P,�?��'!� ���- 7"" d 
�r�,�r;.��.·C.9J�1P..�Y��!!lw.�g�-­
t .  � :Q1J.f.J9l> .. §££�� 
Your immediate boss 
Your fellow workers 
iTh�'r�C'o�Iti6if'ntr"'e .frgaiHvor� L." . • ", .. ,. __  �gg __ , , __ �.jL _ _ . K .. " .. _._ . , _J� ..... ,_ .. .  �" . __ �" . 
The physical work conditions 
mie'at enu"Ii�'lt"-U\�·l5:S{f""�:e�'GoliT>'I(jiIw"r� • . . 'u�_ . •• '_.-•.• _�.SLjt_�J�� .  �J�gl .. _�.,.�" __ .Ya.AJ.r.l._ .... 
The freedom to choose your own method of 
working 
Your hours of work 
Your rate of pay 
Industrial relations between management and 
workers 
1 2 3 
. 804 
.683 
.682 
.497 
t 6 t 6.  I'<.o.-::. ... !"i 
;jjg 
.640 
. 6 19  
f3'6� _'"i. _ 'J �4;rj 
.367 
� ' . � f.3 · 4... ·.�_ ... ... i ��) �J.:;:'''''' 
. 3 13  
.300 
Extraction Method: Maximum Likelihood. Rotation Method: Ob1imin with Kaiser Normalization. 
Rotation converged in 23 iterations. Loadings less than .3 are not shown. 
4 
:r.�2� i:.:'_.�....a 
-.529 
-.469 
Items that cross-loaded or had values 16ss than .3 were removed; item 1 5  was also 
removed because it loaded only with item 1 1 , which was to be removed. The dropped 
items are shaded in grey for ease of identification. In the resulting pattern matrix item 9 
cross-loaded across two factors and item 7 loaded by itself as a factor, so they were also 
removed. The final pattern matrix for the satisfaction scale is shown below in Table 14.  
Table 14: Final Pattern Matrix for the Job Satisfaction Scale 
Scale item Wording Factor Intrinsic Extrinsic 
Job Satisfaction Scale 8 
Job Satisfaction Scale 1 4  
Job Satisfaction Scale 6 
Job Satisfaction Scale 1 0  
Job Satisfaction Scale 1 
Job Satisfaction Scale 3 
Job Satisfaction Scale 5 
Job Satisfaction Scale 2 
Job Satisfaction Scale 1 3  
Your opportunity· to use your abilities 
The amount of variety in your job 
The amount of responsibility you are given 
Your chances of promotion 
The physical work conditions 
Your fellow workers 
Your immediate boss 
The freedom to choose your own method of working 
Your hours of work 
.907 
.706 
.654 
.599 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 5 iterations. Loadings less than .3 are not shown. 
.532 
.474 
.474 
.458 
.322 
The resulting two factors make sense conceptually. The first factor (items 8 ,  14, 6 and 
1 0) relates to respondents' satisfaction with aspects of the actual job they perform, such 
as variety and personal fulfilment. Items in the second factor, on the other hand (items 
1 ,  3 ,  5, 2 and 1 3), relate to aspects of work other than respondents' actual job, i.e., 
interpersonal interactions, hours worked, and the physical work environment. The items 
1 3 1  
Internet Study 
in the two factors relate closely to the ' intrinsic' and ' extrinsic' job satisfaction sub­
scales defined by Warr et al. (1 979) when they initially presented the scale. 
CF A in AMOS: It wa.s hypothesised that extrinsic satisfaction would be antecedent to 
intrinsic satisfaction. The hypothesised relationship between the two satisfaction factors 
is derived from Herzberg's (1 966) two-factor satisfaction theory and Maslow's  ( 1 954) 
need hierarchy theory37. An employee' s  work environment / conditions may be related 
to Herzberg's 'hygiene' or physical factors, and until these, more basic, aspects of work 
are met, the psychological or 'motivating' factors (intrinsic to the work itself) may have 
little salience. The model is also in line with Maslow's ( 1 954) need hierarchy theory. 
According to Maslow, physical needs (such as working conditions) must be met before 
higher order needs (such as a fulfilling work role) will have salience. The resulting 
output did not indicate a particularly good fit of the model to the data (CFI = .9 1 0, PCFI 
= .657, RMSEA = .099). See appendix 1 8d for the complete table of fit indices. Item 2 
(the freedom to choose your own method of working) was regressing on the intrinsic job 
satisfaction factor. Conceptually, of the items in the extrinsic factor, item 2 is the only 
one concerned with the actual job performed, thus it made substantive sense to remove 
this item from the extrinsic factor and, on doing so, the fit was improved (CFI = .940, 
PCFI = .638, RMSEA = .085, refer appendix 1 8e for the complete table of fit indices). 
When the friendship measurement model was confirmed with the full data set the there 
was an excellent fit of the model to the data, with the CFI = .965, the PCFI = .655 and 
RMSEA = .061 (refer appendix 1 8f).  The AMOS diagram for the satisfaction scale is 
shown in Figure 20. 
37 Refer section 2 . 1 . 2  for a description of Ma slow's and Herzberg's theories. 
1 32 
Figure 20: AMOS diagram for the job satisfaction scale 
The items in the satisfaction subscales are as follows: 
Factor 1 :  Satisfaction with aspects of actual job performed (intrinsic). 
8. Your opportunity to use your abilities. 
14.  The amount of variety in your job. 
6. The amount of responsibility you are given. 
1 0. Your chances of promotion 
Internet Study 
Factor 2 :  Satisfaction with interpersonal interactions and workplace (extrinsic); 
1 .  The physical work conditions 
3 .  Your fellow workers. 
5 . Your immediate boss. 
1 3  . Your hours of work. 
1 33 
Internet Study 
Organisational Commitment Questionnaire 
EF A in SPSS: Factor analysing the Organisational Commitment Questionnaire in SPSS 
indicated one factor,. shown clearly in the scree plot below (refer Figure 2 1 ) . 
Confirmatory factor analysis was performed in AMOS to assess the fit of the one-factor 
Organisational Commitment model. 
Scree Plot 
10r---------------, 
" :> 2 � 
Ii -s. $ oL==:=:=����=!�:::::li!::::::fil:d 
1 2 3 .. 5 " 7 8 g 10 11 12 13 14 15 
Factor Number 
Figure 21 : Scree plot for Organisational Commitment Questionnaire 
CF A in AMOS: The AMOS output for the Organisational Commitment Questionnaire, 
including all 1 5  items, resulted in an adequate fit, indicated by a CFI of .908, a PCFI of 
.778 and a RMSEA of .086 (refer appendix 1 8g for the complete table of fit indices). 
On inspection of the modification indices and expected change statistics related to the 
covariances for this model, there was, however, evidence of misspecification, indicated 
by correlated error between items 1 5  and 1 0, items 1 3  and 2 and items 1 4  and 1 1 .  These 
were the largest three modification index values and as such, represent misspecified 
error covariances. As stated above, measurement error covariances represent systematic 
measurement error in item responses, often as a result of a high degree of overlap in 
item content. This seems to be the case for the Organisational Commitment 
Questionnaire as, on inspection of the item pairings, the three pairs of questions do 
indeed tap into similar concepts. 
Items 1 3  and 2 are as follows: 
13. I really care about the fate of this organisation 
1 34 
Internet Study 
2. I am willing to put in a great deal
.
of effort beyond that normally expected in order to 
help this organisation be successful 
Both relate to the degree to which respondents are motivated by, or concerned with, the 
future success of their ·organisation. 
Items 14 and 1 1  are as follows: 
14. For me this is the best of all possible org�nisations for which to work 
1 1. There 's not too much to be gained by sticking with this organisation indefinitely. 
(R) 
Both relate to respondents wanting to stay with their current organisation. 
Items 1 5  and 1 0  are as follows: 
15. Deciding to work for this organisation was a definite mistake on my part. (R) 
1 0. I am extremely glad that I chose this organisation to work for over others I was 
considering at the time I joined. 
Both relate to respondents' satisfaction with their decision to accept their current 
position. 
Post Hoc analysis: Because of the high degree of overlap in item content in the pairs of 
questions described above, the model for the Organisational Commitment Questionnaire 
was respecified with items 2, 1 0  and 1 1  deleted. The resulting fit indices were greatly 
improved, suggesting a good fit of the data to the model, indicated by a CFI of .953,  a 
PCFI of .780 and a RMSEA of .066 (refer appendix 1 8h for the complete table of fit 
indices). When the organisational commitment measurement model was confirmed 
with the full data set the there was an excellent fit of the model to the data, with the CFI 
= .966, the PCFI = .791 and RMSEA = .061 (refer appendix 1 8i). Figure 22 shows the 
AMOS diagram of the Organisational Commitment Questionnaire. 
1 3 5  
Internet Study 
Figure 22: AMOS diagram for the Organisational Comm itment Questionna i re 
Cohesion Scale 
EF A in SPSS: When the cohesion scale was factor analysed in SPSS with the first 
random sample, there were 2 clear factors (this is supported by the scree plot, Figure 
23). 
Scree Plot 
F actor Num ber 
Figure 23: Scree plot for the cohesion scale 
136  
Internet Study 
The social support items ( 1 ,  2, 3) and communication / cooperation items (7, 8, 9) 
loaded together as one factor. Workload sharing items (4, 5 ,  6) loaded as a separate 
factor, this is illustrated below in Table 1 5 . 
Scale Item 
Item 9 
(Communication) 
Item 1 
(Social Support) 
Item 3 
(Social Support) 
Item 8. 
(Communication) 
Item 7 
(Communication) 
Table 15 :  Pattern Matrix for the Cohesion Scale 
Wording Factor Communication I 
Social Support 
Members of my team cooperate to get the 
work done 
.782 
Being in my team gives me the opportunity .75 1 
to work as a team and provide support to 
other team members 
Members of my team help each other out at .728 
work when needed 
Teams enhance the communication among .654 
people working on the same product 
Members of my team are very willing to .606 
share information with other team members 
about our work 
Item 2 My team increases my opportunities for .587 
(Social Support) positive social interaction 
Workload 
SharinK 
Item 4 Everyone on my team does their fair share of -.8 1 6  
(W orkload sharing) the work 
Item 5 No one on my team depends on other team -.796 
(W orkload sharing) members to do the work for them 
Item 6 Nearly all the members of my team -.653 
(W orkload sharing) contribute equally to the work 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 5 iterations. Loadings less than .3 are not shown. 
CF A in AMOS: When the two factor structure, with all nine items was specified in 
AMOS with the second random sample, the output indicated only a moderate fit of the 
model to the data, (CFI = .9 1 0, PCFI = .657, RMSEA = . 1 06; see appendix 1 8j for 
complete table of fit indices). On inspection of the modification indices and expected 
change statistics related to the covariances for this model, there was evidence of 
misspecification associated with the pairings of item 238 with items 1 39 and 940. On 
inspection of the items it is apparent they all deal with teams improving cooperation and 
interaction. 
38 Item 2 of the cohesion scale is "My team increases my opportunities for positive social interaction." 
39 Item 1 of the cohesion scale is "Being in my team gives me the opportunity to work as a team and 
£rovide support to other team members." 
o Item 9 of the cohesion scale is "Members of my team cooperate to get the work done." 
1 37 
Internet Study 
Post Hoc Analysis : On removal of item 2, which had overlapping content with two 
other items (1 and 9) in the social support and cooperation factor, the fit indices were 
greatly improved. The AMOS output for the 8-item cohesion model is shown in 
appendix 18k, the CFI = .977, PCFI = .663 , and RMSEA = .059, indicating an 
excellent fit of the model to the data. When the cohesion model was confirmed with the 
full data set the there was a good fit of the model to the data, with the CFI = .969, the 
PCFI = .65 8 and RMSEA = .072 (refer appendix 181). The two-factor structure of the 
cohesion scale is shown in the AMOS diagram in Figure 24. 
Figure 24: AMOS diagram for cohesion scale 
The items in each sub-scale are as follows: 
Factor 1: Social support and cooperation items 
1. Being in my team gives me the opportunity to work as a team and provide 
support to other team members. 
3 .  Members of my team help each other out at work when needed. 
9. Members of my team cooperate to get the work done. 
8 .  Teams enhance the communication among people working on the same product. 
7. Members of my team are very willing to share information with other team 
members about our work. 
138 
Factor 2 :  Workload sharing items 
4. Everyone on my team does their fair share of the work. 
Internet Study 
5 .  No one on my team depends on other team members to do the work for them. 
6. Nearly all the members of my team contribute equally to the work. 
Intention to Leave 
The analysis of the intention to leave variable proved to be somewhat problematic, as 
because there were only 3 items the scale could not be factor analysed. This latent 
variable will initially be specified in the subsequent causal models with the original 
three items. The items in the scale are: 
1 .  How likely is  it that you will actively look for a new job in the next year? 
2 .  I often think about quitting. 
3 .  I will probably quit my job in the next year. 
Needs Assessment Questionnaire 
EFA in SPSS: The Needs Assessment Questionnaire was factor analysed in spss; the 
initial pattern matrix is shown below in Table 1 6. Items that cross-loaded or loaded on 
the wrong factor were removed (these items are shaded in grey). 
1 39 
Internet Study 
Table 1 6: In itial Pattern Matrix for the Needs Assessment Questionnaire (items to be 
removed are shaded in grey) 
Scale Item Wording Factor Dom All Ach Aut 
Needs Scale 12 dominance 
Needs Scale 20 dominance 
Needs Scale 1 6  dominance 
N�g�]£�t�?:�li!J2� 
Needs Scale 1 8  affiliation 
N�dsTsZ�fe'!I!i�aiimn�'''''' 1 "" �.IJ:.,,� ,,,� � ......... � .. Q!P.M 
Needs Scale 14 affiliation 
Needs Scale 1 0 affiliation 
Needs Scale 6 affiliation 
"--':�- " ':' _�"� '�?",'t'lf"'P')' ·";'.;.":r?-"'!··�"'''i.::"--:'':--� 
N�eas-iS£M�J1§ 'l}\! OJ,QJIl),\ 
Needs Scale 1 achievement 
Needs Scale 9 achievement 
Needs Scale ' 5 achievement 
Needs Scale" 1 3  achievement 
Needs Scale 17  achievement 
I seek an active role in the leadership of a group 
I strive to be "in command" when I am working in a 
group 
I find myself organising and directing the activities of 
others 
r�:��h(f' :lcir.:-dt1i'mS1aii@f'5itroth�""e�1'e ,�R.���A;!�_� ... =_�_,",�,,-=�_P�,�R.� 
I try my best to work alone on a work assignment (R) 
'fl!i',,-,..;r""':;-"'l'!'"i!, ,' . �5""'Ir�l1!?1�'i:W''''''l��'''''!'1?';:'�'J �;my..t,�r.�. �Ql�&�.h.:lli'�9;;...u,s;lm)k � 
I prefer to do my own work and let others do theirs (R) 
When I have a choice, I try to work in a group 
��.��:��2ple" person'!>'i''''''''' ''!i1:ns�."",i'''l !l.-;l��9i,\¥.9.Lj��mJ£l�IQgJQ-p';.Hl$� 
I try to perform my best at work 
It is important to me to do the best job possible 
I am a hard worker 
I push myself to be "all that I can be" 
I try very hard to improve on my past performance at 
work 
.734 
.658 
.545 
.843 
�  
.490 
.40 1 
.348 
�m 
.704 
.702 
.608 
.579 
.435 
Needs Scale 1 1  autonomy I would like to be my own boss .582 ''''7l!rdf '� YT� '' '' ' (r?''' ' .,' ":�',Cj it!�� l\�; ;1� ,, 2!Pl�}l£,C; 
Needs Scale 3 autonomy 
Needs Scale 7 autonomy 
flW5\ila��i(l� li'fiR�;�C�v..�6ef'j:t��'81 �w.""".::H� Ydi�. :"" ' .'�'Y"'J!".",.P--..}l . ....... �s 
I would like a career where I have very little supervision 
I would like a job where I can plan my work schedule 
myself 
� .. �eds,'.�-§.c�8.".·db.Jb.·Tii!ili� �U'f;t'�+""�··��e;r-;;r.·'·""��' -VS���lrrry';'\r�.t:e""'''f'm� _ _ "  - ,., "J¥, __ ....... , �o,q':'J" � ..  �9L!L.,j��an;·gJ. ,�,,,¥ JI!·,.t;J�:J 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 23 iterations. Loadings less than .3 are not shown. 
�t� 
. 5 14  
.420 
The resulting pattern matrix indicated that item 6 had a small factor loading and was 
also removed. The final pattern matrix is below in Table 1 7. 
Table 17 :  Final Pattern Matrix for the Needs Assessment Questionnaire 
Scale item Wording Factor Aff DomAch Aut 
Needs Scale 18  affiliation 
Needs Scale 14 affiliation 
Needs Scale 1 0  affiliation 
I try my best to work alone on a work assignment (R) -.9 1 3  
I prefer to do my own work and let others do theirs (R) -.465 
When I have a choice, I try to work in a group instead of by -.4 1 2  
myself 
Needs Scale 1 2  dominance I seek an active role in the leadership of a group 
Needs Scale 20 dominance I strive to be "in command" when I am working in a group 
Needs Scale 1 6  dominance I find myself organising and directing the activities of others 
Needs Scale 9 achievement It is important to me to do the best job possible 
Needs Scale 1 achievement I try to perform my best at work 
Needs Scale 5 achievement I am a hard worker 
Needs Scale 1 3  achievement I push myself to be "all that I can be" 
Needs Scale 1 7  achievement I try very hard to improve on my past performance at work 
Needs Scale 3 autonomy I would like a career where I have very little supervision 
Needs Scale 1 1  autonomy I would like to be my own boss 
Needs Scale 7 autonomy I would like a job where I can plan my work schedule myself 
.7 12  
.666 
.545 
Extraction Method: Maximum Likelihood. Rotation Method: Oblimin with Kaiser Normalization. 
Rotation converged in 1 1  iterations. Loadings less than .3 are not shown. 
140 
.69 1 
.679 
.639 
.574 
.435 
.6 14  
,547 
.425 
Internet Study 
CFA in AMOS: The four-factor model was specified in AMOS. The AMOS analysis 
supported four-factor structure with PCFI = .67 1 ,  CFI = .860 and RMSEA = .07 1 (refer 
appendix 1 8m for the complete table of fit indices). Although the CFI is a little lower 
than the criterion for good fit (.9) the other two indices indicate good fit. When the 
Needs model was confirmed with the full data set the there was an adequate fit of the 
model to the data, with the CFI = .87,  the PCFI = .68 and RMSEA = .07 (refer 
appendix 1 8n). The measurement model of the Needs Assessment Questionnaire is 
shown in Figure 25. 
Figure 25: AMOS diagram for the Needs Assessment Questionnaire 
The items in each needs factor are as follows: 
Factor 1 :  Affiliation 
1 8 . I try my best to work alone on a work assignment. (R) 
1 4. I prefer to do my own work and let others do theirs. (R) 
1 0. When I have a choice, I try to work in a group instead of by myself 
Factor 2: Dominance 
12 .  I seek an active role in the leadership of a group 
14 1  
20. I strive to be "in command" when I am working in a group 
1 6. I find myself organising and directing the activities of others 
Factor 3 :  Achievement. 
9 .  It  is important to me to do the best job possible 
1 .  I try to perform my best at work 
5 .  I am a hard worker 
1 3 .  I push myself to be "all that I can be" 
1 7 . I try very hard to improve on my past performance at work 
Factor 4: Autonomy 
3 .  I would like a career where I have very little supervision 
1 1 . I would like to be my own boss 
7. I would like a job where I can plan my work schedule myself 
Summary offindings: Measurement models 
Internet Study 
The fit indices for each measurement model are presented in Table 1 8 . The fit indices 
for the Needs Assessment Questionnaire are not shown in Table 1 8, as it is not part of 
the causal model under test. 
Table 18 :  Fit Indices for the Measurement Models: Internet Study 
Scale Number of X df CFI PCFI RMSEA 
actors 
W orkplace friendship scale 2 39.40 
Job satisfaction scale 2 48.42 
Cohesion Scale 2 59.79 
Organisational Commitment Questionnaire 1 1 36.76 
1 3  .97 
19 .97 
19 .97 
54 .96 
.60 
.66 
.66 
.79 
.07 
.06 
.07 
.06 
To further test the validity of the measures and to illustrate the relationships between 
them, Table 1 9  shows the correlations between the composite scores of the items in each 
measure, all correlations are significant and in the expected direction. 
1 42 
Internet Study 
Table 1 9: Correlations between Measured Variables: Internet Study 
Friendship prevalence 
Satisfaction with 
relationships and 
workplace (extrinsic) 
Satisfaction with actual 
job performed (intrinsic) 
Organisational 
Commitment 
Social support and 
cooperation (cohesion) 
Workload sharing 
(cohesion) 
Intention to leave 
Friendship Friendship Satisfaction Satisfaction OCQ Social Workload 
pportunitiesprevalence . with with actual support and sharing 
. .464** 
.4 17**  .293** 
.334** . 1 6 1  **  
.376** .30 1 ** 
.500** .302** 
. 1 53**  .097* 
-.2 17**  -. 1 0 1  * 
interpersonal job cooperation (cohesion) 
relations and performed (cohesion) 
wor lace 
. 5 1 5 ** 
.636** .596**  
.505 **  .374** .394** 
.404**  .222**  .30 1 ** .497**  
- .393**  -.467* - .545** - . 1 84** -. 148** 
* *  Correlation is  significant at  the .01 level (2-tailed) 
* Correlation is significant at the .05 level (2-tailed) 
The kurtosis values of all the items used on all the scales described above were 
acceptable and ranged between .05 and 2.67. Table 20 below shows the alpha of each 
of the subscales used. 
Table 20: Alpha Levels of the Sub-Scales 
Scale I Sub-Scale 
Friendship opportunities 
Friendship prevalence 
Job satisfaction (interpersonal / workplace) 
Job satisfaction (actual job) 
Cohesion (social support/cooperation) 
Cohesion (workload sharing) 
Organisational Commitment Questionnaire 
Intention to leave scale 
4.3.2 Building the Causal Model 
a 
.82 
.7 1 
.73 
.80 
.83 
.8 1 
.9 1 
.87 
Once each measurement model showed good fit, the various latent variables (friendship 
opportunities and prevalence, the two job satisfaction factors, organisational 
commitment, the two cohesion factors and intention to leave), their associated observed 
endogenous items and the hypothesised relationships between them (derived both from 
previous literature, and from the Hospital Study) were modelled and tested using SEM. 
The AMOS graphic output for this complete model is very complex so, in the interest of 
1 43 
Internet Study 
clarity, it is redrawn below in Figure 26, without the endogenous variables (scale items). 
Figure 26 shows only the latent variables (indicated by ovals) and all the tested 
correlations (double headed arrows) and regression weights (single headed arrows). 
The dashed lines indicate non-significant regression weights. 
I 
I 
I 
I 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
'\ 
'\ 
" 
" 
" 
'-
'-
\ 
\ 
\ 
\ 
\ 
\ 
" \ 
" \ 
, \ 
, \ " \ 
, \ , \ , 
, \ 
, \ , \ ", \ 
Figure 26: The causal model tested for the Internet Study (latent variables only) 
As in the Hospital Study, the regression paths from friendship opportunities and both 
cohesion variables to organisational commitment were non-significant. Again j ob 
satisfaction appears to mediate these relationships (which significantly correlated at the 
144 
Internet Study 
bivariate level) .  That is to say, although the friendship opportunities and cohesion 
variables were significantly correlated with organisational commitment (p < 0.0 1 ), the 
relationships between these variables appear to be mediated by job satisfaction. This 
implies that if an individual has a cohesive work environment or perceives their 
workplace to provide friendship opportunities, they will likely be more satisfied with 
their job and, if an increase in satisfaction results from the presence of  cohesion or 
friendship opportunities, organisational commitment will also increase. 
The path from ' satisfaction with interpersonal interactions / work environment' to 
' intention to leave' was also non-significant, suggesting that the relationship between 
these two variables is mediated by the second satisfaction factor (satisfaction with the 
j ob itself). That is, satisfaction with extrinsic aspects of the job will be  significantly 
related to intrinsic job satisfaction, which will, in turn, directly impact on an 
individual ' s  leaving intention. This implies that if an individual reports satisfaction with 
interpersonal interactions / work environment, this may only impact their leaving 
intentions if they are also satisfied with aspects of the job itself. 
In addition, the paths from 'cohesion' to ' friendship prevalence' and from 'friendship 
prevalence' to ' intention to leave' (significant in the Hospital Study) were found not to 
be significant in the current study (although, again, they were significantly correlated at 
the bivariate level). The change in the model when the seven non-significant paths 
described above (and shown with dashed lines in Figure 26) was not significant (!::.X2 ( 7 ) = 
1 1 .6). Non-significant parameters can be considered unimportant to the model, and 
Byme (2001 )  states that, in the interest of scientific parsimony, they should be deleted. 
4.3.3 Statistical Significance of the Model 
Significance of regression weights 
The final causal model, named Model 2, showing only the latent variables and the 
significant regression weights, is presented in Figure 27. In the interest of parsimony 
the model was respecified without the non-significant paths, and the indices of fit 
provided are those resulting from the model having been respecified with the non-
1 45 
Internet Study 
significant parameters deleted. The standardised regresslOn weights or correlation 
coefficients are shown alongside each path (refer appendix 1 9  for the AMOS output of 
these values). All the paths shown are significant at the .05 level. 
Friendship 
opportunities 
.20 
Organisational 
commitment 
Figure 27: Model 2, SEM results of Internet Study 
One way to ascertain significance of a regression path or correlation is by looking at the 
critical ratio ( C . R . )4 1 ; because all the critical ratio values for the regression weights and 
correlations shown are greater than ±1 .96, it is possible to conclude that all the paths 
41 Based on a level of .05, the test statistic needs to be >±1 .96 before we can be sure the relationship 
between two variables is statistically significant (Byrne, 200 1 ). 
1 46 
Internet Study 
shown are significant at the .05 level. The standardised regression weights and critical 
ratios for each regression path and correlation are presented in Table 2 1 ,  below. 
Table 21 : Showing Critical Ratio Values (parameter estimate divided by standard error) 
of the Regression Paths in Model 2 When Specified with Non-Significant Regression 
Paths Removed 
Regression Path I Correlation CR. Estimate S.E. Stan, 
Reg 
wei l 
Friendship Opportunities -7 Friendship Prevalence 9.48 0.97 . 1 0  
Friendship Opportunities -7 Satisfaction with relationships / 3 .79 0.59 . 1 6  
work environment (extrinsic) 
Cohesion (social support) -7 Satisfaction with relationships / 4.69 .74 . 1 6  
work environment (extrinsic) 
Cohesion (workload sharing) -7 Satisfaction with relationships / 3 .34 .33 . 10 
work environment (extrinsic) 
Satisfaction with relationships / -7 Satisfaction with job itself 8 . 1 0  .50 .06 
work environment (extrinsic) (intrinsic) 
Satisfaction with relationships / -7 Organisational Commitment 7.89 .59 .08 
work environment (extrinsic) 
Satisfaction with job itself -7 Organisational Commitment 6.28 .47 .08 
(intrinsic) 
Satisfaction with job itself -7 Intention to leave -4.08 - .53 . 1 3  
(intrinsic) 
Organisational Commitment -7 Intention to leave -5.74 - .56 . 1 0  
Cohesion (social support) f--7 Cohesion (workload sharing) 9.28 .45 .05 
Cohesion (social support) f--7 Friendship Opportunities 7.44 .22 .03 
Cohesion (workload sharing) f--7 Friendship Opportunities 3 .24 . 1 2  .03 
C.R. values >±1 .96 are statistically significant. Parameter estimates and standard errors are also shown, 
along with the standardised regression weights and correlations. 
Statistical significance of the model as a whole 
The hypothesised model can be tested statistically to ascertain whether or not it is 
consistent with the data. If there is an adequate goodness of fit it is possible to argue for 
the existence of the proposed relations between variables. Consistency of the model 
with the data does not constitute proof of a theory (or theoretical model); instead it lends 
support to it (Pedhazur, 1 982). Table 22 shows Chi Square, degrees of freedom and the 
relevant goodness of fit statistics for Model 2. Although these indices of fit have been 
described previously, they are outlined briefly again below, along with an interpretation 
regarding whether the model fits the data. The complete table of fit indices is given in 
appendix 20. 
1 47 
. 56 
.25 
.42 
.24 
.57 
.52 
.37 
-.2� 
-.3� 
.64 
.51 
.20 
Int/ex affit 
Value 
Internet Study 
Table 22: Fit Indices for the Causal Model : Internet Study 
;t t/f CFI PCFI RMSEA Low 90 High 90 Hoelter .01 
1 30 1  6 1 7  .90 .83  .055 .05 1 .059 2 l O  
The comparative indices o f  fit (CFI and PCFI) are based on a comparison o f  the 
hypothesised model against a "baseline model or null model" (Byrne, 2001 ). The CFI is 
a revised version of the normed fit index (NFI) (which, Byrne states, has a tendency to 
underestimate fit in small samples). Unlike the NFI, the CFI takes sample size into 
account. Values for CFI range from zero to 1 .00 with values close to 1 .00 being 
indicative of good fit; a cut-off value of .90 is advised for a model to be considered well 
fitting (Byrne, 200 1 ). As shown in Table 22 the value of the CFI is .90, indicating the 
hypothesised model represents a good fit to the data. 
The PCFI (parsimony-adjusted comparative fit index) addresses the issue of parsimony 
in structural equation modelling. The PCFI takes into account the complexity (number 
of parameters) in the hypothesised model, as well as the goodness of fit, in the 
assessment of the model fit. Typically parsimony-based indexes have lower values than 
are considered acceptable for other normed indices of fit, as such, the value of .83 is 
high, and indicates the model is both well fitting and parsimonious. 
The root square mean error of approximation (RMSEA) is sensitive to the complexity of 
the model and values less than .05 indicate excellent fit, while values less than .08 
represent a good fit. Confidence intervals are used to assess the precision of RMSEA 
estimates; AMOS reports a 90% confidence interval around the RMSEA value (Low 90 
and High 90). Confidence intervals yield information about the precision of the 
estimate. Looking at Table 22 we see that the RMSEA for the model is .056, which 
indicates a good fit. The confidence interval for the model is narrow (ranging from .05 1 
to .059) which indicates that the estimate is precise. 
The last goodness of fit statistic appearing in the AMOS output is Hoelter' s critical N 
(labelled HOELTER). Hoelter' s critical N focuses on the adequacy of the sample size 
rather than on model fit. Its purpose is to estimate the sample size that would be 
sufficient to yield an adequate model fit for a X 2  test. Hoelter ( 1 983;  cited in Byme, 
1 48 
Internet Study 
200 1 )  proposed that a value of 200 is indicative of a model that adequately represents 
the sample data. As shown in Table 22 the .01 critical N value for the hypothesised 
model was 2 1 0, which allows us to conclude that the size of the sample (n = 412) was 
satisfactory. 
Taken together, the goodness of fit statistics reported in this section, indicate that the 
proposed model is a good fit to the data. 
4.3.4 Other Research Questions 
Gender comparisons 
Research Question 4 asked: "Are friendship variables differently correlated with other 
organisational variables, for men and women?" Once each scale was adequately factor 
analysed and showed good fit, Pearson's correlations were calculated between the 
various subscales, first for the whole sample, and then for males and female separately. 
When the correlations between variables were calculated for the whole sample all 
subscales were significantly correlated with all other sub-scales in the expected 
direction. When the correlations between variables were calculated for male and female 
respondents separately, however, some interesting variations emerge. Table 23 shows 
only those correlations where the measured variables were differentially correlated for 
men and women 42 . 
Table 23: Correlations between Subscales, Showing those where Measured Variables are 
Differentially Correlated for Men and Women 
Correlation 
Friendship prevalence with Intention to leave 
Friendship opportunities with Intention to leave 
Cohesion soc�"� o� er�tio�) ,with I�t��tion t� leave 
• <i!'p"!;9'�J. tfj i\ " a£.�f&".lmt 
Cohesion (workload sharing) with Friendship opportunities 
Cohesion (workload sharing) with Friendship prevalence 
Whole Sam le 
Pearson 
correlation 
-. 10 1  (p < .05) 
-.2 1 7  (p < . 0 1 )  
- . 184  (p < .0 1 )  
� 
. l 53 (p < .01 )  
.097 (p < .05) 
Women 
Pearson 
correlation 
-.141 (p < .05) 
-.247 (p < .01) 
-.1.01) 
. 1 1 6  (ns) 
.064 (ns) 
Men 
Pearson 
correlation 
-.008 (ns) 
- . 138  (ns) 
-. 172 (ns) 
IllIHUII 
.213 (p < .01) 
.179 < .01) 
42 All the correlations between measured variables other than the ones shown in table 23 were significant 
to at least a = .05, most were significant to ex = .0 1 .  
1 49 
Internet Study 
On testing for the significance of the difference between these correlations43, only the 
correlation between friendship prevalence and job satisfaction was significantly 
different for men and women, Z = 2.57, P < 0.01 (shaded in grey in table 23). It was 
hypothesised that friendships at work would be more strongly related to some 
organisational outcomes for women than men, and that organisational antecedents to 
friendship would be differently correlated to workplace friendships for men and women. 
On comparing the correlation matrices for the two groups, the following differences 
were apparent: 
1 .  For women, intention to leave is  significantly negatively correlated with 
cohesion (social support and cooperation) (p < .0 1 ), friendship opportunities (p 
< .0 1 )  and friendship prevalence (p < .05); but for men intention to leave is not 
significantly correlated with cohesion or either friendship variable. 
2 .  For men, the workload sharing aspect of cohesion is  significantly correlated to 
friendship opportunities (p < .05) and friendship prevalence (p < .05); but for 
women the workload sharing aspect of cohesion is not significantly correlated 
with either friendship variable. 
3 .  Job satisfaction is significantly correlated with friendship prevalence (p < . 0 1 )  
for men but not for women. 
Hypothesis 4a (that friendships at work would be more strongly related to organisational 
outcomes for women than for men) is only partially supported by the data, with 
cohesion and friendships at work being correlated with intention to leave for women but 
not men. Friendship prevalence, on the other hand, was significantly correlated with job 
satisfaction for men but not women, a finding which does not support hypothesis 4a. In 
addition, it is this correlation between friendship prevalence and job satisfaction that is 
significantly different for men and women (i.e. , not only was there no significant 
correlation between these two variables for women, but also the correlation was 
significantly stronger for men). Hypothesis 4b (that organisational antecedents to 
friendship may be differently correlated to workplace friendships for men and women) 
43 The significance of the difference between the correlations was tested using the Fisher r-to-z 
transformation Zr = (1I2) [loge(1+r) - loge(l-r»). This is automatically calculated, along with p on a 
statistics website administered by Vasser University (http://faculty.vassar.edU/lowry/rdiff.html). 
1 50 
Internet Study 
IS partially supported by the data with the workload sharing aspect of cohesion 
significantly correlated to friendship opportunities and friendship prevalence only for 
men, though again the difference in the size of the correlation between genders is not 
significant. 
Negative relationships 
As in the Hospital Study, Research Question 3 asked, "Do negative relationships in the 
workplace worsen organisational commitment, job satisfaction and intention to 
turnover?" It was hypothesised that individuals who experienced negative relationships 
at work would report decreased organisational commitment and job satisfaction and 
increased intention to turnover. 
To address this research question, an independent samples t-test was performed, 
comparing those who reported having at least one negative peer with those who reported 
having no negative relationships at work44; results are summarised in Table 24. 
Table 24: Independent Samples T-test, Comparing Respondents With and Without 
Ne ative Relationsh ips at Work 
Variable 
Friendship opportunities 
Friendship prevalence 
Satisfaction with workplace 
Satisfaction with actual job 
Organisational Commitment 
Cohesion (social support) 
Cohesion (workload sharing) 
Intention to leave 
t-test statistic df Significance 
-1 .778 
-1 .245 
-7.444 
-2.823 
-5.653 
-3 .302 
-3.989 
2.888 
409 
409 
409 
409 
409 
409 
409 
409 
(2-tailed) 
.076 
.2 14  
.000 
.005 
.000 
.00 1 
.000 
.004 
Significant 
di erence? 
No 
No 
Yes 
Yes 
Yes 
Yes 
Yes 
Yes 
The results support the hypothesis and indicate that those with at least one negative 
relationship at work are significantly less satisfied, report less organisational 
commitment, are part of less cohesive workgroups and are significantly more likely to 
be planning to leave their job. There was no significant difference between the two 
groups in terms of either friendship opportunities or friendship prevalence. 
44 The reason that negative relationships were not tested in the structural equation model is that the data 
gathered, relevant to negative relationships, was dichotomous; respondents either did or did not have 
negative relationships in the workplace. Thus, the data did not lend itself to use in a structural equation 
model. 
1 5 1  
4.4.1 Resume of Fi�dings 
4.4 Discussion 
Internet Study 
An aim of both the Hospital Study and the Internet Study was to investigate the 
relationship between workplace friendships, and organisational and personal outcome 
variables. The overall pattern in the relationships between the variables measured 
suggests that friendships at work do have a significant influence on a range of 
organisational and personal factors. While the Hospital Study took place in one 
organisation, with the participants being almost entirely female, the respondents in the 
Internet Study were from a large number of (predominantly Western) countries and 
were more evenly spread in tenns of gender. Although there were very few respondents 
from non-Western / non English-speaking countries, the theoretical model was 
developed with a Western population in mind, and has proven itself to be applicable in 
this context. 
Most of the linkages between variables in the Internet Study relate very closely to the 
findings in the Hospital Study, demonstrating good support for the theoretical model, 
and indicating that the proposed model is robust. There were some differences between 
the findings from the two studies however. The main differences are: (a) The cohesion 
and job satisfaction variables each have two distinct component factors in the Internet 
Study, while data from the Hospital Study support a single factor structure for each of 
these two variables; (b) the significant regression weight leading from 'friendship 
prevalence' to ' intention to leave' ; found in the Hospital Study, is not present in the 
Internet Study. Overall, however, the models supported by the data in the two studies 
are very similar. 
With respect to (a) above, the analysis of the data in the current study supported a two­
factor model of job satisfaction, one factor relating to individuals' satisfaction with 
aspects of their interpersonal interactions and work environment (e.g., their fellow 
workers, their boss, the physical work place) and one factor relating to satisfaction with 
aspects the actual jobs respondents performed (e.g., the amount of responsibility and the 
1 52 
Internet Study 
variety in an individual ' s  work). While the two cohesion factors appear to be related to 
the other variables in the same way as each other, the satisfaction variables do not. The 
factor relating to ' satisfaction with interpersonal aspects of the job and the workplace' 
(extrinsic satisfaction) appears to mediate the relationship between friendship/cohesion 
factors and the factor relating to ' satisfaction with aspects of the actual job' (intrinsic 
satisfaction). The mediated relationship implies that having a generally friendlier work 
environment (increased opportunities for friendship) will impact directly on employees' 
satisfaction with their workplace and the people they work with, but has no direct 
impact on their satisfaction with the actual job they do . This makes intuitive sense as 
satisfaction with things like the variety in one's job and the opportunity to use your 
abilities is unlikely to be affected by the number or quality of friends at work. Both 
aspects of satisfaction are significantly related to organisational commitment. The 
second satisfaction factor ' satisfaction with aspects of the actual job'  is directly related 
to ' intention to leave, ' and mediates the relationship between ' satisfaction with 
interpersonal aspects of the job and the workplace' and ' intention to leave. ' Taken 
together, however, both cohesion and satisfaction do relate to the other measured 
variables in a similar manner in this study and in the previous, Hospital Study, i .e. ,  the 
friendship opportunities variable is significantly related to cohesion, and both cohesion 
and friendship opportunities are significantly related to satisfaction. Satisfaction, in 
turn, mediates the relationship between friendship and organisational commitment, and 
cohesion and organisational commitment (as in the Hospital Study). 
With respect to (b) above, friendship prevalence was not found to relate to intention  to 
leave in the same way in the Hospital Study and in the current study. In the Hospital 
Study there was a significant relationship between friendship prevalence and intention 
to leave, which was not found in the current study. A possible explanation for the 
difference between the findings of the two studies is the very different samples. 
Respondents in the first study were almost entirely women, all of whom worked in 
health care; a group who are likely to place value on their interpersonal relationships 
(Andrew & Montague, 1 998 ;  Aukett et aI., 1 988; Winstead, 1 986). Respondents in the 
second study were more diverse, both professionally and ethnically, and over 30% were 
male. The finding that having friends at work was directly related to intention to leave 
1 53 
Internet Study 
in the Hospital Study but not this study can be related to the finding in the current study 
that, for women, intention to leave is significantly negatively correlated with friendship 
opportunities and friendship prevalence; but for men intention to leave is not 
significantly correlated with either friendship variable. Thus, the difference between the 
two studies may simply be a gender difference, with women more likely to make a 
leaving decision based on the presence or absence of friendships in the workplace. It is 
worth noting however, that the regression weight found in the Hospital Study, between 
friendship prevalence and intention to leave, though significant, was relatively small (­
. 1 54) and perhaps should be viewed with some caution. 
The difference between friendship prevalence and friendship opportunities 
The analysis of the data in the current study suggests that ' friendship prevalence' and 
'friendship opportunities' are not related to the outcome variables in the same way as 
each other. The AMOS analysis suggests that the friendship opportunities variable, but 
not friendship prevalence, has a direct, positive impact on satisfaction, and a mediated 
relationship with organisational commitment and intention to leave. Given that the two 
friendship constructs are very closely related this, at first, appears somewhat puzzling. 
Unlike friendship opportunities, friendship prevalence does not regress significantly 
onto any of the outcome variables measured (although it is significantly, if weakly, 
correlated with job satisfaction, commitment and intention to leave at the simple 
bivariate level, refer section 4.3 . 1 .) Thus, it initially seems that the prevalence of 
friends in the workplace does not really affect peoples' experience of work, a finding 
that directly contrasts with both previous literature (Nielsen et al., 2000), and the 
qualitative data and comments made by the respondents. In response to the question 
asking how friendships benefited them in the workplace (a question to which over 80% 
of those who answered the qualitative section responded) almost every respondent 
described the positive impact of such relationships, for example: 
#2 I feel that real friendship is what benefits me most. Sometimes also small talk is 
important, but someone who you can trust and say anything to, i.e. a real friend is not so 
easy to find. However once you have one, it really makes your life at work much easier. 
1 54 
Internet Study 
#12  When working with close friends, you know you can trust each other and rely on  each 
other if needed. Wil l be there for each other. Makes you look forward to your day at 
work when you know there is someone there that you can enjoy the time with. 
#19  Having a friend at work can sometimes help you t o  get things done at work faster . . .  It 
makes the workplace environment more relaxed and enjoyable, easier to go about 
things. 
#48 [My friend] made me feel welcome, has given me support and help when I'm unsure 
#60 [My friend] allowed me to develop contacts within the profession to enhance career 
# 1 30 [My friend] makes working more enjoyable 
#132  Laughter shared, teamwork accomplished, sense of  satisfaction achieved. Problems 
halved, better ideas generated. 
#2 1 3  Friendships make me a happier person, therefore I am more productive in the 
workplace. 
#367 [My friend] has definitely made my work a much more enjoyable place and made me 
think of leaving less - I also have people to bounce ideas off, vent to, and discuss work 
problems with which is really important. I also have contacts in different departments, 
which is often really useful for work I am doing. 
A possible explanation for the finding that friendship prevalence is not related to 
satisfaction in the quantitative data, is that the there may be individual differences in the 
ways these two variables (i .e., friendship prevalence and satisfaction) relate to each 
other. While a positive relationship between friendship prevalence and satisfaction may 
exist for some, it is likely that, for others, a positive relationship between dissatisfaction 
and friendship prevalence might exist, i.e. that those who report being dissatisfied with 
their jobs make more friends. The friendship prevalence / dissatisfaction relationship is 
related to findings by Sias and Jablin ( 1 995) and Odden and Sias ( 1 997), described in 
section 2.3 .5, which suggested that feelings of dissatisfaction may lead co-workers to 
1 55 
Internet Study 
fonn and maintain close friendships as a fonn of alliance against the organisation. 
Similarly, Aronson and Cope (1 968) found that a colleague's  hostility towards a 
common enemy (in this case an unpleasant employing organisation) increased others 
liking of them, and w.illingness to work hard for them, i.e., "my enemy's enemy is my 
friend." In addition, in their responses to the qualitative questions in the survey, many 
respondents mentioned strong friendships fonning and helping them, in times of 
adversity. For example: 
#7 In times where there was a lot of stress at work, it is always nice to know you are not 
alone with that. 
# 15  The friendship between myself and my colleague helped to act as
· 
a buffer in what I 
perceived to be a negative team environment. 
# 1 6  Reaffirming values (shared discontent). 
#56 [Friends provide] support and validation of feelings about negative aspects of working. 
#70 It always helps to have a friend at work. It makes you connect with the environment, 
especially making it easier when the times are tough. 
#80 [My friend and 1 ]  were each others sounding boards, sources of information and gave 
each other copious amounts of advice and support - especially when we made mistakes 
or had to deal with frustrating people. We made life easier for each other and made each 
other laugh during the hard times. She ALWAYS understood exactly what I was going 
through. 
# 164 They give you support when you are dealing with difficult things. 
#282 They kept me smiling through the tough times. Working for NCL opened my eyes, 70 -
90 hours a week, seven days a week is tough on anyone. 
#328 A good friendship at work is valuable as it can support one m a negative work 
environment. 
1 56 
Internet Study 
If both the processes described above (i .e., friendship prevalence 7 satisfaction and 
dissatisfaction 7 friendship prevalence) are operating simultaneously, any positive 
relationship that might exist between the satisfaction and friendship prevalence 
variables will be likely to be masked. The friendship opportunities variable, on the 
other hand, is unlikely to have these two processes operating. That is to say, 
dissatisfaction is unlikely to be associated with the perception of a generally friendlier 
workplace (friendship opportunities), in the way that dissatisfaction may be related to 
making and maintaining close friendships (friendship prevalence). 
In addition, the degree to which employees ' give credit' to the organisation for positive 
outcomes related to friendship opportunities and/or friendship prevalence may differ. 
As discussed earlier (section 2.3 .5), although organisations can provide opportunities for 
friendships, it is the employees themselves who make friends. Thus, actually having 
friends at work is not something easily attributed to the organisation, or associated with 
organisational outcomes, in the way that a generally friendlier workplace will be. Thus 
the model shows 'friendship opportunities,' but not 'friendship prevalence,' impacting 
on the organisational outcome variables. 
4.4.2 Links to Existing Research 
Findings related to the role of friendships at work, from the current study, are largely 
consistent with previous literature. For example, in the early 1 970's, as part of the job 
redesign movement, Hackman and Lawler ( 1 97 1 )  found significant positive 
relationships between friendship opportunities and job satisfaction, reflecting findings in 
the current study. More recent research, too, has found that friendship opportunities, 
and the quality of close friendships at work, are associated with job satisfaction 
(Markiewicz et aI. ,  2000; Riordan & Griffeth, 1 995). Positive correlations between 
group cohesion and friendship, found in the current study also reflect previous literature, 
in which these two variables have been found to be related (e.g., Nielsen et aI., 2000; 
Odden & Sias, 1 997). Nielsen et al. (2000) not only reported positive correlations 
between friendship opportunities, friendship prevalence and job satisfaction, but also 
found that those  experiencing friendship at work were less likely to want to leave their 
current jobs. Support for Nielsen et al . ' s  finding in the current study is  mixed, as the 
1 57 
Internet Study 
relationship between friendship prevalence and intention to leave was evident in study 
one, but not in study two. 
Analysis of the data · from these studies suggests that satisfaction is antecedent to 
commitment. The finding that commitment, in part, develops from satisfaction is 
consistent with previous literature, which has suggested that commitment takes longer 
to develop, and is more stable than satisfaction� and that an employee is unlikely to form 
strong organisational commitment if they are not satisfied with their job (e.g., Porter et 
aI . ,  1 974; Williams & Hazer, 1 986). The two satisfaction factors in the current study 
relate closely to the "extrinsic" and "intrinsic" job satisfaction sub-scales defined by 
Warr et al. ( 1 979) when they initially presented the scale. Current findings also support 
the satisfaction-to-commitment relation proposed by Williams and Hazer ( 1 986), who 
used structural equation modelling to re-analyse data from two previous studies 
(Bluedorn, 1 982; Michaels & Spector, 1 982). Williams and Hazer specifically 
examined the causal relationship between satisfaction and commitment. Their analysis 
supported a satisfaction-to-commitment relation, implying that satisfaction is antecedent 
to commitment. 
Much of the research focusing on satisfaction and commitment indicates that both 
satisfaction and commitment contribute independently to the turnover process (refer 
section 2.3 .3) .  This independent-effects model relates to Porter et al. 's ( 1 974) findings 
that job satisfaction and organisational commitment, though related, are distinct 
constructs. Again, results from the current study are consistent with these previous 
findings, as satisfaction and commitment each contributed independently towards 
intention to leave. Further evidence that both satisfaction and commitment contribute to 
intention to leave derives from a study in Tett and Meyer ( 1 993). Tett and Meyer used 
path analysis to re-analyse data from 1 55 different studies. Their meta-analysis 
concluded that job satisfaction and organisational commitment are indeed, independent 
contributors to turnover intention, supporting the independent-effects perspective. 
Riordan and Griffeth (1 995) also provide evidence supporting the independent-effects 
perspective. 
1 58 
Internet Study 
Gender comparisons 
The first finding was that intention to leave is significantly negatively correlated with 
cohesion and both friendship variables for women but not men. This finding suggests 
that women will be directly affected by the presence or absence of close friends at work; 
women may make a leaving decision based, at least in part, by the social opportunities 
their work offers. 
A second finding was that the 'workload sharing' aspect of cohesion is significantly 
correlated with 'friendship opportunities' and 'friendship prevalence' for men but ·not 
significantly correlated with either friendship variable for women. The significant 
relationship between cohesion and the friendship variables for men, may relate to 
previous studies, indicating men achieve and define closeness through sharing activities 
(Odden & Sias, 1 997; Wood & Inman, 1 993), and is in line with other literature in this 
area, which also indicates that men's  friendships tend to be organised around shared 
activities and be action-oriented rather than person-oriented (e.g., Markiewicz et al . ,  
2000; Winstead, 1 986; Wright, 1 988, 1 99 1 )  (refer section 2.2.3). I t  is possible that, for 
men to form friendships at work, they must ' share work' (i.e., perform side by side on a 
task). Thus, it may be possible to interpret the finding that cohesion is significantly 
related to friendships at work, as suggesting that men are more likely to make friends if 
they are in a job involving shared activities with their colleagues than if they are in a job 
that does not offer opportunities to share work. 
A statistically significant difference found between the genders was in the correlation 
between job satisfaction and friendship prevalence. This correlation is significantly 
stronger for men than women. Job satisfaction is not significantly correlated with 
friendship prevalence for women, but is for men. This relates to findings of the Hospital 
Study (Chapter Three) where there was also a finding that friendship prevalence and 
satisfaction were not significantly correlated for a predominantly female sample. A 
possible explanation for this finding is that when women are dissatisfied with their jobs 
they will form strong friendships and lean on their colleagues for social, emotional and 
instrumental support (Carr, 2003). Further support for the suggestion that women make 
friends at work when they are dissatisfied, comes from an American study with a 
1 59 
Internet Study 
sample of teachers, by Odden and Sias ( 1 997). Odden and Sias found that for women, 
special peer relationships were associated · with increased tenure and low 'supervisor 
consideration. ' In other words, the longer a woman was employed in an organisation, 
and the more inconsiderate she perceived her boss to be, the more likely she was to have 
special peer relationships. If women's friendships strengthen in situations where they 
are dissatisfied with their jobs, or unhappy with their boss, a significant positive 
relationship between satisfaction and friendship prevalence will not be found, even if  
having more friends at work improves job satisfaction in other situations, i.e., the two 
processes will cancel each other out. 
Women define and achieve closeness through the sharing of feelings and emotions 
(Odden & Sias, 1 997; Wood & Inman, 1 993). Having an inconsiderate supervisor or 
being very unhappy at work may lead women to share their frustrations, emotions, and 
feelings with their co-workers, enhancing the formation of close friendships. In other 
words, women may use their friends at work as an outlet for venting work-related 
frustrations and emotional expression, while men may use special peers more for 
discussion of other (possibly less negative and/or emotional) topics (Odden & Sias, 
1 997). Thus, in times of adversity, when conditions are bad, women may be more likely 
to make and keep friends than men. Related to this, Ashton and. Fuerhrer ( 1 993) found 
that males are, indeed, less likely than females to seek emotional support. Men are 
perhaps less likely to demonstrate a negative relationship between friendships and job 
satisfaction, because they are less likely than women to seek emotional and social 
support from their colleagues when times are bad. Thus, a positive relationship between 
friendships at work and job satisfaction will be more apparent for men. 
Negative relationships 
On comparing respondents who reported having at least one negative relationship with 
those who did not, the two groups differed significantly on all organisational outcome 
variables. This differs from the findings in the Hospital Study where the only 
significant differences between those who did and those who did not report having 
negative relationships were in terms of friendship prevalence and cohesion (people with 
a negative relationship reported having lower scores on the prevalence of friends scale, 
1 60 
Internet Study 
and lower scores on the cohesion scale; refer section 3 . 3 .4) . It is possible that negative 
relationships make more of a difference to the experience of work for the respondents in 
the current study, but it is probably more likely that the small sample size in the 
Hospital Study meant that between group differences were not statistically significant. 
Data in the current study indicate that those with at least one negative relationship at 
work are significantly less satisfied, report less organisational commitment, are part of 
less cohesive workgroups and are significantly more likely to be planning to leave their 
jobs. These findings are in line with the qualitative data gathered from respondents in 
this study also. Over 80% of respondents answered the question asking how negative 
relationships affected them, and most reported negative relationships adversely 
impacting their experience of work and/or making them more likely to leave. For 
example: 
#5 [the negative l;elationship] made me want to leave, unhappy and unsure of my 
performance 
#7 When I felt observed and being watched with a negative attitude it felt like being 
paralysed. I think if people do not think you will do a good job, it is less likely that you 
actually do. 
# 12  [The negative relationship] creates tension, I look forward to the end of the day so  I 
don't have to be around them, puts me in a bad mood. 
#1 5 I think the negative relationship I had, partiCUlarly with one worker, lead to my 
organisational commitment being undermined, but also my general feelings of 
[dis]empowerment and oftrust. 
#32 When a work relationship has gone bad, it makes it very uncomfortable for all 
concerned, and does affect the enjoyment at work. 
#62 Personal conflict with a work colleague divided the office and caused a lot of stress and 
friction in the work place. The meant that work was not a pleasant place to be at that 
time. 
1 6 1  
Internet Study 
# 120 [The negative relationship meant] I had to leave a job because he was the boss. 
# 1 39 [The negative relationship] generally raised levels of stress. The tension can never be 
contained to just being between the two individuals as other members of the team are 
always involved (willingly or not). This can be a cascade effect where the negative 
relationship with one individual can impact even well established relationships with 
others (either in a positive or a negative way!) .  It can also impact your enthusiasm for 
the project. 
# 149 It brings your motivation levels down. You have to try harder to be positive when 
negative people are around you. 
# 168 Not wanting to go to work because of her constant demoralising, accusatory behaviour. 
#222 Dread going into work. 
#241 Negative relationships particularly with management in conjunction with a lack of 
enjoyment in my role encouraged me to apply for positions elsewhere in the 
organisation. 
The qualitative responses, combined with the t-test results described above, suggest that 
the impact of negative relationships on employees' experience of work may be even 
more substantial than that of friendships, and is something that warrants further 
investigation. 
4.5 Limitations of the Study and Directions for Future Research 
It is possible that the specific demands within a persons' work role and/or the needs of 
respondents may influence the relationships between the measured variables. Individual 
and job differences were not assessed in the analysis of the data in the current study. In 
order to address the question of whether differing work roles or needs of respondents 
will affect the salience ofworkplace friendships, the sample needs to be divided and the 
proposed model tested for group invariance. Invariance testing is achieved by dividing 
1 62 
Internet Study 
the sample into respondents reporting having (for example) jobs high in autonomy 
versus those reporting having jobs high in interdependence. This is the aim of the next 
chapter. 
4.6 Conclusions and Implications 
The prevailing zeitgeist today is, arguably, that organisations must be socially 
responsive, not only providing a job and income for their employees, but also a 
reasonable quality of work life. As mentioned in the introduction (Chapter One), in 
New Zealand's increasingly secular society of primarily nuclear families, there may be 
an increasing reliance on the workplace to give a sense of belonging that people 
formerly drew from their church, community, or extended family. This study sits within 
the quality of work life movement, focusing both on the fulfilment of social needs in the 
workplace and on related organisational outcomes. Overall, the theoretical model was 
supported by the data collected from both the Hospital and the Internet studies. The 
diverse samples give generalisability to the findings, and support to the notion that 
workplace relationships have direct, measurable effects on both organisational and 
personal outcomes - both in New Zealand and overseas. On the basis of these findings, 
it is not unreasonable to conclude that opportunities for friendships at work will have a 
positive effect on employees' experience of work. Research by Barsade (2002) suggests 
that a rippl e  effect, or emotional contagion of positive emotions, will, in the long run, 
improve cooperation, decrease conflict and increase perceived task performance. Thus, 
it is likely that the positive effects of work place relationships will be quite far reaching ­
not only improving employees' experience of work, but also having an effect on the 
financial "bottom line" through factors such as enhanced organisational commitment, 
job satisfaction and reduced intentions to leave. 
Acknowledgements 
Parts of this study have been published in the New Zealand Journal of Psychology: 
1 63 
J 
Internet Study 
Morrison, R. (2004). Infonnal relationships in the workplace: Associations with job 
satisfaction, organisational commitment and turnover intentions. New Zealand Journal 
of Psychology, 33(3), 1 1 4- 1 28.  
Parts of this study have been published in The Bulletin: 
Morrison, R. (2004). Gender comparisons in workplace friendships. The Bulletin( 1 02), 
35-40. 
1 64 
Cross-validating the Causal Model and Testing for Invariance 
Chapter 5 :  Cross-Validating the Causal Model and Testing for Invariance 
Between Sub-groups Differing in Need Strength and Interdependence of Job 
5.1 Introduction 
The previous two chapters outlined results which, overall, supported the notion that 
increased friendship opportunities at work impact positively on organisational outcomes 
such as job satisfaction and organisational commitment, and reduce employees' 
intention to leave an organisation. The findings described in the preceding chapters are 
also in line with previous literature (e.g., Nielsen et aI . ,  2000; Richer et aI., 2002; 
Riordan & Griffeth, 1 995). This study builds on the previous two, by cross-validating 
the causal model and testing for invariance between sub-groups of respondents differing 
in need strength and interdependence of job.  Two additional aims of the current study 
were to outline psychometric properties of the Needs Assessment Questionnaire 
(Heckert et aI . ,  2000) and evaluate the validity of two new interdependence questions 
(refer section 5 .2 .2). 
Although the previous two studies, and much of the literature focusing on workplace 
friendships, suggests that significant linkages exist between workplace friendships and 
organisational outcomes, the question remains: Are there individual or situational 
differences in the way the measured variables relate to one another? It was on the basis 
of four variables (one situational difference and three individual differences) that the 
sample in the current study was divided and tested for group invariance of the model of 
workplace friendships described in the previous chapter. The four variables are: .a) 
interdependence of work role (a situational difference), b) need for affiliation, c) need 
for autonomy and d) need for achievement (individual differences). The impact of 
needs and job type on the linkages between workplace friendships and organisational 
outcomes are enquiries that have yet to receive empirical attention. Highlighting when 
165  
Cross-validating the Causal Model and Testing for Invariance 
and for whom friendships at work are most salient should lead to insights regarding 
applying research, such as the current study, to the work environment. 
5.1 .1  Needs for Affiliation, Autonomy and Achievement 
Personal characteristics, such as work ethic, locus of control and central life interest in 
work have been found to relate to organisational commitment (Meyer & AlIen, 1 99 1 ). 
Similarly, higher order needs are individual difference variables that, while they have 
seldom been studied in this context, might logically relate to the impact of workplace 
friendships. 
The hypotheses associated with Research Question 1 (Chapter Three) are that 
friendships will have significant effects on people' s experience of work (i.e., 
satisfaction, commitment, leaving intentions). It is very possible that the needs of 
respondents may influence the relationships between the measured variables. Although 
in the Hospital Study (Chapter Three), the needs of the respondents were not assessed 
directly, the profession and demo graphics of the sample meant that it was not 
unreasonable to expect that the sample (predominantly female health care professionals) 
would have high affiliation needs; both because of their gender and their profession. 
Women have been found to be generally more relationship focused than men (Andrew 
& Montague, 1 998; Markiewicz et aI . ,  2000; Winstead, 1 986; Wright, 1 988, 1 99 1 )  and 
nurses, as a group, have been found to have a higher than average need for affiliation 
(Medcof & Hausdorf, 1 995). 
Professions and work roles that fulfil particular needs will frequently be filled with 
individuals with those needs. Medcof and Hausdorf ( 1 995) state that a fundamental 
assumption made by researchers in this area is that people gravitate to, and tend to 
remain in, jobs which fulfil their needs. Although not all people will be able  to move to, 
and remain in jobs which satisfy their needs, it is assumed that, as a general rule, a 
significant level of matching of people and jobs does occur (Medcof & Hausdorf, 1 995). 
This is related to the attraction/selection/attrition (ASA)45 model proposed by Schneider 
45 The attraction/selection/attrition (ASA) model is also sometimes tenned the selection/attraction/attrition 
(SAA) model. 
1 66 
Cross-validating the Causal Model and Testing for Invariance 
( 1 987; Schneider et al . ,  1 995; Schneider & Reichers, 1 983). Schneider' s model 
suggests that organisational homogeneity increases over time through three stages. 
Schneider haS reasoned that individuals are attracted to jobs and organisations that fit 
their personalities and from which they can obtain outcomes they desire, particularly 
need satisfaction. Individuals will therefore be engaged in a process of self-selection 
when they decide whether or not to approach a particular organisation for a job. 
Organisations, on the other hand, will invest considerable resources in order to attract 
and select people whose attitudes, values and performance will be consistent with the 
organisation's expectations. In other words, the organisation actively chooses the 
applicants that are perceived to be the most similar to the employees already in the 
organisation. However, because both organisations and individuals tend to present 
themselves in their most favourable light in order to attract as many candidates/job 
offers as possible, this attraction-selection process does not guarantee that no 
"mismatches" will be selected and hired. 
The mismatched individuals will be the first to leave the organisation; provided they 
have alternate job opportunities they perceive to be more consistent with their needs and 
values. Thus, the attrition process will further increase both homogeneity among 
organisation members and person/organisation fit. The result of these three stages 
(selection, attraction and attrition) is that the personal values and preferences of the 
people within an organisation grow more similar over time. 
Indeed, almost all respondents in the Hospital Study felt that informal workplace 
relationships were of importance, which is what one would expect of a sample with high 
needs for affiliation. The Internet based study, however, analyses data from individuals 
with a much wider variety of work roles, so it is not possible to make any assumptions 
about the needs of respondents. For this reason, their needs are assessed directly. 
1 67 
Cross-validating the Causal Model and Testing for Invariance 
Needs for affiliation, autonomy and achievement46 are examined in the current study, 
and are defined by Heckert et al (2000) as follows: 
• The need for af iliation is the desire to interact socially and to be accepted by 
others. 
• The need for autonomy is the desire for self, rather than other, direction. 
• The need for achievement is the desire to excel and involves attempting to 
improve on past performance. 
Needs theory holds that people have different levels of needs for affiliation, dominance, 
autonomy and achievement (Heckert et aI . ,  2000), and will be motivated to fulfil these 
needs. Needs have been studied with a view to assessing their importance as possible 
determinants of person-occupation fit, and it has been shown empirically that the extent 
to which these needs are met predicts job satisfaction and competence in a number of  
occupations (Medcof & Hausdorf, 1 995). Higher order needs have also been found to 
correlate with organisational commitment (Steers, 1 977; Steers & Braunstein, 1 976), 
and opportunities to satisfy needs have been shown to predict job satisfaction and 
tendency to leave (Zinovieva et aI . ,  1 993) as well as organisational commitment 
(Shouksmith, 1 994). These results suggest the possibility that, in a given work 
environment, individual differences in needs may affect whether or not individuals 
report being satisfied with their jobs or committed to their organisation. 
It is probably most useful to consider the interactions between personal dispositions and 
the environment when examining the effects on organisational outcomes. Individuals 
whose work experiences are compatible with their personal dispositions (i.e., provide 
fulfilment of needs) should have a more positive experience of work than those whose 
work experiences are less compatible with their needs (Meyer & Allen, 1 99 1) .  That is 
to say, that the same work environment might engender differing levels of satisfaction 
or commitment in employees, depending on their differing needs. Research in Germany 
into the influence of achievement and affiliation variables on work motivation and job 
satisfaction  suggests that needs for affiliation and achievement, and the degree to which 
46 Also measured by the NAQ, but not used in the current study, as it is not directly relevant to friendship 
issues, is the need for dominance, sometimes called the need for power; it is the desire to influence and 
direct others. 
1 68 
Cross-validating the Causal Model and Testing for Invariance 
these needs are met have a significant influence on employees' experience of work 
(Pifczyk & Kleinbeck, 2000). 
A particular type of "work experience (in this instance, having friends at work) will 
probably influence commitment and/or satisfaction primarily among employees for 
whom it is relevant (for example, those reporting high needs for affiliation). Thus, it is 
likely that people who report high needs for affiliation will be relatively more affected 
by a lack of friends at work, while for those reporting relatively high needs for 
autonomy, workplace friendships will have less salience. In terms of the model being 
tested, it is possible that some of the linkages between variables, or the fit of the model 
as a whole, will differ depending on individual differences in the population under 
investigation. 
Affiliation 
A person with a high need for affiliation enjoys being with friends and people in 
general; accepts people readily and makes efforts to win friendships and maintain 
associations with people (Medcof & Hausdorf, 1 995). Fulfilling one's need for 
affiliation can be seen as obtaining ' social capital. '  Social capital, is about the benefits 
to individuals of initiating and maintaining relationships (Johnson, 2001) .  Informal 
connections among employees are vital in building social capital. Even a small number 
of people who share common experiences can create larger networks within an 
organisation to solve work problems more effectively, share resources and knowledge, 
and master new technology. These ' communities of practice' can be crucial for new 
employees who are looking for well-informed colleagues or mentors within the 
company to help them succeed in their work roles (Johnson, 2001 ). 
In addition, social capital is important for the management of knowledge-based 
organisations. It helps to create the climate of mutual trust, respect, and obligation that 
allows people to readily share documents, contribute knowledge, and understand the 
value of co-workers' contributions. Thus, having informal relationships and friendships 
in the workplace will be strongly related to employees' social capital. 
1 69 
Cross-validating the Causal Model and Testing for Invariance 
According to the conceptualisation on which the present study is based, the degree to 
which jobs provide interpersonal activitIes should relate to job satisfaction and 
organisational commitment more when workers have high desires for the satisfaction of 
social needs and, further, may relate to work motivation or performance when working 
hard on the job can lead to the satisfaction of these needs. This is likely to be 
particularly true for jobs that are designed so that relating meaningfully to others is an 
integral part of doing the job well (Hackman & Lawler, 1 97 1 ). 
Autonomy 
The need for autonomy is the desire to be self-directed. Individuals with a need for 
autonomy would be likely to prefer to work alone rather than in work teams. For 
individuals with high needs for autonomy, acquiring social capital may be seen as less 
important than being in a work role which allows them to work independently of others. 
For people with high needs for autonomy, working in jobs where working closely with 
others is important or essential in doing the job well, might actually decrease 
satisfaction, as it would mean that needs for autonomy could not be met. 
Achievement 
A person with a high need for achievement wants to accomplish difficult tasks, 
maintains high standards and is willing to work toward distant goals. They respond 
positively to competition and are willing to expend effort to attain excellence (Medcof 
& Hausdorf, 1 995) .  Thus, individuals with high needs for achievement will be most 
satisfied with jobs which fulfil this need and the presence or absence or very close 
friends and informal networks may be less important than for individuals with a high 
need for affiliation (unless they have high need for both affiliation and achievement). 
The need for achievement may be related to ' career capital' (Inkson & Arthur, 2001). 
Career capital can be thought of as the skills, experience and relationships individuals 
acquire over time, which will increase an individual' s  employability. Inkson and Arthur 
propose that careers are personal property, and that people are motivated to acquire 
career capital, not just for the good of their employer but by their own self-interest. The 
more career capital an individual acquires, the more he or she is likely to achieve. 
1 70 
Cross-validating the Causal Model and Testing for Invariance 
It is worth noting here that it is incorrect to think of the acquisition of career capital as 
unrelated to individual' s  need for affiliation, because social capital (the building of 
useful relationships) is  certainly an aspect of career capital. This aspect of career capital 
is termed 'knowing whom' by Inkson and Arthur (2001 ), and it includes the 
attachments, relationships, reputation and felt obligations that people gather as they 
pursue their careers (Inkson & Arthur, 2001 ) . 
If individuals are focused on gaining career capital they will be constantly learning and 
acquiring new skills, contacts and knowledge, which will help them in their current and 
future work roles. Table 25 summarises the points made above, focusing on individual 
differences in needs and the presence or absence of both friendship opportunities and 
opportunities to acquire career capital. 
Table 25: Hypothesised Relationship between Occupation, Affi l iation Opportunities and 
Opportunities to Acquire Career Capital. 
Individuals with needs 
for affiliation 
Absence of friendship Low Job Satisfaction / 
opportunities and Increased turnover 
informal relationships 
at work 
Presence of friendship Increased satisfaction, 
opportunities and organisational 
informal relationships commitment and 
at work reduced turnover 
Absence of May or may not affect 
opportunities to satisfaction, 
acquire career capital commitment and 
at work turnover . . .  individual 
differences 
Presence of May or may not affect 
opportunities to satisfaction, 
acquire career capital commitment and 
at work turnover . . .  individual 
differences 
5.1 .2 Interdependence of Work Role 
Individuals with needs 
for autonomy 
May improve 
satisfaction and reduce 
turnover 
May decrease 
satisfaction and 
commitment, and 
increase turnover 
May or may not affect 
satisfaction, 
commitment and 
turnover . . .  individual 
differences 
May or may not affect 
satisfaction, 
commitment and 
turnover. . .  individual 
differences 
Individuals with needs 
for achievement 
May or may not affect 
satisfaction, 
commitment and 
turnover . . .  individual 
differences 
May or may not affect 
satisfaction, 
commitment and 
turnover. . .  individual 
differences 
Decreased satisfaction 
and commitment, and 
increased turnover 
Increased satisfaction, 
organisational 
commitment and 
reduced turnover 
Just as the needs of respondents may influence the relationships between the measured 
variables, so too might the type of job. In some roles, employees may actually require 
higher levels .of social support in order to function effectively because their jobs may be 
1 7 1  
Cross-validating the Causal Model and Testing for Invariance 
designed so that relating to others is an essential part of doing the job well. For 
example, a clinical psychologist may want to debrief after counselling a very violent 
social misfit, or may require advice, mentoring and supervision from her more 
experienced colleagues. In contrast, an accountant who has to handle a particularly 
challenging tax return would be unlikely to rely on social support of this nature, and 
could refer to a text to gain information. As another example, in the computer 
programming profession, asking for assistance and support from colleagues is  
sometimes seen as an inability to effectively perform your job, while individual problem 
solving is highly valued (Kamante, 2002). 
In the case of the accountant or the programmer, informal relationships and social 
support might be an added bonus to their jobs, nice but not essential. While for the 
psychologist they may be crucial to perform her job satisfactorily. This relates to the 
theory of Herzberg et al. ( 1 959) and Herzberg ( 1 966) (described in section 2. 1 .2), which 
postulated that factors intrinsic to the nature and experience of doing work, such as the 
work itself and responsibility, were job satisfiers or 'motivator�' while extrinsic factors, 
including supervision, interpersonal relations and working conditions were job 
'dissatisfiers' or ' hygiene' factors. It is possible that, for people in certain roles, 
interpersonal relationships are intrinsic to the experience of doing work and therefore 
these relationships cease to be hygiene factors (as Herzberg et al. propose) and are, 
instead, motivators. Thus it may be actual job characteristics, as opposed to individual 
differences, which make social support networks and informal relationship structures 
motivators rather than hygiene factors in some industries or organisations. 
For the purposes of this study, the level of interdependence in an individual 's  job is 
conceptualised to be on a continuum from interdependent to autonomous. It is likely 
that, for individuals who report requiring regular communication and interaction with 
colleagues in order to fulfil their duties, informal relationships will generally have more 
salience or importance, than for those who report being able to do their work on their 
own (i.e., who have relatively more autonomous jobs). 
1 72 
Cross-validating the Causal Model and Testing for Invariance 
The notion that relationships in the workplace will have more salience for people in 
particular work roles is supported by the findings of Win stead et al . ( 1 995) who studied 
both general and academic staff in a University context in Virginia (U.S.A). Winstead 
et al. found that the effects of relationship factors (in this case maintenance difficulty) 
on job satisfaction were only apparent for general staff (as opposed to faculty members). 
An explanation Winstead et al. give for this is that, for faculty, the intrinsic qualities of 
their job (teaching, research, supervising students) are more important to them than the 
non task-related aspects; while for general staff, whose work may be more routine or 
work roles more interdependent, the social, non-work aspects are more salient. An 
alternative explanation is that, for academic staff, social needs may be met in the course 
of their job; teaching, supervising and collaborative research all being very social 
activities. For general staff, on the other hand, rewarding social interactions may be 
more rare, and therefore more valued. In addition, faculty have significantly more 
autonomy at work, and more control over their personal space, usually having their own 
office. Consequently academic staff members are able to simply avoid negative 
interactions more easily than general staff, who may have to liase frequently with 
others, or work in an open plan office. Either way therefore, it seems likely that the type 
of work an individual does will interact with the impact of friendships at work, with 
relationships being generally more salient for jobs which are relatively more 
interdependent. 
Research Question 5 
Research Question 5 asks: Is the proposed model invariant across individuals from sub 
groups of the surveyed population? The purposes of this chapter are to : (a) cross­
validate the previously supported model across independent samples from within the 
full sample, and (b) test for the invariance of the specified model across those who 
report working in relatively less or more interdependent jobs, and those who report 
relatively greater and lesser needs for affiliation, autonomy and achievement. The 
hypotheses, related to Research Question 5 are as follows: 
Hypothesis 5. 1 :  That the previously supported model will be invariant across two 
groups of randomly assigned respondents, thereby validating the model . 
1 73 
Cross-validating the Causal Model and Testing for Invariance 
Hypothesis 5.2: That the model will be noninvariant across groups of individuals who 
report having: 
(a) relatively high versus low needs for affiliation, 
Cb) relatively high vers�s low needs for autonomy, 
( c) relati vel y high versus low needs for achievement, 
(d) relatively less or more interdependent jobs. 
5.2.1 Participants 
5.2 Method 
Data from four hundred and forty-five individuals were used in the analysis for this 
study. This total comprised the 412 respondents reported in the previous chapter and an 
additional 3 3  respondents who subsequently responded to the survey. A demographic 
summary is presented in Table 26. Respondents ranged from 1 9  to 64 years in age, with 
a mean age of 3 5  years (s.d. = 1 1 .07). There was variety in the industries/sectors 
respondents reported working in; the l argest reported sector was tertiary education (92 
respondents) followed by health care (including psychology, psychiatry and 
physiotherapy, 53  respondents). People responded to the survey both from within New 
Zealand and internationally, respondents were primarily from Western countries but the 
international mix gives the findings wider generalis ability than previous studies, using 
only American respondents C e.g. , Nielsen et aI . ,  2000; Richer et aI. ,  2002 ; Riordan & 
Griffeth, 1 995;  Winstead et aI . ,  1 995). 
1 74 
Cross-validating the Causal Model and Testing for Invariance 
Table 26: Breakdown of Demographic Data for Respondents to Cross-validation I 
Invariance study 
Variable Freqllency (n) Valitl percent 
Sex (5 missing) 
Males 137 3 1 .2 
Females 303 68.8 
Age (9 missing) 
>20 years 10  2 .3  
20-29 years 168 38 .5 
30-39 years 1 2 1  27.8 
40-49 years 75 17  
50-59 years 57 1 3 .3 
Over 60 years 5 1 . 1  
Country of origin (5 missing) 
New Zealand 294 66.7 
U.S .A. 65 14.8 
United Kingdom 36 8 
Australia 22 4.8 
Canada 5 1 . 1  
Other 1 8  4 . 1  
Note: Values are presented in percentages excluding respondents who declined to answer. 
5.2.2 Materials 
As described in the previous chapter, data were gathered using a self-administered, 
Internet based questionnaire, which was designed to measure the interdependence of 
respondent' s jobs, respondent' s  needs, workplace friendship opportunities, workplace 
friendship prevalence, cohesion, job satisfaction, organisational commitment and 
intention to leave. Most of the instruments and survey questions relevant to the current 
study are fully described in Chapter Three and include the Workplace Friendship Scale 
(Nielsen et aI. ,  2000), the Workgroup Cohesion Scale (Campion et aI. ,  1 993), the 
Organisational Commitment Questionnaire (Mowday et aI. ,  1 979), the Job Satisfaction 
Scale (Warr et aI. ,  1 979), and a measure of intention to leave (Mobley, 1 977). In 
addition the Needs Assessment Questionnaire (Heckert et aI . ,  2000) and two 
interdependence of work role questions were also used, these are described below. 
Needs Assessment Questionnaire 
The Needs Assessment Questionnaire was designed to measure four needs; the needs 
for achievement (nAch), affiliation (nAff), autonomy (nAut), and dominance (nDom) 
(refer appendix 1 6  for the items in the NAQ). 
1 75 
Cross-validating the Causal Model and Testing for Invariance 
The Needs Assessment Questionnaire was developed by Heckert et al. (2000) to 
measure achievement, affiliation, autonomy, and dominance. Although the Manifest 
Needs Questionnaire (Steers & Braunstein, 1 976) exists, Heckert et al. found that prior 
studies reported problematic internal consistency of the Manifest Needs Questionnaire 
scores, with several researchers (e.g., Dreher & Mai-Dalton, 1 983 ; Geiger & Cooper, 
1 995;  Mayes & Ganster, 1 983) obtaining internal consistency values that strongly 
question the reliability of the Manifest Needs Questionnaire. 
To create the Needs Assessment Questionnaire, a 60-item instrument, consisting of the 
20 Manifest Needs Questionnaire items and 40 new items, was administered to 476 
undergraduates. Based on the factor analysis results, Heckert et al. created the 20-item 
Needs Assessment Questionnaire. Internal consistency estimates were higher for the 
Needs Assessment Questionnaire scores than for the Manifest Needs Questionnaire 
scores. Heckert et al. (2000) calculated item-total correlations and coefficient alpha for 
both the original scales and the new scales. The average item-total correlations were 
higher for the new scales than for the original scales (.76 vs . .59 for the nAch scales, .72 
vs . .  53 for the nAff scales, .64 vs.O .58 for the n Aut scales, and .74 vs . .  68 for the nDom 
scales). With all four scales, the scores derived from the new scales outperformed the 
scores derived from the original scales with regard to internal consistency, ranging from 
.65 to .8 1 .  
Evidence of the test-retest reliability of the NAQ was provided by a second study using 
78 undergraduate students. The scales showed relatively stable scores, with test-retest 
reliability estimates of .64, .73, .63 , and .66, respectively, for the nAch, nAff, nAut, and 
nDom scales. 
Taken together, the results from Heckert et al. ' s  studies provide support that the NAQ is 
a reliable and valid alternative to the Manifest Needs Questionnaire, with both student 
and worker samples, for measuring the needs for achievement, affiliation, and 
dominance. The support for the nAut scale was more limited and, as a result, the 
1 76 
Cross-validating the Causal Model and Testing for Invariance 
authors are somewhat cautious in their recommendation for its use for measuring the 
need for autonomy. 
Factor analysis of the· Needs Assessment Questionnaire (refer section 4.3 . 1 )  revealed 
that, for both the affiliation and autonomy subscales, three of  the five items in the 
original scale loaded together as factors, all five of the achievement items loaded 
together. Scores for the questions were combined to give each respondent a composite 
score for the affiliation need, autonomy need and achievement need. The items used are 
presented in Table 27. 
T bl 27 N d A a e ee s ssessment Q f ues lonnalre It ems 
Affiliation Autonomy 
I try my best to work alone on a I would like to be my own boss. 
work assignment. (R) 
I prefer to do my own work and I would like a job where I can 
let others do theirs. (R) plan my work schedule myself. 
U d ·  h C se In t e urrent S d tu Iy 
Achievement 
I try to perform my best at work 
It is important to me to do the best 
job possible 
When I have a choice, I try to I would like a career where I have I am a hard worker 
work in a group instead of by very little supervision. I push myself to be "all that I can myself. be" 
I try very hard to improve on my 
past performance at work 
Interdependence/autonomy questions 
The following two questions were used to assess the level of interdependence in 
respondents' jobs. Items were rated on a 7-point likert type scale ranging from 1 
(strongly disagree) to 7 (strongly agree) . 
1 .  In order to fulfil my duties at work, regular communication and/or interaction 
with my colleagues is important. 
2. The type of work I do can be done satisfactorily on my own, without regular 
interaction and/or communication with my colleagues. 
To assess the validity of the interdependence/autonomy questions, their correlation with 
the measures of cohesion and friendship prevalence was assessed. Those in more 
autonomous jobs will probably not be part of a highly cohesive workgroup and may 
report less need for friendship at work, or perhaps have fewer opportunities to make 
friends. Thus, a negative correlation between the "autonomy of work role" score and a) 
1 77 
Cross-validating the Causal Model and Testing for Invariance 
the social support factor of the cohesion scale, b) the friendship prevalence scale and c) 
the need for affiliation subscale would be expected if the "autonomy of work role" score 
was valid. In addition, according to the attraction/selection/attrition theory (Schneider 
et aI . ,  1 995), discussed in section 5 . 5 . 1 ,  people with relatively low needs for affiliation 
and/or high needs for autonomy are more likely to be in autonomous work roles. Thus, 
respondents indicating that their job was autonomous (by scoring low on Q 1  and high 
on Q2) should also score high on the autonomy sub-scale of the Needs Assessment 
Questionnaire and low on the affiliation sub-scale of the Needs Assessment 
Questionnaire (validity and reliability of the Needs Assessment Questionnaire is 
presented above). 
A measure of respondents' autonomy of work role was calculated by reverse scoring 
question 1 and taking the mean of the two scores (similarly, a measure of the level on 
interdependence of the respondents' jobs may be calculated by reverse scoring the 
second question and taking the mean). To check the psychometric properties of the 
items, their correlation with existing scales was assessed. As expected the ' autonomy of 
work role' score was negatively correlated with the social support factor of the cohesion 
scale (r = -. 1 6, p < . 001), the friendship prevalence scale (r = -. 1 0, p < . 05) and the need 
for affiliation subscale (r = - .23 , p < . 001), providing evidence of divergent validity for 
the two new interdependence/autonomy questions. The 'autonomy of work role' score 
was significantly positively correlated with the 'need for autonomy' subscale (Heckert 
et aI. ,  2000) (r = . 1 7, p < . 001), providing evidence of convergent validity. 
5.2.3 Procedure 
The Internet data collection site remained active following the download of the data 
used in the previous chapter and an additional 3 3  individuals responded to the 
questionnaire. The data were downloaded and used to again create the measurement 
models of the scales and to test the theoretical model. 
1 78 
Cross-validating the Causal Model and Testing for Invariance 
5.3 Analysis and Results 
Analyses were based on the AMOS (Arbuckle, 1 999) program, and were conducted in 
five stages. First, measurement models of the scales used to measure the latent 
variables, derived from the previous chapter, were confirmed with the larger data set. 
Second, the data were assessed for goodness of fit to the proposed model . Third, the 
data were randomly split into two (using SPSS), to form calibration (n = 230) and 
validation (n = 2 1 5) samples. Fourth, the calibration sample was assessed for goodness 
of fit to the proposed model. Fifth, the model was cross-validated by testing for the 
invariance of all parameters across the second independent sample. 
The cross-validation procedure is outlined by Byrne (2001 )  and involves first 
performing an omnibus test, determining the goodness of fit for the two groups in 
combination, and with no equality constraints imposed. Having constrained all 
parameters to be equal across groups it is possible to compare the constrained model 
with the initial multi group model, in which no equality constraints were imposed, to 
determine if the causal structure is invariant. The change in chi-square value (b.X2) 
provides the basis for comparison with the initial multi group model47 . Respondents 
were divided into groups using median splits, and the validated model was then used to 
test for invariance between individuals who reported having relatively less (n = 20 1 )  
and more (n = 244) interdependent work roles, those who indicated high (n = 238) 
versus low (n = 207) needs for affiliation, those who reported high (n = 268) versus low 
(n = 1 77) needs for autonomy and those who reported high (n = 236) versus low (n = 
209) needs for achievement. 
Assessment of model fit was based on multiple criteria, reflecting statistical and 
practical considerations (Byrne, 200 1) ;  these were (a) the chi-square (X2) likelihood 
ratio statistic, (b) the Comparative Fit Index (CSI :  Bentler, 1 990), the Parsimonious 
Normed Fit Index (PCFI: Mulaik et aI . ,  1 989), and (c) the Root Mean Square Error of 
47 If the change in Chi Square is significant the fit of the two data sets to the model can be judged to be 
significantly different. 
1 79 
Cross-validating the Causal Model and Testing for Invariance 
Approximation (RMSEA: Browne & Cudeck, 1 993). These indices are described fully 
in the previous chapter (refer section 4.3 . 1 ) � 
5.3.1 Measurement 'Models of the Scales 
The computer programme AMOS (Arbuckle, 1 999) was used to create measurement 
models of the scales using the data gathered from all 445 respondents. The 
measurement models of the scales used in the current study remained unchanged from 
the previous study. The fit indices for each measurement model show good fit of the 
data to the models, fit indices are presented in Table 28 .  The reliability alpha 
coefficients for each subscale were acceptable, ranging from .70 to .9 1 .  Both the 
cohesion scale and the satisfaction scale were again found to have two factors48. 
Table 28: Fit Indices for the Measurement Models n = 445) 
Scale Number x: df PCFI CFI RMSEA 
0 actors 
W orkplace friendship scale 2 
Job satisfaction scale 2 
Cohesion Scale 2 
Organisational Commitment 1 
Questionnaire 
Needs Assessment Questionnaire 4 
5.3.2 Model of Workplace Relationships 
40.59 1 3  
74. 10  26 
6 1 . 1 0  1 9  
1 53.70 54 
2 16.37 7 1  
.60 
.69 
.66 
.79 
.68 
.98 
.96 
.97 
.96 
.87 
.069 
.065 
.07 1 
.064 
.07 1 
The result of the SEM analysis is shown in Figure 28; there are eight latent variables, 
three hypothesised correlations and nine hypothesised regression paths, indicated by 
arrows. All correlations and regression paths shown are significant (p < .05). In spite of 
the slightly different data set, regression weights between variables were relatively 
unchanged from the previous chapter. Table 29 shows Chi Square, degrees of freedom 
and the relevant goodness of fit statistics for the model under test, again they are largely 
unchanged from the previous chapter and indicate the model is a good fit to the data. 
The complete table of fit indices is given in appendix 2 1 .  
48 The two factors in the satisfaction scale were, (1 ) satisfaction with relationships and workplace 
(extrinsic satisfaction), and (2) satisfaction with aspects of actual job performed; variety/fulfilment 
(intrinsic satisfaction). The two satisfaction factors relate closely to the 'extrinsic satisfaction' and 
' intrinsic satisfaction' clusters of items, identified by Warr et al. ( 1979) . The two cohesion factures were, 
(1 ) social support and cooperation and (2) workload sharing. The workload sharing factor is identical to 
that described by Campion et al ( 1993), while the remaining items loaded together as a single factor, 
combining Campion's 'social support' and 'communication/co-operation' factors. 
1 80 
Cross-validating the Causal Model and Testing for Invariance 
Table 29: Fit Indices for the Causal Model: Based on the Full  Sample (n = 445). 
Index offit df CFI PCFl RMSEA Low 90 High 90 Hoeller . 01 
Value 617 .90 .84 .055 .05 1 .058 2 1 7  
The data supported the notion that friendship opportunities (leading to increased 
friendship prevalence) would be positively correlated with the cohesiveness of a 
workgroup, that friendship opportunities and cohesion would be antecedent to job 
satisfaction, that friendship opportunities and cohesion would impact positively on 
organisational commitment (mediated by satisfaction), and that satisfaction and 
commitment would both be antecedent to intention to leave. For a more detailed 
description ofthe theoretical model see the previous chapter . 
. 21 
,.25 
Figure 28: Results of SEM analysis of the theoretical model. Values represent 
standardised estimates and correlations based on the ful l  sample (n = 445). 
All paths shown are significant ( p  < .05) 
1 8 1  
Cross-validating the Causal Model and Testing for Invariance 
5.3.3 Comparing Groups 
Calibration compared to validation sample 
For the purpose of cross-validation, the hypothesised model was tested for its replication 
across two independent (random) samples, i .e. ,  the calibration and validation groups 
(Byrne, 2001) . The calibration sample showed adequate fit to the model (CFI = .89, 
RMSEA = .05). From an omnibus test, which determines the goodness of fit for the two 
groups in combination, and with no equality constraints imposed, the fit is adequate 
(CFI = .88, RMSEA = .04, X2( 1234) = 22 14.8) see Table 30. Next, to test for invariance 
across groups, equality constraints were specified by labelling all parameters in the 
model equal across the two groups. From Table 30 the change in chi-square with 44 
. .  Jr '!': : . .-
degrees of freedom is 48.5 (�X2(44) = 48.5). Since this test statistic is not stafHticillly 
significant, the model is shown to be invariant across the two groups. 
Table 30: Chi-Square Statistics for Tests of Invariance across Sub-groups of the Sample 
Causal friendship Omnibus test / Comparative model A,i Adf Statistical 
model baseline model (no (Factor loadings, significance 
equality constraints variances and 
imposed) covariances 
constrained equal) 
X2 df X2 df 
Random sample 22 14.8 1 234 2263 .3 1278 48.5 44 ns 
(calibration versus 
validation sample) 
High versus low 2238.3 1 234 2282.8 1 278 44.5 44 ns 
Affiliation needs 
High versus low 2279.5 1 234 2328.2 1 278 48.7 44 ns 
Autonomy needs 
High versus low 2359.0 1 234 24 1 3 . 1  1 278 54. 1 44 ns 
Achievement needs 
High versus low 229 1 .7 1 234 2387.2 1 278 95.5 44 p < .001 
interdependence of 
job 
Needs for affiliation, autonomy and achievement 
The invariance-testing strategy described above was then used to test for the invariance 
of the causal structure for respondents reporting relatively high versus low needs for 
affiliation autonomy and achievement. Table 30  shows that the differences in chi­
square values (L\X2) between the second tests and the omnibus tests were not statistically 
significant (�X2(44) = 44.5, �X2(44) = 48.7 and �X2(44) = 54. 1 for the affiliation, autonomy 
and achievement group comparisons respectively) . Thus, the high autonomy-need 
1 82 
Cross-validating the Causal Model and Testing for Invariance 
group does not differ significantly from the low autonomy-need group in terms of the 
relationships between variables in the model. Similarly the high affiliation-need and 
high achievement-need groups do not differ significantly from the low affiliation-need 
and low achievement-need groups (i.e., the groups are invariant) . Thus, needs seem not 
to influence the way the measured variables in the tested model relate to each other. 
Interdependence of work role 
The same invariance testing strategy was then used to test for the invariance of  
structural paths across groups of  respondents who reported having comparatively less or 
more interdependent jobs. From the omnibus test, the goodness of fit of the model for 
the two groups in combination, and with no equality constraints imposed, is  adequate 
(CFI = .88,  RMSEA = .044, X2( 1234) = 2229 1 .7). 
Next, having constrained all parameters to be equal across groups it is possible to 
compare the results with the initial multi group model to determine if the hypothesised 
causal structure is invariant across the two groups. From Table 30, the change in chi­
square with 44 degrees of freedom is 95.5 (f1X2(44) = 95.5), which is statistically 
significant (p < .00 1 ) ; this indicates that the fit of the data to the model is noninvariant 
(i.e. , different) across the two groups (i.e., those reporting having (a) relatively less and 
(b) relatively more interdependent jobs). 
Given this finding of noninvariance, the next task is to locate the nonequivalent 
parameters in the model. This process involves a series of logically ordered tests for 
invariance, first testing for the equivalence of the factor structure and then for the 
equivalence of the structural model. By doing this, it is possible to determine which 
parameters in the model are different between those in relatively less and more 
interdependent jobs. Table 3 1  shows the results bearing on this series of tests for 
invariance. Each model tested is compared to the baseline model. A significant change 
in Chi-square between the model tested and the baseline model indicates that the two 
are noninvariant (i.e., significantly different). 
1 83 
Cross-validating the Causal Model and Testing for Invariance 
Table 31 : Goodness of Fit Statistics for Tests of Invariance Across Those in Relatively 
Less and More Interdependent Jobs 
Model 
number 
2 
3 
4 
5 
6 
7 
8 
9 
10 
1 1  
12  
1 3  
1 4  
1 5  
1 6  
1 7  
1 8  
MOllel description 
Combined baseline models, high versus low 
interdependence· of job (Model 1 ) 
Factor loadings, variables, regression paths · 
and covariances constrained equal 
Only factor loadings constrained equal 
All factor loadings constrained equal other 
than item 4 (workplace friendship), item 8 
(cohesion) and item 6 (OCQ) which were 
freely estimated. 
t df 
229 1 .7 
2387.2 
2358.6 
2326.9 
1 234 
1 278 
1263 
1 260 
95.5 
66.9 
35 .2 
As 4 but with variables constrained equal also 2330.2 1 263 38 .5 
As 5 but with all regression paths and 2355.4 1275 63.7 
covariances constrained equal also 
As 5 but with only the path between 
friendship opportunities and friendship 
prevalence constrained equal also 
As 7 but with path between friendship 
opportunities and extrinsic satisfaction 
constrained equal also 
As 8 but with path between cohesion (social 
support) and extrinsic satisfaction constrained 
equal also 
As 9 but with path between cohesion 
(workload sharing) and extrinsic satisfaction 
constrained equal also 
2332.2 
2333.4 
2333.7 
2336 . 1  
1 264 40.5 
1265 4 1 .7 
1 266 42 
1 267 44.4 
As 10 but with covariance between friendship 2344.4 1 268 52.7 
opportunities and cohesion (social support) 
constrained equal also 
As 10 but with co variance between cohesion 2339.3 1268 47.6 
(workload sharing) and cohesion (social 
support) constrained equal also 
As 1 2  but with covariance between friendship 2339.8 1 269 48.7 
opportunities and cohesion (workload 
sharing) constrained equal also 
As 13  but with path between extrinsic 
satisfaction and intrinsic satisfaction 
constrained equal also 
As 14 but with path between extrinsic 
satisfaction and organisational commitment 
constrained equal also 
As 1 5  but with path between intrinsic 
satisfaction and organisational commitment 
constrained equal also 
As 16 but with path between intrinsic 
satisfaction and intention to leave constrained 
equal also 
As 17 but with path between organisational 
commitment and intention to leave 
2340.2 
2344.5  
2345.6 
2345 .8  
2346.0 
1 270 1 270 
127 1  52.8 
1272 53.9 
1 273 54. 1 
1 274 54.3 
t1df Statistical 
44 
29 
26 
29 
4 1  
30 
3 1  
32 
33 
34 
34 
35 
36 
37 
38 
39 
40 
sigllificance 
p < .00 1 
p < .00 1 
ns 
ns 
p < .05 
ns 
ns 
ns 
ns 
p < .05 
ns 
ns 
ns 
ns 
ns 
ns 
ns 
constrained equal also 
Note. !!.X2, difference in X2 values; !!.df, difference in degrees of freedom. All models are compared witl 
Model 1 .  
1 84 
Cross-validating the Causal Model and Testing for Invariance 
As shown previously in Table 30, there was a significant difference between the 
baseline model and the fully constrained model for the high versus the low 
interdependence groups; this information is also displayed in the first two lines of Table 
3 1 .  Once the noninvadance of a model has been established the next step is to test for 
the equivalence of the factor-loading pattern across the two groups.  The third line in 
Table 3 1  shows the result of this test. The significantly different Chi-square indicates 
that not all factor loadings are invariant across groups .  A noninvariant factor-loading 
pattern necessitates the use of partial measurement invariance in testing for the equality 
of regression paths and covariances (which are the parameters of interest in this case). 
To ascertain which aspects of the factor-loading pattern vary across the two groups, 
each measurement model was tested for invariance49. Table 32 shows the result of these 
analyses. The statistical significance of the change in Chi-square for the cohesion, 
friendship and organisational commitment · measurement models indicates that those 
who report being in jobs which are relatively less or more independent, differ 
significantly in terms of the way these three measures fit. 
Next, every factor within the cohesion, friendship and the organisational commitment 
measurement models was systematically tested for invariance in order to identify the 
noninvariant parameters. The results bearing on these series of analyses indicated that 
only one item from each scale was ,noninvariant. The noninvariant items were item 8 of 
the cohesion scale, item 4 of the workplace friendship scale and item 6 of the 
organisational commitment questionnaire. The complete series of tests is not displayed 
here but the relevant results are shown in Table 32 .  The significance of the change in 
Chi-square when all factor loadings are constrained equal, along with the non­
significance of the change in chi-square when each of these three items was freely 
estimated, indicates that they were the sole source of invariance in each of the 
measurement models .  
49 The measurement model for Intention to  Leave was not tested here as i t  has only three items and 
therefore 0 df. 
1 85 
Cross-validating the Causal Model and Testing for Invariance 
Table 32: Goodness of Fit Statistics for Tests of Measurement Model Invariance Across 
Those in Relatively Less and More Interdependent Jobs 
Model description i cif Lli Lldf Statistical 
siani lcance 
Cohesion Scale 
Cohesion measurement model (unconstrained) 1 7 1 .7 52 
Cohesion measurement model (fully constrained) 202 .8 62 3 1 . 1  1 0  P < .00 1 
All items except item 8 constrained equal (item 8 1 84.7 6 1  1 3  9 ns 
freely estimated) 
Friendship Scale 
Friendship measurement model (unconstrained) 65.4 26 
Friendship measurement model (fully constrained) 8 1 .6 32 1 6.2 6 p < .05 
All items except item 4 constrained equal (item 4 76. 1 3 1  1 0.7 5 ns 
freely estimated) 
Organisational Commitment Questionnaire 
OCQ measurement model (unconstrained) 239.3 108 
OCQ measurement model (fully constrained) 263.6 1 1 9 24.3 1 1  p < .05 
All items except item 6 constrained equal (item 6 253.5 1 1 8 1 4.2 1 0  ns 
freely estimated) 
Job Satisfaction Scale 
Satisfaction measurement model (unconstrained) 100.4 52 
Satisfaction measurement model (fully constrained) 1 1 3.6 60 1 3 .2 8 ns 
Note. /}.X?, difference in X2 values; /}.df, difference in degrees of freedom. Each constrained model is compar 
with the unconstrained model for the same measurement model. 
When the full model is tested for invariance, allowing only the three scale items 
identified as being noninvariant to be freely estimated, there is no longer significant 
difference in the factor structure (as shown in line 4 of Table 3 1 ) .  The invariance of the 
regression paths and covariances (the parameters of interest) across high and low 
interdependence groups can now be tested. 
The testing of invariance hypotheses involves increasingly restrictive models (Byrne, 
200 1 ). In Table 3 1 ,  the model tested in line five is more restrictive than the one above it 
because, in addition to the partial equality constraint being imposed on the factor 
variances, equality constraints are also maintained for the variables (i.e., friendship 
opportunities, job satisfaction, etc.). The non-significant change in Chi-square indicates 
that the model remains invariant across the two groups (i.e., any difference that exists is 
not due to difference in the variables constrained as being equal). Model six (line six of 
Table 3 1) shows the change in Chi-square when, in addition to the constraints described 
1 86 
Cross-validating the Causal Model and Testing for Invariance 
above (i.e., factor variances and variables), the regression paths and covariances are also 
constrained equal. 
Because the difference in the Chi-square value between Model six and Model one is  
statistically significant (�X2(23) = 40.0), the hypothesis of invariance can now be 
rejected. The next step is to ascertain which regression paths or covariances are 
contributing towards this inequality. To do this, it is necessary to test for the invariance 
of each parameter individually, while continuing to hold constrained all parameters 
found to be cumulatively invariant across the two groups. 
The change in Chi-square for model seven, which has only the path from friendship 
opportunities to friendship prevalence constrained equal is non-significant (�X2 ( l 2 ) = 
1 7 .7). This means this path is invariant across the two groups. Thus, the parameter 
from friendship opportunities to friendship prevalence is held invariant whilst the next 
path (from friendship opportunities to extrinsic satisfaction) is tested for invariance 
(model eight, Table 3 1 ), again the non-significant change in Chi-square indicates that 
this path is invariant across groups. Based on this general procedure of cumulatively 
maintaining equality constraints only for invariant elements, the next two parameters 
were tested and were also found to be invariant. 
On testing for the invariance of the covariance between friendship opportunities and 
cohesion (social support), the change in chi-square (�X2 ( l 6) = 28.8) was significant 
(Model 1 1 , Table 3 1 ). Thus, this parameter is noninvariant (different) across the tWQ 
groups. The equality constraint for the covariance between friendship opportunities and 
cohesion (social support) was therefore released for all subsequent models, none of 
which resulted in a significant change in Chi-square (refer lines 1 2- 1 8, Table 3 1 ). 
To sum up, the testing of invariance hypotheses shown in Table 3 1  indicated that there 
was one main difference in the structural relations among the variables measured in the 
current study. Other than a single item in each of three measurement models 
(friendship, cohesion and organisational commitment), the factor-structure related to the 
measurement models is equivalent across the two groups. There are, however, 
1 87 
Cross-validating the Causal Model and Testing for Invariance 
significant group differences with respect to the covanance between friendship 
opportunities and the social support aspect of cohesion. 
Table 33 shows the correlation between the cohesion (social support) and friendship 
opportunity variables for high and low interdependence groups, along with the critical 
ratio values. Although both correlations are statistically significant, the analyses 
described above indicate that they are significantly different. It seems that the 
relationship between cohesion (social support) and friendship opportunities is 
significantly stronger for the Low Interdependence group. This implies that, for those in 
relatively less interdependent (more autonomous), jobs the social support aspect of 
cohesion and opportunities for friendship are correlated significantly more strongly, 
than for those in more interdependent jobs. 
Table 33: Showing Correlation Coefficients and Critical Ratio Values (parameter estimate 
d ivided by standard error) of the Correlation Between the Cohesion (Social Support) and 
Friendship Opportunity Variables for High and low Interdependence Groups 
High Interdependence Group 
Cohesion (social support) 
Low Interdependence Group 
Cohesion (social support) 
CR. Estimate S.E. Correlation 
f--7 Friendship Opportunities 4 .24 . 1 3  .03 .37 
f--7 Friendship Opportunities 6.22 .29 .05 .63 
C.R. values >±1 .96 are statistically significant. Parameter estimates and standard errors are also shown, 
along with the correlation coefficient. 
Finally the indices of fit were compared for the low and high interdependence groups. 
The non-equivalence of the causal structure suggests that the model may be better fitting 
for one group compared to another. Table 34 shows the goodness of fit indices when 
the model is tested separately for the high interdependence group compared to the low 
interdependence group. For the high interdependence group the CFI ( .90) and the 
RMSEA (.058) both meet the criteria for a well fitting model, while for the low 
interdependence group the CFI (.85), does not meet the criteria for good fit (i.e. ,  >.9), 
the RMSEA too,  is higher than for the high interdependence group (RMSEA = .067), 
suggesting that the data fit the model less well . These results indicate that the causal 
model, showing the impact of workplace friends on organisational outcomes, is better 
fitting for those in highly interdependent work roles. 
1 88 
Cross-validating the Causal Model and Testing for Invariance 
Table 34: Goodness of Fit Statistics for the H igh and Low Interdependence Groups 
Scale r df CEI PCEI RMSEA 
High Interdependence 1 33 1 .79 690 .90 .82 .058 
Low Interdepentlence 133 1 .22 690 .85 .79 .067 
5.4 Discussion 
There is support for hypothesis 5 . 1 ;  the structure was invariant across two groups of 
randomly assigned respondents, thereby cross-validating the model (hypothesis 5 . 1 ) .  
Hypotheses 5.2a, 5 .2b and 5 .2c were not supported however. Findings indicate 
invariance in the causal model when the samples compared were divided on the basis of 
needs for affiliation (hypothesis 5.2a), autonomy (hypothesis 5 .2b) or achievement 
(hypothesis 5 .2c). These are somewhat unexpected findings, given that it seems 
reasonable to expect that data gathered from individuals with high needs for affiliation, 
and autonomy, particularly, would differentially fit a causal model of friendships 
compared to those reporting relatively lower needs. The non-invariance of the model 
when comparing individuals with relatively high or low needs for achievement is not as 
surprising, as the need for achievement is likely to be less directly linked to friendships 
at work (refer Table 25, section 5 . 1 . 1 ). 
A possible explanation for the unexpected findings may be  that individuals expressing 
higher order needs are having them fulfilled outside the workplace. This relates to the 
concept of 'compensation' from the work-family balance literature (Campbell-Clark, 
200 1 ;  Lambert, 1 990; Sumer & Knight, 2001 ). For example, if employees expressing 
high needs for affiliation have their needs met at home, they may be less likely seek to 
fulfil them at work; thus the absence of friends at work will be unlikely to be any less or 
more salient for these individuals, than for colleagues who have low needs for 
affiliation. 
The model was noninvariant (different) across the two groups reporting having 
relatively less or more interdependent jobs (hypothesis S.2d). Specifically, one item in 
each of the measurement models for friendship opportunities, cohesion and 
organisational commitment differed, along with the correlation between friendship 
1 89 
Cross-validating the Causal Model and Testing for Invariance 
opportunities and the social support aspect of cohesion. The finding that the correlation 
between friendship opportunities and the social support aspect of cohesion is stronger 
for those who report being in relatively less interdependent jobs seems somewhat 
counterintuitive. A possible explanation is that those in very interdependent jobs will 
have opportunities for friendship, regardless of the perceived cohesion In their 
workplace (so a significant correlation · will not be  found) while those In very 
autonomous jobs will only experience increased opportunities for friendship if they also 
perceive themselves to be socially supported by their colleagues. 
When the findings of group invariance between those with high and low needs are 
considered alongside the finding that the model was noninvariant when the sample was 
divided on the basis of the level of interdependence of individual' s  work roles an 
interesting pattern emerges. It seems that the degree of interdependence in an 
individual' s  job influences the relationships between the measured variables, while the 
subjective needs of employees for autonomy, affiliation and achievement will not. 
Findings that the model (featuring friendships as antecedent to organisational outcomes 
such as commitment and job satisfaction) is better fitting for those in interdependent 
jobs is consistent with the hypothesis that those in interdependent jobs will be more 
affected by friendship opportunities than those in autonomous jobs, and also with prior 
research by Winstead et al . ( 1 995) . . Thus, it seems that the actual job someone does, and 
whether or not it is necessary to work with others in order to perform one's job, will 
affect the salience of informal interpersonal relationships at work, while whether or not 
individuals self-report having needs for autonomy and affiliation, will not. 
5.5 Conclusion 
This study cross-validated a model of friendships in the workplace, and results suggest 
that the proposed model is robust, evidenced by the invariance of the model on four out 
of five tests for invariance. In addition, the findings demonstrate that there is a 
difference between those occupying relatively less or more interdependent jobs in terms 
of how the variables in the model relate to each other. 
1 90 
.:; 
Cross-validating the Causal Model and Testing for Invariance 
It was found that the correlation between social support and friendship opportunities is 
significantly stronger for those in less interdependent jobs. A likely explanation for this 
finding is that, for those who report being in highly interdependent work roles, 
friendship opportunities will exist regardless of the reported group cohesion. For those 
in very autonomous work roles, on the other hand, friendship opportunities will 
probably be more dependent on the perceived social support in the work environment. 
In addition, findings suggest that the relationship between opportunities for friendships 
in the workplace and job satisfaction, organisational commitment and intention to leave 
are not the same for all employees. While it makes logical sense that having more 
friends at work will make one's work day more pleasant, it seems that, in terms of 
organisational outcomes, friendships have less impact on those in relatively more 
autonomous jobs (evidenced by the relatively poor fit of the model for this group). This 
finding supports that of Fine ( 1 986), who claims that those in occupations with 
relatively more autonomy will have less need for solidarity with their peers. Second, it 
seems that the reported needs of employees for affiliation, autonomy or achievement do 
not really affect the relationships between measured variables; regardless of 
respondents' reported needs, the impact of friends in the workplace remained 
considerable. 
Acknowledgements 
Parts of this study were presented in the PhD forum of the Fourth Annual International 
Conference of Emotion in the Workplace. 
Morrison, R. (2004). Friendships at work, job type and needs:  The impact on 
organisational outcomes. Paper presented at the PhD Forum of the Fourth Annual 
International Conference of Emotion in the Workphice, London. 
1 9 1  
Synthesis and Implications 
Chapter 6 :  Synthesis and Implications 
6.1 Summary of Major Findings 
The central question of this thesis is whether or not friendships at work (a social factor) 
are linked to the job satisfaction, organisational commitment and turnover intentions of  
employees (i.e. , organisational issues). This focus on informal relationships in 
organisations places the thesis  within the realms of both social and organisational 
psychology. 
A model of workplace relationships and organisational outcomes, derived from previous 
research and theory, was outlined in Chapter Two. The theoretical model hypothesised 
links between relationship factors (workgroup cohesion, friendship opportunities, 
friendship prevalence) and organisational outcomes (organisational commitment, j ob 
satisfaction and intention to leave). 
6.1 . 1  Hospital Study 
The first attempt to test the theoretical model, in a hospital setting within New Zealand, 
was described in Chapter Three (Hospital Study). Three research questions, derived 
from the hypothesised model, were posed. Research Question 1 asked: Do the 
opportunities for, and prevalence of, friendships within a hospital have an impact 
(either mediated or direct) on job satisfaction, organisational commitment, and 
intention to leave? It was hypothesised that opportunities for friendships would have a 
significant effect on how satisfied people are with their jobs, that having more 
opportunities for friendship would increase organisational commitment, and that both 
opportunities for friendship and the prevalence of friends at work would reduce 
turnover. 
1 92 
Synthesis and Implications 
As expected, friendship opportunities did have a positive impact on job satisfaction. 
Friendship opportunities also predicted organisational commitment but the link was 
mediated by job satisfaction. As expected, a relationship between friendship 
opportunities and intention to leave was also found; this linkage too was mediated by 
friendship prevalence and job satisfaction. The friendship prevalence variable was not 
hypothesised to impact directly on job satisfaction or commitment, but was 
hypothesised to relate directly to intention to leave. The data from the Hospital Study 
supported this hypothesis, as there was a significant relationship between friendship 
prevalence and intention to leave. 
Research Question 2 asked: Is the perceived level of cohesion related to opportunities 
for friendship, prevalence of friendship, job satisfaction, organisational commitment 
and intention to leave? It was hypothesised that cohesion would be positively related to 
opportunities for, and prevalence of, friendship in the workplace, and would be  
associated with an increase in job satisfaction and commitment and a decrease in  
intention to  leave. Cohesion was found to relate directly to friendship prevalence and 
friendship opportunities, and to relate to the organisational outcome variables in the 
same way that friendship opportunities did, i.e., directly with job satisfaction and 
indirectly with commitment and intention to leave. 
Research Question 3 asked : Do negative relationships in the workplace worsen 
organisational commitment, job satisfaction and intention to leave? It was 
hypothesised that individuals who experienced negative relationships at work would 
report decreased organisational commitment and job satisfaction and increased intention 
to turnover. It was found that people who reported having at least one negative peer had 
a significantly lower score on the prevalence of friendships at work scale. In addition, 
people who reported having at least one negative peer had a significantly lower score on 
both cohesion sub-scales (social support / communication and workload sharing). Thus, 
although there were no significant differences in satisfaction or commitment between 
those who did, and those who did not, report having at least one negative relationship, 
the data from this study suggested that individuals who report negative relationships at 
work will report being part of a comparatively less cohesive team. 
1 93 
Synthesis and Implications 
The analysis of the data in the Hospital Study indicated that there was good support for 
the proposed model. The results suggested that friendships at work have a significant 
impact on respondents' satisfaction with their jobs, commitment to their organisation 
and intention to leave. In addition, the cohesiveness of employees' workgroups was 
significantly related to both opportunities for, and prevalence of, friends at work. 
6.1 .2 Internet Study 
In order to assess how robust the proposed model was across different contexts, and to 
deal with some of the limitations of the first study, a larger Internet-based study was 
conducted and was described in Chapter Four. In addition to assessing the 
generalisability of the findings from the Hospital Study, a fourth research question, 
focusing on gender differences, was posed. Research Question 4 asked: Are friendship 
variables differently correlated with other organisational variables for men and 
women? 
The larger data set made it possible to test a full structural equation model (SEM); the 
AMOS computer programme was used to carry out the SEM analysis. Taken together, 
the goodness of fit statistics reported in the Internet Study indicated that the proposed 
model was a good fit to the data. Most of the linkages between variables in the Internet 
Study related very closely to the findings in the Hospital Study, demonstrating good 
support for the theoretical model, and indicating that the proposed model was robust. 
In this study, the effect of negative relationships was found to be far greater than in the 
initial study. The results supported the hypothesis and indicated that those with at least 
one negative relationship at work were significantly less satisfied, reported l ess 
organisational commitment, were part of less cohesive workgroups and were 
significantly more likely to be planning to leave their job . 
With respect to the research question focusing on gender differences, the findings 
indicated that the presence of friends and opportunities for friendships at work are more 
strongly related to the leaving intentions of women than of men, suggesting that women 
1 94 
Synthesis and Implications 
will be more directly affected by the presence or absence of close friends at work than 
men; it seems women may make a leaving decision based, at least in part, by the social 
opportunities their work offers . A second finding was that the workload sharing aspect 
of cohesion was significantly correlated with friendship opportunities and friendship 
prevalence for men but not for women, suggesting that women will form friendships at 
work whether or not they share work, but men are more likely to make friends if they 
are in a job involving shared activities with their colleagues. A third difference found 
between the genders was that job satisfaction was not significantly correlated with 
friendship prevalence for women, but was for men. This may relate to a tendency for 
women to make friends at work when they are dissatisfied in their jobs or, alternatively, 
may relate to Herzberg's ( 1 966) two-factor theory of satisfaction (discussed further 
below). 
Another finding, which came through in both the Hospital Study and Internet Study, 
was the difference in the way 'friendship prevalence' (the actual number of friends an 
employee has in the workplace) and 'friendship opportunities' (a generally friendlier 
work environment) relate to the other measured variables. While friendship 
opportunities related significantly to outcome variables such as job satisfaction and 
organisational commitment, friendship prevalence was not found to relate to satisfaction 
or commitment in either study, and was only weakly (though significantly) related to 
intention to leave in the Hospital S�dy. 
6.1 .3 Invariance Testing Study 
The third study, described in Chapter Five, built on the previous two, by cross­
validating the causal model and testing for invariance between sub-groups of 
respondents differing in need strength and interdependence of job. It was on the basis 
of four variables that the sample was divided and tested for group invariance of the 
model of workplace friendships. The four variables were: a) interdependence of work 
role, b) need for affiliation, c) need for autonomy and d) need for achievement. 
It was found that the degree of interdependence in an individual' s  job influenced the 
relationships between the measured variables, while the subjective needs of employees 
1 95 
Synthesis and Implications 
for autonomy, affiliation and achievement did not. Thus, it seems that the actual job 
someone does, and whether or not i t  is necessary to work with others in order to perfonn 
one's  job, will affect the salience of informal interpersonal relationships at work, while 
whether or not individuals self-report having needs for autonomy and affiliation, will 
not. 
6.2 Synthesis 
On reflection of the findings described above and in the body of this thesis, it is possible 
to offer three main conceptual points. 
1 .  The 'friendship prevalence' and ' friendship opportunities' variables seem to 
operate differently from each other; with friendship opportunities being 
linked more strongly to organisational outcomes than friendship prevalence. 
Friendship prevalence should therefore be conceptually distinguished from 
friendship opportunities by researchers, practitioners and within 
organisations. 
2. The interdependence / autonomy of an employee's work role affects how 
applicable the theoretical model is, while the needs of employees do not; 
situations, therefore, appear to affect the salience of relationships at work, 
while individual differences in needs do not. 
3 .  There are gender differences in  the way friendships at work are related to  job 
satisfaction, with friendships being significantly more strongly related to the 
job satisfaction of men. 
6.2.1 Friendship Opportunities versus Friendship Prevalence 
The first conceptual point focuses on the finding that a generally friendlier workplace 
(friendship opportunities) is something which employees seem to both value and to give 
credit to their employing organisation for. Friendship opportunities are a situational 
variable which, when provided by an organisation, are associated with increased job 
satisfaction, organisational commitment and reduced intention to leave. Friendship 
prevalence, on the other hand, was not found to be significantly related to job 
satisfaction, organisational commitment or intention to leave. 
1 96 
Synthesis and Implications 
Several possible explanations for the difference between 'friendship opportunities ' and 
'friendship prevalence' were discussed in the body · of the thesis, but the two most 
plausible explanations are: While friendship opportunities (a generally friendlier 
workplace) will result in increased commitment and satisfaction, friendship prevalence 
(the friends an individual has made) will not, because (i) friendships are initiated and 
maintained regardless of whether or not people are generally satisfied with their jobs or 
committed to the organisation, and (ii) individuals do not generally attribute to the 
organisation positive outcomes gained from the actual friends that they have made. 
Related to the first point, an explanation for the finding that friendship prevalence seems 
unrelated to job satisfaction or organisational commitment is that there may be 
individual or situational differences in the ways friendship prevalence and satisfaction 
relate to each other. While a positive relationship between friendship prevalence and 
satisfaction may exist for some, it is likely that, for others, a positive relationship 
between dissatisfaction and friendship prevalence might exist, i .e . ,  that those who report 
being dissatisfied with their jobs make more friends. The dissatisfaction -7 friendship 
prevalence relationship is related to findings by Sias and Jablin ( 1 995) described in 
section 2.3 .5 ,  which suggested that feelings of dissatisfaction may lead co-workers to 
form and maintain close friendships as a form of alliance against the organisation. It is 
possible that more friendships may be made when people feel dissatisfied in ' solidarity 
against the organisation, ' because they need the increased social support friends can 
provide when they are experiencing unhappiness or stress. Thus, a situation where 
friendships form when employees are dissatisfied may be as likely as one where 
satisfaction increases as a result of having more friends at work. Friendships therefore, 
may form regardless of the satisfaction of employees. 
The second explanation for the difference between friendship prevalence and friendship 
opportunities, is that organisational members may be unlikely to 'give credit' to their 
employing organisation for the friends they have made at work. Some people will make 
friends more easily or will maintain their relationships more effectively than others, and 
this individual difference will likely exist regardless of the situation the individuals are 
1 97 
Synthesis and Implications 
In. In addition, if an employee does not identify strongly with their organisational role, 
they may see their friendships at work as separate from their relationship with the 
organisation (Amason & AlIen, 1 997) and, as a result, friendships will be unlikely to 
have organisational outcomes. Friendship opportunities, on the other hand can be  
attributed to the organisation, as a generally friendly work environment i s  something 
employees can easily see their employing organisation as having provided, and it is 
therefore likely to improve commitment. The extent to which organisational members 
identified with their employing organisation, or attributed positive outcomes of 
friendship to their organisation were not specifically investigated in the current studies, 
so the possible effect of these variables can only be speculative at this stage. 
It is possible that, while ' friendship opportunities' is an environmental / situational 
variable (those in a friendly workplace will score higher on the friendship opportunities 
scale), ' friendship prevalence' might be a personal / dispositional variable (friendly 
people will have more friends, and will score higher on the friendship prevalence scale). 
The two variables are of course closely related, both because friendship opportunities 
may be a necessary condition for friendships to form and because, related to Schneider' s 
( 1 987) attraction/selection/attrition model (refer section 5 . 1 . 1 ) ,  given the opportunity, 
individuals with high needs for affiliation (a personal variable) are likely to select, and 
remain in, friendly workplaces (a situational variable). 
6.2.2 Interdependence of Work Role versus Individual Differences in Needs 
The situational / dispositional differentiation also comes through in the second major 
conceptual point, which derives from the finding that the interdependence of an 
employee's work role affects the linkages between relationship factors and 
organisational outcomes, while the needs of employees will not. Again the situation, in 
this case the type of work role, impacts on the salience of friendships at work, with the 
model fitting better for those in an interdependent work role. On the other hand, the 
personal variables, in this case individual differences in needs for affiliation, autonomy 
and achievement, did not seem to have a significant effect on the way friendship 
variables related to organisational outcome variables. 
1 98 
Synthesis and Implications 
6.2.3 Gender Differences 
Do any human characteristics have an effect on the impact of friendships in the 
workplace? One construct that did seem to make a difference was the gender of 
employees. The third conceptual point focuses on gender differences outcomes of 
organisational friendships, suggesting that men and women differ in the ways they 
approach friendships at work. 
The finding, discussed above, that 'friendship prevalence' and ' friendship opportunities' 
operate differently for people in organisations, can be related to the finding that job 
satisfaction was not significantly correlated with friendship prevalence for women, but 
was for men. It is possible that women, particUlarly, will form strong friendships and 
lean on their colleagues for social, emotional and instrumental support when they are 
dissatisfied with aspects of their work environment. If women's  friendships strengthen 
in situations where they are unhappy or dissatisfied with their jobs, a significant positive 
relationship between satisfaction and friendship prevalence will not be found, even if 
having more friends at work does improve job satisfaction, i .e. ,  the two processes 
(friendships being associated with an increase in satisfaction, and friendships also being 
associated with an increase in dissatisfaction) will cancel each other out. In times of 
adversity, when conditions are bad, women may be more likely to make and keep 
friends than men. Men are perhaps less likely to demonstrate a negative relationship 
between friendships and job satisfaction, because they are less likely than women to 
seek emotional and social support from their colleagues when times are bad (Ashton & 
Fuerhrer, 1 993). Thus, a positive relationship between friendships at work and job 
satisfaction will be more apparent for men. 
The current study is not the first to find friendships at work significantly correlated with 
organisational outcome variables for men but not women. In a study set in a university 
and two engineering firms in the United States, Amason and Allen ( 1 997) found that, 
regardless of the quality of their relationships with colleagues, women's  perceptions of 
organisational support 50 did not change. On the other hand, males reporting positive co­
worker relationships also indicated higher perceived organisational support. It seems 
so Perceived organisational support is closely related to organisational commitment (Alien, 1992). 
1 99 
Synthesis and Implications 
that males were more likely than females to feel that their organisation cared about them 
when they had more and better friends at work (Amason & Allen, 1 997). In other 
words, males ' credited' the organisation for their friendships in a way that females did 
not. Females probably got as much pleasure from their workplace friendships, but were 
less likely to attribute these positive outcomes to the organisation. Again, the degree to 
which respondents in the current study attributed positive outcomes of interpersonal 
relationships to the organisation was not directly measured, but is a direction for future 
research, both in the context gender differences and individual differences. 
Identifying why the friendship prevalence 7 job satisfaction relationship is significant 
for men but not women5 1  can only be speculative here, as this was not a primary focus 
of the current study. Perhaps, as discussed above, women will make more friends than 
men when they experience dissatisfaction. Alternatively, previous research on 
relationships suggests that men derive satisfaction and identity from being part of a team 
(Amason & Allen, 1 997; Markiewicz et aI . ,  2000; Winstead, 1 986). So perhaps when 
men have friends at work, compared to when they do not, they will work better and 
more successfully within the team, achieve more goals and thereby derive satisfaction 
from their job.  Then again, it might be, as Amason and Allen ( 1 997) propose, that 
women do not identify as strongly with their organisational role as men do, they may 
see their friendships at work as more separated from their relationship with the 
organisation, thus friendships will not have organisational outcomes for women. 
Finally, it is possible that women simply expect to have friendships at work. Thus, in 
the absence of friends, women will be dissatisfied, but the presence of friendships will 
not have positive outcomes any more than other expected outcomes of working will (for 
example, their wages). In other words, women may perceive friendship as a necessary 
aspect of work, whereas men may see their organisational friendships as an added 
bonus. This final point relates to Herzberg' s ( 1 966) well-known two-factor theory of 
satisfaction (refer section 2 . 1 .2) .  
5 1  Note: The gender of the person whom respondents were friends with was not assessed, either in  the 
current study, or in Amason and Allen ( 1 997). The friendship dyads discussed will likely include both 
same sex and cross sex relationships. 
200 
Synthesis and Implications 
Herzberg ( 1 966) postulated that factors intrinsic to the nature and experience of doing 
work were job satisfiers or 'motivators' while extrinsic factors, including interpersonal 
relationships, were job 'dissatisfiers' or 'hygiene' factors. In Chapter Five (section 
5 . 1 .2), the possibility' was discussed that, for people in certain roles, interpersonal 
relationships are intrinsic to the experience of doing work, and therefore interpersonal 
relationships cease to be hygiene factors (as Herzberg et al. propose) and are, instead, 
motivators. There was support in the current study for the idea that the degree of 
interdependence in a work role may cause informal relationships to be more salient. 
In terms of the gender findings, it seems that the gender of employees may be related to 
whether or not interpersonal relationships are motivators or hygiene factors. It was 
found that, for men, there was a significant positive correlation between satisfaction and 
friendship prevalence, implying that friendships at work operate as a ' satisfier' or 
'motivator' for men, improving job satisfaction. For women, on the other hand, not 
only was there no significant relationship between satisfaction and friendship 
prevalence, but there was a significant negative correlation between intention to leave 
and all the relationship measures (cohesion, friendship opportunities, friendship 
prevalence). This implies that, for women, friendship acts as a 'dissatisfier' or a 
'hygiene' factor, inasmuch as women will be more likely to be intending to leave their 
job if they report having few or no friends at work, and will be more likely to be 
intending to stay if they report hav.ing more friends at work. It is possible that having, 
or not having, friends at work may be enough to influence female employees' leaving 
decisions. In other words, friendships in the workplace are motivators for men; they 
will improve satisfaction if they are present; while for women, the absence of friends at 
work will cause dissatisfaction; acting as a hygiene factor as Herzberg proposed. 
Although there has been historical criticism of Herzberg's two-factor theory (Hackman 
& Lawler, 1 97 1 ;  Hulin & Smith, 1 967; King, 1 970) it is  often used to provide a 
framework within which to interpret job satisfaction research (e.g. , Adigun & 
Stephenson, 1 992; Furnham et al. ,  1 999; Knoop, 1 994; Maidani, 1 99 1 ;  Yamashita, 
1 995) (refer section 2. 1 .2 for a discussion of the criticism and a description of more 
recent studies utilising the Herzberg's two-factor theory). 
20 1 
6.3 Future Directions 
Synthesis and Implications 
Three directions for future research are (i) investigating the effects of friendships at 
work on performance, (ii) examining the impact of negative relationships in the 
workplace and (iii) investigating the effect of factors such as organisational 
identification and/or attribution on the relationship between friendship prevalence and 
organisational outcomes. The first two enquiries have been discussed within the body 
of the thesis so are outlined only briefly below. The third direction for future research 
arose from attempts to explain some of the findings in the current studies; these 
hypotheses have not been discussed previously and so are described in more detail here. 
6.3.1 Performance 
It would be interesting to examine the possible links between friendship opportunities, 
friendship prevalence and performance. Performance was not examined in the current 
study, and findings from previous research have been somewhat mixed. Friendships in 
the workplace have been associated both positively (Campion et aI. ,  1 993 ; Jebn & Shah, 
1 997; Ross, 1 997) and negatively (Bramel & Friend, 1 987; Eisenberg, 1 994) with job 
performance (refer section 2.3 .3) .  Thus, It would be worthwhile to examine whether (or 
in what circumstances) friendship prevalence or opportunities have an effect on job 
performance. 
6.3.2 Negative Relationships 
Negative relationships are defined and discussed in section 2.3 .7. Although some data 
on negative relationships were gathered in the current study, the impact of these 
relationships was not really within the scope of this thesis. Preliminary findings do 
suggest, however, that the effect of 'enemies' on individual' s  experiences of work can 
be profound. Examining how negative relationships form, looking at the effect of 
negative relationships in the workplace and determining how they might be managed 
are certainly areas that warrant further research. 
202 
Synthesis and Implications 
6.3.3 Organisational Identity and Attributions of Responsibility 
The links from 'friendship prevalence' to 'job satisfaction' and 'organisational 
commitment' are aspects of the theoretical model that warrant further research. Much 
of the qualitative data with a focus on these variables indicates that people believe 
friendships improve their experience of work a great deal. Indeed, it makes intuitive 
sense that, if friendships make us happier in our workday, this positive affective 
response will cause us to be more satisfied with our job and, consequently, more 
committed to the organisation. Yet a friendship prevalence � satisfaction link was not 
convincing in the studies described in this thesis. Two possible variables, which may 
affect any relationship friendship prevalence might have with satisfaction and 
commitment, are (i) the degree to which employees identify with their organisation, and 
(ii) the degree to which employees will attribute positive outcomes of their friendships 
to the organisation. 
Defining organisational identity 
The social identity perspective of organisational identity defines it as the perception of 
sharing experiences of a focal group and sharing characteristics of the group' s  members 
(Ashforth & Mael, 1 989; Mael & Tetrick, 1 992). In recent years interest in the topic of 
organisational identity has increased dramatically; there have been special issues 
devoted to the topic (Albert, Ashforth, & Dutton, 2000; van Knippenberg & Hogg, 
200 1 ), and the number of articles whose titles and/or abstracts make reference to the 
term 'organisational identity' has risen exponentially since 1 988  (Haslam, Postmes, & 
Ellemers, 2003). In the workplace, employees might define themselves in terms of 
team, departmental, organisational or professional identities (Haslam et aI. ,  2003 ; van 
Knippenberg & van Schie, 2000). In addition, other types of social identity can also 
become salient in organisations (e.g., based on gender, class, and ethnicity). 
Irrespective of the specific components defining these forms of organisational identity, 
each will define the individual in terms of a social identity that is shared with other 
members of an in-group but not with members of an out-group (Haslam et aI ., 2003). 
It is  hypothesised that organisational members who identify strongly with the 
organisation (i .e., see themselves as part of the 'organisational in-group' )  may see their 
203 
Synthesis and Implications 
friendships at work as being connected to their relationship with the organisation, thus 
friendships will be more likely to have organisational outcomes for those with a strong 
organisational identity. If an employee views their -identity as wholly separate from 
their work identity, on the other hand, they are unlikely to define their relationships 
within the organisational context (Amason & Allen, 1 997) and may not, therefore, give 
credit to the organisation for the positive outcomes of their friendship. 
Defining attribution 
There are different types of attribution, and attribution theory can be applied to (i) 
understanding the causality for an event, (ii) assessing responsibility for a particular 
outcome, or (iii) assess the personal qualities of those involved in an event (Lord & 
Smith, 1 983). Relevant to the current enquiry, is the second type of attribution; 
assessing responsibility for an outcome. In this case, the focus is the presence of 
friendships at work, and whether or not employees will attribute positive outcomes of 
these friendships to the organisation. Although attributions concerning responsibility 
assessments are usually examined in the context of administering blame or sanctions 
(e.g., Hamilton, 1 980; Hamilton & Sanders, 1 992; Sanders, Hamilton, & Yuasa, 1 998), 
attributions have also been examined in the context of giving credit or reward for 
positive outcomes or behaviours (e.g., Allen & Rush, 1 998; Allen, Russell, & Rush, 
1 994; Crant & Bateman, 1 993 ; Johnson, Erez, Kiker, & Motowidlo, 2002). 
Generally the attributions for responsibility are studied in the context of an individual 
(e.g. the behaviour of a leader, manager or employee) rather than an employing 
organisation. However, it may be possible to examine the degree to which employees 
attribute positive phenomena, such as friendships, back to the organisation (i.e., 'the 
organisation has made it possible or me to have these friends ') ; versus attributing the 
presence of friends to their own effort (i.e., '1 am a friendly person and 1 have put a lot 
of energy into developing and maintaining my friendships; 1 would have made these 
friends regardless of the organisational practices ') . It is hypothesised that if an 
employee attributes positive outcomes of their workplace friendships to the 
organisation, there will be an increase in organisational outcomes such as job 
satisfaction or organisational commitment. If, however, the individual does not 'give 
204 
Synthesis and Implications 
credit' to the organisation for their workplace friendships, the friendships are unlikely to 
result in an increase in job satisfaction or commitment. 
Adding attribution and identification to the theoretical model 
Attribution and identification can be added to the existing theoretical model as potential 
mediators. Figure 29 shows a hypothesised relationship between friendship prevalence 
and job satisfaction. Figure 29 shows that friendship prevalence will be associated with 
increased job satisfaction, but only if an employee either identifies strongly with their 
organisation or attributes the positive outcomes of friendships at work to their 
organisation. The relationships between job satisfaction and commitment, with 
intention to leave remain unchanged from the original model as these linkages were well 
supported by the current studies and are also in line with previous research (e.g., Porter 
et al . ,  1 974; Riordan & Griffeth, 1 995; Tett & Meyer, 1 993). 
Friendship 
prevalence 
Attribution of 
positive outcomes 
of friendships to the 
organisation 
Identification with 
the employing 
organisation 
Organisational 
commitment 
Job satisfaction 
+--.... � Intention to leave 
Figure 29: The hypothesised mediating effect of identification and attribution, on the 
relationship between friendship prevalence and job satisfaction.  
An alternative relationship between the variables is shown in Figure 30.  Here, whether 
or not an employee attributes positive outcomes of their friendships to the organisation 
is modelled as being a consequence of the degree to which they identify with their 
205 
Synthesis and Implications 
employer. According to this model, friendship prevalence will result in satisfaction if 
(and only if) the employee attributes positive outcomes of their friendships to the 
organisation, but the level of attribution is hypothesised to be dependent upon whether 
or not the individual identifies strongly with their organisation. 
Friendship 
prevalence 
Attribution of 
positive outcomes 
of friendships to the 
organisation 
Identification with 
the employing 
organisation 
Organisational 
commitment 
Job satisfaction 
+-__ � Intention to leave 
Figure 30: Alternative hypothesised mediating effect of identification and attribution, on  
the relationship between friendship prevalence and job satisfaction. 
The link between ' identification with their organisation' and 'attributions of positive 
outcomes of friendship' can, perhaps, be related to Pettigrew ( 1 979) who described the 
"ultimate attribution error." The ultimate attribution error is the tendency to internally 
attribute bad out-group and good in-group behaviour, and externally attribute good out­
group and bad in-group behaviour (Pettigrew, 1 979). A person who identifies strongly 
with their employing organisation (i.e., has strong organisational identity) will be likely 
to consider other organisational members, and the organisation itself, as aspects of their 
own in-group (Haslam et al. ,  2003). Although Pettigrew focused on behaviour, and did 
not specifically discuss attributions for friendships, it is possible to take the theory a step 
further and, following Pettigrew's logic, hypothesise that any positive outcomes of 
belonging to the in-group will be attributed to internal causes (in this case, positive 
206 
Synthesis and Implications 
outcomes of friendships at work being attributed to a "good" organisation). On the 
other hand, an individual who does not identify strongly with the organisation (i.e., does 
not consider themselves to be part of the organisational in-group) will be less likely 
attribute positive outcomes of their friendships to the organisation. Instead, positive 
outcomes will be attributed to phenomena external to the organisation (most likely they 
will give themselves the credit for the positive outcomes of the friendships they have 
formed and maintained) . 
It is also possible that organisational identification acts as a moderator between 
friendship prevalence and job satisfaction; perhaps, as individuals identify more 
strongly with their organisation, the friendship prevalence � job satisfaction relation 
becomes stronger. It would be interesting, in the future, to empirically test these 
hypotheses, perhaps measuring 'organisational identification' and ' attribution for 
friendships' along with relationship factors and organisational outcomes. 
6.4 Implications 
It is worth reflecting on how findings described in this thesis might be used by 
employing organisations, employees andlor organisational psychologists. First, can the 
findings be used to enhance the quality of work life? Second, are there some situations 
where the findings might have more relevance? Third, are there any possible negative 
applications? 
6.4.1 Enhancing the Quality of Work Life 
Because findings in the current study implied that the situation, work environment 
and/or type of job affects the salience of friendships at work more than personal 
variables or individual differences, it may be that organisational psychologists aiming to 
engender organisational change in the area of interpersonal relationships should be 
turning their focus to situational / organisational variables rather than personal / 
individual ones. Organisational psychologists should not entirely disregard the 
traditional psychological focus on the individual, but perhaps they would be wise to 
maintain an awareness that, to increase the likelihood of positive outcomes both for 
organisations and individuals, they should look at situational variables such as friendly 
207 
Synthesis and Implications 
work environments, rather than focusing specifically on trying to increase the number of 
friends employees have. Making friends is something that employees will do for 
themselves if they want to, and may have little to do with characteristics of their 
occupation, organisation or job. 
Similarly, if organisational psychologists attempt to ascertain in which circumstances 
friendships in the workplace are most salient, they might do well to look not at 
individual differences in employees (such as needs), but rather at differences in the type 
of job people do. In other words, organisational psychologists should perhaps focus 
their attention on whether the role is interdependent (where relating meaningfully to 
others in an integral part of doing the job well), versus autonomous (where jobs can be 
done satisfactorily on one's own), when planning interventions related to workplace 
relationships, either for management or to improve individuals' work environments. 
The importance of situational factors is good news for management as, although there 
are limits to what an organisation can do in terms of planning interventions aimed at 
improving informal relationships at work, one thing that organisations can do is alter the 
work environment and associated 'office policy' or 'house rules;'  specifically by 
increasing opportunities for friendships to form. The following items are the 
behavioural facets underlying the 'friendship opportunities' construct : 
.. Communication among employees is encouraged by my organisation. 
.. In my organisation I have the opportunity to talk informally and visit with 
others. 
.. I have the opportunity to get to know my co-workers. 
.. I am able to work with my co-workers to collectively solve problems. 
Thus, any intervention aimed at increasing opportunities for friendships in the 
workplace could logically include changes that would increase the likelihood of 
employees agreeing with the above statements. 
208 
Synthesis and Implications 
6.4.2 Situations Where the Findings Might have More Relevance 
Given that it is important (in terms of satisfaction and organisational commitment) for 
organisations to provide friendship opportunities, will some organisations be likely to 
benefit more that others from organisational change that focuses on interpersonal 
relationships? The results of the third study indicate that those organisations with many 
organisational members working in highly interdependent work roles will benefit most. 
In addition, the third study implied that there is less need to focus on the individual 
differences or the needs of employees, as individual differences in needs did not seem to 
affect the linkages between friendship factors and organisational outcomes. For 
example, selecting employees based on their subjective needs may not be as productive 
as analysing the job characteristics, i.e., focusing on the situation or work environment. 
Another aspect of the work environment, that may affect how the findings reported in 
this thesis might be applied, is the gender composition of the workplace. One 
implication of the findings, suggesting that there are gender differences in the 
organisational outcomes of friendships, is that it may be useful to conceptualise 
relationship factors differently in predominantly male workplaces than in predominantly 
female ones. In male dominated work environments friendship prevalence is likely to 
be associated with increased job satisfaction. Therefore, an increased number of 
friendships will probably have positive organisational outcomes. In addition, because 
there were findings that sharing work with colleagues will be associated with an 
increase in friendships for men, jobs involving group tasks and teamwork are likely to 
increase friendships (and therefore satisfaction) in these male dominated work 
environments. 
For women, on the other hand, friendship prevalence is associated, not with an increase 
in satisfaction, but with a decrease in turnover. Where staff retention is an aim, 
organisations that are predominantly female could encourage the formation of 
friendships among the staff. In the context of Herzberg' s  two-factor theory, friendships 
seem to operate as 'hygiene' factors for women. Herzberg maintained that hygiene 
factors cannot produce pleasure but, by their absence, will create dissatisfaction and 
consequently turnover. By this logic, although there is a link between satisfaction and 
209 
Synthesis and Implications 
fiiendship for men but not women, it is perhaps more essential to ensure that female 
employees have satisfying interpersonal relationships at work as, without them, they 
will be comparatively dissatisfied and may leave their jobs. 
Another aspect of an employees work environment, which was found to affect 
respondent' s  experience of work, was the presence of negative relationships. Findings 
in the Internet Study (Chapter Four), that employees with at least one negative 
relationship (compared to those not reporting any negative relationships) reported 
significantly lower satisfaction and commitment and significantly increased intention to 
leave, imply that the impact of negative relationships on employees' experience of work 
may be  even more substantial than that of fiiendships. Targeting interventions, aimed at 
improving workplace relationships, towards workgroups or dyads where negative 
interactions such as concealment, manipulation, conflict, disrespect, disagreement 
and/or animosity are frequent may be a way to improve morale and commitment. In 
addition, in their qualitative responses, respondents often mentioned the presence of a 
negative relationship as a reason they chose to leave their jobs; so attempting to reduce 
negative interactions between employees, perhaps by creating a more friendly and 
supportive work environment, may also be an effective way to reduce turnover. 
6.4.3 Possible Negative Implications 
In Industrial Organisational Psychology it is not always clear who the client is; is it the 
individual or the organisation? Who is served by research and subsequent interventions 
or application of theory? As ethical researchers we must ask ourselves if there are any 
negative or dangerous implications to our findings; this is pursuant to aspects of 
Principle rv 5 2  of the Code of Ethics (2002). Can our research be misused? Can it be 
applied by organisations so that the outcome is detrimental to the individual? Although 
it is difficult to imagine a situation where application of the findings described in this 
52 4. 1. Welfare of society. Psychological knowledge will be increased, and psychology will be practised, 
in such ways as to promote the welfare of society. 
4.3. Benefit to society. Psychologists strive to ensure that psychological knowledge, when used in the 
development of social structures and policies, will be used for beneficial purposes. 
4.4. Accountability, standards and ethical practice. Psychologists strive to ensure the appropriate and 
relevant use of psychological knowledge, practices and structures, and to avoid their misuse ("Code of 
Ethics for Psychologists Working in AotearoalNew Zealand," 2002). 
2 1 0  
Synthesis and Implications 
thesis could be misused, there is a link between a probable antecedent and a 
consequence of female friendships at work, which could perhaps go some way towards 
explaining why some women stay in aversive work environments. 
In light of the evidence that friendship prevalence may increase when employees are 
dissatisfied or unhappy in their jobs (Carr, 2003 ; Sias & Jablin, 1 995), it is possible that, 
although friendships at work can provide emotional support and enjoyment for 
employees, a high proportion of close relationships in the workplace may actually be 
symptomatic of problems within the organisation. When this possibility is  considered 
alongside a finding that women are less likely to be planning to leave their jobs if they 
have friendships at work, it brings up an interesting ethical dilemma. Given the 
probability that women will form close friendships in poor work environments, coupled 
with findings that they will be inclined to stay in their jobs if they have friends at work, 
it is possible to imagine a situation where this link can be abused. If an organisation 
provides opportunities for friendships to form, in an . otherwise unsupportive or 
unpleasant work environment (two factors likely to increase the actual prevalence of 
friends) perhaps female employees will continue to work there, even though it is, 
perhaps, in their best interests to leave. Where is the incentive for management to 
improve conditions if the women will stay anyway, so long as they have friends? 
Although perhaps not directly applicable to the sample of respondents in the current 
study, it is possible to speculate about the processes described above operating in 
sweatshops and sewing factories. Given that at least some of them must have a choice, 
why do women put up with such terrible working conditions? In addition to sheer 
economic necessity, perhaps it is also partly the camaraderie and social support the 
women receive from their fellow employees; the close friendships they form may 
indeed be enough to reduce turnover. Recent works of fiction, based on real-life 
experiences of women in sewing factories, such as Brick Lane (Ali, 2003) and Four 
Fires (Courtenay, 200 1 )  have documented the lives of these women. Brick Lane, for 
example, gives a poignant account of the experiences of  a woman in India, working in 
exploitative conditions which would be illegal in most countries, but who regards her 
time in the sewing factory as one of the happiest in her life, primarily because of her 
2 1 1 
Synthesis and Implications 
relationship with her "sisters" in the factory. A negative implication might therefore be, 
that an organisation can provide poor work conditions and, up to a point, bad quality 
supervision to a predominantly female staff and, so long as they make friends, they will 
remain in the organisation and turnover will be reduced. 
6.5 Conclusion 
This study sits within the quality of work life movement, focusing both on the fulfilment 
of social needs in the workplace and on related organisational outcomes. Overall, 
findings indicate that improving the quality of work life by creating the opportunity for 
the fulfilment of social needs will have a profound effect, both on individuals and within 
organisations. It seems women, particularly, perceive their friendships at work to be an 
essential aspect of their work life. Individuals working in highly interdependent jobs 
are also likely to be affected by the presence of a friendly work environment. This may 
be solely due to aspects of the job, or may be because (related to Schneider' s ( 1 987) 
attraction/selection/attrition model), individuals who are more relationship focused may 
select more interdependent work roles, and those who do not have a relationship focus 
will be likely to be the first to leave a very interdependent job. Finally, although 
specific 'workplace interventions' are unlikely to increase the number of friends people 
actually make, management should perhaps not be overly concerned, as the provision of 
a generally friendly workplace may be what improves organisational commitment and 
job satisfaction. 
2 1 2  
References 
References 
Adams, J. S. ( 1 963). Towards an understanding of inequity. Journal of Abnormal & 
Social Psychology, 67(5), 422-436. 
Adams, R. G. ,  & Blieszner, R. ( 1 994). An integrative conceptual framework for 
friendship research. Journal of Social and Personal Relationships, 11 ,  1 63- 1 84. 
Adigun, 1. 0.,  & Stephenson, G. M. ( 1 992). Sources of Job Motivation and Satisfaction 
among British and Nigerian Employees. Journal of Social Psychology, 132(3), 
369-376. 
Ajzen, I . , & Fishbein, M.  ( 1 980) .  Understanding attitudes and predicting social 
behavior. Englewood Cliffs, NJ : Prentice-Hall .  
Albert, S . ,  Ashforth, B .  E. ,  & Dutton, J .  E. (2000). Organizational identity and 
identification: Charting new waters and building new bridges. Academy of 
Management Review, 25( 1 ), 1 3- 1 7. 
Ali, M.  (2003) .  Brick Lane. London: Doubleday. 
Allen, E. ( 1 995). Should you be friends with your boss? Working Woman, 20( 1 1 ), 72-
74. 
Allen, M. W. ( 1 992). Communication and organizational commitment: Perceived 
organizational support as a mediating factor. Communication Quarterly, 40, 
357-367. 
Allen, N. J . ,  & Meyer, J .  P .  ( 1 990). The measurement and antecedents of affective, 
continuance and normative commitment to the organization. Journal of 
Occupational Psychology, 63(1 ), 1 - 1 9 . 
Allen, T. D . ,  McManus, S .  E., & Russell, J. E. A. ( 1 999). Newcomer socialization and 
stress: Formal peer relationships as a source of support. Journal of Vocational 
Behavior, 54(3), 453-470. 
Allen, T. D . ,  & Rush, M. C. ( 1 998). The effects of organizational citizenship behavior 
on performance judgments: A field study and a laboratory experiment. Journal 
of Applied Psychology, 83(2), 247-260. 
2 1 3  
References 
AlIen, T. D., Russell, l. E. A., & Maetzke, S. B .  ( 1 997). Formal peer mentoring -
Factors related to proteges' satisfaction and willingness to mentor others. Group 
& Organization Management, 22(4), 488-507. 
AlIen, T. D., Russell, l.  E. A., & Rush, M. C. ( 1 994). The Effects of Gender and Leave 
of Absence on Attributions for High-Perfonnance, Perceived Organizational 
Commitment, and Allocation of Organizational Rewards. Sex Roles, 3 1  (7 -8), 
443-464. 
Amason, P . ,  & Allen, M. W. ( 1997). Intraorganizational communication, perceived 
organizational support, and gender. Sex Roles, 3 7( 1 1 - 1 2), 955-977. 
Anderson, C. M., & Hunsaker, P. L. ( 1 985). Why there's romancing at the office and 
why it's everybody's business. Personnel, February, 57-63 . 
Anderson, C.  M.,  & Martin, M. M. ( 1 995). Why employees speak to coworkers and 
bosses: Motives, gender, and organizational satisfaction. The Journal of 
Business Communication, 32(3), 249. 
Andrew, A., & Montague, 1. ( 1998). Women's friendship at work. Women 's Studies 
International Forum, 21(4), 355-36 1 .  
Angle, H. L., & Perry, 1 .  L. ( 1 98 1 ). An empirical assessment of organizational 
commitment and organizational effectiveness. Administrative Science Quarterly, 
26( 1 ), 1 - 1 4. 
Arbuckle, 1. L. ( 1 999). Amos 4. 0 [Computer software) . Chicago: Smallwaters. 
Aronson, E. ,  & Cope, V. ( 1 968). My enemy's enemy is my friend. Journal of 
Personality and Social Psychology, 8, 8- 1 2. 
Ashforth, B .  E. ,  & Mael, F. ( 1 989). Social identity theory and the organization. 
Academy of Management Review, 14( 1 ), 20-39. 
Ashton, W. A., & Fuerhrer, A. (1 993). Effects of gender and gender role identification 
of participant and type of social support resource on support seeking. Sex Roles, 
28(7-8), 46 1 -476. 
Aston, L. , & Molassiotis, A. (2003). Supervising and supporting student nurses in 
clinical placements: the peer support initiative. Nurse Education Today, 23(3), 
202-21 0. 
Aukett, R., Ritchie, l.,  & Mill, K. ( 1 988). Gender differences in friendship patterns. Sex 
Roles, 19( 1 /2), 57-66. 
2 1 4  
References 
Barhayim, A., & Berman, G. S. ( 1 992). The dimensions of organizational commitment. 
Journal of Organizational Behavior, 13(4), 379-387. 
Baron, R. A. ( 1 989). Personality and organizational conflict: Effects of the Type A 
behavior pattern and self monitoring. Organizational Behavior and Human 
Decision Processes., 44, 2 8 1 -296. 
Barsade, S. G. (2002). The ripple effects: Emotional contagion and its influence on 
group behavior. Administrative Science Quarterly, 47(4), 644-675 .  
Baxter, L .  ( 1 990). Dialectical contradictions in relationship development. Journal of 
Social and Personal Relationships, 7, 69-88. 
Baxter, L. A. ( 1 988). A dialectical perspective on communication strategies in 
relationship development. In S. Duck, & Hay, D. F., et al (Ed.), Handbook of 
personal relationships: Theory, research and interventions. (pp. 257-273). 
Chichester, England UK.: John Wiley & Sons. 
Beans, B. ( 1 999). Proteges want mentors "sympatico" not paternal. APA Monitor 
On line, 30(1 0) .  
Becker, H. S. ( 1 960). Notes on the concept of commitment. American Journal of 
Sociology, 66, 32-40. 
Bedeian, A. G. (1 995). Workplace envy. Organizational Dynamics, 23(4) . 
Bedeian, A. G., & Wren, D. A. (2001 ). Most influential management books of the 20th 
century. Organizational Dynamics, 29(3) .  
Bentler, P.  M. ( 1 990). Comparative fit  indices in structural models. Psychological 
Bulletin, 107, 238-246. 
Bentler, P. M., & Bonett, D. G. ( 1 980). Significance tests and goodness of fit in the 
analysis of covariance structures. Psychological Bulletin, 88, 588-606. 
Betz, E., L. ( 1 982). Need fulfilment in the career development of women. Journal of 
Vocational Behaviour, 20, 60-6l .  
Betz, E .  L. ( 1 984). 2 tests of Maslow theory of need fulfilment. Journal of Vocational 
Behavior, 24(2), 204-220. 
Blau, G. ( 1 999). Testing the longitudinal impact of work variables and performance 
appraisal satisfaction on subsequent overall job satisfaction. Human Relations, 
52(8). 
2 1 5  
References 
B lau, G. ,  & Boal, K. ( 1 989). Using job involvement and organizational commitment 
interactively to predict turnover. Journal of Management, 15( 1 ), 1 1 5- 1 27. 
Blegen, M. A., & Mueller, C.  W. (1 987). Nurses' job satisfaction: a longitudinal 
analysis. Research in Nursing and Health, 10, 227-237. 
B lieszner, R., & Adams, R. ( 1 992). Adultfriendship. Newbury Park, CA: Sage. 
Bluedorn, A. C. ( 1 982). A unified model of turnover from organizations. Human 
Relations, 35, 1 35- 1 53 .  
Bramel, D., & Friend, R. ( 1 987). The work group and its vicissitudes in social and 
industrial psychology. Journal of Applied Behavioral Science, 23(2), 233-253 .  
Bramson, K .  ( 1 98 1 ) . Coping with difficult people. New York: Ballantine Books. 
Bridge, K., & Baxter, L. A. ( 1 992). Blended relationships :  Friends as work associates. 
Western Journal o/Communication, 56(3), 200-225 .  
Browne, M.  W.,  & Cudeck, R .  ( 1 993). Alternative ways of assessing model fit .  In K. A. 
Bollen & J. S .  Long (Eds.), Testing Structural Equation Models (pp. 445-455). 
Newbury Park, CA: Sage . .  
Bryman, A. ( 1 989). Research methods and organization studies ( 1  st ed. Vol. 20). 
London: Unwin Hyman Ltd. 
Bui, K. T., Peplau, L. A., & Hill, C. T. ( 1 996). Testing the Rusbult model of 
relationship commitment and stability in a 1 5-year study of heterosexual 
couples. Personality and Social Psychology Bulletin, 22, 1 244- 1 257.  
Bukowski, W. M.,  Nappi, B .  J . ,  & Hoza, B. ( 1 987). A test of Aristotle's model of 
friendship for young adults' same-sex and opposite-sex relationships. Journal 0/ 
Social Psychology, 127(6), 595-603 . 
Buunk, B. P . ,  Doosje, B .  J . ,  Liesbeth, G., Jans, J. M.,  & Hopstaken, L. E. M. ( 1 993) .  
Perceiyed reciprocity, social support, and stress at  work: The role of exchange 
and communal orientation. Journal 0/ Personality and Social Psychology(65), 
801 -8 1 l .  
Buunk, B. P. ,  & Prins, K. S .  ( 1 998). Loneliness, exchange orientation, and reciprocity in 
friendships. Personal Relationships, 5(1),  1 - 1 4. 
Byrne, B .  M.  ( 1 994). Burnout: Testing for the validity, replication, and invariance of 
causal structure across elementary, intermediate and secondary teachers. 
American Educational Research Journal, 31(3), 645-673 .  
2 16 
References 
Byme, B .  M. (2001) .  Structural equation modeling with AMOS. Basic concepts, 
applications and programming. New Jersey: Lawrence Erlbaum Associates, Inc. 
Campbell-Clark, S. (2001 ). Work cultures and worklfamily balance. Journal 0/ 
Vocational Behavior, 58(3), 348-365. 
Campion, M. A., Medsker, G. J . ,  & Higgs, A. C. ( 1 993). Relations between work group 
characteristics and effectiveness: Implications for designing effective work 
groups. Personnel Psychology, 46(4), 823-850. 
Carr, S .  C.  (2003). Social psychology. Context, communication and culture. Milton, 
Qld: John Wiley & Sons Australia. 
Chang, E. ( 1 999). Career commitment as a complex moderator of organizational 
commitment and turnover intention. Human Relations, 52(1 0), 1 257- 1 277. 
Chapman, J. ( 1 993). Collegial support linked to reduction of job stress. Nursing 
Management, 24(5), 52-56. 
Chen, R. (2003). An SAS/IML procedure for maximum likelihood factor analysis. 
Behavior research methods instruments computers, 35(2), 3 1 0-3 1 7. 
Cherniss, C. ( 1 99 1 ). Career commitment in human service professionals. A biographical 
study. Human Relations, 44(5), 41 9-437.  
Clark, M.  S. ,  & Mills, J.  ( 1 979). Interpersonal attraction in exchange and communal 
realtionships. Journal of Personality and Social Psychology, 3 7, 12-23 . 
Cochrane, D. K., & Jowett, S. A. ( 1 994). Preparing for trust status investigation into the 
morale of nurses undergoing change. Journal of Nursing Management, 2, 223-
227. 
Code of Ethics for Psychologists Working in Aotearoa/New Zealand. (2002). Prepared 
by the Code of Ethics Review Group, a joint working party of the NZ 
Psychological Society, the NZ College of Clinical Psychologists and the NZ 
Psychologists Board. 
Cohen, A. ( 1 993). Organizational commitment and turnover: A metaanalysis. Academy 
o/Management Journal, 36(5), 1 1 40- 1 1 57. 
Cohen, A. ( 1 996). On the discriminant validity ofthe Meyer and Allen measure of 
organizational commitment: How does it fit with the work commitment 
construct? Educational and Psychological Measurement, 56(3), 494-503 . 
2 1 7  
References 
Cohen, A. ( 1 999). Relationships among five forms of commitment: an empirical 
assessment. Journal o/Organizational Behavior, 20(3), 285-308. 
Cohen, A. ,  & Hudecek, N. ( 1 993). Organizational commitment/turnover relationship 
across occupational Groups. A metaanalysis. Group & Organization 
Management, 18(2), 1 88-2 1 3 .  
CoIling, K., Grabo, T., Rowe, M., & Straneva, J .  ( 1 998). How to develop and sustain a 
peer-mentored research work group. Journal 0/ Professional Nursing, 14(5), 
298-304. 
Corner, D. R. ( 1 99 1 ). Organizational newcomers' acquisition of information from peers. 
Management Communication Quarterly, 5(1 ), 64-89. 
Courtenay, 8. (2001 ). Four fires. Ringwood, Vic: Viking. 
Crant, J .  M.,  & Bateman, T. S. ( 1 993). Assignment of Credit and Blame for 
Performance Outcomes. Academy 0/ Management Journal, 36( 1 ), 7-27. 
Cronbach, L., & Meehl, P. ( 1 955). Construct validity in psychological tests. 
Psychological Bulletin, 52(4), 28 1 -302. 
Cudeck, R., & O'Dell, L. L. ( 1 994). Applications of Standard Error Estimates in 
Unrestricted Factor Analysis: Significance Tests for Factor Loadings and 
Correlations. Psychological Bulletin, 1 15(3) .  
Cunningham, J. 8. ,  & MacGregor, J .  (2000). Trust and the design of work: 
Complementary constructs in satisfaction and performance. Human Relations, 
53(12), 1 575-1 591 .  
Dillard, J .  P . ,  & Fritz, J .  ( 1 995). Interpersonal relationships and organizational 
performance. Paper presented at the Annual Convention of the Western Speech 
Communication Association., Portland. OR. 
Dreher, G. F., & Mai-Dalton, R. R. ( 1 983). A note on the internal consistency of the 
Manifest Needs Questionnaire. Journal 0/ Applied Psychology, 68, 1 94- 1 96.  
Duck, S . ,  & Perlman, D.  ( 1 985). Understanding personal relationships: An 
interdisiplinary approach. London: Sage. 
Duck, S . ,  & Wright, P. H. ( 1 993). Reexamining gender differences in same-gender 
friendships: A close look at two kinds of data. Sex Roles, 28( 1 1 1 1 2), 709-727. 
2 1 8  
References 
Dumelow, c., Littlejohns, P., & Griffiths, S. (2000). Relation between a career and 
family life for English hospital consultants : qualitative, semistructured interview 
study. British Medical Journal, 320(7247), 1 437- 1440. 
Eisenberg, E. M. ,  & Goodall, H. L., Jr. ( 1 997). Organizational communication: 
Balancing creativity and constraint (2nd ed.) . New York: St. Martin's Press. 
Eisenberg, S. A. J. ( 1 994). Friendship or fraternization? Credit Union Management, 
1 7(7), 22-23 . 
Ellingwood, S .  (200 1) .  The collective advantage. Gallup Management Journal (Vol. 1 ,  
Issue 3). Retrieved Feb, 2002, from the World Wide Web: 
http://www.gallupjoumal.comlGMJarchive/issue31200 19 1 5c.asp 
Ensher, E. A., & Murphy, S. E. ( 1 997). Effects of race, gender, perceived similarity, and 
contact on mentor relationships. Journal o/Vocational Behavior, 50(3), 460-
48 1 .  
Ensher, E. A.,  Thomas, C. ,  & Murphy, S .  E. (200 1 ) .  Comparison oftraditional, step­
ahead, and peer mentoring on proteges' support, satisfaction, and perceptions of 
career success: A social exchange perspective. Journal 0/ Business and 
Psychology, 15(3), 419-438.  
Ferris, K. R., & Aranya, N. (1 983). A comparison oftwo organizational commitment 
scales. Personnel Psychology, 36( 1 ), 87-98. 
Fine, G. ( 1 986). Friendships in the workplace. In V. J. Derlega, & Winstead, B .  A. 
(Ed.), Friendship and Social Interaction (pp. 1 85-206). New York: Springer­
Verlag. 
Fisher, C. D.  (2002). Antecedents and consequences of real-time affective reactions at 
work. Motivation and Emotion, 26( 1 ), 3 -30. 
Foote, N. N. ( 1 985). The/riendship game. New York: Image Books. 
Fritz, J. H. ( 1 997). Men's and women's organizational peer relationships: A comparison. 
The Journal o/Business Communication, 34(1) , 27-46. 
Froebe, D., Deets, C. ,  & Knox, S. ( 1 983) .  What motivates nurses to to join and remain 
with an organization. Nursing Leadership, 6, 22-33 .  
Froman, R.  D.  (2001 ). Elements to consider in planning the use of factor analysis. 
Southern Online Journal o/Nursing Research, 2(5). 
2 1 9  
References 
Furnham, A., Forde, L., & Ferrari, K. ( 1 999). Personality and work motivation. 
Personality and Individual Differences, 26(6), 1 035- 1 043 . 
Gaertner, S. (1 999). Structural determinants of job satisfaction and organizational 
commitment in- turnover models. Human Resource Management Review, 9(4), 
479-493 . 
Gant, L. M., Nagda, B. A., Brabson, H. V.,  Jayaratne, J .  S . ,  Chess, W.  A. ,  & Singh, A. 
(1 993). Effects of social support and undermining on African American workers' 
perceptions of coworker and supervisor relationships and psychological 
wellbeing. Social Work, 38, 1 58- 1 64. 
Geiger, M. A., & Cooper, E. A. ( 1995). Predicting academic performance: The impact 
of expectancy and needs theory. Journal of Experimental Education, 63, 25 1 -
262. 
Gilbert, N. M. ,  & Sneed, J. ( 1 992). Organizational culture. Does it affect employee and 
organizational outcomes. Journal of the Canadian Dietetic Association-Revue 
De L Association Canadienne Des Dietetistes, 53(2), 1 55 - 1 58 .  
Grove, A .  S .  ( 1 993). Getting past personality conflicts. Working Woman, 18, 34-35 .  
Hackett, R .  D . ,  & Lapierre, L .  M.  (200 1 ). Understanding the links between work 
commitment constructs . Journal of Vocational Behavior, 58(3), 3 92-4 1 3 .  
Hackman, J .  R. ,  & Lawler, E .  E. ( 197 1 ) . Employee reactions to job characteristics. 
Journal of Applied Psychology, 55, 259-286. 
Hamilton, V. L. ( 1 980). Intuitive P$ychologist or Intuitive Lawyer - Alternative Models 
of the Attribution Process. Journal of Personality and Social Psychology, 39(5),  
. 767-772. 
Hamilton, V. L. ,  & Sanders, J .  (1 992). Responsibility and Risk in Organizational 
Crimes of Obedience. Research in Organizational Behavior, 1 4, 49-90. 
Haslam, S. A., Postmes, T., & Ellemers, N. (2003) .  More than a metaphor: 
Organizational identity makes organizational life possible. British Journal of 
Management, 14(4), 3 57-369. 
Hayes, J .  L. ( 1 982). Management by involvement. Credit & Financial Management, 
84(4), 43-44. 
Heckert, T. M. ,  Cuneio, G., Hannah, A. P . ,  Adams, P. J . ,  Droste, H. E . ,  Mueller, M. A.,  
Wallis, H. A., Griffin, C. M., & Roberts, L. L. (2000). Creation of a new needs 
220 
References 
assessment questionnaire. Journal of Social Behavior and Personality, 15( 1 ), 
1 2 1 - 1 36. 
Herzberg, F .  ( 1 966). Work and the nature of man. Cleveland: World. 
Herzberg, F . ,  Mausner;B. ,  & Snyderman, B. ( 1 959). The motivation to work. New 
York: Wiley. 
Higgins, M. C. ,  & Kram, K. E. (200 1 ). Reconceptualizing mentoring at work: A 
developmental network perspective. Academy of Management Review, 26(2), 
264-288.  
Hulin, C. ,  L., & Smith, P. ,  A. ( 1 967). An empirical investigation of two implications of 
the two-factor theory of job satisfaction. Journal of Applied Psychology, 51(5), 
396-402. 
Ibarra, H. ( 1 993). Personal networks of women and minorities in management: A 
conceptual framework. Academy of Management Review, 18( 1 ), 56-87. 
Impett, E .  A. ,  Beals, K. P. ,  & Peplau, L. A. (200 1 ) . Testing the investment model of 
relationship commitment and stability in a longitudinal study of married couples. 
Current Psychology, 20(4), 3 1 2-326. 
Inkson, K. , & Arthur, M. B. (2001) .  How to be a successful career capitalist. 
Organizational Dynamics, 30( 1 ), 48-61 . 
Irvine, D .  M. ,  & Evans, M. G. ( 1 995). Job satisfaction and turnover among nurses: 
Integrating research findings across studies. Nursing Research, 44, 246-253 .  
Irving, P .  G. ,  & Meyer, J .  P .  ( 1 994). Reexamination of the met expectations hypothesis. 
A longitudinal analysis. Journal of Applied Psychology, 79(6), 937-949. 
Jacelon, C. S . ,  Zucker, D. M., Staccarini, J.  M., & Henneman, E. A. (2003). Peer 
mentoring for tenure-track faculty. Journal of Professional Nursing, 19(6), 335-
3 3 8. 
Jehn, K. A. ,  & Shah, P. P .  ( 1 997). Interpersonal relationships and task performance: An 
examination of mediating processes in friendship and acquaintance groups.  
Journal of Personality and Social Psychology, 72(4), 775-790. 
Johnson, D. (2001) .  Investing in human relations. FJ,lturist, 35(4), 9-1 1 .  
Johnson, D .  E. ,  Erez, A., Kiker, D. S . ,  & Motowidlo, S .  J.  (2002). Liking and 
attributions of motives as mediators of the relationships between individuals' 
22 1 
References 
reputations, helpful behaviors, and raters' reward decisions. Journal of Applied 
Psychology, 87(4) ,  808-8 1 5 . 
Kaldenberg, D. 0.,  Becker, B. W., & Zvonkovic, A. ( 1 995). Work and commitment 
among young professionals. A study of male and female dentists. Human 
Relations, 48( 1 1 ), 1 3 55-1 377. 
Kalliath, T. J., & Beck, A. (2001 ). Is the path to burnout and turnover paved by a lack of 
supervisory support? A structural equations test. New Zealand Journal of 
Psychology, 30(2), 72-78 . 
Kamante, J. (2002). Personal communication, 8th August. Auckland. 
Kao, H. S . ,  & Sek-Hong, N. ( 1 993). Organisational commitment: From trust to altruism 
at work. Psychology and Developing Societies, 5( 1 ), 43-60. 
Kenny, D. A., & Kashy, D. A. ( 1 994). Enhanced co-orientation in the perceptions of 
friends: A social relations analysis. Journal of Personality and Social 
Psychology, 67(6), 1 024- 1033 .  
King, N. ( 1 970). Clarification and evaluation of the two factor theory of  job satisfaction. 
Psychological Bulletin, 74( 1 ), 1 8-3 1 .  
Knoop, R. ( 1 994). Work Values and Job-Satisfaction. Journal of Psychology, 128(6), 
683-690. 
Krackhardt, D. ,  & Stem, R. N. (1 988). Informal networks and organizational crises: An 
experimental simulation. Social Psychology Quarterly, 51(2), 1 23 - 140.  
Kram, K., & Isabella, L. ( 1 985). Mentoring alternatives: The role of peer relationships 
in career development. Academy a/Management Journal, 28(1 ), 1 1 0- 1 32 .  
Kram, K.  E. ( 1 983). Phases of the mentor relationship. Academy of Management 
Journal, 26(4), 608-625. 
Kramer, M. W. ( 1 996). A longitudinal study of peer communication during job 
transfers. The impact of frequency, quality, and network mUltiplexity on 
adjustment. Human Communication Research, 23(1 ), 59-86. 
Lambert, S .  J .  ( 1 990). Processes linking work and family: A critical review and research 
agenda. Human Relations, 43(3), 239-257. 
Larnm, H.,  & Wiesmann, U. ( 1 997). Subjective attributes of attraction: How people 
characterize their liking, their love, and their being in love. Personal 
Relationships, 4(3), 27 1 -284. 
222 
References 
Largent, R. N. ( 1 987). The relationship of friendship with a sLlpervisor to job 
satisfaction and satisfaction with the supervisor. Unpublished Masters Thesis, 
University of North Dakota, Grand Forks, North Dakota. 
LaRocco, J. M.,  & Jones, A. P .  ( 1 978). Co-worker and leader support and moderators of 
stress-strain relationships in work situations. Journal of Applied Psychology, 63, 
629-634. 
Lawley, D. N. ( 1 940). The estimation of factor loadings by the method of maximum 
likelihood. Proceedings of the Royal Society of Edinburgh, 60, 64-82. 
Lee, K., Allen, N. J . ,  & Meyer, J. P.  (2001) .  The three component model of 
organisational commitment: An application to South Korea. Applied 
Psychology-An International Review-Psychologie Appliquee-Revue 
Internationale, 50(4), 596-6 14.  
Lee, K.,  Carswell, J .  J . ,  & AlIen, N. J .  (2000). A metaanalytic review of occupational 
commitment: Relations with person and work related variables. Journal of 
Applied Psychology, 85(5), 799-8 1 1 .  
Lee, S. M. ( 1 97 1 ). An empirical analysis of organizational identification. Academy of 
Management Journal, 14(2), 2 1 3-226. 
Levy, P. L. (2003). Industrial / organizational psychology. Understanding rhe 
workplace. Boston: Houghton Mifflin Company. 
Lord, R. G., & Smith, J. E. ( 1 983). Theoretical, Information Processing, and Situational 
Factors Affecting Attribution Theory Models of Organizational Behavior. 
Academy of Management Review, 8(1 ), 50-60. 
Loscocco, K. A., & Spitze, G. ( 1 990). Working conditions, social support, and the well­
being of female and male factory workers. The Journal of Health and Social 
Behavior, 31(4), 3 1 3-328. 
Louis, M.  R., Posner, B. Z., & Powell, G. N. ( 1 983). The avaliability and helpfulness of 
socialization practices. Personnel Psychology, 36, 857-866. 
Lozada, M. ( 1 996). Social misfits, workplace outcasts. Vocational Educational Journal, 
71 . 
Lu, K. Y., Lin, P .  L., Wu, C. M.,  Hsieh, Y. L., & Chang, Y. Y. (2002). The relationships 
among turnover intentions, professional commitment, and job satisfaction of 
hospital nurses. Journal of Professional Nursing, 18(4), 2 14-2 19 .  
223 
References 
Mael, F. A., & Tetrick, L. E. ( 1 992). Identifying Organizational Identification. 
Educational and Psychological Measurement, 52(4), 8 1 3-824. 
Maidani, E. A. ( 1 99 1 ). Comparative-Study of Herzberg 2-Factor Theory of Job­
Satisfaction among Public and Private Sectors. Public Personnel Management, 
20(4), 441 -448. 
Marelich, W. D. ( 1 996). Can we be friends? Hr Focus, 73(8), 1 7- 1 8 . 
Markiewicz, D. ,  Devine, 1 . ,  & Kausilas, D. (2000). Friendships of women and men at 
work: Job satisfaction and resource implications. Journal of Managerial 
Psychology, 15( 1 -2), England: MCB Univ Press Ltd. 
Maslow; A. H. ( 1 954). Motivation and personality. New York: Harper & Row. 
Mathews, K. A. ( 1 982). Psychological perspectives on the type A behavior pattern. 
Psychological Bulletin, 91, 293-323. 
Mathieu, J .  E., & Zajac, D.  M. ( 1 990). A review and meta-analysis of the antecedents, 
correlates, and consequences of organizational commitment. Psychological 
Bulletin, 1 08(2). 
Mayes, B. T., & Ganster, D. C. ( 1 983). A multitrait-multimethod matrix analysis ofthe 
PRF and MNQ scales. Journal o/Management, 9, 1 1 3- 1 26. 
Mayo, E. ( 1 945). The social problems of industrial civilisation. Harvard: Harvard 
University Press. 
McClelland, D. C. ( 1 96 1 ) .  The achieving society. Princeton, NJ: Van Nostrand. 
McDougall, M. ,  & Beattie, R. S. ( 1 997). Peer mentoring at work - The nature and 
outcomes of non-hierarchical developmental relationships. Management 
Learning, 28(4), 423-437. 
McEvoy, G. M. ,  & Cascio, W. F. ( 1 987). Do good or poor performers leave. A 
metaanalysis of the relationship between performance and turnover. Academy 0/ 
Management Journal, 30(4), 744-762. 
Medcof, J. W. ,  & Hausdorf, P. A. ( 1 995). Instruments to measure opportunities to 
satisfy needs, and degree of satisfaction of needs, in the workplace. Journal 0/ 
Occupational & Organizational Psychology, 68(3), 1 93 .  
Meyer, J .  P . ,  & Allen, N. 1 .  ( 1 991) .  A three component conceptualization of  
organizational commitment. Human Resource Management Review, 1 ( 1 ), 6 1 -89. 
224 
References 
Meyer, J. P . ,  & AlIen, N. 1. (1 997). Commitment in the workplace : theory, research, 
and application. Thousand Oaks, California: Sage Publications. 
Meyer, J. P� ,  AlIen, N. J . ,  & Smith, C. A. ( 1 993). COrnlnitment to organizations and 
occupations: Extension and test of a three component conceptualization. Journal 
of Applied Psychology, 78(4). 
Meyer, J. P., Stanley, D.  J. ,  Herscovitch, L., & Topolnytsky, L. (2002). Affective, 
continuance, and normative commitment to the organization: A metaanalysis of 
antecedents, correlates, and consequences. Journal afVocational Behavior, 
61 ( 1 ) , 20-52. 
Michaels, C. E., & Spector, P.  E. (1 982). Causes of employee turnover: A test of the 
Mobley, Griffeth, Hand and Meglino model . Journal of Applied Psychology, 67, 
53-59. 
Middleton, J. (2002, Feb. 9). Profiting from friendships. New Zealand Herald, pp. C l .  
Mitchel, J .  O. ( 198 1 ). The effect o f  intentions, tenure, personal, and organizational 
variables on managerial turnover. Academy of Management Journal, 24(4), 742-
75 1 .  
Mobley, W. H. (1 977). Intermediate linkages in the relationship between job satisfaction 
and employee turnover. Journal of Applied Psychology(62), 237-240. 
Mobley, W. H., Homer, S. 0., & Hollingsworth, A. T. (1 978). An evaluation of 
precursors of hospital employee turnover. Journal of Applied Psychology, 63, 
408-4 14. 
Monroe, C., Borzi, M.  G.,  & DiSalvo, V. S. ( 1 992). Managerial strategies for dealing 
with difficult subordinates. Southern Communication Journal, 58, 247-254. 
Morrison, E. W. (1 993). Longitudinal study of the effects of information seeking on 
newcomer socialization. Journal of Applied Psychology, 78(2), 1 73- 1 83 .  
Morrow, P .  C. ,  & McElroy, J. C .  ( 1 986). On assessing measures of work commitment. 
Journal of Occupational Behaviour, 7(2) , 1 39- 145 .  
Mowday, R .  T . ,  Steers, R .  M. ,  & Porter, L .  W.  ( 1 979). The measurement of 
organizational commitment. Journal of Vocational Behavior, 14, 224-227. 
Mulaik, S. A., James, L. R., Van Alstine, J . ,  Bennet, N. ,  Lind, S . ,  & Stilwell, C. D. 
( 1 989). Evaluation of goodness of fit indices for structural equation models. 
Psychological Bulletin, 1 05(430-445). 
225 
References 
Murphy, S. E. ,  & Ensher, E. A. ( 1 999). The effects of leader and subordinate 
characteristics in the development of leader-member exchange quality. Journal 
of Applied Social Psychology, 29(7), 1 37 1 - 1 394. 
Murstein, B. ,  & Spitz, L ( 1 974). Aristotle and friendship: A factor analytic study. 
Interpersonal Development, 4, 2 1 -34. 
Myers, S .  A., Knox, R. L., Pawlowski, D. R., & Ropog, B. L. ( 1 999). Perceived 
communication openness and functional communication skills among 
organizational peers. Communication Reports, 12(2), 7 1 -8 1 . 
Neubert, M. J .  ( 1 999). Too much of a good thing or the more the merrier? Small Group 
Research, 30(5), 635-647 . 
Nielsen, I. K.,  Jex, S. M., & Adams, O. A. (2000). Development and validation of 
scores on a two dimensional Workplace Friendship Scale. Educational & 
Psychological Measurement, 60(4), 628-643 . 
Odden, C. M. ,  & Sias, P. M. ( 1 997). Peer communication relationships, psychological 
climate, and gender. Communication Quarterly, 45, 1 53- 1 66. 
Parkes, L. P . ,  Bochner, S. ,  & Schneider, S .  K. (200 1) .  Person-organisation fit across 
cultures: An empirical investigation of individualism and collectivism. Applied 
Psychology, 50( 1 ), 8 1 - 109. 
Pedhazur, E. J. ( 1 982). Multiple regression in behavioral research (2nd ed. ) .  New 
York: Holt, Rinehart & Winston. 
Pettigrew, T. F. ( 1 979). The ultimate attribution error: Extending Allport's cognitive 
analysis of prejudice. Personality & Social Psychology Bulletin, 5(4), 46 1 -476. 
Pifczyk, A., & Kleinbeck, U. (2000). The influence of achievement and affiliation 
variables on work motivation and work satisfaction in a social work 
environment. ZeitschriJt Fur Arbeits-Und Organisationspsychologie, 44(2), 57-
68. 
Porter, L. W. ,  Steers, R. M. ,  Mowday, R. T., & Boulian, P .  V. ( 1 974). Organizational 
commitment, job satisfaction, and turnover among psychiatric technicians. 
Journal of Applied Psychology, 59, 603-609. 
Quinn, R. E. ( 1 977). Coping with cupid: The formation and impact of romantic 
relationships in organizations. Administrative Science Quarterly, 22( 1 ), 30-45. 
226 
References 
Raabe, B. ,  & Beehr, T. A. (2003). Formal mentoring versus supervisor and coworker 
relationships: differences in perceptions and impact. Journal of Organizational 
Behavior, 24(3), 27 1 -293 . 
Rambur, B . ,  Palumbo, 'M. V., McIntosh, B . ,  & Mongeon, J. (2003) .  A statewide analysis 
ofRNs' intention to leave their position. Nursing Outlook, 51(4), 1 82- 1 88 .  
Randall, D. M.  ( 1 993). Cross-cultural research on organizational commitment: A review 
and application ofHofstede's value survey module. Journal of Business 
Research, 26( 1 ), 9 1 - 1 1 0. 
Reichers, A. E .  ( 1 987). An interactionist perspective on newcomer socialization rates .  
Academy of Management Review, 12(2), 278-287. 
Reise, S. P . ,  WaIler, N. G., & Comrey, A.  L. (2000). Factor analysis and scale revision. 
Psychological Assessment, 12(3), 287-297. 
Rentsch, J. R., & Steel, R. P. ( 1 992) . Construct and concurrent validation of the 
Andrews and Withey job satisfaction questionnaire. Educational & 
Psychological Measurement, 52(2), 357-367. 
Richer, S .  F., Blanchard, U., & Vallerand, R. J .  (2002). A motivational model of work 
turnover. Journal of Applied Social Psychology, 32(1 0), 2089-2 1 1 3 .  
Ringer, T .  M. ,  & Robinson, P .  ( 1 996). Focus and strategic action in management: using 
a systemic model of organisational culture to inform managerial actions, Work 
Study, 45(6), 5 - 1 6 . 
Riordan, C.  M. ,  & Griffeth, R. W. ( 1 995). The opportunity for friendship in the 
workplace: An underexplored construct. Journal of Business & Psychology, 
10(2), 1 4 1 - 1 54. 
Rook, K. S. ( 1 984). The negative side of social interaction: Impact on psychological 
well being. Journal of Personality and Social Psychology, 46, 1 097- 1 1 08 .  
Ross, J .  A. ( 1 997). Human resources: Does friendship improve job performance? 
Harvard Business Review, 75(2), 8-9. 
Rubin, R. B . ,  Perse, E. M. ,  & Barbato, C. A. ( 1 988). Conceptualization and 
measurement of interpersonal communication motives. Human Communication 
Research, 14(4), 602-628 .  
Rubin, z. ( 1 970) .  Measurement of  romantic love. Journal of Personality and Social 
Psychology, 16, 265-273.  
227 
References 
Rusbult, C. E. ( 1 980). Commitment and satisfaction in romantic associations: A test of 
the investment model . Journal of Experimental Social Psychology, 1 6, 1 72- 1 86 .  
Sager, E. K. , Griffeth, R. W. ,  & Horn, P .  W. ( 1 998). A comparison of  structural models 
representing turnover cognitions. Journal o/Vocational Behavior, 53(2), 254-
273 . 
Sanders, J. ,  Hamilton, V. L., & Yuasa, T. ( 1 998). The institutionalization of sanctions 
for wrongdoing inside organizations: Public judgments in Japan, Russia, and the 
United States. Law & Society Review, 32(4), 87 1 -929. 
Sapadin, L. A. ( 1 988).  Friendship and gender: Perspectives of professional men and 
women. Journal o/Social and Personal Relationships, 5, 387-403 . 
Schneider, B .  ( 1 987). The people make the place. Personnel Psychology, 40, 437-453 .  
Schneider, B . ,  Goldstein, H .  W., & Smith, D .  B.  ( 1 995). The ASA framework: An 
update. Personnel Psychology, 48(4), 747-773. 
Schneider, B., & Reichers, A. E. ( 1 983). On the etiology of climates. Personnel 
Psychology, 36( 1 ), 1 9-39. 
Schultz, V. (2003). The sanitized Workplace. Yale Law Journal, 1 12(8), 206 1 . 
Sheldon, M.  E. ( 1 97 1 ). Investments and involvements as mechanisms producing 
commitment to the organization. Administrative Science Quarterly, 1 6(2), 1 43-
1 50.  
Shouksmith, G. ( 1 994). Variables related to organizational commitment in health­
professionals. Psychological Reports, 73(3), 707 -7 1 l .  
Sias, P .  M. ,  & Cahil l ,  D .  J .  ( 1 998). From coworkers to friends: The development o f  peer 
friendships in the workplace. Western Journal o/Communication, 62(3), US: 
Western States Communication Assoc. 
Sias, P.  M. ,  & Jablin, F. M. ( 1 995). Differential superior / subordinate relations, 
perceptions of fairness, and coworker communication. Human Communication 
Research, 22(5-38). 
Siu, O. L. (2002). Occupational stressors and well-being among Chinese employees: 
The role of organisational commitment. Applied Psychology-An International 
Review-Psychologie Appliquee-Revue Internationale, 51(4), 527-544. 
Steers, R. M. ( 1 977). Antecedents and outcomes of organizational commitment. 
Administrative Science Quarterly, 22( 1 ) . 
228 
References 
Steers, R. M., & Braunstein, D. N. ( 1 976). A behaviorally based measure of manifest 
needs in work settings. Journal of Vocational Behavior, 9, 25 1 -266. 
Stumpf, S. A., & Hartman, K. ( 1 984). Individual exploration to organizational 
commitment or-withdrawal. Academy of Management Journal, 2 7(2), 308-329. 
Sumer, H. C., & Knight, P.  A. (2001 ) . How do people with different attachment styles 
balance work and family? A personality perspective on work-family linkage. 
Journal of Applied Psychology, 86(4), 653-663 . 
Tao, M., Takagi, H., Ishida, M., & Masuda, K. ( 1 998). A study of antecedents of 
organizational commitment. Japanese Psychological Research, 40(4), 1 98-205 . 
Tett, R. P., & Meyer, J. P. ( 1993). Job satisfaction, organizational commitment, turnover 
intention, and turnover: Path analyses based on metaanalytic findings. Personnel 
Psychology, 46(2), 259-293.  
Tovey, E. J . ,  & Adams, A.  E .  (1 999). The changing nature of nurses' job satisfaction: an 
exploration of sources of satisfaction in the 1 990s. Journal of Advanced Nursing, 
30( 1 ), 1 50- 1 58. 
Tumulty, G., Jernigan, I. E., & Kohut, G. F. (1 995). The Impact of Perceived Work­
Environment on Job-Satisfaction of Hospital Staff Nurses. Applied Nursing 
Research, 7(2), 84-90. 
van Knippenberg, D.,  & Hogg, M. A. (2001 ). Social identity processes in organisations. 
Group Processes and Intergroup Relations, 4(3), Special Issue. 
van Knippenberg, D., & van Schie, E. C. M. (2000). Foci and correlates of 
organizational identification. Journal of Occupational and Organizational 
Psychology, 73(2), 1 37-1 47. 
Vandenberghe, C. ( 1 999) . Organizational culture, person-culture fit, and turnover: a 
replication in the health care industry. Journal of Organizational Behavior, 
20(2), 1 75- 1 84. 
Vaughan, G. M., & Hogg, M. A. ( 1 998). Introduction to social psychology (Second ed.). 
Sydney: Prentice Hall. 
Vecchio, R. P . ,  & Bullis, R. C. (200 1 ). Moderators of the influence of supervisor / 
subordinate similarity on subordinate outcomes. Journal of Applied Psychology, 
86(5), 884-896. 
229 
References 
Veniegas, R. C., & Peplau, L. A. (1997). Power and the quality of same-sex friendships. 
Psychology of Women Quarterly, 21(2), 279-297. 
Vinokur, A. D., Schul, Y., & Caplan, R. D. (1987). Detenninants of perceived social 
support: Interpersonal transactions, personal outlook, and transient affect states. 
Journal of Personality and Social Psychology, 53, 1 1 37-1145 .  
Vroom, V. H .  ( 1 964). Work and motivation. New York, NY: lohn Wiley & Sons. 
Vuorinen, R., Tarkka, M. T., & Meretoja, R. (2000). Peer evaluation in nurses' 
professional development: a pilot study to investigate the issues. Journal of 
Clinical Nursing, 9(2), 273-281. 
Wahba, M. A., & Bridwell, L. G. (1976). Maslow reconsidered - review of research on 
need hierarchy theory. Organizational behavior and human performance, 15(2), 
212-240. 
WaIT, P. B. ,  Cook, 1. ,  & Wall, T. D. (1979). Scales for the measurement of some work 
attitudes and aspects of psychological well-being. Journal of Occupational 
Psychology, 52, 129- 1 48 .  
Wesolowski, M. A. ,  & Mossholder, K.  W. (1997). Relational demography in 
supervisor-subordinate dyads: Impact on subordinate job satisfaction, burnout, 
and perceived procedural justice. Journal of Organizational Behavior, 18(4), 
35 1 -362. 
Whitley, B. E. (2002). Principles of research in behavioral science. Second edition. 
Boston: McGraw Hill. 
Williams, A. W., Ware, 1. E., & Donald, C. A. ( 198 1 ). A model of mental health, life 
events and social support applicable to general populations. Journal of Health 
and Social Behavior, 22, 324-336. 
Williams, L. 1.,  & Hazer, 1. T. ( 1986) . Antecedents and consequences of satisfaction and 
commitment in turnover models: A reanalysis using latent variable structural 
equation methods. Journal of Applied Psychology, 71 , 2 1 9-23 1 .  
Winstead, B. A. (1 986). Sex differences in same sex friendships. In 1. Derlelga, & 
Winstead, B. A. (Ed.), Friendship and Social Interaction. (pp. 8 1 -99). New 
York: Springer-Verlag. 
230 
References 
Winstead, B .  A., Derlega, V. J . ,  Montgomery, M. J. ,  & Pilkington, C. ( 1 995). The 
quality of friendship at work and job satisfaction. The Journal of Social and 
Personal Relationships, 12(2), 1 99-2 1 5 . 
Wood, J. T. ,  & Inrnan, C .  C. ( 1 993). In a different mode: Masculine styles of 
communicating closeness. Journal of Applied Communication Research, 21, 
279-295. 
Wright, P .  H. ( 1 969). A model and a technique for studies of friendship. Journal of 
Experimental Social Psychology, 5, 295-309. 
Wright, P. H. ( 1 974). The delineation and measurement of some key variables in the 
study of friendship. Representative Research in Social Psychology, 5, 423-436. 
Wright, P .  H. ( 1 984). Self-referent motivation and the intrinsic quality of friendship. 
Journal of Social and Personal Relationships, 1 ( 1 1 5- 1 30) .  
Wright, P .  H. ( 1 988). Interpreting research on gender differences in friendship: A case 
for moderation and a plea for caution. Journal of Social and Personal 
Relationships, 5, 367-373. 
Wright, P. H., & Scanlon, M. B .  ( 1 99 1 ) . Gender role orientations and friendship: Some 
attenuation, but gender differences abound. Sex Roles, 1 4, 55 1 -566. 
Yamashita, M. ( 1 995). Job-Satisfaction in Japanese Nurses. Journal of Advanced 
Nursing, 22( 1 ), 1 58- 1 64. 
Yousef, D. A. (2003). Validating the dimensionality of Porter et al . 's measurement of 
organizational commitment in a non-Western culture setting. International 
Journal of Human Resource Management, 14(6), 1 067- 1 079. 
Zajonc, R. B. (2001 ). Mere exposure: A gateway to the subliminal. Current Directions 
in Psychological Science, 10(6), 224-228. 
Zinovieva, 1. L. ,  ten Horn, L.  A., & Roe, R. A. ( 1 993) .  Work motivation in "post­
socialist" industrial organizations. European Work and Organizational 
Psychologist, 3(3), 25 1 -264. 
Zorn, T. ( 1 995). Bosses and buddies: Constructing and performing simultaneously 
hierarchical and close friendship relationships. In J. T. Wood & S. Duck (Eds.), 
Understudied relationships. Off the beaten track. (pp. 1 23 - 147). London: Sage. 
23 1 
Appendices 
Appendices 
Appendices for the Literature Review (Chapter 2) 
Appendix A :  Riordan and Griffeth's (1995) Latent Variable Representation of 
Friendship Opportunity Model 
Friendship 
Opportunities 
Job 
Involvement 
Job 
Satisfaction 
Intention to 
Turnover 
232 
Organisational 
Commitment 
Appendices 
Appendix B: Richer, Blanchard and Vallerand's (2002) Model of Work Turnover 
233 
Appendices 
Appendix C: Antecedents, correlates and consequences to organisational 
commitment. 
From Mathieu and Zajac ( 1 990) 
A n te c e d e n t s  
Personal 
Characteristics 
Age 
Sex 
Education 
Marital status 
Position tenure 
Organisational tenure 
Perceived personal 
competence 
Ability 
Salary 
Protestant work ethic 
Job level 
Role states 
Role Ambiguity 
Role Conflict 
Role Overload 
Job Characteristics 
Skil l  variety 
Task autonomy 
Challenge 
Job Scope 
Group Leader 
Relations 
Group cohesiveness 
Task 
interdependence 
Leader initiating 
structure 
Leader consideration 
Leader 
communication 
Participative 
leadership 
Organisational 
Characteristics 
Organisation size 
Organisational 
centralisation 
234 
C o r r e l a t e s  
Motivation 
Overall 
Internal 
Job involvement 
Stress 
Occupational 
commitment 
Union Commitment 
Job Satisfaction 
Overall 
I ntrinsic 
Extrinsic 
Supervision 
Co-workers 
Promotion 
Pay 
Work itself 
C o n s e q u e n c e s  
Job Performance 
Other's ratings 
Output measures 
Perceived job 
a lternatives 
Intention to search 
Intention to leave 
Attendance 
Lateness 
Turnover 
ORGANISATIONAL COMMITMENT 
Appendices 
Appendix D: Irvine and Evans ' (1995) Model of Nurse Turnover Behaviour. 
235 
Appendices 
Appendices for the Hospital Study (Chapter 3) 
Appendix 1 :  Information Sheet 
[Massey University letterhead] 
An Investigation of Informal Relationships in Organisations 
Information Sheet 
Invitation 
Introducing the 
researcher 
Why we are 
doing this 
research 
Your input 
You are invited to take part in a study investigating informal interpersonal 
relationships in organisations. 
The research is a joint undertaking by Rachel Morrison, Massey 
University and Waitemata DHB. 
You have the right to not participate if you do not wish to. 
My name is Rachel Morrison, and I am interested in the effects of 
informal relationships on outcomes such as Organisational Commitment 
and Job Satisfaction. I am completing this research as part of my doctoral 
degree. 
All contact details are given below. 
The study of relationships in the workplace is important because informal 
social relations may offer significant and rewarding benefits to 
individuals. Friendships can provide increased communication, trust, 
respect, co-operation, support and security that, in turn, can influence 
work related attitudes and behaviours. On the other hand, there may be 
negative consequences of close friendships. 
The information obtained will be used to better understand the impact 
informal relationships on individuals and organisations. 
You can participate in the research by completing this set of 
questionnaires and surveys. The questionnaires will take 15-25 minutes 
to complete (depending on the amount of detail you choose to include in 
the open ended questions at the end.) 
236 
Protecting your 
privacy 
Your rights 
Appendices 
NO material that could personally identify you will be used in any of my 
reports, ensuring that the infonnation you provide is confidential and 
completely anonymous. 
No respondents will be identifiable by the organisation. 
Data will be archived and destroyed after a period of 5 years. 
You have the right to: 
• Decline to participate, 
• Refuse to answer any particular questions 
• Ask any questions about the study 
• Be given access to a summary of the findings of the study when it is 
concluded 
• Withdraw from the study at any time up until you have returned your 
anonymous set of questionnaires and surveys. 
Contact details Please feel free to contact my supervisor or me if you have any concerns 
or questions about this study. 
Who Massey Phone email 
University 
Rachel School of 02 1 1 1 42800 rachel .molTison. 3 @uni .massey.ac.DZ 
Momson Psychology 
(Researcher) 
Stuart Carr School of 09 443 9700 S.C.Carr@massey.ac.nz 
(Supervisor) Psychology extn 9073 
Approval This project has been reviewed and approved by the Massey University 
Regional Human Ethics Committee, Albany Campus, Protocol MUAHEC 
02/005 . If you have any concerns about the conduct of this research, 
please contact Associate-Professor Kerry Chamberlain, Chair, Massey 
University Regional Human Ethics Committee, Albany, ph 09 443 9799, 
email K.Chamberlain@massey.ac.nz 
237 
Appendices 
Appendix 2 :  The Workplace Friendship Scale 
Listed below are a series of statements that represent opinions people might have about 
friendships in their organisation. Indicate the extent to which you agree with the statements 
about the organisation where you are now working (by circling the numbers), using the 
scale at the top of the Table. 
Items measuring the 
Friendship Opportunity 
Dimension 
I have the opportunity to get to 
know my co-workers 
I am able to work with my co­
workers to collectively solve 
problems 
In my organisation I have the 
opportunity to talk informally 
and visit with others 
Communication among 
employees is encouraged by my 
organisation 
I have the opportunity to develop 
close friendships at my 
workplace 
Informal talk is tolerated by my 
organisation as long as the work 
is completed 
Items measuring the 
Friendship Prevalence 
Dimension 
I have formed strong friendships 
at work 
I socialise with co-workers 
outside the workplace 
I can confide in people at work 
I feel I can trust many co­
workers a great deal 
Being able to see my co-workers 
is one reason I look forward to 
my job 
I do not feel that anyone I 
work with is a true friend 
(R) 
Strongly 
disagree 
1 
1 
1 
1 
1 
1 
Strongly 
disagree 
1 
1 
1 
1 
1 
1 
238 
Disagree 
2 
2 
2 
2 
2 
2 
2 
2 
2 
2 
2 
2 
Neither 
agree nor 
disagree 
3 
3 
3 
3 
3 
3 
Neither 
agree nor 
disagree 
3 
3 
3 
3 
3 
3 
Agree 
4 
4 
4 
4 
4 
4 
Agree 
4 
4 
4 
4 
4 
4 
Strongly 
agree 
5 
5 
5 
5 
5 
5 
Strongly 
agree, 
5 
5 
5 
5 
5 
5 
Appendix 3 :  Relationships in the workplace 
1 .  Please read the definition of a SPECIAL PEER: 
Appendices 
You consider this person il best friend. You woultl be friends with this person even if you didn 't work 
together. You consider this person much more than merely a co-worker and feel you know each other 
very well. 
Are there any people who you work with who you would consider to be a special peer? 
(please circle) 
Yes No 
If yes, how many? 
2 .  Please read the definition of a COLLEGIAL PEER: 
This person is a work buddy. You might not share every detail of your life with this person, but this 
person is more than merely an acquaintance. You may consider this person a friend or a colleague 
and interact with this person fairly regularly on an equal basis. 
Are there any people who you work with who you would consider to be a collegial peer? 
(please circle) 
Yes �o 
If yes, how many? (approximately) ______ _ 
3 .  Please read the definition of an  INFORMATION PEER: 
You do not know this person very well or feel very close to this person. You consider this person an 
acquaintance more than a friend. You do interact with this person on a fairly regular basis but you 
would probably not continue the relationship if you did not work here. 
Are there any people who you work with who you would consider to be an information peer? 
(please circle) 
Yes No 
If yes, how many? (approximately) _____ _ 
4.  Please read the definition of a NEGATIVE RELATIONSHIP: 
This person is not one of your friends. You do interact with this person on a fairly regular basis but 
you would definitely not continue the relationship if you did not work here. Your interaction with this 
person is characterised by disrespect, disagreement, dislike, conflict and/or animosity. You would 
rather not have to interact with this person. 
Are there any people who you work with who you would consider to have a negative relationship with? 
(please circle ) 
Yes No 
If yes, how many? 
239 
Appendix 4 :  Dual Role Tension 
1. Do you have a close friend (Specia l Peer) at (Plt�a se circle) 
work that is also someone who you work with? 
Appendices 
YES NO 
If YESRlease complete the following questionnaire with reference to the close fri(md . . 
2. Have you have previously had a close (Please circle) 
friendship with a work-mate? 
YES 
If YES please complete the following questionnrure with reference to the close. friend . . J f  
NO you do not rieed to · complete this section: ' ·� . 
Listed below are a series of statements that represent possible feelings people might have 
about a friendship at work. Indicate the extent to which you agree with the statements (by 
circling the numbers), using the scale at the top of the Table, and referring to the friend you 
mention in question 1. or 2. (top of this page). 
;" . .  Overall, the friendship half and the work 
half of our relationship interfere with each 
other, creating problems for us. 
Our relationship would be a lot easier if we 
were only friends or only work associates 
instead of being both. 
Our work relationship and our friendship 
are often in conflict with one another. 
Problems rarely arise because our 
friendship and our work relationship are 
so much a part of each other. (R) 
It requires extra effort to maintain both the 
friendship side and the work side of our 
relationship. 
My friend and I have lowered our 
expectations about what we expect as both 
friend and co-worker in order to maintain 
our relationship. 
240 
Strongly . Dis,a.gree 
:distigree .. 
1 2 
1 2 
1 2 
1 2 
1 2 
1 2 
Neither· 
agree 
. nor 
. .... disag1'ie 
3 
3 
3 
3 
3 
3 
.Agree . 
4 5 
4 5 
4 5 
4 5 
4 5 
4 5 
Appendix 5 :  Measure .ofWorkgroup Cohesion 
Do you work as part of a work group or team? 
If YES please complete the following questionnaire, 
if NO you do not need to complete this section. 
(Please circle) 
Appendices 
YES NO 
Listed below are a series of statements that represent possible feelings people might have 
about their work group or team. Indicate the extent to which you agree with the statements 
with reference to your own work group or team (by circling the numbers), using the scale at 
the top of the page. 
Social Suppofi 
Being in my team gives me the opportunity to 
work as a team and provide support to other 
team members. 
My team increases my opportunities for 
positive social interaction. 
Members of my team help each other out at 
work when needed. 
Everyone on my team does their fair share of 
the work. 
No one on my team depends on other team 
members to do the work for them. 
Nearly all the members of my team contribute 
equally to the work. 
Members of my team are very willing to share 
information with other team members about 
our work. 
Strongly Disagree Neither 
disagree agree 
1 2 
1 2 
1 2 
1 2 
1 2 
1 2 
1 2 
nor 
d. I l.Sagree 
3 
3 
3 
3 
3 
3 
3 
Agree 
4 
4 
4 
4 
4 
4 
4 
Strongly 
agree 
5 
5 
5 
5 
5 
5 
-=�----�----�------�--�--------------------------------------------=---Teams enhance the communication among 1 2 3 4 5 
people working on the same product. 
���--��--------------------�------------------------------------�---Members of my team cooperate to get the 1 2 3 4 5 
work done. 
--------------------------------------------------------------------------
241 
Appendices 
Appendix 6: The Organisational Commitment Questionnaire 
1 2 3 4 5 6 
STRONGLY DISAGREE SLIGHTLY NEITHER SLIGHTLY AGREE 
7 
STRONGLY · 
AGREE DISAGREE DISAGREE AGREE 
Listed below are a series of statements that represent possible feelings people might have 
about the organisation for which they work. Indicate the extent to which you agree with the 
statements about the organisation where you are now working (by circling the numbers), 
using the scale at the top of the page. 
I describe this organisation to my friends as a great 
organisation to work for. 
I am willing to put in a great deal of effort beyond 
that normally expected III order to help this 
organisation be successful. 
I feel very little loyalty to this organisation. (R) 
I would accept almost any sort of job assignment in 
order to remain at this organisation. 
I find that my values and the organisation's values 
are very similar. 
I am proud to tell others that I am part of this 
organisation. 
I could just as well 'be working for a different 
organisation as long as the type of work was 
similar. (R) 
This organisation really inspires the very best in me 
in the way of job performance. 
It would take very little change in my present 
circumstances to cause me to leave this 
organisation. (R) 
I am extremely glad that I chose this organisation to 
work for over others I was considering at the time I 
joined. 
242 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
[The Organisational Commitment Questionnaire] (continued) 
There's not too much to be gained by sticking 
with this organisation indefmitely. (R) 
Often, I find it difficult to agree with this 
organisation's  policies on important matters 
relating to its employees. (R) 
I really care about the fate of this organisation. 
For me this is the best of all possible organisations 
for which to work. 
Deciding to work for this organisation was a 
definite mistake on my part. (R) 
243 
1 
1 
1 
1 
1 
2 3 
2 3 
2 3 
2 3 
2 3 
Appendices 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
Appendices 
Appendix 7: Job Satisfaction Questionnaire 
This questionnaire deals with your satisfaction with various aspects of your present job. 
Please indicate how satisfied or dissatisfied you feel with each of these aspects using the 
key below. Simply circle the number beside each job aspect that matches your answer. 
1 2 3 4 5 7 
Very Dissatisfied 
Dissatisfied 
Slightly Not Sure 
Dissatisfied 
Slightly 
Satisfied 
6 
Satisfied Very 
Satisfied 
1 .  The physical work conditions. 1 
2 .  The freedom to choose your own 1 
method of working. 
3 .  Your fellow workers. 1 
4. The recognition you get for good 1 
work. 
5 .  Your immediate boss. 1 
6. The amount of responsibility you are 1 
given. 
7 .  Your rate of pay. 1 
8 .  Your opportunity to use your abilities. 1 
9 .  Industrial relations 
management and workers 
between 1 
1 0. Your chances of promotion. 1 
1 1 .  The way the company is managed. 1 
1 2 . The attention paid to suggestions 1 
you make. 
1 3  . Your hours of work. 1 
1 4. The amount of variety in your job. 1 
1 5 . Your job security. 1 
16. Now, taking everything into 
consideration, how do you feel about 1 
your job as a whole? 
244 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
2 3 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
4 5 6 7 
Appendices 
Appendix 8: Measure of Intention to Leave 
Listed below are a series of statements that describe your intentions in the near future. 
Please show your intentions for your present job using the scales below. 
Remember, all responses are confidential. 
1 .  How likely is it that you will actively look for a new job  in  the next year? 
(Please circle the number of your response) 
1 2 3 4 5 
NOT AT ALL · · SLIGHTLY · SOMEWHAT NEITHER QUITE 
LIKELY LIKELY LIKELY 
2. I often think about quitting. 
(Please circle the number of your response) 
1 2 3 4 
STRONGLY DIsAGREE SLIGHTLY . NEITHER 
STR()NGLY 
DISAGREE DISAGREE' 
3.  I will probably quit my job in  the next year. 
(Please circle the number of your response) 
245 
LIKELY 
5 
SLlGfitLY 
.AGREE 
. 6 
HIGHLY 
LIKELY 
6 .. 
.. �GREE 
. �.� 
.7 
EXTREMELY 
LIKELY 
7 .": 
. .  
AGREE 
Appendices 
Appendix 9 :  The impact of informal relationships at work 
(Open ended Questions) 
Use the definition� for friendships and negative relationships (on the second page of 
this questionnaire) to help you answer these questions. If you feel that the 
definitions I have provided do not adequately define the friendships and negative 
relationships you have experienced, please feel free to change or comment on the 
definitions I have provided. 
1 .  Please briefly outline ways in which a friendship with one or more people with 
whom you work(ed) have benefited you in the workplace. 
2. Please briefly outline ways in which a friendship with one or more people with 
whom you work(ed) have made your work more difficult. 
3 .  Please briefly outline how a negative relationship with someone with whom you 
work( ed) has made your work more difficult. 
4. Please briefly outline how your work environment has facilitated or in some 
way enriched a friendship. 
5 .  Please briefly outline how your work environment has strained a workplace 
friendship. 
6. Please briefly outline how your work environment has exacerbated (made 
worse) a negative relationship you have had at work. 
246 
Appendices 
Appendix 10:  Demographic Information 
Appendix lOa: Demographic Information Sheet 
The following questions are about you. As with all the infonnation requested in this questionnaire, your 
response is confidential to Rachel Morrison. Only summary information will be made available. 
Therefore, no information will be fed back to Waitemata DRB which can identify any individual. 
1. Age 
_________ years 
2. Gender Male Female 
3. What Ethnic group do you identify with? (e.g., N.Z. Maori, European, Indian etc) 
Please state ---------------------
4. How long have you worked in this 
organisation? 
5. Which best describes your role? 
(Please circle) 
____ years 
Clerical 
Teclmical 
Clinical 
Management 
Months ----
Other (Please state) __________ _ 
8. Do you work a team I work group I unit? Yes 
9. If yes, How many people are in your team I workgroup I unit? 
No 
2-4 5-9 10- 14  1 5-20 More than 20 
10. What is the n ame of your team I (Please state) ___________ _ 
I unit? 
Thank you for your help. 
Please check back through your answers to made sure that you have not missed any out. As mentioned before, 
the responses to these questionnaires are confidential and will be collated before use. No information that will 
in any way identify individuals will be made available to Waitemata DHB. 
r.K.aclie{ 1vf.omson 
247 
Appendices 
Appendix lOb:  Pie Charts show�ng Demographics of sample for the Hospital 
Study 
Women 
95% 
Men 
5% 
Figure 1 Ob.1 : Gender of sample for the Hospital Study 
60-68 years 
1 0% 
50-59 years 
20% 
22-39 years 
1 3% 
40-49 years 
44% 
30-39 years 
1 3% 
Figure 1 0b.2: Age of sample for the Hospital Study 
Other 
2% 
Oinical 
79% 
Oerical 
Technical 
3% 
Figure 10b.3: Work role of respondents to the Hospital Study 
248 
Maori 
5% 
Pacific Island 
3% Chinese 
1 %  
Appendices 
Canadian 
1 %  
NZ European 
or "New 
Zealander" 
89% 
Phil ippine 
1 %  
Figure 1 0b.4: Ethnicity of sample for the Hospital Study 
60-68 years 
1 0% 
50-59 years 
20% 
22-39 years 
1 3% 
40-49 years 
44% 
30-39 years 
1 3% 
Figure 1 0b.S: Tenure of sample for the Hospital Study 
249 
Appendices 
Appendix 1 1 :  Reminder letter · 
Dear Waitemata DHB employee, 
The recent 
questionnaire 
Thank-you 
A plea 
What if I have 
mislaid my 
questionnaire? 
Contact me 
To return the 
questionnaire 
Recently you were ra:ndomly selected from employees of Waitemata DHB to receive a 
questionnaire on workplace relationships and were invited to take part in a study 
investigating informal interpersonal relationships in organisations. 
I would like to take this opportunity to thank those of you who have already returned the 
questionnaire; this type of research would not be possible without your assistance. 
For those of you who have not yet returned the questionnaire but do wish to participate, it 
is not too late, I am still receiving documents and every response is of value. 
The information obtained will be used to better understand the impact informal 
relationships have on individuals and organisations. 
If you no longer have the questionnaire but wish to participate, please contact Shelley 
Hughes and she will put a new one in the internal mail for you. 
Shelley Hughes, Management Secretary Community and Disability Services, Health West 
Building, Waitakere Hospital, Waitemata DHB. 
(or phone 8376632). 
If you have any questions or concerns please don't hesitate to contact me 
at rachel.morrison@xtra.co.nz or 021 1 1 42800. 
Either put the completed questionnaire in the return envelope provided or, if you prefer, 
you can mail it directly to me at the address below: 
Rachel Morrison (PhD candidate) 
School of Psychology 
Massey University (Albany Campus) 
Private Bag 102 904 
North Shore MSC 
Auckland 
Thanks again for your time. 
(j(acfie[ ?rf. o rris o n  
250 
Appendices 
Appendix 12 :  Scree Plots of factor analyses for the Hospital Study 
Scree P lot 
7�--------------------------------------------� 
6 
5 
4 
3 
2 
3 5 7 9 1 1  1 3  1 5  1 7  1 9  2 1  
Factor Number 
Figure 1 2.1 : In itial scree plot of the factor analysis of the friendship, cohesion and dual 
role tension measures 
Scree P lot 
7�--------------------------------------------� 
6 
5 
4 
3 
2 
2 3 4 5 6 7 8 9 1 0  1 1  1 2  1 3  1 4  1 5  1 6  17 1 8  1 9  20 
Factor Number 
Figure 1 2.2:  Final scree plot of the factor analysis of the friendship, cohesion and dual  
role tension measures 
25 1 
Appendices 
Scree Plot 
1 2�--------------------------------------------� 
1 0  
8 
6 
4 
3 5 7 9 1 1  1 3  1 5  1 7  1 9  2 1  23 25 27 29 31 33 
Factor Number 
Figure 1 2.3: Initial scree plot of the factor analysis of the satisfaction, commitment and 
leaving intention measures 
Scree Plot 
1 0�--------------------------------------------� 
8 
6 
4 
2 3 4 5 6 7 8 9 1 0  1 1  1 2  1 3  1 4  1 5  1 6  1 7  1 8  19 2 0  
Factor Number 
Figure 1 2.4: Final scree plot of factor analysis of the satisfaction,  commitment and 
leaving intention measures 
252 
Appendices 
Appendix 13 :  Factor analysis of all the measures together 
To further confinn that the scales used in the questionnaire were conceptually distinct, 
the six variables used in the path analysis were factor analysed together. As stated in 
Chapter Three, it is possible that the scales would tap into similar constructs, so the 
analysis is carried out to ensure that the scales are internally coherent and also distinct 
from the other. measures. The 54 items (i.e., a 1 2  item Friendships scale, a 9 item 
Cohesion scale, a 1 5  item Organisational Commitment scale, a 1 5  item Job Satisfaction 
scale and a 3 item Intention to Leave scale) were factor analysed with a maximum 
likelihood extraction and direct oblimin rotation. The solution was set to six factors to 
represent the six components (i .e., friendship opportunities, friendship prevalence, 
cohesion, organisational commitment, job satisfaction and intention to leave). See 
Figure 1 3 . 1  at the end of this section for the scree plot of this analysis. 
The Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a value of .8 1 .  The 
resulting matrix accounted for 49.0% of the variance. Most of the items loaded on their 
hypothesised factors (refer Table 1 3 . 1 ) . 
of the a l l  the measures . .-----------------�----.-�--.-�-.--� 
253 
Appendices 
lWorkplace Friendship Scale 5 .656 .47J 
IW orkplace Friendship Scale 1 0  .623 
tworkplace Friendship Scale 1 1  .56C 
Y{orkplace Friendship Scale 1 2  .501 " 
Workgroup cohesion 2 .39� 31 � 
ntention to leave 3 .965 
ntention to leave I .949 
ntention to leave 2 .509 
�ob Satisfaction Scale 9 .356 .340 
�ol:) Satisfaction Scak 6 
ptgilhisauoilal G6mmitm:erit: 7: 
� cia Satisfa,ctiou.-Scale ,J 4 
�igarus;iition:al " CoIitiliiimerit' 3i 
tw orkgroup cohesion 9 .780 
IW orkgroup cohesion 7 .696 
IW orkgroup cohesion 6 .634 
IW orkgroup cohesion 4 .593 
W orkgroup cohesion 5 .58..: 
w.(;rl<gr.ouji �oliesio&s: �_5J ;tllQ. 
Workgr()Ui': Gohesi6ii�3' :'i{8 
WorJfgrquP. cQh"e"�i6ji t 41:3 ;� J� 
.to9�S;atisf�tiioi Lseare�3 ;'3"35 :4�4 
�orkplace Friendship Scale 3 .73 1 
\Vorkplace Friendship Scale 2 .694 
\Vorkplace Friendship Scale 4 .647 
lWorkplace Friendship Scale 6 .474 
IWbj-JCP14p:e$.li�d§1ih;;: stale; 11 :4� :tl® 
�ob Satisfaction Scale 1 6  .335 
�li'$;�ti�lactiQ1i'$�at6:;,� 
�ob Satisfaction Scale 5 .584 
�Qf) S'apsfadlQn ... ScaICb 'j � �J� �"9.� 
[ob Satisfaction Scale I I  .473 
�ob Satisfaction Scale 1 3  .452 
�ob Satisfaction Scale 4 .406 
Extraction Method: MaXImum LIkelIhood. Rotation Method: Oblimm WIth KaIser Normalization. 
Rotation converged in 1 7  iterations. Loadings less than .3 are not shown 
All items which had absolute values less than .3 ,  or which were cross loading across 
more than one factor, were removed. These are shaded in grey in Table 13.1, above. 
A second factor analysis, with the items shaded in grey not included, was performed to 
see how well the remaining items loaded on each factor. It accounted for 54.9% of the 
variance, and revealed six distinguishable factors representing the six scales. The 
Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a value of .82.  The pattern 
matrix for the second analysis is shown in Table 13.2. See Figure 13.2 at the end of this 
section for the scree plot of this analysis. 
254 
Table 1 3.2:  S 
Appendices 
econ d P tt a ern M t ' f th f a fiX 0 e actor analysIs 0 f th 1 1  he measures. e a  t 
I:'acto 
1 2 3 4 5 P �rganisational Commitment 6 .867 
Orga:nisational Commitment 1 .838 
Qt'g!lnisational Commitment 1 4  .760 
Organisational Commitment 10  .726 
Organisational Commitment 5 .694 
Q!ganisational Commitment 1 5  . 5 1 7  
Organisational Commitfuent g :� 1 5  .428 
Organisational Commitment 2 .483 
Organi$atiCmal Commitmeiit ;4. .394 .308 
Organisational Commitment 9 . 384 
prganisational Commitment 1 3  .3 1 6  
ntention to leave 3 .953 
�ntention to leave 1 .884 
�ntention to leave 2 .45 1 
1W0rkplace Friendship Scale 7 .802 
1W0rkplace Friendship Scale 8 .74 1 
Workplace Friendship Scale 9 .688 
Wot.ICPl.���: 'F tleh�fsliiii-S�a�e tQ ;57� :3SC 
Workplace FriendshiI>_ Scale 1 1  .563 
Workplace Friendship Scale 1 2  .539 
VI ork�roup cohesion 7 .732 
W orkgroup cohesion 5 .669 
W orkgroup cohesion 4 .666 
IWorkgroup cohesion 9 .653 
IW orkgroup cohesion 6 .633 
1W0rkplace Friendship Scale 3 .747 
lWorkplace Friendship Scale 4 .698 
1W0rkplace Friendship Scale 2 .659 
1W0rkplace Friendship Scale 6 .446 
�pb:;S'atistlihliQq;S'c�l�� ;3�3 
�ob Satisfaction Scale 5 .666 
�ob Satisfaction Scale 1 1  .60 1 
�ob Satisfaction Scale 4 .55C 
�ob Satisfaction Scale 9 .50 
�ob Satisfaction Scale 1 3  . 4 1  
Extraction Method: Maximum Likehhood. Rotation Method: Obhmm With Kaiser Nonnalization. 
Rotation converged in 9 iterations. Loadings less than .3 are not shown. 
Three final items were cross loading with other scales so were also removed; these are 
shaded in grey in Table 1 3 .2 above. 
A final factor analysis, with the solution still set to six factors to represent the six 
components accounted for 55 .8% of the variance, and revealed six distinguishable 
factors representing the six scales (refer Table 1 3 .3). Each of the items loaded on the 
expected factor. The scree plot for the final factor analysis is shown at the end of this 
255 
Appendices 
section as Figure 1 3 . 3 .  The Kaiser-Meyer-Olkin Measure of Sampling Adequacy had a 
value of .826. 
Table 1 3.3:  5 econ d P tt a ern M t ' f th f t a nx o e ac or analysIs 0 f th 11 the measures. e a  
facto 
1 2 3 � 5 . � �ganisational Commitment 6 .889 
K:>rganisational Commitment 1 .841 
Organisational Commitment 14  .777 
Organisational Commitment 10  .741 
Organisational Commitment 5 .7 1 5  
Organisational Commitment 1 5  .538 
Organisational Conunitment 2 .520 
Organisational Commitment 9 .382 
Organisational Conunitment 13  .34 1 
ntention to leave 3 .95� 
ntention to leave 1 .874 
Ilntention to leave 2 .438 
lWorkplace Friendship Scale 7 .844 
lWorkplace Friendship Scale 8 .721  
Workplace Friendship Scale 9 .623 
Workplace Friendship Scale 1 2  .580 
Workplace Friendship Scale 1 1  . 5 19  
W orkgroup cohesion 7 .7 10  
W orkgroup cohesion 4 .676 
W orkgroup cohesion 9 .674 
W orkgroup cohesion 5 .666 
IW orkgroup cohesion 6 .654 
IWorkplace Friendship Scale 3 .787 
lWorkplace Friendship Scale 4 .72 1 
lWorkplace Friendship Scale 2 .652 
lWorkplace Friendship Scale 6 .432 
�ob Satisfaction Scale 5 .683 
�ob Satisfaction Scale 4 .604 
�ob Satisfaction Scale 1 1  . 587 
�ob Satisfaction Scale 9 .558 
�ob Satisfaction Scale 1 3  . 4 1 3  
Extraction Method: MaXimum Likelihood. Rotahon Method: Oblirrun With Kaiser Normalization. 
Rotation converged in 7 iterations. Loadings less than .3 are not shown. 
The items remaining in Table 1 3 .3 are almost identical to those remaining after the two 
separate factor analyses described in Chapter Three. The main difference between this 
factor analysis and the one described in Chapter Three are the job satisfaction items. 
While the items remaining after the analysis above are 4, 5, 9, 1 1  and 1 3 ;  the remaining 
items in the analysis described in Chapter Three are 4, 5, 6, 7, 9, 1 1  and 1 5 .  Other than 
this discrepancy the two analyses yielded remarkably similar results. 
256 
Cl) 
.2 ('IJ > c Cl) Cl 
W 
Appendices 
Scree Plot 
1 6�--------------------------------------------� 
1 4  
1 2  
1 0  
8 
6 � 
4 
2 
4 7 1 0  1 3  16 1 9  22 25 28 31 34 37 40 43 46 49 52 
Factor Number 
Figure 1 3.1 : First scree plot of the factor analysis of all six measures in the 
questionnaire 
Scree P l o t 
1 2.--------------------------------------------. 
1 0  
8 
6 
4 
3 5 7 9 1 1  1 3  1 5  1 7  1 9  21 23 25 27 29 31 33 35 
Factor Number 
Figure 1 3.2: Second scree plot of the factor analysis of al l  six measures in the  
questionnaire 
257 
Appendices 
Scree Plot 
1 0
.-
----------------------------------------------. 
8 
6 
4 
3 5 7 9 1 1  13 15 17 1 9  21 23 25 27 29 31 
Factor Number 
Figure 1 3.3: Final scree plot of the factor analysis ..of all six measures 
258 
Appendix 14 :  AMOS output: the Hospital Study 
Appendix 14.1 
Standardised Regression Weights 
Standardi zed Regression Weight s :  
Appendix 14.2 
JS SUM <----- F 0 SUM 
JS SUM <----- COH SUM 
F P SUM <--�- F 0 SUM 
F P SUM <-- - - COH SUM -
OCQ SUM <---- COH SUM - -
OCQ SUM <----- JS SUM - -
OCQ SUM <---- F 0 SUM -
ITL  SUM <---- F P SUM -
I TL SUM <----- JS SUM - -
I TL SUM <---- OCQ SUM - -
ITL  SUM <---- COH SUM 
Estimat e  
0 . 2 7 1  
0 . 2 8 7  
0 . 3 0 7  
0 . 2 3 0  
-0 . 0 3 9  
0 . 5 8 6  
0 . 1 65 
-0 . 1 5 4  
-0 . 2 97 
-0 . 2 4 2  
-0 . 0 0 1  
Appendices 
95.0% Standardised (Beta) Weights and confidence intervals (bias corrected 
percentile method) 
Standardized ( Bet a )  Weight s : 
Lower Upper  
Bound Bound p 
JS SUM <----- F 0 SUM 0 . 0 9 8  0 . 4 4 4  0 . 0 1 1  
JS SUM <----- COR SUM 0 . 02 9  0 . 4 8 4  0 . 0 1 3  
F P SUM <---- F 0 SUM 0 . 07 8  0 . 4 5 4  0 . 0 4 6  
F P SUM <---- COH SUM 0 . 03 0  0 . 4 5 9  0 . 02 5  
· ·�HW.1\'"�� m��n.r.'i�;�§.@iR��,d"Y�Nlii\'Iii� g€;��!ifi �7tiI �iIBi!t.�.:ttflM.�ii. . � � 
OCQ SUM < - -- - - JS SUM 0 . 4 3 9 0 . 7 0 8  0 . 0 1 9  
� t __ _ 
ITL  SUM <---- F P SUM -0 . 3 2 2  - 0 . 0 0 9  0 . 0 4 6  
I TL SUM <----- JS SUM -0 . 5 0 5  - 0 . 1 07 0 . 00 4  - -
I TL SUM <---- OCQ SUM -0 . 4 0 8  - 0 . 0 6 1  0 . 0 3 4  
[1li:YW$15t;;(���tiiQ�j,m��.!'1�'i'{)M�lf?� ��J,l!.lI,)\.�� i".:Lt�'· .�; .. d,:.a.. ��i?ii�i£.�� . .J. &f ��..-}� �;t1rL -£rlG ;��'4.:�  
259 
Appendices 
Appendix 14.3 
Regression weights and critical ratio values, (shading indicates a non-significant 
regression) 
Maximum Likelihood Estimates  
Regres sion Weight s : Estimate  S . E .  C . R . 
JS SUM <----- F 0 SUM 0 . 8 4 6  0 . 27 0  3 . 1 3 6  
JS SUM <----- COH SUM 0 . 67 8  0 . 2 0 3  3 . 3 3 5  
F P SUM <--- - F 0 SUM 0 . 602  0 . 1 7 2  3 . 4 97 
F P SUM <--- - COH SUM 0 . 3 4 3  0 . 1 3 0  2 . 6 4 5  
9G:g7$.:Q��,r-S-':{,�RK��m1T:. �;:� ·.,::::Q�IA4=4.L·:2:9.7�17.���{�!.E9�fi"� 
OCQ_SUM <----- JS SUM 0 . 7 87  0 . 1 0 5  7 . 4 63 
OCQ_SUM <---- F 0 SUM 0 . 7 09  0 . 32 6  2 . 1 7 9  
ITL  SUM <--- - F P SUM -0 . 158  0 . 08 4  - 1 . 8 8 6  
ITL SUM <----- JS SUM -0 . 1 90 0 . 0 67 -2 . 8 2 8  
I TL  SUM <--- - OCQ SUM -0 . 1 1 8 0 . 0 4 7  -2 . 4 8 5  
��tL?;$bM <��.::::' j:j2�:<ST�E:' ;:: . : :� : : . �Q �-g:lT5'�> ::?Ejg:;�'t��-!{�;:t:�'0�7:qfl1�:� 
Appendix 14.4 
Indices of fit for Model 1 
Summary of models 
Model NPAR CMIN DF P CMIN/DF  
---------------- --------- --------- ---------
Rachel ' s  model 15  6 . 663  6 0 . 3 5 3  1 . 1 1 0  
Saturated model 2 1  0 . 0 00  0 
Independence model 6 67 . 8 2 4  1 5  0 . 0 00  4 . 52 2  
Zero model 0 3 6 3 . 000  2 1  0 . 0 00  1 7 . 2 8 6  
Model RMR GFI  AGFI PGFI 
---------------- ---------- ---------- ---------- ----------
Rache l ' s model 1 .  4 1 5  0 . 9 8 2  0 . 9 3 6  0 . 2 8 0  
Saturated model 0 . 0 00  1 .  000  
Independence model  20 . 590  0 . 8 1 3  0 . 7 3 8  0 . 58 1  
Zero model 3 0 . 2 98 0 . 0 0 0  0 . 0 00  0 . 00 0  
DELTA1 RH01 DELTA2 RH02 
Model NFI RFI I FI TLI  CFI 
---------------- ---------- ---------- --------- - ---------- -------
Rachel '  5 model 0 . 902  0 . 7 5 4  0 . 98 9  0 . 9 6 9  0 . 9 8 7  
Saturated model 1 .  000  1 .  0 00  1 .  0 0 0  
Independence model 0 . 000  0 . 0 00  0 . 000  0 . 00 0  0 . 0 0 0  
Model PRATIO PNFI PCFI 
---------------- ---------- ---------- ----------
Rachel '  5 model 0 . 4 00 0 . 3 61 0 . 3 95 
Saturated model 0 . 000  0 . 0 0 0  0 . 000  
Independence model 1 . 0 00  0 . 0 0 0  0 . 0 00  
260 
Appendices 
Model NCP LO 90 HI 90  
---------------- ---------- ---------- ----------
Rache l ' s model 0 . 6 6 3  0 . 000  1 1 . 3 1 2  
Saturated model 0 . 000  0 . 000  0 . 0 00  
Independence model 52 . 824  3 0 . 9 4 6  8 2 . 2 4 5  
Model FMIN FO LO 90 HI 9 0  
---------------- ---------- ---------- ---------- ----------
Rache l ' s model 0 . 0 55  0 . 0 0 5  0 . 00 0  0 . 0 93 
Saturated model 0 . 000 0 . 000  0 . 0 00  0 . 0 0 0  
Independence model 0 . 5 61 0 . 4 3 7  0 . 2 5 6  0 . 6 8 0  
Model RMSEA LO 90  HI  90  PC;LOSE 
---------------- ---------- ---------- ---------- ----------
Rachel '  5 model 0 . 030  0 . 0 00  0 . 1 2 5  0 . 53 6  
Independence model o . 17l  0 . 1 3 1  0 . 2 1 3  0 . 0 0 0  
Model AIC BCC BIC  CAIC  
---------------- ---------- ---------- ---------- ----------
Rachel '  5 model 3 6 . 663  3 8 . 505  1 0 5 . 5 99  9 3 . 7 2 3  
Saturated model 4 2 . 000 4 4 . 57 9  1 3 8 . 5 1 1  1 2 1 . 8 8 4  
Independence model  7 9 . 8 24 8 0 . 561  1 0 7 . 3 99  1 02 . 64 8  
Zero model 3 63 . 000 3 63 . 000  3 6 3 . 0 00  3 63 . 0 0 0  
Model ECVI LO 90 HI 90  MECVI 
---------------- ---------- ---------- ---------- ----------
Rachel '  5 model 0 . 3 03 0 . 2 98 0 . 3 9 1  0 . 3 1 8  
Saturated model 0 . 3 4 7  0 . 3 4 7  0 . 3 4 7  0 . 3 68 
Independence model 0 . 6 60 0 . 4 7 9  0 . 90 3  0 . 6 6 6  
Zero model 3 . 000  2 . 520  3 . 5 4 2  3 . 0 0 0  
HOELTER HOELTER 
Model . 05 . 0 1 
---------- ------ ---------- ----------
Rachel '  5 model 229  3 06  
Independence model 4 5  5 5  
Zero model 1 1  1 3  
261 
Appendices 
Appendix 15 :  Summary of answers to qualitative questions: the Hospital Study 
1 .  Please briefly outline ways in which a friendship with one or more people with whom you 
work( ed) have benefited you in the workplace. 
Answer Number of 
respondents answering 
in this way 
Makes work more pleasant / fun / look forward to/enjoy coming to work. 24 
Compassion / consideration / confiding / listenin� 21  
Social support 2 1  
Learning from more experienced colleagues / advice / mentoring 20 
Improves communication / information exchange between colleagues / honesty / trust 20 
Professional /practical help and support / workload sharing 20 
Bounce ideas off / shared problem solving / sounding boards 1 3  
Sharing celebrations / laughs 8 
Being able to rely on / count on colleagues 7 
Leads to friendships / social activities out of work time. 7 
Enhanced feeling of acceptance / feel part of team / improves teamwork 6 
Helped when things are negative/busy at work / made bad situation bearable 6 
Improves mood / makes me feel better 5 
Reduces stress 5 
Improves confidence in ability 5 
Improved job satisfaction 4 
Enhances standard of work / professional development 2 
Career advancement 1 
Choose NOT to have particular bonds / friends at work. 1 
2. Please briefly outline ways in which a friendship with one or more people with whom you 
work( ed) have made your work more difficult. 
Difficult to criticise friends / wont complain when they are friends, more likely to put up with 6 
a bad situation. 
Distraction from work related tasks 3 
Conflict of interest / don't want to do or say what should be done, in case I �set my friend 3 
Friends expecting special treatment, favours 3 
Having to take a peace-maker role between friend and management / between two �eers 2 
Difficult to be objective / biasedp_erformance reviews 2 
Truthful friends at work can seem like criticism / difficult to receive criticism from friends 2 
Friendship breaking down and becoming verbally violent / difficult to trust 2 
Difficult to have a close friendship with someone you manage when addressing performance 2 
issues / peer reviews 
Always keep work and personal relationships �arate so NO conflict arises 2 
Having to "tip toe" around friend/workmate in a bad mood 1 
Confidentiality issues 1 
Gossip between workmates concerning me 1 
Friend not getting along with others in team 1 
Negativity / dissatisfaction / negative opinions of a friend can influence me 1 
Being "tarred with the same brush" as a dissatisfied / trouble making friend � the boss 1 
Having to phrase instructions / teaching so as not to sound condescending to a friend / not to 1 
offend 
262 
Appendices 
Can create unfairness 
People at work knowing aspects of my personal life 
Jealousy between friends 
Friend decided to keep work and personal life entirely separate, creates tension, friendship 
ended 
Causing others to feel left out 
When one friend is promoted in the place of another 
Getting intimately / romantically involved 
3 .  Please briefly outline how a negative relationship with someone with whom you work(ed) 
has made your work more difficult. 
Feeling anger / discomfort / stress / unhappiness 
More difficult to communicate ideas and needs / communication breakdown / withholding 
information 
Reluctance to talk with / interact with the person / avoiding the person 
Demotivates other team members / interferes with team work / weakens sense of team 
Disrupts work / breaks concentration /�et less work done 
Creates an un�leasant work environment / not wanting to come to work 
Changing bookings / schedules to avoid the person, interferes with getting my work done. 
Reduces job satisfaction 
The person influencing other team members, making relationships with them difficult also. 
Find it difficult to be positive / effective in other aspects of my j ob if I am already frustrated 
by this person 
Lack of trust in people and work environment 
Gossip / back-biting is destructive 
Decreases my confidence 
Feel I cannot speak up against the person who acts in a way that is detrimental to our client's 
well-being 
Being embarrassed by the person's 'behaviour 
The person receives special treatment and favours, the rest of us do her work 
The person setting me up as not performing 
Lack of confidence in my initiative, over-supervises me 
Waste of time dealing with that person's actions 
1 
1 
1 
1 
1 
1 
1 
20 
17 
9 
8 
7 
5 
5 
3 
3 
3 
3 
2 
2 
1 
1 
1 
1 
1 
1 
4. Please briefly outline how your work environment has facilitated or in some way enriched a 
friendship. 
Met / made friends as a result of us working together / proximity / team work 25 
Shared work experiences give an extra bond / common ground 14 
Work environment allows us to talk and socialise / friendly environment 8 
Offering I receivin� social / professional support to workmates has facilitated friendship 8 
Open plan / shared office allows lots of contact with others 6 
We are able to take tea breaks together, gives opportunity to catch up / debrief 5 
Allowed me to meet a variety of people / people from different backgrounds 4 
Working as a team / team work encourages communication 4 
Professional respect for my colleagues 3 
Similar ideas / goals leads to appreciation of other's achievements 2 
Spending time together on "away days" team building 2 
Our work sUEEorts families, shared family experiences creates a bond 2 
Night shift provides a quiet place to talk, get to know colleagues 1 
263 
Appendices 
Through being open and being able to discuss things with team members, builds relationships 1 
Learning tolerance of other ways of life 1 
We don't see much of each other in my team so when we do we make the most of this 1 
precious time 
Management doesn't listen so we turn to each other 1 
Our manager encourages teamwork / innovation 1 
Poor work conditions mean we rely on each other more 1 
The supervisor scheme partnered me with a great supervisor who has become a good friend / 1 
support 
5 .  Please briefly outline how your work environment has strained a workplace friendship. 
Heavy workload / unable to find time for breaks / unrealistic expectations of management 3 
When the case load is heavy and a peer doesn't help share others workload / others not 2 
pulling their weight 
If one person is promoted or received duties another wants / j ealousy 2 
Separate offices split up teams / friendship not maintained when I stopped sharing office 2 
space with them 
Having to discipline them at work has strained the friendship 1 
New structures / policies making teamwork impossible 1 
Lack of support by senior management 1 
Peer appraisal system, a peer wrote about me negatively, ruined the friendship 1 
When a colleague has failed to provide a needed service 1 
DurinK busy / stressful times colleagues unfriendly and snap 1 
When working closely together, you must be seen to be treating all team members equally 1 
some team members think you are showing favouritism and get annoyed 
Open plan environment, noisy and can't get away from each other 1 
Having different goals or objectives has sometimes strained a friendship 1 
Not giving a friend a promotion she thought she deserved ended the friendship. 1 
When friends are dissatisfied with work, results in strain 1 
6. Please briefly outline how your work environment has exacerbated (made worse) a negative 
relationship you have had at work. 
Have to work in a close confined space with people I have a negative relationship with / 8 
havin� to work closely / regularly with the person 
Heavy workload / fatigue 3 
Being busy and short staffed adds to pressure, leaves no time to deal properly with conflicts 2 
Workplace politics 2 
The team grouping together to discuss negative aspects of a team member exacerbates the 2 
situation / gossip 
Lack of communication 2 
Inability of management to effectively deal with a negative issue - making one side do all the 2 
mediation 
Senior staff unwilling to support nurses during personal unhappy times 1 
Unclear roles and communication from management mean relationships are ambiguous 1 
Slow organisational change, slow to act on recommendations 1 
Relationship not being handled well by management 1 
Being told to take an issue (being accused of something I didn't do) no further by 1 
management has meant that person still believes I am wrong 
Lack of academic colleagues to talk to 1 
264 
Appendices 
Having to correct an individual without them taking it personally 1 
Disagreements about care delivery 1 
Inappropriate person promoted to senior role 1 
Poor morning tea / lunch facilities 1 
265 
Appendices 
Appendices for the Internet Study (Chapter 4) 
Appendix 16 :  Needs Assessment Questionnaire 
This scale contains 20 statements that may describe you and the types of things you may 
like to do. Indicate the extent to which you agree with each statement using the scale at 
the top of the Table. 
Strongly ·Disagree Neither ,Agree. ;.��ngJ� . : 
disagree agree nor . . . : ·agree ': . 
disa�ee 
I try to perfonn my best at work. (nAch) 1 2 3 4 5 
I spend a lot of time talking to other people. (nAff) 1 2 3 4 5 
I would like a career where I have very little 1 2 3 4 5 
supervision. (nAut) 
I would enjoy being in charge of a project. (nDom) 1 2 3 4 5 
I am a hard worker. (nAch) 1 2 3 4 5 
I am a "people" person. (nAff) 1 2 3 4 5 
I would like a job where I can plan my work schedule 1 2 3 4 5 
myself. (nAut) 
I would rather receive orders than give them. (R) 1 2 3 4 5 
(nDom) 
It is important to me to do the best job possible. 1 2 3 4 5 
(nAch) 
When I have a choice, I try to work in a group instead 1 2 3 4 5 
of by myself. (nAff) 
I would like to be my own boss. (nAut) 1 2 3 4 5 
I seek an active role in the leadership of a group. 1 2 3 4 5 
(nDom) 
I push myself to be "all that I can be." (nAch) 1 2 3 4 5 
I prefer to do my own work and let others do theirs. 1 2 3 4 5 
(R) (nAff) 
I like to work at my own pace on job tasks. (nAut) 1 2 3 4 5 
I fmd myself organizing and directing the activities of 1 2 3 4 5 
others. (nDom) 
I try very hard to improve on my past perfonnance at 1 2 3 4 5 
work. (nAch) 
I try my best to work alone on a work assignment. (R) 1 2 3 4 5 
(nAff) 
In my work projects, I try to be my own boss. (nAut) 1 2 3 4 5 
I strive to be "in command" when I am working in a 1 2 3 4 5 
group. (nDom) 
266 
Appendices 
Appendix 17:  Respondents' occupations (Internet Study) 
Respondents 1-120 
1 Project Manger 41 81 Project Advisor 
2 Business Analysis 42 Project coordinator on 82 Probation officer 
a study 
3 Marketing Co-ordinator I Administrator 43 Senior Accountant 83 Teacher 
(Qualified) 
4 PhysiotherapisUacupuncturist 44 Sales Rep 84 Account executive 
5 Manager 45 Toolmaker 85 HR AdviserlOffice 
Manager 
6 Office-all-round employee, to finance my 46 Change Manager 86 Registered nursel staff 
2. studies at university nurse 
7 Senior lecturer 47 Lifeguard 87 Administration 
8 Professor 48 Sandwich Artist 88 North Island Customer 
Services Manager 
9 Research assistant 49 Account Manager 89 Brand Manager 
10  Senior Consultant 50 Statistician 90 Research fellow 
1 1  Customer Service Consultant 51 Account Manager 91 Office Manager 
12  Teacherlresearcher 52 Editor & Copywriter 92 Telephone interviewer 
13  Senior development officer 53 Supervisor, Medical 93 Server 
Staff Services 
14 Assessment Coordinator 54 Design engineer 94 Office Manager 
15  Lecturer 55 Programme co- 95 Sponsorship manager 
ordinator 
16  Teller 56 96 Education Officer 
17  Human Resources Consultant 57 Systems admin 97 PhysiotheraQist 
1 8  Research fellow 58 Management 98 Statistician 
19  Teacher 59 Physiotherapist 99 Account Manager 
20 Psychologist 60 Lending Supervisor 1 00 Consultant 
21 Senior Lecturer 61 Case Manager 101  AnalysUconsultant 
22 Lecturer 62 Consultant 102 Consultant 
23 Commanding officer 63 Consultant 103 Digital Media Specialist 
24 Health worker 64 Advertising 104 Communications 
coordinator manager 
25 Lecturer 65 Interim Manager 105 Medical writer 
26 Psychologist 66 Senior Clinical 106 Publicist 
Psychologist 
27 Education Facilitator 67 Consultant 107 Nurse 
28 Project manager 68 PAlSecretary 108 Clerical receptionist 
29 Software analyst 69 Personal Assistant 109 Sales Consultant 
30 Partner 70 Senior Systems 1 1 0  Nurse 
Consultant 
31 E-Commerce Project Manager 71 Small Business 1 1 1  Counsellor 
Manager 
32 Research Associate, Search Team 72 Case Manager 1 1 2  Service Del ivery 
Manager 
33 Head of development 73 Graduate student 1 1 3  Principal of a K-8 school 
34 Consultant 74 Retailer Sales 1 1 4  Shipping 
Manager 
35 MarketinQ Executive 75 Registered nurse 1 1 5  User Support Manager 
36 Telemarketer 76 Interviewer 1 1 6  Software Development 
Consultant 
37 Consultant 77 Marketing Co- 1 1 7  Clerical 
ordinator supporUCustomer 
Service 
38 Senior Tutor 78 GM Finance 1 1 8  Sales associate 
39 Investigator 79 Marketing & 1 1 9  Senior lecturer 
Communications 
Adviser 
40 HR Advisor 80 Project Manager 120 Senior lecturer 
267 
Appendices 
Appendix 17 (continued) 
R d t 121 240 es pon en s -
12 1  Accounts/Debt controller 1 61 I ntern Architect 201 Sales Secretary 
1 22 Research ergonomics 1 62 Doctor 202 Partner 
1 23 Interviewer 1 63 Psychologist 203 Account Manager 
1 24 Senior lecturer, curriculum 1 64 Staff scientist 204 Lab technician 
leader, programme leader 
1 25 Pharmacist 1 65 Veterinarian 205 Manager 
1 26 Reception / Customer 1 66 General Manager 206 Manager 
Services 
1 27 Educator 1 67 Nurse educator 207 Sales/pre-sales 
1 28 Lecturer 1 68 Human Resources 208 Customer Relations 
Director Manager 
1 29 Evaluations Manager 1 69 Managing Director 209 sales assistant 
1 30 Operations Manager 1 70 Detachment 2 1 0  
Commander 
1 31 E learning manager 1 7 1  Psychologist 2 1 1  Account Manager 
1 32 Advisor 1 72 Massage therapist 2 12  Administration Assistant 
1 33 Coordinator 1 73 Appraisals 2 13  SEO 
Administrator 
1 34 Consultant 1 74 Publicity & Promotions 214  Chiropractor 
1 35 N/A 1 75 Scientist 2 15  Sales administrator 
1 36 Planner 1 76 Manag_er 216  Account manager 
1 37 IT Manager 1 77 HR Manager 2 17  Adjunct professor 
1 38 Researcher 1 78 Employment 218  e Leaming Advisor 
Consultant 
1 39 I ntermediate/Senior foreign 1 79 Technical Officer 219  ACCOUNTS 
exchange consultant ADMIN ISTRATION 
1 40 Customer Retention ManaQer 1 80 Technician 220 Front desk 
1 41 Marketing Manager 1 8 1  Commercial Traffic Co- 221 Administrator 
ordinator 
1 42 Staff member ( salesperson, 1 82 Ass!. Manager 222 Swimming Lesson 
secretarial work too) Instructor/ Lifeguard 
1 43 Surgeon 1 83 Registered Nurse 223 Support Officer 
1 44 Credit scoring analyst 1 84 Scientist 224 PA 
1 45 Claims officer 1 85 Teacher/Media 225 Regional On Premise 
Specialist Manager 
1 46 Claims officer 1 86 Teacher/Media 226 Advertising Co-ordinator 
Specialist 
1 47 Claims officer 1 87 Communications 227 Psychologist (that gives it 
coordinator away!) 
1 48 Claims officer 1 88 228 Personal Assistant 
1 49 Statistician 1 89 Analyst 229 Communications 
Manager 
1 50 Statistician 1 90 Designer 230 Social Worker 
1 51 Marketing Executive 1 91 New Business & 231  Director 
Marketing Manager 
1 52 Project Engineer 1 92 Subeditor 232 Legal Secretary 
1 53 it manager 1 93 Ride OperatorlTeam 233 Knowledge Manager 
Lead 
1 54 Account Manager 1 94 Supervisor 234 Registered Nurse 
1 55 $tatistician 1 95 Product Development 235 
Engineer 
1 56 Graphic Designer 1 96 Data Entry Operator 236 Training Manager 
1 57 Senior graphic designer/art 1 97 Copyeditor/book 237 Officer 
director designer 
1 58 Resident doctor 1 98 Executive assistant 238 Community Librarian 
(manager) 
1 59 IT Manager 1 99 Account Manager 239 Registrar/Director of 
Admissions 
1 60 Web developer 200 Corporate Sales 240 Planner 
Manager 
268 
Appendices 
Appendix 17 (continued) 
R d t 141 360 es �on en s -
241 PhD student / tutor 281 Supervisor 321 Researcher 
242 Corporate Human Resources 282 Librarian 322 Data Services Manager 
Advisor 
243 Parks Policy & Planning 283 Clinical psychologist 323 Project manager 
ManaQer 
244 Lecturer 284 Croupier/Dealer 324 Senior researcher 
245 Project Manager 285 325 Research fellow 
246 Prolect Coordinator 286 326 Research Scientist 
247 287 ReQional Manager 327 Research technician 
248 Manager 288 Clinical Nurse 328 Research Co-ordinator 
Specialist 
249 Parks Policy & Planning 289 Typist 329 Assistant research 
Advisor fellow/PhD student 
250 Sales Representative 290 Centre ManaQer 330 PhD student / technician 
251 Trade Research Executive 291 Health Educator 331 
252 Assistant Manager 292 Guest Services 332 Project manager 
Representative 
253 Corporate account support 293 Supervisor 333 Marketing Product 
Specialist 
254 Field Service Rep 294 Senior software 334 Senior lecturer 
enQineer 
255 Account Executive 295 Test Analyst 335 Teacher 
256 Lease Administrator/Assistant 296 Software Consultant 336 Doctor 
to Special Projects Manager 
257 Asst. Professor 297 Senior Engineer 337 Case manaQer 
258 Administrator 298 Account manager 338 Nurse ManaQer 
259 Change Director 299 339 Lecturer 
260 Supervisory Epidemiologist 300 Research assistant 340 Business Development 
manager 
261 Researcher 301 Senior Account Director 341 Secretary 
262 Director 302 Web Developer 342 Managing small national 
service 
263 Associate Dean 303 Accounting clerk 343 General Manager 
Operations 
264 Instructor 304 Lecturer 344 Research nurse 
265 PsycholoQist 305 AccountinQ technician 345 Teacher 
266 Clerk 306 Research 346 Accountant 
Assistant/Marker 
267 Senior Lecturer 307 Senior Accountant 347 Senior lecturer 
268 Pacific co-ordinator 308 Sales Representative 348 Hon. Ass. Research Fellow 
269 UnderwritinQ ManaQer 309 Network Engineer 349 Student Services Manager 
270 SUQervisor 31 0 Marketirlg Assistant 350 Senior Lecturer 
271 Data cleansing 31 1 PhD student (doing 351 Office Manager 
research) 
272 Administrative 31 2 Project Manager 352 Webmaster 
273 Account Manager 31 3 Chief EnQineer Officer 353 Research/ Admin istration 
274 Waitress 314 Administrator 354 Research fellow 
275 Head of Technical Training 31 5 Operations Manager, 355 Research assistant 
Project Office secretary Animal Ethics 
Committee 
276 Assistant office manaQer 31 6 Post-doctoral fellow 356 PA / Administrator 
277 Undergraduate Assistant 31 7 Duty Manager/ 357 WP operator 
Waitperson 
278 Operational/Managerial 318 358 Solicitor 
279 Stockroom manager/Human 31 9 Research administrator 359 Legal Secretary 
resources 
280 Human Resources 320 Office Tea Girl 360 Teacher 
269 
Appendices 
Appendix 17 (continued) 
R d t 361 412 es pon en s -
361 Group Business ManaQer 381 MarketinQ Co-ordinator 401 Cook 
362 Service co-ordinator (six 382 Senior Product 402 Police inspector 
services under one umbrella) Manager 
363 Solicitor 383 Secretary 403 Lecturer 
364 Sol icitor 384 Research 404 Member Services 
manager/officer Administrator 
365 Solicitor 385 Senior Lecturer 405 Director 
366 Administrative Assistant 386 Stockbroker 406 Personal Assistant 
367 Sol icitor 387 Reception 407 
368 Social worker 388 Commercial manager 408 HR Rep/Office 
Manager 
369 Departmental Administrator 389 I nternet developer 409 Sr. Network EnQineer 
370 Admin 390 Director Global Supply 410  Scientist 
Chain 
371 Psychologist 391 Clinical Cytogenetics 41 1 Insurance/Bil ling 
T echnoloQist 
372 Consultant / senior lecturer 392 Secretary 412  Ph.D. student 
373 Reception (Relief) 393 Dietician 
374 Senior Lecturer, Paediatrics 394 Operations manager 
375 Research Nurse 395 Lawyer 
376 Research technician 396 Teacher 
377 Quality Assurance Officer 397 Core faculty & director of 
fieldwork 
378 Clinical Director 398 Manager 
379 Senior research fellow 399 Staff physiotherapist 
380 WP Operator 400 Graduate Student 
270 
Appendices 
Appendix 18:  Indices of fit for the Internet Study 
Appendix 18a:  Fit indices for workplace friendship scale with items 6, 9 and 10 
removed (Random 2) 
Summary of models  
Model  
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
S aturated model 
Independence model 
Model 
----------------
Default model 
Saturated model  
Independence model  
Model 
----------------
Default model  
Saturated model 
Independence model  
NPAR CMIN OF 
1 9  
4 5  
9 
123 . 0 63 
0 . 000  
7 82 . 07 1  
2 6  
0 
3'6  
RMR GFI 
---------- ----------
0 . 07 8  0 . 8 7 8  
0 . 0 00  1 .  000  
0 . 3 62  0 . 4 0 1  
DELTA1 RH01 
NFI RFI 
---------- ----------
0 . 8 4 3  0 . 7 8 2  
1 .  0 00  
0 . 0 0 0  0 . 000  
PRATIO  PNFI 
---------- ----------
0 . 7 22  0 . 60 9  
0 . 0 00  0 . 0 00  
1 .  0 00  0 . 0 00  
NCP LO 90  
---------- ----------
97 . 0 6 3  6 6 . 2 4 6  
0 . 0 0 0  0 . 000  
7 4 6 . 07 1  658 . 8 0 6  
FMIN FO 
---------- ----------
0 . 62 8  0 . 4 9 5  
0 . 0 00  0 . 000  
3 . 9 90  3 . 8 0 6  
P CMIN/DF 
--------- ---------
0 . 000  4 . 7 33  
0 . 0 0 0  2 1 . 7 2 4  
AGFI  PGFI 
---------- ----------
0 . 7 8 8  0 . 507  
0 . 2 5 1  0 . 3 2 1  
DELTA2 RH02 
I FI TLI 
---------- ----------
0 . 8 7 2  0 . 82 0  
1 .  0 0 0  
0 . 0 00  0 . 0 00  
PCFI 
----------
0 . 6 2 8  
0 . 0 0 0  
0 . 0 0 0  
H I  9 0  
----------
1 35 . 4 1 9  
0 . 0 00  
8 4 0 . 7 5 1  
LO 90  HI  90  
---------- ----------
0 . 33 8  0 . 69 1  
0 . 0 0 0  0 . 0 00  
3 . 3 6 1  4 . 2 90 
Model RMSEA LO 90  HI  9 0  PCLOSE 
Default model 0 . 1 3 8  0 . 1 1 4  0 . 1 63 0 . 0 00  
Independence model 0 . 32 5  0 . 30 6  0 . 3 4 5  0 . 000  
27 1 
CFI 
--------
0 . 8 7 0  
1 .  0 0 0  
0 . 0 00  
Model ATC BCC 
---------------�  ---------- ----------
Default model 161 . 0 6 3  1 6 3 . 1 0 6  
Saturated model. 90 . 0 0 0  94 . 8 3 9  
Independence model 800 . 07 1  801 . 039  
Model ECVI LO 90  
---------------- ---------- ----------
Default 
Saturated 
Independence 
model 
model 
model 
0 . 8 2 2  f 0 . 665  
0 . 4 5 9 0 . 4 5 9  
4 . 0 8 2  3 . 637  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 62 7 3  
I ndependence model 1 3  1 5  
Appendices 
BIC CAIC  
---------- ----------
2 65 . 191  2 4 2 . 4 4 4  
3 3 6 . 619  2 82 . 7 4 4  
8 4 9 . 395  8 3 8 . 620  
HI  90  MECVI 
---------- ----------
1 .  017  0 . 8 32  
0 . 4 5 9  0 . 4 8 4  
4 . 5 65 4 . 0 8 7  
Appendix ISb: Fit indices for workplace friendship scale with items 3 ,  5 ,  6, 9 and 
1 0  removed (Random 2) 
Summary of models 
Model 
----------------
Default model 
Saturated model  
I ndependence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
NPAR 
15  
2 8  
7 
CMIN 
32 . 7 68  
0 . 000  
5 1 8 . 0 1 6  
RMR 
OF 
1 3  
0 
2 1  
GFI 
---------- ----------
0 . 0 5 5  
0 . 0 00  
0 . 3 7 1  
DELTAl 
NFI  
----------
0 . 937  
1 .  0 00  
0 . 0 00  
0 . 955  
1 .  0 00  
0 . 4 6 9 
RHOl 
RFI 
----------
0 . 8 98 
0 . 0 00  
272 
P CMIN/DF  
--------- ---------
0 . 002  2 . 52 1  
0 . 0 0 0  2 4 . 6 67 
AGFI PGFI 
---------- ----------
0 . 902  0 . 4 4 3  
0 . 2 9 2  0 . 3 52  
DELTA2 RH02 
I FI TLI CFI 
---------- ---------- --------
0 . 9 6 1  0 . 93 6  0 . 9 6 0  
1 . 00 0  1 .  0 0 0  
0 . 0 00  0 . 00 0  0 . 0 00  
Model 
---
---
---
---
---
-
Defaul t  model 
Saturated model 
Independence  model 
Model 
----------------
Default  model 
Saturated model 
Independence model  
Model 
----------------
Default  model 
Saturated model 
Independence model 
Model 
--------- -------
Default model 
Independence model  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model  
----------------
Default  model 
S"aturated model 
Independence model 
PRATIO PNFI 
---------- ----------
0 . 61 9  0 . 5 80  
0 . 000  0 . 0 00  
1 .  000  0 . 0 00  
NCP LO 90  
----------
----------
1 9 . 7 68 6 . 602  
0 . 000  0 . 000  
4 97 . 016  4 2 6 . 572  
FMIN FO 
----------
----------
0 . 1 67 0 . 1 0 1  
0 . 000  0 . 000  
2 . 64 3  2 . 53 6  
RMSEA LO 90  
----------
----------
0 . 088  0 . 0 51  
0 . 347  0 . 322  
AIC BCC 
----------
----------
62 . 7 68  64 . 0 4 5  
5 6 . 000  58 . 3 83  
532 . 0 1 6  532 . 6 1 1  
ECVI LO 90 
----------
----------
0 . 320 0 . 2 53  
0 . 2 8 6  0 . 2 8 6  
2 . 7 14  2 . 3 55  
HOELTER HOELTER 
Model  . 05 . 0 1 
Default  model 134  1 6 6  
Independence model 13  1 5  
273 
Appendices 
PCFI 
----------
0 . 5 9 4  
0 . 0 00  
0 . 0 00  
HI  90  
----------
4 0 . 603  
0 . 000  
5 7 4 . 8 7 8  
LO 90 HI 90 
----------
----------
0 . 03 4  0 . 207  
0 . 0 00  0 . 0 00  
2 . 17 6  2 . 933  
HI  90  PCLOSE 
----------
----------
0 . 1 2 6  0 . 0 4 7  
0 . 37 4  0 . 0 00  
BIC CAIC  
----------
----------
1 4 1 .  205  1 2 7 . 0 1 6  
2 0 2 . 4 1 5  1 7 5 . 930  
5 6 8 . 619  5 6 1 . 998  
HI  90  MECVI 
----------
----------
0 . 427  0 . 327  
0 . 2 8 6  0 . 2 98 
3 . 1 12  2 . 7 1 7  
Appendices 
Appendix 18c:  Fit indices for workplace friendship scale with items 3, 5, 6, 9 and 
10 removed (Whole Sample) 
Summary of models  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default  model 
Saturated model 
Independence model  
Model  
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default  model  
Saturated model 
Independence model  
Model 
----------------
Default model  
Saturated model 
I ndependence model 
NPAR CMIN  DF  
15  
28  
7 
3 9 . 3 97 
0 . 0 0 0  
1 03 7 . 0 6 2  
1 3  
0 
2 1  
RMR GFI 
---------- ----------
0 . 04 6  0 . 9 7 5  
0 . 000  1 .  000  
0 . 367  0 . 4 8 9  
DELTA1 RH01 
NFI RFI 
----------
----------
0 . 962  0 . 93 9  
1 .  0 0 0  
0 . 000  0 . 0 00  
PRATIO PNFI 
---------- ----------
0 . 619  0 . 5 96  
0 . 0 00  0 . 0 00  
1 .  000  0 . 0 00  
NCP LO 90  
----------
----------
2 6 . 397  1 1 . 2 0 9  
0 . 0 00  0 . 0 00  
10 1 6 . 062  9 1 4 . 3 1 7  
FMIN FO 
--
--
--
--
-- ----------
0 . 0 9 6  0 . 0 6 4  
0 . 000  0 . 00 0  
2 . 523  2 . 4 7 2  
Model RMSEA LO 90  
Default model  0 . 0 7 0  0 . 0 4 6  
Independence model 0 . 34 3  0 . 325  
274 
P CMIN/DF 
--------- ---------
0 . 000  3 . 03 1  
0 . 000  4 9 . 3 8 4  
AGFI PGFI 
---------- ----------
0 . 9 4 6  0 . 4 5 3  
0 . 3 1 9  0 . 367  
DELTA2 RH02 
I FI TLI 
---------- ----------
0 . 97 4  0 . 958  
1 .  000  
0 . 0 00  0 . 0 00  
PCFI 
----------
0 . 603  
0 . 000  
0 . 000  
HI  90  
----------
4 9 . 2 02  
0 . 000  
1 1 2 5 . 1 9 8 
LO 90  HI  90  
---------- ----------
0 . 027  0 . 1 2 0  
0 . 0 00  0 . 0 00  
2 . 2 25  2 . 7 3 8  
HI  9 0  PCLOSE 
0 . 0 96  0 . 0 83  
0 . 3 6 1  0 . 0 00  
CFI 
--------
0 . 97 4  
1 . 00 0  
0 . 0 00  
Model  AIC BCC 
---------------- ---------- ----------
Default model  69 . 397  6 9 . 9 92  
Saturated model 5 6 . 000  5 7 . 1 1 2  
Independence model  1 051 . 0 62 1 0 5 1 . 3 4 0  
Model  ECVI LO 90  
---------------- ---------- -----�----
Default 
Saturated 
Independence 
model 
model 
model  
0 . 1 6 9 0 . 1 3 2  
0 . 1 3 6  0 . 1 3 6  
2 . 557  2 . 3 1 0  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 234  2 8 9  
Independence model 13  1 6  
Appendices 
BIC CAIC 
---------- ----------
1 5 8 . 9 0 1  1 4 4 . 7 1 2  
2 23 . 07 4  1 9 6 . 5 8 9  
1 0 92 . 8 30  1 0 8 6 . 2 0 9  
HI  90  MECVI 
---------- ----------
0 . 22 4  0 . 1 7 0  
0 . 1 3 6  0 . 1 3 9  
2 . 8 23  2 . 5 5 8  
Appendix 18d:  Fit Indices for the Job Satisfaction scale with 4 ,  7 ,  9, 1 1, 12  and 1 5  
removed (Random 2) 
Summary of models  
Model 
----------------
Default model  
Saturated mode l  
Independence mode l  
Model  
----------------
Default mode l  
S aturated model  
Independence model 
Model  
----------------
Default model  
Saturated model  
Independence model  
Model  
----------------
Default model  
Saturated model  
Independence model  
NPAR CMIN DF  
19  
45  
9 
7 6 . 1 0 1  
0 . 0 00  
5 92 . 7 3 9  
2 6  
0 
3 6  
RMR GFI  
---------- ----------
0 . 173  0 . 9 1 9  
0 . 0 00 1 .  0 0 0  
0 . 94 8  0 . 4 6 9  
DELTA1 RH01 
NFI RFI  
---------- ----------
0 . 8 72  0 . 8 2 2  
1 .  0 0 0  
0 . 000  0 . 0 0 0  
PRATIO PNFI 
---------- ----------
0 . 722  0 . 62 9  
0 . 000  0 . 0 00  
1 . 000  0 . 0 0 0  
275 
P CMIN/DF  
--------- ---------
0 . 000  2 . 92 7  
0 . 0 00  1 6 . 4 65 
AGFI  PGFI 
---------- ----------
0 . 8 5 9  0 . 5 3 1  
0 . 33 7  0 . 37 5  
DELTA2 RH02 
IFI  TLI CFI  
---------- ---------- --------
0 . 9 1 2  0 . 8 7 5  0 . 91 0  
1 .  0 00  1 .  0 0 0  
0 . 0 00  0 . 0 00  0 . 00 0  
PCFI 
----------
0 . 657  
0 . 000  
0 . 000  
Appendices 
Model NCP .10 90 H I  9 0  
----------------
---------- ---------- ----------
Default model  50 . 1 0 1  2 7 . 694  8 0 . 1 4 2  
Saturated model 0 . 0 00  0 . 0 00  0 . 00 0  
Independence model 556 . 7 3 9  4 8 1 . 5 4 8  639 . 3 6 0  
, 
Model FMIN FO LO 9 0  H I  9 0  
----------------
---------- ----------
---------- --
--
--
--
--
Default model 0 . 3 8 8  0 . 2 5 6  0 . 1 4 1  0 . 4 0 9  
Saturated model 0 . 00 0  0 . 000  0 . 0 0 0  0 . 0 00  
Independence model 3 . 02 4  2 . 8 4 1  2 . 4 5 7  3 . 2 62 
Model RMSEA LO 90 HI 90  PCLOSE 
Default model 0 . 0 99  0 . 07 4  0 . 1 2 5  0 . 00 1  
I ndependence model 0 . 2 8 1  0 . 2 6 1  0 . 3 0 1  0 . 00 0  
Model 
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model  
AIC  BCC 
---------- ----------
1 1 4 . 1 0 1  1 1 6 . 1 4 4  
90 . 0 0 0  94 . 8 3 9  
6 1 0 . 7 3 9  611 . 7 0 6  
ECVI LO 90 
----------
----------
0 . 582  0 . 4 68 
0 . 4 5 9  0 . 4 5 9  
3 . 1 1 6  2 . 732  
HOELTER HOELTER 
Model . 05 . 01 
Default  model  1 0 1  1 1 8  
I ndependence model  1 7  20  
B I C  CAI C  
----------
----------
2 18 . 2 2 9  1 9 5 . 4 8 2  
33 6 . 6 1 9  2 8 2 . 7 4 4  
6 60 . 0 6 2  64 9 . 2 8 7  
HI  90  MECVI 
----------
----------
0 . 7 3 5  0 . 5 9 3  
0 . 4 5 9 0 . 4 8 4 
3 . 5 3 8  3 . 1 2 1  
Appendix 18e: Fit Indices for the Job Satisfaction scale with 2, 4, 7, 9, 11 ,  12 and 
15 removed (Random 2) 
Summary  of  models  
Model NPAR 
Default model 
Saturated model  
Independence model 
17  
36  
8 
CMIN 
4 5 . 830  
0 . 000  
4 7 4 . 4 90 
276 
OF P CMIN/DF  
---------
---------
1 9  0 . 0 0 1  2 . 4 12 
0 
2 8  0 . 000  1 6 . 9 4 6  
Appendices 
Model RMR GFI AGFI PGFI 
---------------- ---------- ---------- ---------- ----------
Default model 0 . 1 5 6  0 . 9 47  0 . 900  0 . 5 00  
Saturated model  0 . 0 00 1 .  000  
Independence model  0 . 9 32 0 . 523  0 . 3 8 6  0 . 4 0 7  
DELTA1 RHOl DELTA2 RH02 
Model NFI RFI I FI TLI  CFI  
---------------- ---------- ---------- ---------- ---------- --------
Default model  0 . 903 0 . 8 5 8  0 . 9 4 1  0 . 9 1 1  0 . 9 4 0  
Saturated model 1 . 000 1 .  0 0 0  1 . 0 0 0  
Independence model  0 . 000  0 . 000  0 . 0 00  0 . 0 0 0  0 . 0 0 0  
Model PRATIO PNFI PCFI 
---------------- ---------- ---------- ----------
Default model 0 . 67 9  0 . 6 13  0 . 63 8  
Saturated model  0 . 000 0 . 000  0 . 0 0 0  
Independence model  1 . 000  0 . 000  0 . 0 0 0  
Model NCP LO 90 HI  90  
---------------- ---------- ---------- ----------
Default model  2 6 . 8 30  1 0 . 7 2 9  5 0 . 6 2 1  
Saturated model 0 . 0 00 0 . 000  0 . 0 00  
Independence model 4 4 6 . 490  37 9 . 5 70  5 20 . 8 4 3  
Model E11IN FO LO 90  HI 90 
---------------- ---------- ---------- ---------- ----------
Default model 0 . 2 34 0 . 137  0 . 0 5 5  0 . 2 5 8  
Saturated model  0 . 000 0 . 000  0 . 0 0 0  0 . 0 00  
Independence model 2 . 4 21  2 . 27 8  1 . 937  2 . 657  
Model RMSEA LO  90 HI  90  PCLOSE 
Default model  0 . 0 85  0 . 054  0 . 1 1 7  0 . 0 3 5  
Independence model 0 . 2 85  0 . 2 63 0 . 3 08  0 . 0 00  
Model AIC BCC BIC  CAIC 
---------------- - --------- ---------- ---------- ----------
Default model  7 9 . 8 30 8 1 . 4 6 6  1 7 0 . 9 95  1 52 . 64 4  
Saturated model  7 2 . 0 00 7 5 . 4 65 2 6 5 . 0 5 5  2 2 6 . 1 9 5  
Independence model  4 9 0 . 490  4 9 1 . 2 60 5 3 3 . 3 9 1  5 2 4 . 7 5 6  
Model ECVI LO 90  HI  90  MECVI 
---------------- ---------- ---------- ---------- ----------
Default model  0 . 4 07 0 . 325  0 . 52 9  0 . 4 1 6  
Saturated model 0 . 3 67 0 . 367  0 . 3 6 7  0 . 3 8 5  
Independence model 2 . 5 03 2 . 1 61  2 . 8 8 2  2 . 50 6  
277 
Model 
Default model  
Independence model 
HOELTER HOELTER 
. 05 . 0 1 
1 2 9  155  
18  2 0  
AppenOlces 
Appendix 18f: Fit Indices for the Job Satisfaction scale with 2, 4, 7, 9, 1 1 ,  12 and 
15 removed (Whole sample) 
Summary of models 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model  
Independence model  
Model  
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default  model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
NPAR CMIN OF 
17  
36  
8 
4 8 . 4 1 9  
0 . 000  
8 7 1 . 1 2 6  
1 9  
0 
2 8  
RMR GFI 
----------
----------
0 . 1 1 0  0 . 97 1  
0 . 000  1 .  0 00  
0 . 827  0 . 558  
DELTA1 RH01 
NFI RFI 
----------
----------
0 . 9 4 4  0 . 91 8  
1 .  0 0 0  
0 . 000  0 . 0 00 
PRATIO  PNFI 
----------
----------
0 . 67 9  0 . 64 1  
0 . 0 00  0 . 000  
1 . 0 0 0  0 . 000  
NCP LO 90 
----------
----------
2 9 . 4 1 9  1 2 . 6 1 8  
0 . 000  0 . 0 00  
8 4 3 . 12 6 7 50 . 4 7 5  
FMIN FO 
----------
----------
0 . 1 1 8  0 . 07 2  
0 . 000  0 . 00 0  
2 . 1 2 0  2 . 05 1  
278 
P CMIN/DF  
---------
---------
0 . 000 2 . 54 8  
0 . 0 00  31 . 1 1 2  
AGFI PGFI 
----------
----------
0 . 9 45  0 . 5 1 3  
0 . 431  0 . 4 3 4  
DELTA2 RH02 
IFI TLI  CFI 
----------
----------
- -------
0 . 9 65 0 . 9 4 9  0 . 9 65  
1 . 000 1 .  0 00  
0 . 000 0 . 0 00  0 . 000  
PCFI 
----------
0 . 655 
0 . 000 
0 . 000 
HI 90 
----------
53 . 8 93 
0 . 000  
943 . 179  
LO 90 H I  90  
----------
----------
0 . 031  0 . 1 3 1  
0 . 000 0 . 0 00  
1 .  82 6 2 . 2 95  
Model 
Default model 
Independence model 
Model 
----------------
Default model 
Saturated model  
Independence model  
Model 
----------------
Default model 
Saturated model 
Independence model 
RMSEA LO 90 
0 . 0 61 0 . 0 4 0  
0 . 2 7 1  0 . 2 55  
AIC BCC 
---------- ----------
8 2 . 4 1 9  8 3 . 1 8 0  
7 2 . 000  7 3 . 6 12  
8 8 7 . 1 2 6  8 8 7 . 4 85  
ECVI LO 90 
---------- ----------
0 . 2 0 1  0 . 1 60 
0 . 17 5  0 . 17 5  
2 . 15 8  1 . 933  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 2 5 6  3 08  
Independence model 2 0  2 3  
H I  9 0  
0 . 0 8 3  
0 . 2 8 6  
SIC 
----------
1 8 6 . 1 2 7  
2 9 1 . 6 1 7  
935 . 930  
HI  9 0  
----------
0 . 2 60 
0 . 1 7 5  
2 . 4 0 2  
PCLOSE 
0 . 1 7 5  
0 . 0 00  
CAIC  
----------
1 67 . 7 7 6  
2 52 . 7 57 
927 . 2 9 4  
MECVI 
----------
0 . 2 02  
0 . 1 7 9  
2 . 1 5 9  
AppenOlces 
Appendix 18g: Fit indices for Organisational Commitment Questionnaire with all 
15 items (Random 1)  
Summary  o f  models 
Model 
----------------
Default model 
Saturated model 
Independence model  
Model  
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
NPAR 
30  
1 2 0  
1 5  
CMIN 
2 3 1 . 6 8 8  
0 . 0 00  
1 64 3 . 1 63 
RMR 
OF 
90  
0 
1 0 5  
GFI 
---------- ----------
0 . 1 3 6  0 . 872  
0 . 0 00  1 .  0 00  
1 . 1 12 0 . 2 60 
DELTA1 RH01 
NFI  RFI 
---------- ----------
0 . 8 5 9  0 . 8 35  
1 .  0 00  
0 . 0 0 0  0 . 000  
279 
P CMIN/DF  
--------- ---------
0 . 000  2 . 5 7 4  
0 . 0 0 0  1 5 . 64 9  
AGFI  PGFI 
---------- ----------
0 . 8 2 9  0 . 6 5 4  
0 . 1 5 4  0 . 2 2 7  
DELTA2 RH02 
I FI TLI CFI 
---------- ---------- --------
0 . 90 9  0 . 8 9 3  0 . 90 8  
1 .  0 0 0  1 . 00 0  
0 . 00 0  0 . 00 0  0 . 00 0  
Model PRATIO PNFI PCFI 
----------------
----------
---------- ----------
Default model � . 8 5 7  0 . 7 3 6  0 . 7 7 8  
Saturated model 0 . 0 00  0 . 000  0 . 000  
Independence model 1 .  000  0 . 0 00  0 . 000  
Model NCP LO 90  HI  90  
----------------
----------
- - --------
--
--
--
--
--
Default model  14 1 .  6 88  1 0 0 . 52 6  1 90 . 52 9 
Saturated model 0 . 0 00  0 . 0 00  0 . 000  
Independence model 1 538 . 1 63 1 4 1 0 . 6 09  1 67 3 . 107  
Model FMIN FO La 90  H I  9 0  
----------------
---------- ----------
----------
----------
Default model 1 .  0 8 3  0 . 6 62 0 . 4 7 0  0 . 8 9 0  
Saturated model 0 . 000  0 . 0 00  0 . 0 00  0 . 000  
Independence model 7 . 67 8  7 . 1 8 8  6 . 592  7 . 8 1 8  
Model RMSEA LO 90  H I  9 0  PCLOSE 
Default model 0 . 0 8 6  0 . 07 2  0 . 099  0 . 000  
Independence model 0 . 262  0 . 2 5 1  0 . 2 73  0 . 000  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default  model 
' Saturated model 
Independence model 
AIC BCC 
---------- ----------
291 . 6 88  2 9 6 . 5 3 6  
2 4 0 . 000  2 5 9 . 3 9 4  
1 673 . 1 63 1 6 7 5 . 5 8 7  
ECVI  La 90  
--
--
--
--
-- ----------
1 .  363  1 . 17 1  
1 . 12 1  1 . 12 1  
7 . 8 1 9  7 . 2 22  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 1 0 5  1 1 5  
Independence model 1 7  1 9  
280 
BIC CAIC  
---------- ----------
4 7 4 . 0 4 9  4 22 . 807  
9 69 . 4 4 3  7 6 4 . 4 7 7  
1 7 64 . 3 43  1 73 8 . 722  
HI  90  MECVI 
----------
--
--------
1 .  5 9 1  1 .  3 8 6  
1 . 1 2 1  1 .  2 1 2  
8 . 4 4 9  7 . 8 30  
Appendices 
Appendices 
Appendix ISh:  Fit indices for Organiiational Commitment Questionnaire with 
items 2, 10 and 1 1  deleted (Random 1) 
Summary of  models 
-----------------
Model NPAR CMIN OF P CMIN/DF  
---------------- --------- --------- ---------
Default mode l  2 4  104 . 654  54  0 . 0 00  1 . 9 3 8  
Saturated mode l  7 8  0 . 0 00  0 
Independence model 12  1 1 4 0 . 653  66  0 . 0 00  1 7 . 2 8 3  
Model RMR GFI  AGFI PGFI 
---------------- ---------- ---------- ---------- ----------
Default model  0 . 1 1 7  0 . 9 1 8  0 . 8 8 1  0 . 6 3 5  
Saturated model  0 . 000  1 . 0 0 0  
Independence model 1 .  0 6 9  0 . 3 1 8  0 . 1 9 4  0 . 2 6 9  
DELTA1 RH01 DELTA2 RH02 
Model NFI RFI I FI TLI  CFI 
---------------- ---------- ---------- ---------- ---------- --------
Default model 0 . 908  0 . 8 8 8  0 . 953  0 . 9 4 2  0 . 953  
Saturated model 1 .  000 1 .  0 00  1 .  0 00  
Independence model 0 . 00.0 0 . 0 00  0 . 0 00  0 . 0 0 0  0 . 0 0 0  
Model PRATIO PNFI PCFI 
---------------- ---------- ---------- ----------
Default model 0 , 8 1 8  0 . 7 4 3  0 . 7 8 0  
Saturated model 0 . 000  0 . 00 0  0 . 000  
Independence model 1 .  000  0 . 0 0 0  0 . 0 00  
Model NCP LO 9 0  H I  9 0  
---------------- ---------- ---------- ----------
Default model 50 . 654  2 5 . 4 8 8  8 3 . 61 7  
Saturated model 0 . 000  0 . 00 0  0 . 000  
Independence model 1 07 4 . 653  9 6 8 . 8 3 2  1 1 8 7 . 8 8 4  
Model FMIN FO  LO 90  HI  9 0  
---------------- ---------- ---------- ---------- ----------
Default model 0 . 4 8 9  0 . 2 3 7  0 . 1 1 9  0 . 3 9 1  
Saturated model  0 . 0 00  0 . 0 0 0  0 . 0 00  0 . 00 0  
Inc;:l.ependence model 5 . 3 30  5 . 02 2  4 . 52 7  5 . 5 5 1  
Model RMSEA LO 9 0  H I  9 0  PCLOSE 
Default model 0 . 0 66  0 . 0 4 7  0 . 0 85  0 . 0 8 0  
Independence model 0 . 2 7 6  0 . 2 62 0 . 2 9 0  0 . 0 0 0  
28 1 
Model AIC BCC 
---------------- ----------
----------
Default model 152 . 6 5 4  155 . 7 58  
Saturated model 156 . 0 0 0  1 6 6 . 0 9 0  
Independence model 1 1 6 4 . 65 3  1 1 66 . 2 0 6  
Model ECVI LO 90  
---------------- ----------
----------
Default 
Saturated 
Independence 
model 
model 
model 
0 . 7 1 3  0 . 5 9 6  
0 . 7 2 9  0 . 7 2 9  
5 . 4 4 2  4 . 9 4 8  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 1 4 8  1 6 6  
Independence model 1 7  1 8  
Appendices 
BIC  CAIC  
----------
----------
.2 9 3 . 1 8 7  257 . 5 4 9  
6 12 . 7 3 2  4 9 6 . 9 1 0  
1234 . 92 0  1 2 17 . 1 0 1  
HI  90  MECVI 
----------
----------
0 . 8 67  0 . 7 2 8  
0 . 7 2 9  0 . 7 7 6  
5 . 97 1  5 . 4 50 
Appendix 18i: Fit indices for Organisational Commitment Questionnaire with 
items 2, 10 and 11 deleted (Whole sample) 
Summary of mode ls  
Model NPAR CMIN DF 
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
I ndependence model 
Model 
----------------
Default model 
Saturated model 
I ndependence model 
Model 
----------------
De fault model 
Saturated model 
Independence model 
24  
78  
12 
1 3 6 . 757  
0 . 0 00  
2 53 1 . 7 72  
5 4  
0 
66  
RMR GFI 
------ ---- ----------
0 . 1 0 1  0 . 9 4 6  
0 . 0 0 0  1 . 0 00  
1 . 3 0 8  0 . 2 7 9  
DELTA1 RH01 
NFI RFI 
----------
----------
0 . 9 4 6  0 . 93 4  
1 .  0 0 0  
0 . 0 0 0  0 . 0 00  
PRATIO  PNFI 
---------- ----------
0 . 8 1 8  0 . 7 7 4  
0 . 0 00  0 . 000  
1 .  000  0 . 000  
282 
P CMIN/DF 
--------- ---------
0 . 00 0  2 . 5 33  
0 . 0 0 0  38 . 3 60 
AGFI PGFI 
---------- ----------
0 . 92 2  0 . 655  
0 . 1 4 8  0 . 2 3 6  
DELTA2 RH02 
IFI  TLI  CFI 
---------- ----------
--------
0 . 9 67 0 . 9 59  0 . 9 6 6  
1 .  0 0 0  1 .  0 00  
0 . 0 0 0  0 . 0 0 0  0 . 0 0 0  
PCFI 
----------
0 . 7 91 
0 . 0 0 0  
0 . 0 0 0  
, 
Model  NCP 10 90  HI  90  
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Default model 8 2 . 757  52 . 0 66  1 2 1 . 132  
Saturated model 0 . 0 00 0 . 0 00  0 . 0 00  
Independence model 2 4 6 5 . 7 72  2 30 4 . 938  2 633 . 94 0  
Model FMIN FO LO 90 HI 9 0  
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Default model 0 . 3 33 0 . 2 0 1  0 . 1 2 7  0 . 2 9 5  
Saturated model  0 . 000  0 . 00 0  0 . 000  0 . 0 00  
Independence model 6 . 1 60 5 . 9 99  5 . 608  6 . 4 0 9  
Model RMSEA LO 9 0  H I  9 0  PCLOSE 
Default  model 0 . 061  0 . 0 4 8  0 . 0 7 4  0 . 07 3  
Independence model  0 . 3 01  0 . 2 9 1  0 . 3 12  0 . 0 0 0  
Model AIC BCC 
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Default  model  1 8 4 . 7 57 1 8 6 . 325  
Saturated model 1 5 6 . 000 1 6 1 . 095  
Independence model 2 5 5 5 . 772  2 5 56 . 555  
Model  ECVI LO 90  
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Default 
Saturated 
Independence 
model 
model  
model  
0 . 450  0 . 3 7 5  
0 . 3 80  0 . 3 8 0  
6 . 218  5 . 827  
HOELTER HOELTER 
Model . 05 . 01 
Default  model  217  2 4 4  
Independence model  14  1 6  
BIC CAIC  
- - - - - - - - - - - - - - - - - - - -
3 4 0 . 8 9 9  305 . 2 6 2  
6 63 . 4 63 5 4 7 . 6 4 0  
2 633 . 8 4 3  2 6 1 6 . 0 2 4  
HI  90  MECVI 
- - - - - - - - - - - - - - - - - - - -
0 . 5 43  0 . 4 5 3  
0 . 380  0 . 3 9 2  
6 . 62 8  6 . 22 0  
Appendices 
Appendix 18j :  Fit indices for the cohesion scale with all 9 items (Random 1)  
Summary of models 
Model 
- - - - - - - - - - - - - - - -
Default model 
Saturated model  
Independence model 
NPAR 
1 9  
4 5  
9 
CMIN 
8 8 . 327  
0 . 000  
7 2 6 . 6 4 3  
283 
OF P CMIN/OF  
- - - - - - - - - - - - - - - - - -
26  0 . 0 00  3 . 3 9 7  
0 
3 6  0 . 0 00  2 0 . 1 8 5  
Model 
----------------
Default model 
Saturated model  
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
Default model 
Independence model 
Model  
----------------
Default model  
Saturated model 
Independence model  
Model  
--
-
--
--
-
--
--
--
--
Default model 
Saturated model 
Independence model  
, 
RMR 
----------
0 . 057  
0 . 000  
0 . 2 8 8  
DELTA1 
NFI 
----------
0 . 8 7 8  
1 .  000  
0 . 000  
PRATIO 
----------
0 . 7 22  
0 . 000  
1 .  000  
NCP 
----------
62 . 327  
0 . 0 00  
690 . 6 43  
FMIN 
----------
0 . 4 1 3  
0 . 0 00  
3 . 3 96  
RMSEA 
0 . 1 0 6  
0 . 2 99  
AIC  
----------
12 6 . 327  
90 . 0 00  
7 4 4 . 64 3  
ECVI 
----------
0 . 5 9 0  
0 . 4 2 1  
3 . 4 8 0  
GFI 
----------
0 . 9 10  
1 .  0 00  
0 . 4 4 9  
RH01 
RFI 
----------
0 . 832  
0 . 0 00  
PNFI 
----------
0 . 63 4  
0 . 0 00  
0 . 000  
LO 90  
----------
37 . 4 8 9  
0 . 0 00  
60 6 . 7 3 8  
FO 
----------
0 . 2 9 1  
0 . 0 00  
3 . 22 7  
LO 9 0  
0 . 0 82  
0 . 2 8 1  
BCC 
----------
1 2 8 . 1 90 
94 . 4 1 2 
7 4 5 . 5 25  
LO 90  
----------
0 . 4 7 4  
0 . 4 2 1  
3 . 0 8 8  
284 
AGFI 
- ---------
0 . 8 4 4  
0 . 3 12  
DELTA2 
I FI 
----------
0 . 9 1 1  
1 .  0 00  
0 . 000  
PCFI 
- ---------
0 . 657  
0 . 0 00  
0 . 000  
HI  90  
----------
94 . 7 6 4  
0 . 000  
7 8 1 . 965  
LO 90  
----------
0 . 1 7 5  
0 . 000  
2 . 8 35  
HI  9 0  
0 . 1 3 1  
0 . 3 1 9  
BIC  
----------
232 . 1 1 6  
3 40 . 5 5 4  
7 9 4 . 753  
HI  90  
----------
0 . 7 4 2 
0 . 4 2 1  
3 . 9 0 6  
Appendices 
PGFI 
----------
0 . 52 6  
0 . 3 5 9  
RH02 
TLI 
----------
0 . 8 7 5  
0 . 0 00  
HI 90 
----------
0 . 4 4 3  
0 . 0 00  
3 . 654  
PCLOSE 
0 . 0 00  
0 . 0 00  
CAIC  
----------
209 . 3 6 9  
28 6 . 67 9  
7 83 . 9 7 8  
MECVI 
----------
0 . 5 99  
0 . 4 4 1  
3 . 4 8 4  
CFI 
--------
0 . 9 1 0  
1 .  0 00  
0 . 0 00  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 95  1 1 1  
Independence model 1 6  1 8  
Appendices 
Appendix 18k: Fit indices for cohesion scale with item 2 removed (Random 1) 
Summary  o f  models  
Model NPAR CMIN OF 
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
De fault model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
1 7  
36  
8 
33 . 1 7 1  
0 . 0 00  
633 . 633  
19  
0 
2 8  
RMR GFI 
---------- ----------
0 . 0 37  0 . 9 59  
0 . 000  1 : 0 00  
0 . 3 04  0 . 4 5 6  
DELTA1 RH01 
NFI RFI 
---------- ----------
0 . 9 4 8  0 . 923  
1 .  0 00  
0 . 000  0 . 000  
PRATIO PNFI 
---------- ----------
0 . 67 9  0 . 6 4 3  
0 . 000  0 . 000  
1 .  000  0 . 0 00  
NCP LO 90  
---------- ----------
1 4 . 17 1  1 .  9 4 7  
0 . 000  0 . 000  
605 . 633  527 . 4 1 4  
FMIN FO 
---------- ----------
0 . 155  0 . 0 6 6  
0 . 000  0 . 000  
2 . 9 6 1  2 . 8 30  
285 
P 
---------
0 . 02 3  
0 . 0 00  
AGFI  
----------
0 . 923  
0 . 3 0 1  
DELTA2 
I F! 
----------
0 . 97 7  
1 .  0 00  
0 . 0 00  
PCFI 
----------
0 . 6 63  
0 . 0 0 0  
0 . 0 00  
H I  9 0  
----------
34 . 22 9  
0 . 0 0 0  
6 9 1 . 2 7 1  
LO 9 0  
----------
0 . 00 9  
0 . 0 00  
2 . 4 6 5  
CMIN/DF  
---------
1 . 7 4 6  
22 . 630  
PGFI 
----------
0 . 50 6  
0 . 3 55  
RH02 
TLI CFI  
---------- --------
0 . 9 6 6  0 . 9 7 7  
1 .  0 0 0  
0 . 0 00  0 . 0 0 0  
H I  9 0  
----------
0 . 1 60 
0 . 00 0  
3 . 2 30  
Model 
Default  model  
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model  
Model 
----------------
Default  model  
Saturated model 
Independence model 
RMSEA 
0 . 0 59  
0 . 318  
AIC 
----------
67 . 1 7 1  
7 2 . 000  
64 9 . 633 
ECVI 
----------
0 . 314  
0 . 33 6  
3 . 0 36  
. LO 90  
0 . 022  
0 . 2 9 7  
BCC 
----------
68 . 6 63  
7 5 . 1 61 
6 50 . 3 3 5  
LO 90  
----------
0 . 2 5 7  
0 . 3 3 6  
2 . 67 0  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 195  2 3 4  
Independence model 14  1 7  
HI 9 0  
0 . 0 92 
0 . 3 4 0  
BIC 
----------
1 5 9 . 822  
2 6 8 . 203  
6 93 . 233  
HI  90  
----------
0 . 4 0 8 
0 . 3 3 6  
3 . 4 3 6  
PCLOSE 
0 . 2 99  
0 . 000  
CAIC  
----------
1 4 1 .  4 7 2  
2 2 9 . 3 4 3  
68 4 . 598  
MECVI 
----------
0 . 32 1  
0 . 3 51  
3 . 0 3 9  
Appendices 
Appendix 181: Fit indices for cohesion scale with item 2 removed (Whole Sample) 
Summary of  models  
Model 
----------------
Default  model 
Saturated model 
Independence model  
Model 
----------------
Default model 
Saturated model 
Independence model 
NPAR 
1 7  
3 6  
8 
CMIN  
5 9 . 7 9 9  
0 . 000  
1 3 4 7 . 4 0 1  
RMR 
OF  
1 9  
0 
2 8  
GFI  
---------- ----------
0 . 042  0 . 9 62 
0 . 000  1 .  0 0 0  
0 . 333  0 . 4 3 2  
286 
P CMIN/DF  
--------- ---------
0 . 000  3 . 1 4 7  
0 . 000  4 8 . 12 1  
AGFI PGFI 
---------- ----------
0 . 92 9  0 . 508  
0 . 2 7 0  0 . 3 3 6  
Model  
----------------
Default model 
Saturated mode l  
Independence model .  
Model 
----------------
Default model  
Saturated model  
Independence model 
Model 
----------------
Default model  
Saturated model  
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model  
Model 
----------------
Default model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
I ndependence model 
DELTA1 RH01 
NFI RFI 
----------
----------
0 . 95 6  0 . 9 3 5  
1 .  0 0 0  
0 . 000  0 . 0 0 0  
PRATIO PNFI 
---------- ----------
0 . 67 9  0 . 6 4 8  
0 . 0 00  0 . 0 00  
1 .  000  0 . 0 0 0  
NCP LO 9 0  
---------- ----------
4 0 . 7 9 9  2 1 . 1 7 1  
0 . 000  0 . 00 0  
13 1 9 . 4 0 1 1 202 . 92 5  
FMIN FO 
----------
----------
0 . 1 4 5  0 . 0 9 9  
0 . 000  0 . 0 0 0  
3 . 2 7 8  3 . 2 1 0  
RMSEA LO 9 0  
---------- ----------
0 . 072  0 . 0 5 2  
0 . 339  0 . 3 2 3  
AIC BCC 
---------- ----------
93 . 7 99  94 . 5 60 
7 2 . 000  7 3 . 6 1 2  
1 3 63 . 4 0 1  1 3 63 . 7 5 9  
ECVI LO 9 0  
--
--
--
--
--
----------
0 . 228  0 . 1 8 0  
0 . 17 5  0 . 1 7 5  
3 . 317  3 . 0 3 4  
HOELTER HOELTER 
Model . 05 . 0 1 
Default model 208  2 4 9  
Independence model 13  1 5  
287 
Appendices 
DELTA2 RH02 
IF! TLI  CFI  
----------
------- ---
--------
0 . 969  0 . 954  0 . 9 6 9  
1 .  000  1 .  0 0 0  
0 . 000  0 . 0 0 0  0 . 0 00  
PCFI 
----------
0 . 658  
0 . 000  
0 . 000  
HI  90  
----------
6 8 . 0 4 1  
0 . 000  
1 4 4 3 . 2 58  
LO 90  HI  90  
---------- ----------
0 . 0 52  0 . 1 6 6  
0 . 0 00  0 . 0 00  
2 . 92 7  3 . 5 1 2  
H I  9 0  PCLOSE 
----------
----------
0 . 0 9 3  0 . 0 3 6  
0 . 3 5 4  0 . 00 0  
BIC CAIC 
----------
----------
1 9 7 . 507  17 9 . 1 5 7  
2 9 1 . 6 1 7  2 52 . 7 57  
1 4 1 2 . 2 0 5  1 4 03 . 5 6 9  
HI  90  MECVI 
---------- ----------
0 . 2 9 5  0 . 2 3 0  
0 . 1 7 5  0 . 17 9  
3 . 6 1 9  3 . 3 1 8  
Appendices 
Appendix I8m: Fit indices for the Needs Assessment Questionnaire (Random 2) 
Summary of models  
Model NPAR CMIN DF P CMIN/ DF 
---------------- --------- --------- ---------
Default model 3 4  1 4 1 .  690  7 1  0 . 0 00  1 .  9 9 6  
Saturated model 105  0 . 000  0 
Independence model 1 4  595 . 8 0 0  9 1  0 . 0 00  6 . 5 4 7  
Model RMR GFI AGFI  PGFI 
---------------- ---------- ---------- ---------- ----------
Default model 0 . 0 4 7  0 . 9 1 1  0 . 8 6 9  0 . 6 1 6  
Saturated model 0 . 0 0 0  1 .  000  
Independence model 0 . 1 3 5  0 . 659  0 . 6 0 6  0 . 57 1  
DELTA1 RH01 DELTA2 RH02 
Model NFI  RFI I FI TLI  
---------------- ---------- ---------- ---------- ----------
Default model 0 . 7 62 0 . 6 95  0 . 8 6 5  0 . 8 2 1  
Saturated model  1 .  000  1 .  000  
Independence model 0 . 0 00  0 . 0 00  0 . 000  0 . 0 00  
Model PRATIO PNFI PCFI 
---------------- ---------- ---------- ----------
Default model 0 . 7 8 0  0 . 5 95  0 . 67 1  
Saturated model 0 . 0 00  0 . 0 00  0 . 000  
Independence model 1 .  0 0 0  0 . 0 00  0 . 000  
Model NCP LO 90  HI  90  
---------------- ---------- ---------- ----------
Default  model 7 0 . 6 90  4 0 . 618  1 0 8 . 5 4 7  
Saturated model  0 . 0 0 0  0 . 000  0 . 000  
Independence model 50 4 . 8 0 0  4 3 1 .  3 3 2  5 85 . 7 5 6  
Model FMIN FO LO 90  HI 90 
---------------- ---------- ---------- ---------- ----------
Default model 0 . 72 3  0 . 3 6 1  0 . 2 07  0 . 5 5 4  
Saturated model 0 . 00 0  0 . 0 00  0 . 0 00  0 . 0 0 0  
Independence model 3 . 0 4 0  2 . 57 6  2 . 2 0 1  2 . 9 8 9  
Model RMSEA LO  90  HI  90  PCLOSE 
Default model 0 . 07 1  0 . 0 54  0 . 0 8 8  0 . 0 2 3  
Independence model 0 . 1 6 8  0 . 1 56  0 . 1 8 1  0 . 0 00  
288 
CFI 
--------
0 . 8 60 
1 .  000  
0 . 0 00  
Model AIC BCC 
----------------
----------
----------
Default model 2 0 9 . 690  2 15 . 32 6  
Saturated model 2 1 0 . 000  2 27 . 4 03  
Independence model 623 . 8 00 626 . 1 2 1  
Model ECVI LO 90 
---------------- ---------- ----------
Default 
Saturated 
Independence 
model 
model 
model 
1 .  070  0 . 9 1 6  
1 . 071  1 . 07 1  
3 . 183  2 . 8 0 8  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 127  1 4 1  
Independence model 38  42  
Appendices 
BIC  CAIC  
----------
----------
4 1 1 . 0 4 7  355 . 3 1 9  
8 3 1 . 837  6 59 . 7 3 6  
7 0 6 . 7 l2 6 83 . 7 6 5  
H I  9 0  MECVI 
- ---------
----------
1 .  2 63  1 .  0 9 9  
1 .  0 7 1  1 . 1 6 0  
3 . 5 96  3 . 1 9 4  
Appendix 1 8n:  Fit indices for the Needs Assessment Questionnaire (Whole 
sample) 
Summary of models  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model  
Independence model 
Model 
----------------
Default model  
Saturated model 
Independence model 
Model 
NPAR CMIN  OF  
3 4  
1 05  
1 4  
2 l6 . 3 67 
0 . 0 00  
1203 . 5 35  
7 l  
0 
91  
RMR GFI 
----------
----------
0 . 0 43  0 . 9 3 1  
0 . 000  1 .  000  
0 . 1 3 4  0 . 6 4 7  
DELTA1 RH01 
NFI RFI 
---------- ----------
0 . 820  0 . 7 7 0  
1 .  000  
0 . 000  0 . 0 00  
PRATIO PNFI 
---------- ----------
0 . 7 8 0  0 . 64 0  
0 . 000  0 . 00 0  
1 . 000  0 . 00 0  
NCP LO 90  
289 
P CMIN/DF  
---------
---------
0 . 0 00  3 . 0 4 7  
0 . 000  1 3 . 2 2 6  
AGFI PGFI 
----------
----------
0 . 8 98 0 . 6 30  
0 . 593  0 . 5 6 1  
DELTA2 RH02 
I FI TLI  CFI  
----------
----------
--------
0 . 8 7 2  0 . 8 3 3  0 . 8 6 9  
1 . 00 0  1 .  0 0 0  
0 . 000  0 . 0 0 0  0 ; 0 0 0  
PCFI 
- ---------
0 . 67 8  
0 . 0 0 0  
0 . 0 00  
H I  90  
----------------
----------
---------- ---------.-
Default model 1 4 5 . 367  10 4 . 8 8 4  1 9 3 . 4 7 9  
Saturated model 0 . 00 0  0 . 000  0 . 0 00 
Independence model  1 1 1 2 . 5 3 5  1 004 . 2 56  1 2 2 8 . 232  
Model FMIN FO LO 90 
---------------- --
--
--
--
--
---------- ----------
Default model 0 . 52 6  0 . 3 54 0 . 2 55  
Saturated model 0 . 00 0  0 . 0 00 0 . 0 00  
Independence model 2 . 92 8  2 . 707 2 . 4 4 3  
Mode l RMSEA LO 9 0  H I  90 
De fault model 0 . 0 7 1  0 . 0 60 0 . 0 8 1  
Independence model 0 . 1 7 2  0 . 1 64 0 . 1 8 1  
Model AIC BCC 
---------------- ---------- ----------
Default model 2 8 4 . 367  2 8 6 . 9 4 3  
Saturated model 2 1 0 . 0 00  217 . 955 
Independence model  1 2 3 1 . 535  1 2 32 . 5 96  
Model ECVI LO 90 
---------------- ---------- ----------
Default 
Saturated 
Independence 
model 
model 
model 
0 . 692  0 . 593  
0 . 51 1  0 . 51 1  
2 . 9 9 6  2 . 7 33 
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 1 75  1 9 4  
Independence model 4 0  4 3  
290 
BIC  
----------
5 10 . 8 1 0  
9 0 9 . 3 0 8  
1 3 2 4 . 7 7 6  
HI  90  
----------
0 . 8 0 9  
0 . 5 1 1  
3 . 27 8  
HI 9 0  
----------
0 . 4 7 1  
0 . 0 00  
2 . 98 8  
PCLOSE 
0 . 0 0 1  
0 . 000  
CAIC 
----------
4 5 5 . 082  
7 37 . 207  
1 3 0 1 . 830  
MECVI 
----------
0 . 698  
0 . 530  
2 . 9 9 9  
Appendices 
Appendices 
.. 
Appendix 19:  Regression weights for causal model: Internet Study 
Maximum Likelihood Estimates 
Regression Weights : 
Extrinsic JS < - cohesion workload sh 
Extrinsic JS < - cohesion_social supp 
Extrinsic  JS < - - - - - - - - - - - - - - - - - - WFO 
Intrinsic  JS < - - - - - - - - - Extrins ic JS 
OCQ < - - - - - - - - - - - - - - - - - - Extrinsic  JS 
OCQ < - - - - - - - - - - - - - - - - - - Intrinsic JS 
WFP < - - - - - - - - - - - - - - - - - - - - - - - - - - - WFO 
ITL < - - - - - - - - - - - - - - - - - - - - - - - - - - - OCQ 
ITL < - - - - - - - - - - - - - - - - - - Intrins ic JS 
Standardized Regression Weights : 
Extrinsic JS < - cohes ion workload sh  
Extrinsic  JS  < - cohes ion_social supp 
Extrinsic  JS < - - - - - - - - - - - - - - - - - - WFO 
Intrinsic  JS <- - - - - - - - - Extrinsic JS 
OCQ < - - - - - - - - - - - - - - - - - - Extrinsic  JS 
OCQ < - - - - - - - - - - - - - - - - - - Intrinsic JS 
WFP < - - - - - - - - - - - - - - - - - - - - - - - - - - - WFO 
ITL < - - - - - - - - - - - - - - - - - - - - - - - - - - - OCQ 
ITL <- - - - - - - - - - - - - - - - - - Intrinsic  JS 
Covariances :  
cohesion social <> cohesion workloa 
WFO <- - - - - - > cohesion_social support WFO <- - - - - - > cohesion workload share 
Correlations : 
cohesion social <> cohesion workloa 
WFO < - - - - - - > cohesion_social support WFO < - - - - - - > cohesion workload share 
29 1 
Estimate 
-- ---- - -
0 . 331  
0 . 735  
0 . 58 9  
0 . 503  
0 . 5 8 8  
0 . 4 7 2  
0 . 969  
- 0 . 5 55  
-0 . 530  
Est imate 
0 . 2 4 0  
0 . 4 1 7  
0 . 2 52  
0 . 57 2  
0 . 520  
0 . 3 67 
0 . 5 5 9  
-0 . 3 7 6  
- 0 . 27 9  
Estimate  
- - - - - - - -
0 . 4 52  
0 . 2 15 
0 . 1 0 6  
Estimate 
0 . 63 6  
0 . 5 1 3  
0 . 1 9 7  
S . E .  C . R .  
- - - - - - - -- --- - -
0 . 0 9 9  3 . 3 36  
0 . 1 5 7  4 . 68 8  
0 . 1 5 5  3 . 7 93 
0 . 0 62 8 . 102  
0 . 0 7 5  7 . 8 92 
0 . 07 5  6 . 283  
0 . 1 0 2  9 . 4 7 7  
0 . 0 9 7  - 5 . 7 4 0  
0 . 1 3 0  - 4 . 07 9  
S . E .  C . R .  
- --- - -- - - - - - --
0 . 0 4 9  9 . 2 8 3  
0 . 02 9  7 . 4 3 5  
0 . 03 3  3 . 2 3 8  
Appendix 20: Fit indices for causal model: Internet Study 
Summary of models  
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
I ndependence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model  
Saturated model  
Independence model 
Model 
----------------
Default model  
Saturated model  
Independence model 
NPAR CMIN DF  
---------
8 4  1301 . 3 8 7  5 8 2  
o 
630  
6 6 6  0 . 0 00  
36  8014  . 6 14  
RMR 
----------
0 . 1 38  
0 . 0 00  
0 . 7 7 0  
DELTA1 
NFI 
----------
0 . 8 3 8  
1 .  0 00  
0 . 0 00  
PRATIO 
----------
0 . 924  
0 . 0 00  
1 .  0 00  
NCP 
----------
7 1 9 . 3 8 7  
0 . 0 00  
7 3 8 4 . 61 4  
FMIN 
- ---------
3 . 1 6 6  
0 . 0 00  
1 9 . 500  
GFI  
----------
0 . 8 4 7  
1 .  0 00  
0 . 2 3 5  
RHOl 
RFI 
----------
0 . 82 4  
0 . 0 0 0  
PNFI  
----------
0 . 7 7 4  
0 . 0 00  
0 . 00 0  
LO 90  
----------
618 . 6 4 1  
0 . 0 00  
7 0 9 9 . 1 8 7  
FO 
----------
1 .  7 5 0  
0 . 000  
1 7 . 9 67 
Model  RMSEA LO 90 
Default model  0 . 05 5  0 . 051  
Independence model 0 . 1 6 9  0 . 1 6 6  
292 
P 
0 . 0 00  
0 . 000  
AGFI 
----------
0 . 8 2 5  
0 . 1 92  
DELTA2 
I FI 
----------
0 . 903  
1 .  000  
0 . 000  
PCFI 
----------
0 . 8 3 4  
0 . 0 00  
0 . 000  
H I  9 0  
----------
827 . 8 4 4  
0 . 0 00  
7 67 6 . 4 9 4  
LO 90 
----------
1 . 505  
0 . 000  
1 7 . 273  
H I  90  
0 . 059  
0 . 17 2  
CMIN/DF 
2 . 2 3 6  
1 2 . 7 22  
PGFI 
----------
0 . 7 4 0  
0 . 2 23  
RH02 
TLI 
----------
0 . 8 9 5  
0 . 000  
HI  90  
----------
2 . 0 1 4  
0 . 0 00  
1 8 . 6 7 8  
PCLOSE 
0 . 023  
0 . 000  
Appendices 
CFI  
--------
0 . 9 03  
1 .  0 0 0  
0 . 0 0 0  
Model AIC BCC 
---------------- ---------- ----------
Default  model 1 4 69 . 3 8 7  1 4 8 6 . 0 07  
Saturated model  1 3 32 . 000  1 4 63 . 7 7 5  
Independence model 8 0 8 6 . 6 1 4  8 0 93 . 7 37  
Model ECVI LO 90  
---------------- ---------- ----------
Default  
Saturated 
Independence 
model 
model 
model 
3 . 57 5  3 . 3 30  
3 . 2 4 1  3 . 2 4 1  
1 9 . 675  1 8 . 9 81  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Defaul t  model 2 02 2 1 0  
Independence model 3 6  3 7  
293 
Appendices 
BIC  CAIC  
---------- ----------
2 1 0 8 . 1 69  1 8 9 1 . 153  
63 9 6 . 625  4 67 6 . 002  
8 3 60 . 3 7 8  8 2 67 . 37 1  
HI  90  MECVI 
---------- ----------
3 . 8 3 9  3 . 6 1 6  
3 . 2 4 1  3 . 5 6 1  
2 0 . 3 8 6  1 9 . 693  
Model AIC BCC 
---------------- ---------- ----------
Default model 1 610 . 2 3 6  1 62 6 . 3 35  
Saturated model 1 4 0 6 . 000  1 537 . 5 9 6  
Independence model '  9059 . 8 68  9 0 6 6 . 7 9 4  
Model ECVI LO 90  
---------------- ---------- ----------
Default  
Saturated 
Independence 
model 
model 
model 
3-. 627  3 . 38 6  
3 . 1 67 3 . 1 67 
20 . 4 05  1 9 . 7 23  
HOELTER HOELTER 
Model . 0 5 . 0 1 
Default model 209  2 17  
Independence model 3 6  3 8  
295 
Appendices 
BIC CAIC 
---------- ----------
2 27 3 . 2 10  20 4 8 . 6 7 1  
6 8 25 . 4 22  4 9 8 9 . 94 6 
93 4 5 . 1 0 1  9 2 4 8 . 4 97 
HI  90  MECVI 
---------- ----- -----
3 . 8 8 4  3 . 6 63 
3 . 1 67 3 . 4 63 
2 1 . 102  2 0 . 4 2 1  
Appendices 
Appendices for the Invariance Testing Study (Chapter 5) 
Appendix 2 1 :  Fit indices for the causal model: Full data set (n=445) 
Summary of models 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model  
- ---------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
Independence model 
Model 
----------------
Default model 
Saturated model 
I ndependence model 
NPAR CMIN DF 
---------
8 6  1 4 3 8 . 2 3 6  6 1 7  
o 
6 6 6  
7 0 3  0 . 0 00  
37  8 9 8 5 . 8 68 
RMR GFI  
---------- ----------
0 . 1 3 1  0 . 8 4 8  
0 . 000  1 . 0 00  
0 . 7 90 0 . 2 2 4  
DELTAl RHOl 
NFI  RFI  
---------- ----------
0 . 8 4 0  0 . 8 2 7  
1 .  000  
0 . 0 00  0 . 0 0 0  
PRATIO  PNF I  
---------- ----------
0 . 92 6  0 . 7 7 8  
0 . 0 00  0 . 0 0 0  
1 .  0 0 0  0 . 0 0 0  
NCP LO 9 0  
-
-
-
-
-
-
-
-
-
-
----------
821 . 2 3 6  7 1 4 . 4 7 4  
0 . 0 00  0 . 00 0  
8 3 19 . 8 6 8  8 0 1 7 . 0 8 8  
FMIN FO 
----------
--
--
--
--
-
-
3 . 2 3 9  1 .  8 5 0  
0 . 00 0  0 . 0 0 0  
2 0 ·. 2 3 8  1 8 . 7 3 8  
P CMIN/DF 
0 . 000  2 . 3 3 1  
0 . 000  1 3 . 4 92 
AGFI PGFI 
----------
-
-
-
-
-
-
-
-
-
-
0 . 8 2 6  0 . 7 4 4  
0 . 1 8 1  0 . 2 1 2  
DELTA2 RH02 
IFI  TLI  
----------
----------
0 . 902  0 . 8 9 3 
1 . 00 0  
0 . 0 0 0  0 . 0 0 0  
PCFI 
----------
0 . 8 3 5  
0 . 0 00  
0 . 0 0 0  
H I  9 0  
----------
9 35 . 687  
0 . 0 0 0  
8 62 9 . 0 9 1  
LO 90  HI  9 0  
---------- ----------
1 .  6 0 9  2 . 1 0 7  
0 . 0 0 0  0 . 00 0  
1 8 . 0 5 7  1 9 . 4 3 5  
Model RMSEA LO 9 0  H I  9 0  PCLOSE 
Default model 0 . 0 5 5  0 . 0 5 1  0 . 05 8  0 . 0 1 7  
Independence model 0 . 1 68 0 . 1 65 0 . 1 7 1  0 . 0 0 0  
294 
CFI  
--------
0 . 9 0 1  
1 . 00 0  
0 . 00 0