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. 
 



The Extent and Measurement of 
VCR Time Shifting 

A thesis presented in partial fulfilment 
of the requirements for the degree of 

Master of Business Studies 
in Marketing 

at Massey University 

Penelope Jane Turner 

1997 



ii 

ABSTRACT 

The validity and accuracy of television programme ratings are critical to media planners 

and broadcasters alike. Media planners use ratings to select programmes that will satisfy 

certain reach and frequency objectives, which in tum contribute to an advertising 

campaign's overall goals. Broadcasters deliver audiences to advertisers, and base 

programme scheduling and pricing decisions on ratings data. 

At present, ratings are delivered within 24 hours of viewing and do not include 

adjustments for time shift viewing. Time shifting occurs when a television programme is 

videotaped and replayed at a later date. Given that VCR penetration in New Zealand has 

increased to over 75 percent of households, it is clear that programme ratings may be 

higher than the current overnight ratings suggest. 

This thesis explored the extent and measurement of time shifting in New Zealand. More 

specifically, it used AGB McNair's people meter data to examine: the scale of time 

shifting, the current methods of measuring time shifting, and future methods of 

estimating time shift viewing. The study aimed to identify whether patterns of time 

shifting behaviour exist, and whether these patterns could be used to model more 

inclusive overnight ratings. 

The findings suggest that, although the overall effect of time shifting on programme 

ratings is small, some programmes have very high levels of time shift viewing, prompting 

the need to include time shift viewing in the overnight ratings. The main constraint 

impeding the inclusion of VCR ratings in the overnight ratings is the difficulty in 

estimating time shift audiences overnight. 

This study proposed a number of methods of estimating VCR ratings overnight, 

including the recording level adjustment method, the same day playback adjustment 

method, and the genre/station correction method. While further research is required to 



iii 

compare the predictive ability of the methods, in the meantime implementing any of the 

methods is likely to provide more accurate overnight estimates of total audiences. 



iv 

ACKNOWLEDGEMENTS 

First, I would like to thank my supervisor, Dr Janet Hoek, for her professional guidance 

and especially her encouragement at the times when my motivation was lacking or I was 

having a crisis. I also gratefully acknowledge the advice and support of Professor Philip 

Gendall, especially during the data analysis stage of the study. I would like to thank Don 

Esslemont for his advice which was given willingly, with patience and understanding. 

Thanks also to Lisa Emmerson who provided constructive criticism throughout the 

writing of this thesis. 

Without the co-operation of AGB McNair this study would not have been possible. I 

greatly appreciate the support and encouragement of Nick Jones. I also thank Mike 

Frisby for the hours he spent preparing the data and the prompt answers he gave to my 

many questions. 

Funding for the project was generously provided by the Massey University Graduate 

Research Fund (GRF) and the Department of Marketing at Massey University. 

Most importantly I would like to thank Brent for his continual support, understanding, 

and knowing exactly the right words to say in my many moments of stress. Thank you 

to Mum, Dad, and Paul for their interest, and also the support of Jason, who knew 

exactly what I was going through. Without these people the past year would have been 

much harder to cope with. 



TABLE OF CONTENTS 

ABSTRACT n 
ACKNOWLEDGEMENTS 1v 
TABLE OF CONTENTS v 
LIST OFT ABLES vu 
LIST OF FIGURES ix 

CHAPTER 1. INTRODUCTION 1 

1.1 VCR Penetration 1 
1.2 VCR Time Shifting 2 
1.3 Time Shifting and Programme Ratings 3 
1.4 Time Shifted Commercials 5 
1.5 Summary 7 

CHAPTER 2. THE EXTENT OF TIME SHIFTING AND INFLUENCING 
FACTORS 9 

2.1 VCR Use: Time Shift Viewing versus Pre-recorded Viewing 9 
2.2 Time Shift Viewing versus Live Viewing 13 
2.3 Time Shift Viewing across Programme Genres 15 
2.4 Time Shift Viewing across Stations 18 
2.5 Time Shift Viewing across Times and Days 21 
2.6 Time Shift Viewing across Demographic Groups 23 
2.7 Summary 24 

CHAPTER 3. MEASURING TIME SHIFT VIEWING 25 

3.1 Levels of Recording and Playback 25 
3.2 Frequency of Playback 30 
3.3 The "Viewing Lag" 32 

3.3.1 Proportion of Recordings 32 
3.3.2 Proportion of Playbacks 35 
3.3.3 Variations in Viewing Lags 37 

3.4 Summary 39 

CHAPTER 4. RESEARCH OBJECTIVES AND METHODOLOGY 40 

4.1 Research Objectives 40 
4.1.1 The Extent of Time Shifting 40 
4.1 .2 Variations in Time Shift Viewing Levels 41 
4.1.3 Measuring Time Shift Viewing 42 

4.2 Research Methodology 44 



CHAPTER 5. THE EXTENT OF TIME SHIFTING: 
RESULTS AND DISCUSSION 45 

5.1 VCR Use: Home Recorded versus Pre-recorded Viewing 45 
5.2 Time Shift Viewing versus Live Viewing 48 

CHAPTER 6. VARIATIONS IN TIME SHIFT VIEWING LEVELS: 
RESULTS AND DISCUSSION 55 

6.1 Time Shift Viewing across Programme Genres 55 
6.2 Time Shift Viewing across Stations 58 
6.3 Time Shift Viewing across Times of the Day 60 
6.4 Time Shift Viewing across Days of the Week 63 
6.5 Time Shift Viewing across Demographic Groups 65 

CHAPTER 7. MEASURING TIME SHIFT VIEWING: 
RESULTS AND DISCUSSION 68 

7 .1 Current measures 
7 .1.1 Levels of Recording and Playback 
7 .1.2 The "Viewing Lag" 

7 .2 Estimating VCR ratings 
7.2.1 Recording Level Adjustment Method 
7 .2.2 Same Day Playback Adjustment Method 
7.2.3 Genre/Station Correction Method 
7.2.4 Regression Analysis 

CHAPTER 8. CONCLUSIONS 

8.1 Conclusions 
8.2 Implications 
8.3 Limitations of the Research 
8.4 Direction for Future Research 

REFERENCES 

APPENDICES 

Appendix A. Technical Appendix 
A.1 Calculation of Ratings 
A.2 Calculation of VCR Rating Contributions 
A.3 Calculation of Correction Factors 

Appendix B. VCR Use 
Appendix C. Viewing lags of Genres 
Appendix D. Case Numbers 
Appendix E. Multiple Comparison Tests 

68 
68 
71 
75 
75 
77 
78 
83 

85 

85 
86 
87 
88 

90 

94 

95 
95 
97 
98 

100 
101 
102 
105 



LIST OF TABLES 

Table 1. Household VCR penetration by country 2 

Table 2. Proportion of the sample by age group who view coded and 
uncoded tapes per day (10 March - 6 April 1996) 11 

Table 3. Recording versus playback of programme genres 16 

Table 4. Proportion of playbacks across programme genres 17 

Table 5. Live viewing and recording shares (all people 5+) 19 

Table 6. Proportions of recording minutes across days of the week 22 

Table 7. Average daily minutes per VCR home (24 hour basis) 23 

Table 8. Average minutes recorded and played back per week in 
VCR households 28 

Table 9. Replay delay of recordings 33 

Table 10. Percentage of recordings by elapsed days 34 

Table 11. VCR Playback (1/7/91 - 28/7/91) 36 

Table 12. Percentage of recordings replayed during diary week by 
programme type 37 

Table 13. Percentage of recordings played back during diary week 
by recording mode and cable status 38 

Table 14. Playing of time shifted material versus pre-recorded material 
by all panel members per week 46 

Table 15. Proportions of time shifted viewing versus pre-recorded 
viewing by gender 46 

Table 16. Proportions of time shifted viewing versus pre-recorded 
viewing by age groups 47 

Table 17. Proportion of television viewing constituted by time shift viewing 48 

Table 18. Top VCR rating programmes 51 

Table 19. Ratings by programme genres 56 

Table 20. Ratings by stations 59 

Table 21. Ratings for weekdays versus weekends 64 

Table 22. Ratings by day of the week 64 

Table 23. Ratings by demographic groups 66 

Table 24. Average number of programmes recorded and played back 
by households per week by programme genres 70 

Table 25. Delay between recording and playback over 28 days 72 



Table 26. Programme genre recording: playback factors 76 

Table 27. Programme genre viewing lag multipliers 77 

Table 28. Ratings by programme genres 79 

Table 29. Live correction factors: genre by station 81 

Table 30. Adjusted genre/station live correction factors 83 

Table Bl . Playing of time shifted versus pre-recorded material by age groups 100 

Table Cl. Proportion of programmes played back over first seven days 
by genre (for all programmes replayed over 28 days) 101 

Table Dl. Number of programmes in each programme genre 102 

Table D2. Number of programmes in each station 102 

Table D3. Number of programmes in each day of the week 103 

Table D4. Number of programmes viewed predominantly by each 
gender group 103 

Table D5. Number of programmes viewed predominantly by each 
age group 103 

Table D6. Number of programmes: genre by station 104 

Table El. Correction factors of programme genres 105 

Table E2. Correction factors of sports programmes across stations 105 

Table E3. Correction factors of children's programmes across stations 106 

Table E4. Correction factors of documentaries across stations 106 

Table ES. Correction factors of current affairs across stations 106 

Table E6. Correction factors of soap operas across stations 107 



LIST OF FIGURES 

Figure 1. Penetration of the VCR and colour television in New Zealand l 

Figure 2. Viewing lag of time shifted programmes (July 1991) 5 

Figure 3. Live viewing, recording and playback shares by stations in 
VCR homes (29/7/91 - 4/8/91) 20 

Figure 4 . Proportion of hours taped by day part 21 

Figure 5 . Playback frequency (minutes 2/9/91 - 22/9/91) 30 

Figure 6. Playback frequency of households and children 31 

Figure 7. VCR ratings versus live ratings 53 

Figure 8. Live ratings across hours of the day 61 

Figure 9. VCR ratings across hours of the day 61 

Figure 10. VCR rating contributions across hours of the day 62 

Figure 11. Cumulative proportions of programmes played back over 
28 days by programme genres 74 



1 

CHAPTER 1. INTRODUCTION 

1.1 VCR Penetration 

New Zealanders have been quick to adopt many new technologies, among them video 

cassette recorders, or VCRs. Current data suggests that over 75 percent of New 

Zealand households own a VCR and this is expected to continue slowly rising until 

nearly all homes with a television also have a VCR (see figure 1). Many households are 

also acquiring additional VCRs, corresponding with the increase in the number of 

television sets per home. 

Figure 1. Penetration of the VCR and colour television in New Zealand 

"if. 
C 
0 
:;:: 
f! -Q) 
C 
Q) 
ll. 
'0 
0 
.c 
Q) 
0 
:::J 
0 :c 

100 

90 

80 

70 

60 

50 

40 

30 

20 

10 

0 +--+----+--+---+---+--+--~--+--+---+-----! 
1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 

Year 

-+-TV (owned) 
-a-VCR (owned) 

Source: Department of Statistics. NZ household expenditure and income survey section. 

Compared to other countries, New Zealand has one of the highest levels of VCR 

penetration in the world (see table 1). The United States has had the fastest adoption 

rate of VCRs, as it has with many other media products, such as televisions. VCR 



2 

penetration and viewing trends in New Zealand have generally followed trends of the 

United States. 

Table 1. Household VCR penetration by country 

Country 

United States 

United Kingdom 

New Zealand 

Japan 

Irish Republic 

Sweden 

Netherlands* 

France 

Germany 

Finland 

Household VCR 
penetration 

% 
81 

76 

75 

74 

73 

72 

68 

66 

63 

63 

Source: Robbins (1996) and New Zealand Department of Statistics 
All data 1995 except * 1994 

The high level of VCR penetration in New Zealand, almost certainly affects television 

viewing. This study examines the effect of VCR time shifting on programme audiences, 

an issue which is of importance to broadcasters and media planners alike. 

1.2 VCR Time Shifting 

VCRs facilitate more flexible television viewing since they allow programmes to be 

taped, stored and viewed at some future time. This behaviour, known as time shifting, 

permits viewers to rearrange the programme schedule devised by broadcasters to suit 

their own lifestyle and to avoid conflicts in programme scheduling. Viewers' enhanced 

ability to control the time at which they view particular programmes means that more 

viewers may be watching television programmes. 



3 

VCR time shifting may increase individual's television viewing levels in three ways: some 

programmes which might otherwise be missed due to competing activities can be saved 

on tape; viewers are able to watch two different programmes even though they are 

broadcast simultaneously; and repeat viewing of the same tape is possible (Levy, 1980; 

Levy, 1981; Kirk.ham, 1982). To support the claimed increase in television viewing, 

Levy ( 1981) stated that his findings "strongly suggest that the earliest adopters of home 

video cassette recorders use their VCRs as a complement to, and not as a replacement 

for, established patterns of broadcast exposure " (p.405). 

Some researchers disagree that time shift viewing increases viewing levels and instead 

claim that time shift viewing appears to substitute normal live viewing hours. For 

example, Darkow (1984) found that "the time that people with access to a video 

machine spend in looking at television programmes is the same as that spent by people 

without videos, although about 16% is time-shifted" (p. 29). 

Whether time shifting increases total audiences or not is a difficult issue to quantify. It is 

known, however, that time shift viewing contributes additional viewers to the live 

audiences of programmes. Therefore, the current system of audience measurement, 

which delivers overnight ratings, may underestimate real programme audiences. 

1.3 Time Shifting and Programme Ratings 

The validity and accuracy of television programme ratings are critical to media planners 

and broadcasters alike. Media planners use ratings to select programmes that will satisfy 

certain reach and frequency objectives which, in tum, contribute to an advertising 

campaign's overall goals. Broadcasters deliver audiences to advertisers, and base 

programme scheduling and pricing decisions on ratings data. 



4 

Television ratings are assumed to provide a measure of the total audience to a broadcast 

programme; arguably, ratings should, therefore, include both live and time shifted 

viewing. At present, New Zealand's overnight ratings, collected by AGB McNair, do 

not include adjustments for time shifted viewing1
• AGB McNair do, however, measure 

VCR recording and playback, although this behaviour is not incorporated in programme 

ratings for a number of reasons. 

The first reason is that the overall effect of time shifting on programme ratings has not 

yet been clearly established. While some figures suggest that playback constitutes only 

2.8 percent of each individual's total viewing hours (Robbins, 1996), other studies have 

suggested that, for some programmes, time shift viewing may account for a quarter of all 

household viewing (Mojo, 1994 ). Until the likely size and composition of time shift 

audiences is more precisely documented, there seems little point in enhancing the 

software which calculates ratings. 

The second reason is that ratings are provided overnight, whereas playback of time 

shifted programmes from the time of taping is usually greater than one day (see Figure 

2). To overcome this problem, two sets of data could be released: live ratings within 24 

hours, and, after a given period of time, live ratings augmented by VCR playback2
• 

However, media planners and advertisers often need to make decisions from day to day 

and viewing data a week or more old may be of little assistance. Alternatively, if stable 

patterns in time shifting behaviour exist, it may be possible to develop factors which 

could be used to adjust the overnight ratings and thereby model the likely final audience 

of programmes. 

1. A.C. Nielsen, an American rating service, has included VCR recordings in its ratings since 1978 (Goodman, 
1996, email). However, this method overstates the size of the broadcast audience, since some recordings are not 
played back until after the ratings period, and some recordings are never played back. 

2 . TN AGB, the ratings service in the United Kingdom, uses this method, in fact, most of the ratings data that they 
publish include time shift viewing which takes place within 7 days of recording (Taylor, 1996, email). 



5 

Figure 2. Viewing lag of time shifted programmes (July 1991) 

40% 
35% 

Ill 
.II: u 

30% <a 
.0 
>-
<a 
ii 
0 20% c 
GI u ... 
GI 
ll. 10% 

0% 

same 2-3 4-6 7-14 14-21 21+ 

Delay between recording and playback (days) 

Source: AGB McNair 

A third constraint associated with incorporating VCR playback in the television ratings is 

the controversy over the effect of time shifting on commercials, which is outlined in more 

detailed below. 

1.4 Time Shifted Commercials 

Television advertisers are concerned that the inclusion of time shifting in the ratings 

could result in double jeopardy: they may have to pay more for time slots , yet their 

advertisements would almost certainly be prone to commercial avoidance behaviours, 

such as 'zipping' or 'zapping' or to time shift delay. 

Zipping occurs when viewers fast-forward commercial breaks during VCR playback. 

"At best, zipped advertising will have no audio material, lower visual quality, and 

faster information flow" (Abernethy and Rotfeld, 1991, p.14). Zapping involves 

deleting advertisements during the recording process3
• 

3. In television research literature, zapping also refers to changing channels to avoid advertisements. 



6 

Researchers have tried to estimate the extent of zipping; Cronin and Menelly (1992) 

found that six out of ten commercials were completely or partially zipped. A study by 

Nielsen in 1988 found that between 60 and 80 percent of commercial minutes are zipped 

during playback (Gilmore, Secunda, and Warrens, 1991). Other studies in the late 1980s 

found rates of commercial avoidance via zipping ranging from 50 to 65 percent of VCR 

persons (Yorke and Kitchen, 1985; Metzger, 1986) or VCR playbacks (Kaplan, 1985a; 

Papazian, 1986; Reiss, 1986). These results suggest that time shifted commercials are 

more likely to be zipped than viewed. 

However, zipped advertisements are not completely worthless, for a number of reasons. 

Firstly, Gilmore and Secunda (1993) argued that "zipped commercials can produce 

retrieval and reinforcement of previously learned information. Consequently, zipped 

commercials can function effectively as reminder ads" (p.28) (also supported by 

Metzger, 1986; Reiss, 1986; Johnson, 1988; Gilmore, Secunda, & Warrens, 1991). 

Secondly, zippers may stop fast-forwarding to view a commercial; "18% of zippers 

named at least one TV commercial that caused them to stop fast-forwarding and watch 

the commercial" (Reiss, 1986, p.3). Thirdly, advertisers can restrict the effects of 

zipping by having a stationary shot of the brand, for an extended period of time, on the 

television screen (Stout & Burda, 1989). The underlying idea is that if a stationary 

picture emerges on the screen from a mass of rapidly moving pictures, there is more 

opportunity to notice the product. 

Zapping of commercials appears to be increasing but is still less prevalent than zipping. 

Metzger ( 1986) reported that only one percent of viewers zapped while recording. Ten 

years later, a Nielsen survey found that 25 percent of VCR owners omit the commercials 

when recording programmes (McDonald, 1996). Even though there is no opportunity 

for time shift viewers to see advertisements that are zapped during recording, the people 

who delete the commercials may actually be the most attentive advertisement viewers, as 

zappers have to watch the commercials with extra concentration in order to start the 

VCR after the commercial break (Metzger, 1986; Singer, 1986). 



7 

The delay of commercials due to time shifting is also a concern for advertisers, as some 

commercials may not be effective after a period of time. As an extreme example, an 

advertisement for an acclaimed musician's last concert will not be very effective when 

the time shifter sees it the day after the concert. 

Given the avoidance issues related with time shifted commercials, the rating services 

ideally would be able to measure how many people see advertisements during time 

shifted programmes. However, commercial avoidance has long been a problem for live 

viewing also; "The impact of people leaving the room during commercials and of 

people using remote controls to switch channels during commercial breaks, is far 

greater than zapping and zipping that's done with VCRs" (Singer, 1986, p.67). This 

wider issue of commercial avoidance behaviour remains a highly contentious topic as far 

as live viewing is concerned and this debate would inevitably extend to attempts to 

include VCR playback in the ratings. 

1.5 Summary 

Time shifting occurs when a television programme is videotaped and replayed at a later 

date. Given that VCR penetration in New Zealand has increased to over 75 percent of 

households, it is clear that programme audiences may be higher than the current 

overnight ratings suggest. It seems that the two issues impeding the inclusion of time 

shift viewing in the ratings are the lack of thorough knowledge of the effect of time 

shifting on programme ratings, and the time shift viewing lag which makes it impossible 

to include real time shift data in the overnight ratings. Before addressing each of these 

issues in the objectives of this research, it is logical to examine the research literature, to 

ascertain whether any clear patterns in time shifting behaviour have yet been established. 

Chapter two investigates the extent of time shifting and factors that influence time shift 

levels. Chapter three examines previous literature relating to the measurement of time 

shift viewing. The objectives of this study and the research process undertaken to 



8 

address the objectives are detailed in chapter four. The results are presented and 

discussed in chapters five to seven, followed by the conclusions which are contained in 

chapter eight. 



CHAPTER 2. THE EXTENT OF TIME SHIFTING AND 
INFLUENCING FACTORS 

9 

The New Zealand rating service does not include VCR ratings' in the overnight ratings in 

the belief that the extent of time shift viewing is minimal. To estimate the effect of time 

shifting on programme audiences, researchers have studied the extent of time shifting and 

whether certain factors influence time shift viewing levels. The extent of time shift 

viewing has been quantified by investigating the use of VCRs for pre-recorded versus 

time shifted material and by comparing time shift audiences to live programme audiences. 

Variations in time shift viewing levels have been examined across programme genres, 

stations, times of the day, days of the week, and across demographic variables. The 

remainder of this chapter summarises and discusses the findings that have emerged to 

date. 

2.1 VCR Use: Time Shift Viewing versus Pre-recorded Viewing 

Researchers have compared the extent to which viewers use their VCRs to play home 

recorded or pre-recorded2 material. In the early 1980's, VCRs were predominantly used 

to play home recorded material; however, there is increasing evidence that viewers now 

play a greater proportion of pre-recorded material (Levy, 1980; Levy, 1981 ; Kirkham, 

1982; Stoessl, 1982; Fiddick, 1984; Darkow, 1984; Metzger, 1986; Sims, 1989; Warrens 

and Thompson, 1992; Mojo, 1994; Jones, 1996). This trend is causing concern to 

broadcasters and advertisers alike, as pre-recorded tapes contain no advertising and 

detract viewers from live broadcast audiences, and therefore programme and commercial 

audiences3 may be diminishing. 

1. VCR ratings represent an estimate of the proportion of the population who view a programme via time shifting. 

2. Pre-recorded tapes are those which are not recorded on household VCRs. For example, this may include rented 
tapes, or pre-recorded tapes purchased from shops. 

3. Due to this concern, advertisers are beginning to advertise on pre-recorded tapes. 



10 

Studies in the early 1980s indicate that viewers used VCRs to play more home recorded 

tapes rather than pre-recorded material (Levy, 1980; Levy, 1981; Kirkham, 1982; 

Stoessl, 1982; Darkow, 1984). Levy (1980) found that, on the average day, 3.8 percent 

of American households played home recordings and only 0.6 percent played pre­

recorded tapes. In a more comprehensive study, based on the same data but a different 

measure, Levy (1981) reported that, on average, households played 3.4 home-made 

tapes compared to only 0.5 pre-recorded tapes per day. Similarly, Kirkham (1982) 

found that 36 minutes were spent on pre-recorded viewing per week, compared to three 

hours and 44 minutes on home recorded viewing. It is unclear from Kirkham's (1982) 

article whether he based his results on all households, VCR households only, individuals, 

or some other measure. Darkow's (1984) study found that of all VCR tapes played by 

850 German households in May and October 1993, 25 percent were pre-recorded and 75 

percent were home recorded. 

In the mid-eighties, studies began reporting that pre-recorded viewing was rapidly 

increasing (Fiddick, 1984; Kitchen and Yorke, 1985; Metzger, 1986). In 1984, an AGB 

study found that in the United Kingdom, VCR homes spent 2½ hours per week viewing 

pre-recorded material, and 3½ hours viewing home recordings (Fiddick, 1984). Metzger 

(1986) estimated, that of VCR tapes watched, 50 percent were home recorded and 49 

percent were either pre-recorded or were home movies made with a video camera; he did 

not state the classification of the remaining one percent of tapes. 

By the late eighties, American viewers were playing a greater proportion of pre-recorded 

material than home recorded programmes. A study by Sims (1989), based on AGB data, 

found that households played three hours and 12 minutes of pre-recorded material per 

week compared to only one hour of home recordings per week4. Similarly, an A.C. 

Nielsen study found pre-recorded play (two hours and 26 minutes per week) to be higher 

than home recorded play ( one hour and 20 minutes per week) (Warrens and Thompson, 

1992). While there has been no more recently published research from the United States, 

4. Sims (1989) also compared VCR use within four demographic subgroups and found in each case that VCRs were 
predominantly used to view pre-recordings. 



11 

Levy (1996) suggests that "time shifting has largely dropped out of VCR use in the US. 

What we are using VCRs for is almost exclusively as a playback device for rented 

movies" (email 7 July). 

In New Zealand, VCR penetration has been about five years behind the trends evident in 

the United States; consequently our VCR viewing trends also lag behind the United 

States patterns. AGB McNair (1991) found that, on average, there were 16 minutes of 

coded (home recorded) playback per day, compared to 13 minutes of uncoded (pre­

recorded) playback5 per day (Mojo, 1994). In 1994, Mojo reported that coded versus 

uncoded playback was roughly 50/50. More recently pre-recorded viewing has become 

more prominent than home recorded viewing; that is, on an average day 0.4 percent of 

the population watch uncoded tapes compared to 0.3 percent who view coded tapes 

(Jones, 1996). This trend does not yet appear to be evidenced across all demographic 

groups. For example, New Zealanders aged 50 plus still watch more home recorded 

programmes than pre-recorded tapes (Mojo, 1994; Jones, 1996), as shown in Table 2. 

Table 2. Proportion of the sample by age groups who view coded 
and uncoded tapes per day (10 March - 6 April 1996) 

Age Group 

All 5+ 5-17 18-49 50+ 
Viewing % % % % 

Coded 0.3 0.2 0.3 0.4 

Uncoded 0.4 0.5 0.5 0.2 

Source: Adapted from Jones (1996) 

VCR use in the USSR, Middle East, and Third World countries has not followed the 

trend of Western countries; instead, VCRs have been used exclusively for playing pre­

recorded tapes, with time shifting being virtually non-existent (Boyd and Straubhaar, 

1985; Boyd, 1987; Cohen, 1987; Levy, 1989; Dobrow, 1990). A large number of the 

5. 'Coded' and 'uncoded' tapes are terms associated with FingerPrinting technology, a method for collecting time 
shifting data. See chapter 4 for more information. 



12 

pre-recorded tapes played are from the Western world and, in some countries, such as 

the USSR, Soviet-Bloc countries and Arabian Gulf states, there are large black markets 

for such tapes (Boyd, 1987; Boyd, 1989; Dobrow, 1990). The high levels of pre­

recorded viewing in the Third World may also be due to the lack of broadcasts available. 

Researchers have found variations in the use of VCRs due to other variables. When 

Dobrow (1990) compared the VCR use of light and heavy viewers in Boston in 1986, he 

found that "while heavy viewers used their VCRs primarily to time shift, lighter viewers 

used the technology primarily to view pre-recorded video cassettes" (p.76). Sims 

(1989) found that non-cable homes devoted more time to playing pre-recorded tapes 

than pay cable homes (4 hours versus 2.9 hours), while pay cable households watched 

more home recorded tapes than non-cable homes (1.4 hours versus 1.1 hours). These 

figures suggest that as the penetration of cable increases, pre-recorded viewing is likely 

to decrease, home recorded viewing is likely to increase, but pre-recorded viewing will 

still be the dominant form of VCR viewing. 

Previous research suggests that the ratio of pre-recorded viewing to time shift viewing in 

New Zealand will continue to increase, following the trend of the United States. To 

understand the implications of the this trend, VCR viewing levels need to be compared to 

live viewing levels. For example, if VCR viewing is increasing compared to live viewing, 

time shift viewing may be increasing compared to live viewing. On the contrary, if VCR 

viewing is decreasing compared to live viewing, the question of whether ratings should 

include time shift viewing will diminish in importance. 

The following section compares levels of time shift viewing and live viewing to 

investigate whether time shift viewing has the potential to affect programme audiences 

even though VCRs are no longer predominantly used to play time shifted material. 



13 

2.2 Time Shift Viewing versus Live Viewing 

Although programme audiences are made up of live viewers and time shift viewers, at 

present the New Zealand ratings measure only live viewers. Research has consolidated 

live and time shift audiences to explore the proportion of the total ratings made up of 

VCR ratings. Estimates of time shifting behaviour have varied. Recent studies from the 

United States, and the United Kingdom have estimated the proportion of time shifting to 

be two to three percent of a programme's total audience (Sims, 1989; Robbins, 1996). 

In New Zealand, Jones (1996) estimated that time shift viewers constitute 2.7 percent of 

the average television audience. An earlier study suggested, however, that up to 10 

percent of total television viewing could be comprised of time shift viewing (Kirkham, 

1982), although the calculations used to arrive at this estimate appear to be flawed, as 

discussed below. 

At the time of Kirkham's (1982) study, VCR households made up only seven percent of 

the UK population. Kirkham (1982) estimated that VCR households played back about 

three and three-quarter hours of time shift material per week. He concluded that this 

was "equivalent to about 10 percent of the total viewing of the UK population" (p. 

405). The flaw in his calculation was that the time shift viewing measure was averaged 

across VCR households not the whole population. Kirkham' s estimate of time shift 

viewing contributing 10 percent to total viewing hours should, therefore, only be applied 

to VCR households, and not used as an estimate for the whole population. Re­

calculating his estimates reveals that time shifting comprised only 0.7 percent6 of total 

viewing - somewhat less than the more recent estimates which reflect higher VCR 

penetration rates. 

6. Contribution of VCR homes to population viewing: 

7% of population x 90% live viewing => 
93% 

7% of population x 10% time shift viewing =>0.7% 

For total population: Time shift viewing= 0.7% 

Live viewing = 99.3% 

Contribution of non-VCR homes to population viewing: 

6.3% 93 % of population x 100% live viewing=> 



14 

Even though the average time shift audience of programmes is estimated to be much 

smaller (3%) than the live audience (97%), these generic proportions may mask 

variations which occur across individual programmes. Recent studies from the United 

Kingdom have reported that some programmes have very high proportions of time shift 

viewing, that is, the top twenty VCR rating programmes for November 1995 had VCR 

ratings ranging from 3.2 percent for Cracker to 1.7 percent for Coronation Street 

(Robbins, 1996). These ratings are considerably higher than average VCR ratings which 

are around 0.3 percent. As another example, 2.334 million time shifters watched 

Wallace and Gromit (an animation) in December 1995 in the United Kingdom; time shift 

viewers comprised 22 percent of the total audience for the programme (AGB Television, 

1995). 

A recent study in New Zealand found similar results to United Kingdom studies. That is, 

of programmes that were screened in April 1994, Rain Man (a movie) had the largest 

time shift audience of 32,863 households which comprised 24 percent of the total 

audience, and time shift viewing of NBA basketball constituted 21 percent of the 

household audience (Mojo, 1994). It is unclear from Mojo's (1994) report whether the 

analysis included all playbacks of tapes by different household members or only the first 

playback for the household. If all playbacks were included this would inflate the 

proportion of household time shift viewing, as a programme can only be viewed live 

once by the household but it can be played back many times by different individuals 

within the household. Even so, these results and the United Kingdom studies suggest 

that some programmes have much higher proportions of time shift viewing than others, 

accounting for thousands of programme viewers. 

Factors that are believed to influence time shift viewing levels are detailed in the 

remaining sections of this chapter. 



15 

2.3 Time Shift Viewing across Programme Genres 

Overall time shifting appears to make up only a small proportion (3%) of total viewing, 

however, within certain programme genres time shift viewing has been found to be much 

higher. In general, movies have the highest levels of time shift viewing, followed by soap 

operas, dramas, and comedies (Arbitron, 1979; Levy, 1980; Agostino et al, 1980; Levy, 

1981; Kirkham, 1982; Byington, 1983; Darkow, 1984; Fiddick, 1984; Singer, 1986; 

Gunter and Svennevig, 1988; Gunter and Wober, 1989; Mojo, 1994; Robbins, 1996); 

whereas sports, documentaries, and children's programmes have much lower levels of 

time shift viewing (Levy, 1980; Kirkham, 1982; Byington, 1983; Darkow, 1984; Fiddick, 

1984; Kaplan, 1985; Gunter and Wober, 1989; Robbins, 1996). The researchers which 

have quantified levels of time shifting across programme genres have used three 

measures7
: percent of recordings, percent of replays, and VCR contributions to 

programme ratings. 

Kirkham (1982) found that the majority of programmes recorded were 'feature films' 

(45%), while sports programmes were least recorded (3%). Byington (1983) found 

similar results with movies consisting of 33 percent of recordings, followed by TV series 

(25%) and soaps (20%); sporting events (2%) and children's programmes (3%) had the 

lowest levels of recording. 

Since all recordings are not played back, results based on recordings may give inaccurate 

levels of time shifting across programme genres - if recordings of some programme 

genres are more likely to be played back than others. Levy (1980a) found that the genres 

that were most frequently recorded and played back had very similar proportions of 

recording and play back. Darkow (1984) reported only "slight differences" between the 

proportions of recording and playback across programme genres; however, he included 

playbacks of pre-recorded tapes in the feature films category. Recalculating Darkow's 

(1984) figures to exclude pre-recordings, decreased the reported proportion of feature 

7. Not all of the research referenced in this paragraph has quantified levels of time shifting across programme 
genres, some have only listed genres with higher and lower levels of time shifting. 



16 

film playbacks from 61 percent to 38 percent. Darkow's recalculated figures are 

presented in Table 3. 

Table 3. Recording versus playback of programme genres 

Genre Recording Playback1 

% % 
Feature films 50 38 
Television series 22 30 
Light entertainment 15 17 
Information 9 10 
S2ort 4 5 
Total 100 100 
Source: Darkow (1984) 
1. Playback figures adjusted to exclude playback of pre-recordings. 

After recalculating Darkow's figures, feature films still had higher levels of recording and 

playback than the other programme genres. However, comparing recording to playback 

(see Table 3), a larger proportion of feature films were recorded than played back, 

whereas a larger proportion of television series were played back than recorded. 

Levy and Gunter (1988) also measured both recording and playback of programme 

genres, but found different results to other studies. Instead of movies having the highest 

levels of time shifting, US drama series were most likely to be recorded and UK drama 

series were most likely to be played back. Gunter and Wober (1989) claimed that Levy 

and Gunter's study found different results to previous studies because their study was 

based on diary data whereas earlier studies were based on questionnaire surveys. 

However, this does not seem to be the case as Darkow's (1984) study was based on 

diary data. There does not seem to be any obvious reason for the different results. Levy 

and Gunter's ( 1988) study did find that some programme genres were more likely to be 

recorded than played back, but the genres that this affected were different from 

Darkow's (1984) study. 



17 

There is no conclusive evidence that VCR recordings are an inappropriate measure of 

time shifting across genres, although actual VCR playback is obviously a more realistic 

measure. Levy (1980) used a diary survey to analyse playback within United States 

households in 1979 - the results are presented in Table 4. Although his results are almost 

twenty years old, the programme genre rankings are very similar to those of recent 

studies (Mojo, 1994; Robbins, 1996)8
, with movies, comedies, and soap operas making 

up reasonably large proportions of all playbacks. 

Table 4. Proportion of playbacks across programme genres 

Programme genre Playback 
% 

Movie 23.15 

Situation comedies 14.56 

Soap operas 11.69 

Entertainment series 8.71 

Entertainment specials 5.73 

"Kids-vid" 5.01 

Police / Detective 4.89 

News / Public affairs 4.18 

Dramatic series 4.18 

Talk shows 3.93 

Mini-series 3.22 

Sports 3.10 

Science Fiction 2.63 

Miscellaneous 2.37 

Unclassifiable 2.65 
Source: Levy (1980) 

Warrens and Thompson ( 1992) also quantified levels of time shifting across programme 

genres. They calculated the contribution of VCR ratings to total ratings for different 

8. These studies did not report the actual proportions of playback across genres, only the most preferred genres. 



18 

programme genres, a measure which estimates the effect of time shifting on programme 

ratings. Only a couple of programme genres were reported and some were classified by 

day part. Daytime dramas had the highest VCR contribution to ratings of 12 percent; 

that is, 12 percent of the total audience to the average daytime drama was made up of 

time shifters. Feature films had the next highest VCR contribution of five percent, 

followed by prime time dramas (3%) and prime time sitcoms (2%). 

To summarise, some researchers have used recording levels to estimate playback levels 

( or time shift viewing) across programme genres. Logically then, playback levels are a 

more precise measure of variations in time shifting across genres. Playback levels, 

however, do not explain how variations in genre time shift levels affect programme 

ratings. To date, only one study has examined VCR rating contributions across a few 

programme genres. Further research into VCR contributions variations across 

programme genres is needed to gain an estimate of the omitted viewing audience within 

current ratings, especially within the movie genre. 

2.4 Time Shift Viewing across Stations 

The television stations on which programmes are broadcast also appears to influence 

time shift viewing levels. Generally programmes are more likely to be time shifted if they 

broadcast on stations with high viewing shares (Kirkham, 1982; AGB McNair, 1991; 

Mojo, 1994; Robbins, 1996). New Zealand has three national free-to-air stations 

(TVNZl, TVNZ2 and TV3), five Sky TV (pay) stations, and a number of local stations. 

TVNZl and TVNZ2 have the highest viewing shares, followed by TV3 which is the 

most recently introduced free-to-air national station; Sky TV and local stations have very 

low channel shares. A recent study found that 35 percent of recordings were of TVNZl 

programmes, 37 percent of TVNZ2, 21 percent of TV3, and the remaining seven percent 

of recordings were of Sky TV and other stations (Jones, 1996). 



19 

A study by Mojo ( 1994) also compared recording shares across stations and then 

compared them to current live viewing shares - these results are presented in Table 5. 

Table 5. Live viewing and recording shares (all people 5+) 

Station Live viewing Recording 
% % 

TVNZl 43 39 

TVNZ2 37 39 

TV3 18 21 

Sky TV / Other 2 2 

Source: Mojo (1994) 

TVNZ 1 had a higher live viewing share than TVNZ2, but they had similar recording 

shares. The Mojo ( 1994) study suggested that this was due to programme genre effects, 

for example, movies are one of the most likely genres to be time shifted and TVNZ2 

broadcasts more movies than TVNZl. 

Table 5 shows that Sky TV and the other stations accounted for 2 percent of viewer's 

recording. Sky Movies was the most likely Sky channel to be taped, which provides 

further evidence that movies have high time shift viewing levels. In Jones' (1996) later 

study, the recording share of Sky TV and other stations had increased to seven percent; 

this coincides with the increase in Sky TV penetration and the increase in the number of 

local stations within this period. Therefore, the recording shares of Sky TV and other 

stations may continue to increase as the penetration of these stations increases. 

Levels of playback across stations seem to be slightly different to levels of recording. 

AGB McNair (1991) compared live viewing, recording and playback shares across New 

Zealand's three national free-to-air stations, as shown in Figure 3. 



Figure 3. 

Cl) ... 
a, 
.c ,,, 
Cl) 
ti) 

.s 
C 
Cl) 

~ 
Cl) 
D. 

Live viewing, recording and playback shares by station 
in VCR homes (29/7/91 - 4/8/91) 

100% 

80% 

60% 41% 

40% 

20% 

0% 
Live viewing 

44% 

29% 

Recording 

Activity 

50% 

Playback 

DTV3 

DTVNZ2 

DTVNZ1 

Source: AGB McNair (1991) 
Note: Analysis based on 28 days from recording. 

20 

The recording and playback shares in Figure 3 both show that stations with larger live 

audiences also have larger time shift audiences; however, measuring time shifting by 

recording appears to distort the time shift viewing shares because the stations have 

different playback rates. TVNZ 1 and TV3 had higher levels of recording than playback 

compared to TVNZ2, while TVNZ2 had comparatively higher levels of playback than 

recording. There was also a larger difference between playback shares of TVNZl and 

TVNZ2 than between the recording shares of these stations. 

Since stations with higher live audiences also have higher time shift audiences, it seems 

that the effect of time shift viewing on total audiences may be quite similar between 

stations. For example, if one station had large audiences of 10,000 live viewers and 

1,000 time shift viewers, and another station had small audiences of 100 live viewers and 

10 time shift viewers, the contribution of time shift viewing to total viewing for both 

stations would be the same (10%). To date, research has not compared live and VCR 

ratings across stations; these details are needed to estimate whether the proportions of 

programme audiences not included in the current ratings differ across stations. 



21 

2.5 Time shift Viewing across Times and Days 

The time of day and day of week that a programme is broadcast may also influence time 

shift viewing levels. Researchers have examined recording levels across times of the day 

and days of the week. As previously discussed, recording levels are not the best measure 

of time shifting, because not all recordings are played back, and playback rates may differ 

across different times of the day and days of the week. However, considering that there 

are no studies that have compared playback levels or ratings, recording levels can be 

used as an estimate of time shifting levels. 

Recording levels across times of the day generally follow live viewing levels, with the 

majority of recording occurring at prime time (that is, between approximately 7pm and 

11pm, depending on the researcher) (Agostino et al, 1980; Levy, 1980; Metzger, 1986; 

Mojo, 1994; Robbins, 1996, Jones, 1996). Metzger ( 1986) reported that 52 percent of 

recorded hours were taped in prime time (see figure 4). 

Figure 4. Proportion of hours taped by day part 

Prime (7-11pm) 
52% 

Source: Metzger (1986) 

Late 
10% Weekday/day 

24% 

Weekend/day 
14% 

Furthermore, AGB McNair found that ten percent of homes recorded programmes 

between 8.15pm and 8.45pm (Mojo, 1994). Jones (1996) found "a definite lunchtime 



22 

peak with a smaller less defined peak in the morning" during weekdays (p.1). Levy 

(1980) also found a lunchtime peak on Saturdays. 

Across days of the week, research has found that weekends rather than weekdays have 

higher levels of recording (Metzger, 1986; Mojo, 1994). Mojo (1994) found this to be 

the case on all New Zealand channels. During the weekend, studies have found Sundays 

to have the highest level of recording (Sims, 1989; AGB McNair, 1991; Jones, 1996). 

Estimates of recording levels on Saturdays have varied: Sims (1989) found the lowest 

level of recording for all days of the week occurred on Saturdays, whereas AGB McNair 

(1991) found Saturdays to have the second highest level of recording (see Table 6). 

Sims' (1989) results only included prime time recording, which is a probable explanation 

for the differences in the results. 

Table 6. Proportions of recording minutes across days of the week 

Sims AGB McNair 
Day (1989) (1991) 

% % 
Monday 15 13 

Tuesday 13 10 

Wednesday 12 15 

Thursday 14 11 

Friday 11 12 

Saturday 9 17 

Sunday 26 21 

Source: Sims (1989) and AGB McNair (1991) 

Although the majority of time shifted programmes are recorded in prime time and on 

weekends, it is unknown how many of these recordings are played back. Perhaps 

programmes recorded on weekdays are more likely to be played back, resulting in higher 

time shifting levels compared to weekends. Further research is needed to investigate 

whether different broadcast times and days cause variations in playback levels. 



23 

2.6 Time Shift Viewing across Demographic Groups 

Various studies have found that women are more likely to time shift programmes than 

men (Lindstrom, 1989; Gunter and Wober, 1989; Sims, 1989; Robbins, 1996). Gunter 

and Wober (1989) found that, on average, women played 1.7 home recordings per week, 

whereas men played one recording per week. Lindstrom's (1989) study of heavy 

recording households found that women (aged 18+) taped 69 percent of recordings, 

while men (aged 18+) recorded 31 percent. Sims (1989) studied a number of 

demographic groups - his results are presented in Table 7. 

Table 7. Average daily minutes per VCR home (24 hour basis) 

Demographic group 

Men 18+ 

Women 18+ 

Teens 12-17 

Children 2-11 

Source: Sims (1989) 

Playback 
minutes 

3.4 

5.1 

3.4 

4.3 

As well as women playing more home recording minutes than men, Sims ( 1989) also 

found that children had higher levels of time shifting than men ( aged 18+) and teenagers. 

Furthermore, Jones (1996) studied playback rates of slightly different age groups, and 

found that home recorded tapes had "a high likelihood of being watched by those over 

50 than those under 17 with I 8-49's being very similar to all 5+" (p.1). 

The research suggests that women, children, and people aged 50 plus are more likely to 

time shift programmes. To relate these findings back to variations in the time shift 

viewing of programmes, further research should investigate whether programmes that 

are predominantly watched by women, children, or people aged 50 plus have higher 

VCR ratings. 



24 

2.7 Summary 

The research to date has found that, overall, pre-recorded viewing has replaced time 

shifting as the predominant use of VCRs, and that the extent of time shift viewing is low. 

However, time shift ratings have been found to be as high as 3.2 and, considering that a 

high live rating is around 20, it is clear that overnight ratings may underestimate the total 

viewing audience of some programmes. The extent of this problem remains unquantified 

because of the limited work undertaken thus far. 

Although it has been documented that some variables that may lead to higher time shift 

levels, such as, the programme's genre, the broadcast channel, the time and day of 

broadcast, and the demographics of the audience, the specific effects of these variables 

on programme ratings remain less clear. If stable patterns in time shifting behaviour 

exist, it may be possible to develop factors which could be used to adjust the overnight 

ratings and thereby model the likely final audience of programmes. 



25 

CHAPTER 3. MEASURING TIME SHIFT VIEWING 

Researchers have studied a number of important issues relating to the measurement of 

time shifting behaviours, namely, whether to use VCR recording or VCR playback to 

measure time shifting, whether to include the frequency of playback in VCR ratings, and 

how long it takes for VCR playback to occur from the time of recording. This chapter 

summarises and discusses the findings that have emerged to date. 

3.1 Levels of Recording and Playback 

Rating companies around the world use either VCR recording or VCR playback to 

measure time shifting behaviour. Methodologies which document VCR playback, count 

a programme as time shifted when it is actually played back. Although this is the more 

logical and direct measure of time shifting behaviour, it can not be included in the 

overnight ratings. 

Alternatively, methodologies which document VCR recordings, count a programme as 

time shifted when it is recorded. As a result, an estimate of time shift viewing can be 

included in the overnight ratings. However, this method assumes that all recordings are 

played back; that the correlation between playback and recording is one. Yet research 

has found that the level of VCR recording is typically much higher than the level of 

playback (Levy, 1981; Stoessl, 1982; Levy, 1983; Hunt, 1987; Sims, 1989; AGB 

McNair, 1991; Warrens and Thompson, 1992; Mojo, 1994), thus seriously questioning 

the validity of VCR recording as a measure of time shifting behaviour. Researchers have 

used three variables to compare VCR recording and playback: percent of programmes, 

percent of households, and percent of minutes. This variation in the measures used 

makes direct comparisons difficult, although overall patterns can be explored. 



26 

The first measure used by researchers to examine recording and playback levels 

compared the proportion of programmes; the findings suggest that more programmes are 

recorded than played back. Levy (1983) estimated that the average VCR household 

recorded 3.31 programmes a week and played back 2.42 programmes a week, which is 

equivalent to a 73 percent playback rate. An AGB (UK) survey claimed that "of every 

JOO programmes recorded, 13 were never played back, or not within the period of the 

study"' (Fiddick, 1984, p.14). In a more recent study, Mojo (1994) examined nine 

specific programmes and found that the proportion of playbacks varied across 

programmes. For example, 64 percent of the recordings of World Around Us were 

played within 4 weeks, while only 34 percent of the recordings of Rugby - NZ Big Game 

were played back within this period. As yet, the extent to which playback levels vary 

across programme genres is not clear. 

Secondly, researchers have compared the recording and playback levels of households. 

Levy (1981) analysed the proportion of VCR households that recorded and played back 

prograrnrnes within one week during 1979. He found that "during the diary week 

80.8% of VCR households recorded at least one broadcast ... and 74.7% replayed one 

or more tapes of previously recorded programmes" (p.402). This finding complements 

the previously mentioned studies, that is, more broadcasts are recorded than played back 

(Levy, 1981; Levy, 1983; Fiddick, 1984; Mojo, 1994). Other studies have measured the 

number of minutes recorded and played back. 

Research based on minutes is difficult to compare as several factors, which are generally 

not discussed by the researchers, may influence some results more than others. For 

example, if the number of minutes recorded is greater than the number of minutes played 

back, researchers have assumed that not all programmes have been played back, although 

other evidence relating to viewers' playback behaviour suggests that such differences are 

almost certainly attributable to fast forwarding (Yorke & Kitchen, 1985; Kaplan, 1985; 

Papazian, 1986; Reiss, 1986; Metzger, 1986; Gilmore, Secunda & Warrens, 1991; 

I . Fiddick (1984) did not state the time period of the study. 



27 

Cronin & Menelly, 1992). In addition, some researchers have included multiple 

playbacks of a taped programme in their playback estimates, while others have only 

included the first playback of a programme, ignoring any subsequent playbacks. 

Studies that excluded multiple replays found more minutes were recorded than played 

back. A.C. Nielsen drew a convenience sample from their people meter panel, to gain 

some preliminary insights into panel members' VCR viewing behaviour (Hunt, 1987). 

The minutes were reported on a household basis and excluded recordings that were 

viewed live, on the grounds that these were less likely to be played back (as found by 

Levy (1983)). Hunt's (1987) study found that within 28 days, 57 percent of recorded 

minutes were played back in the April/May study, and in the July study, only 44 percent 

of recorded minutes were played back. The difference in these results may be explained 

by the different seasons; that is, respondents may have spent more time outdoors and less 

time watching television in July, as July is warmer than April and May in the United 

States. 

In a New Zealand study, AGB McNair (1991) compared the number of minutes recorded 

and played back by VCR households and by individuals in the VCR households. The 

study found that, on average, individuals recorded 87 minutes in a week but of those 

minutes only 29 minutes were played back within 28 days, that is, only 33 percent of 

recorded minutes were played back over 28 days. Similarly, for households only 34 

percent of recorded minutes were played back over 20 days. Some of the recorded 

minutes which were not played back in the study may have been zipped during 

commercial breaks, even so, there would still be a large proportion of recorded 

programme minutes not played at all. 

The previously mentioned studies by A.C. Nielsen and AGB McNair excluded multiple 

replays, that is, the analyses only included the first playback of recordings, and found that 

more minutes were recorded than played back. In contrast, other studies have included 



28 

multiple replays in their results and have found that more minutes are played back than 

recorded (Metzger, 1986; Lindstrom, 1989). 

Metzger (1986) reported that, on average, people from VCR households taped 12 

minutes and played back 19 minutes of television on weekdays, and during the weekend 

taped 15 minutes and watched 25 minutes of playbacks per day. He estimated that there 

are 1.6 hours of playback per one hour of taping. A limitation of his study was that each 

respondent had to account for the entire family's VCR activity, which may have required 

a certain amount of guesswork, and thus introduced unquantifiable error to the results 

(Hunt, 1987). 

Lindstrom (1989) found similar results to Metzger (1986). He measured the number of 

minutes recorded and played back between 1985 and 1987, the results are presented in 

Table 8. 

Table 8. Average minutes recorded and played back per week in VCR 

households 

Year 

1985 1986 1987 

Minutes recorded 171 172 154 

Minutes played back 253 273 258 

Playback : recording 1.5:1 1.6: 1 1.7: 1 

Source: Lindstrom (1989) 

The increase in the ratios of playback to recording between 1985 and 1987 seem to 

suggest that the number of multiple replays per tape are increasing. This raises some 

interesting questions about the relationship between time shifting behaviour and 

programme frequency. The frequency of playback and its relation to programme ratings 

is discussed in more detail in section 3.2. 



29 

Other researchers have not stated whether they included multiple playbacks or only the 

first playback of recordings in their analyses. Therefore, it is difficult to interpret the 

results and compare whether their results support or contradict the previously mentioned 

studies. These studies are briefly discussed below; all found that more minutes were 

recorded than played back. 

Kirkham ( 1982) found that there were four hours and eight minutes of recording per 

week compared to three hours and 44 minutes of viewing per week, that is, 90 percent of 

recorded minutes were played back. Kirkham (1982) described very little about his 

survey methods, it is unclear whether his figures are based on households or individuals 

or whether multiple playbacks were included. Sims (1989) conducted two studies, one 

in January and the other in April. He found that, on average, VCR homes devoted 2.4 

hours per week to recording in both months. Although, in January, VCR homes played 

back 1.2 hours per week, and in April, they played back one hour per week. January 

may have had slightly more hours of playback because it is a colder month and people 

are more likely to watch television. Warrens and Thompson's (1992) study of A.C. 

Nielsen data found 1.6 hours of recording and 1.2 hours of playback per week, that is, 76 

percent of recorded minutes were played back. 

Researchers who excluded multiple replays found that many recordings are not played 

back; this suggests that VCR playback is the more logical and conservative measure of 

time shifting behaviour. Rating services that use VCR recording to estimate time shifting 

behaviour may be overestimating time shift audiences by up to 65 percent. Studies that 

have included multiple replays have found recordings that are played back are often 

played more than once. This is an important factor to consider when measuring VCR 

playback. Research has investigated the frequency of playback, the results are presented 

in the following section. 



30 

3.2 Frequency of Playback 

Before VCRs were introduced programmes could only be viewed once, unless the 

broadcast was repeated. With the introduction of VCRs individuals are able to view 

broadcasted programmes many times. A number of studies have explored the number of 

times that home recorded tapes are played back. 

The following studies, which are based on three different measures, suggest that the 

majority of tapes are played back only once. Arbitron's (1979) study of early VCR 

usage, found that of the three most recent recordings, 48 percent were played back once, 

nine percent were played back twice, five percent were played back three or more times, 

and 38 percent had not yet been replayed. Metzger (1986) found that the majority (64 

percent) of respondents replayed tapes only once, 18 percent twice and 18 percent of 

people watched tapes three times or more. AGB McNair (1991) investigated the 

frequency of minutes replayed over 20 days within VCR households, and found that the 

majority of minutes are replayed only once - the results are presented in Figure 5. 

Figure 5. Playback frequency (minutes 2/9/91 - 22/9/91) 

100% 
88% 

Ill 80% 
s 
:::, 
C e 60% -0 
C 
0 
'f 40% 
8. 
0 ... 
0.. 20% 

9% 
3% 

0% 

once twice more than three 

Playback frequency 

Source: AGB McNair (1991) 



31 

A.C. Nielsen found that the frequency of playback varied across demographic groups 

(Hunt, 1987). Specifically, a large majority (90%) of males and females viewed home 

recorded tapes only once. Compared to adults, children were much more likely to view 

multiple replays. Figure 6 presents the frequency of playback for children and 

households (which included all family members). 

Figure 6. Playback frequency of households and children 

C 
100% 

! 90% :!2 
:c 80% C, ... 
0 70% fll 
't, 

0 
.s::. 

60% 
QI 

50% fll 
::, 
0 

40% .s::. 
0 

30% C 
0 
t: 20% 
8. 

10% 0 ... 
0. 

0% 

once 

Source: Hunt (1987), July study 

twice 

Playback frequency 

more than three 

•Households 

El Children 

The accumulated evidence suggests that some home recorded tapes are played back 

more than once (especially by children). This raises the issue of whether multiple 

playbacks should be included in VCR ratings. Multiple playbacks increase the frequency 

of programmes, whereas ratings are a measure of programme reach. AGB McNair, for 

example, describe ratings as: "a survey estimate of the size of an audience expressed as 

a percentage of the total group sampled or potential" (p.18). Multiple playbacks of 

programmes by individuals should, therefore, be discarded from the calculations of VCR 

ratings, as including them would inflate the actual time shift audience of programmes. 

It maybe useful for advertisers, if the rating companies also reported the average 

frequency of programmes, given that multiple playbacks of programmes increase the 



32 

frequency of advertisement exposure. For example, it was estimated that 19 percent of 

children who watch a time shifted programme watch it at least three times and, therefore, 

they may also see the advertisements at least three times, assuming that the 

advertisements are not zipped or zapped. 

Another issue to consider in the measurement of VCR ratings is the time period within 

which recordings are played back. 

3.3 The "Viewing Lag" 

Time shift viewing is not included in overnight ratings, due to the delay between 

recording and playback, or the viewing lag. Given that the viewing lags of some 

programmes may be very long, the ratings services need to know the time frame within 

which the majority of recorded programmes are played back, so that VCR ratings can be 

released within a reasonably short period of time. 

To examine average viewing lags, some researchers have studied the proportion of 

recordings played back over a certain period, while others have limited their study to 

include only programmes that are actually played back, that is, ignoring recordings that 

are not played back. The following sections detail the results fro both types of analysis, 

and then examine factors that appear to influence viewing lags. 

3.3.1 Proportion of Recordings 

Levy (1981) found that "almost 80 percent of all recordings were replayed within the 

diary week" (p.402). His data was collected from a one week diary survey; it is unclear 

whether "all recordings" included only recordings made within the diary week or 

whether he also included recordings made outside the diary period. Levy (1981) also 



33 

studied the proportion of tapes replayed over other time periods - his results are 

presented in Table 9. 

Table 9. Replay delay of recordings 

Tapes replayed on: Recordings 
% 

same day 29.9 

day after 29.1 

2-3 days after 13.0 

4-7 days after 7.6 

8-30 days after 9.4 

1 month or more 11.0 

Total 100.0 

Source: Adapted from Levy (1981) 

The results in table 9 are dubious for a number of reasons. Firstly, replays were 

measured 8-30 days and 1 month or more after recording, even though the sample period 

was only one week. Secondly, the results indicate that all recordings were played back, 

even though previous research has found that many recordings are not played back 

(Levy, 1981; Levy, 1983; Fiddick, 1984; Hunt, 1987; AGB McNair, 1991; Mojo, 1994). 

As a possible explanation it seems that Levy's (1981) results were actually based on the 

proportion of playbacks which occurred over the one week diary period. Obviously his 

results need further explanation before they can be accurately interpreted and compared 

to other research. 

In another study, Levy (1983) estimated that 52.7 percent of all programmes recorded 

during one week were also played back during the same week. This measure, however, 

is likely to be influenced by the playback patterns of programmes recorded at the 

beginning of the week, as these recordings had six days to be played back, whereas 

recordings at the end of the week had less than one day to be played back. In a later 

article, Levy and Fink (1984) recalculated Levy's (1983) figures (by averaging the 



34 

proportion of playbacks for each elapsed day from recording), and found that 58.2 

percent of recordings were played back within one week of their broadcast (see Table 

10). 

Table 10. Percentage of recordings by elapsed days 

Elapsed days Recordings 
% 

0-1 23.8 

1-2 19.7 

2-3 7.4 

3-4 3.2 

4-5 2.2 

5-6 1.3 

6-7 0.6 

0-7 58.2 

Source: Levy and Fink (1984) 

Table 10 shows that over seven days the highest proportion of recordings (24%) were 

played back in less than one day from broadcast, followed by rapidly declining 

proportions as the number of elapsed days increased. Of the remaining 41.8 percent of 

programmes not replayed within seven days, Levy and Fink ( 1984) expected that a small 

proportion would be replayed but most recordings would not be replayed at all. 

In another study, Robbins (1996) reported that "90 percent of all recorded material is 

played back, if viewed at all, within the seven-day period" (p.15). Compared to other 

studies, this estimate seems to be very high; however its accuracy remains unclear as 

Robbins (1996) did not report the research procedure. 

Although analysing the proportion of recordings replayed over time provides some 

insight into time shift viewing lags, the proportion of playbacks is more relevant to the 



35 

measurement of VCR ratings, as this measure excludes recordings that are not played 

back, that is, are not time shifted. 

3.3.2 Proportion of Playbacks 

Research which has analysed the proportion of playbacks, estimates that approximately 

half of all playbacks occur within two days of recording, and a large proportion are 

played back within one week from recording (Fiddick, 1984; Hunt, 1987, AGB McNair, 

1991). Hunt (1987) found that 99 percent of all playbacks were played within one week 

from recording, with 60 percent of playbacks occurring on the same day or the day after 

recording. 

AGB McNair (1991) found that a smaller proportion, that is, 75 percent, of playbacks 

occurred within one week. AGB McNair (1991) probably found lower proportions of 

playbacks compared to Hunt (1987), because Hunt's (1987) study was based on 

recordings played back within 28 days from the day of broadcast, whereas playbacks in 

the study by AGB McNair (1991) had a longer period to be played back2
• The results of 

the AGB McNair (1991) study are presented in Table 11. The highest proportion of 

playbacks were played on the day after recording. 

2. In AGB McNair's (1991) study, the introduction ofFingerPrinting technology in New Zealand determined the 
earliest time that playbacks could have been recorded; see chapter five for details on FingerPrinting technology. 



36 

Table 11. VCR playback (1/7/91 - 28/7/91) 

No. of days recorded Percent 
prior to playback % 

Same 16 

1 35 

2-3 15 

4-6 9 

7-14 7 

14-21 3 

21 Plus 15 

Total 100 

Source: AGB McNair (1991) 

The research suggests that time shift data collected two days after the day of broadcast 

would include half of all playbacks ( or 60 percent of the playbacks which are recorded 

and played back within 28 days), whereas waiting until one week after the day of 

broadcast would include three quarters of all playbacks, ( or 99 percent of playbacks 

which are recorded and played within 28 days). To report VCR ratings, ratings services 

must, therefore, make a compromise between the delay in the release of the ratings and 

the proportion of playbacks included. In the United Kingdom, BARB (the Broadcasting 

Audience Research Board) define time shifting as "the cumulative audience who view a 

particular broadcast as video playback within 163 hours of its transmission" (Robbins, 

1996, p.15). In this case the release of VCR ratings would be delayed by one week, but 

the ratings would include a large majority of the time shifted programmes. 

Research has also found viewing lags to vary across programme genres, recording 

modes, and household channel availability. These variations may create opportunities for 

the earlier release of VCR ratings. 



37 

3.3.3 Variations in Viewing Lags 

Across programme genres, research has found that soap operas, dramas, and current 

affairs are played back faster than other genres, whereas movies have longer viewing lags 

(Levy, 1981; Levy, 1983; Fiddick, 1984; Mojo, 1994). The results of Levy's (1983) 

study are presented in Table 12. He suggested that movies have longer viewing lags, 

because they are often recorded and then stored in personal video libraries (Levy, 1983). 

Table 12. Percentage of recordings replayed during 
diary week by programme type 

Programme type Proportion replayed 
% 

All programmes 52. 7 

Soap operas 81. 7 

Dramatic series 77. 3 

Situation comedies 61.3 

Police/ detective 60.6 

Entertainment series 48.9 

Sports 47.9 

Children's programmes 45.5 

Broadcast movies 45.5 

Pay-TV movies 33.3 

Source: Levy (1983) 

In another study, Levy (1981) found that 88 percent of soap operas and 50 percent of 

news and current affair programmes were played back on the day of broadcast, while 

only 30 percent of all playbacks were played on this day. Considering that such a high 

proportion of soap operas are played back on the day of broadcast, it is possible that 

VCR ratings of soap operas could be included in the overnight ratings. 

Levy and Fink (1984) investigated the viewing lags of daily, weekly and one-shot 

programmes. Of all recordings played within one week, 79 percent were daily 



38 

programmes, 59 percent were weekly programmes, and 43 percent were one-shot 

programmes. As expected, the majority (55%) of daily programmes were played back 

on the same day of broadcast, while most weekly (25%), and one-shot programmes 

(18%), were played back on the day after broadcast. As a result, VCR ratings of daily 

programmes could be reported earlier than the VCR ratings of other programmes. This 

has potentially important implications for media planners and advertisers, who generally 

need to make faster decisions on daily programme slots than weekly programme slots. 

Levy (1983) investigated two other factors that appear to influence viewing lags, namely, 

the activity of viewers while recording is taking place and the number of channels 

available in the household. The results are presented in Table 13. 

Table 13. Percentage of recordings played back during diary week 
by recording mode and cable status 

Proportion replayed 
Recording mode % 

Unattended recording 56.8 

View one, record another 48.0 

View and record 34.3 

Cable status of household 

Cable 

Not cable 

Source: Levy (1983) 

45.4 

58.1 

The proportions of replays by recording mode seem quite self-evident. Unattended 

recordings were most likely to be played back within a week, followed by recordings that 

were taped while another live programme was viewed off-air; in this situation viewers 

are likely to view their favourite programme off-air and tape their second choice. 

Recordings that were viewed as they were recorded were least likely to be played back 

within a week. The proportion of recordings that are not played back in each of the 



39 

recording modes are likely to confound the viewing lag results, as it seems that "view 

and record" recordings would have a lower playback rate than "unattended" recordings. 

Cable households were less likely than non-cable households to replay recordings within 

the diary week. Levy's (1983) explanation was that cable households have a greater 

variety and number of programmes to choose from, and that greater choice reduced their 

need to play back programmes quickly. However, as Levy's (1983) results represent the 

proportion of all recordings played back, cable homes may, in fact, play back fewer 

recordings, rather than take longer to play back tapes. The possible implications of 

Levy's (1983) results are: as the number and variety of stations in New Zealand increase, 

either the viewing lags of time shift programmes will also increase, or fewer recordings 

will be played back. 

3.4 Summary 

It is important that researchers who investigate time shift viewing behaviours, or rating 

companies who report VCR rating are aware of three methodological issues relating to 

the measurement of time shift viewing. The first of these issues is: given the 

accumulated evidence suggesting that a large proportion of VCR recordings are not 

played back, VCR playback is the more logical and conservative measure of time shifting 

behaviour. Secondly, multiple replays of programmes by individuals should not be 

included in VCR ratings or other time shift audience measures as they inflate the actual 

time shift audience. Lastly, since VCR recordings may take months to be played back, a 

time period of one week seems to be the most satisfactory period to measure VCR 

playback as research has estimated that approximately three quarters of all playbacks are 

played within this time. This study investigates these issues further, as discussed in the 

research objectives in the following chapter. 



CHAPTER 4. RESEARCH OBJECTIVES AND 
METHODOLOGY 

4.1 Research Objectives 

40 

The main objectives of this study were to examine the effect of VCR time shifting on 

television audiences and investigate the measurement of VCR ratings. Research to 

investigate these issues was carried out in three stages: the first stage examined the 

extent of time shift viewing in New Zealand, the second stage investigated variations in 

time shift viewing across a number of variables, and, lastly, the study examined issues 

relating to the current and future measurement of time shift viewing. 

4.1.1 The Extent of Time Shifting 

The study examined the extent of time shift viewing in New Zealand to ascertain whether 

it had the potential to greatly affect programme ratings. In particular, the following 

objectives were explored: 

• to compare the average weekly viewing of time shifted material to the viewing 

of pre-recorded material. 

• to estimate the proportion of average weekly television viewing comprised of 

time shifted material. 

The first objective investigated whether the predominant use of VCRs is changing from 

time shift viewing to pre-recorded viewing. In the early eighties, VCRs were 

predominantly used to play home recorded material; however, there is increasing 

evidence that viewers now play a greater proportion of pre-recorded material (Levy, 

1980; Levy, 1981; Kirkham, 1982; Stoessl, 1982; Fiddick, 1984; Darkow, 1984; 

Metzger, 1986; Sims, 1989; Warrens and Thompson, 1992; Mojo, 1994; Jones, 1996). 



41 

Although this trend is evident in New Zealand, New Zealanders aged 50 plus still watch 

more home recorded programmes than pre-recorded material (Mojo, 1994; Jones, 1996). 

This study extends previous research by continuing to monitor the ratio of pre-recorded 

viewing to home recorded viewing. To understand the implications of the ratio, pre­

recorded viewing levels were compared to live viewing levels. The study also examined 

whether the ratio varies across age and gender groups. 

Objective two gives an intimation of the extent of time shift viewing in New Zealand. 

Recent studies suggest that, on average, time shift viewing makes up two to three 

percent of a programme's total audience (Sims, 1989; Robbins, 1996; Jones, 1996). This 

may seem to be only a small proportion of viewing, unlikely to affect overnight ratings 

substantially; however, these generic proportions appear to mask variations which occur 

across individual programmes (Mojo, 1994; AGB Television, 1995; Robbins, 1996). 

This study, therefore, examined the proportion of television viewing constituted by time 

shift viewing, and how the proportion varies across individual programmes. 

4.1.2 Variations in Time Shift Viewing Levels 

Research has found that time shift viewing levels vary across programme genres, 

broadcast channels, times and days of broadcast, and the demographics of the audience. 

The specific effects of these variables on VCR ratings remain less clear; therefore, this 

study investigated the following objective: 

• to examine variations in VCR ratings and VCR rating contributions across 

programme genres, stations, hours of the day, days of the week, and demographic 

groups. 

Firstly, this objective investigated the relationship between the variables and variations in 

VCR ratings. As this study was based on non-experimental data, it was not possible to 

determine whether the relationships were causal, that is, whether the variables caused 

variations ·in VCR ratings. Secondly, the study compared variations in VCR rating 



42 

contributions across each of the variables. VCR rating contributions measure the 

proportion of a programme's audience which is made up of time shift viewers. If 

reasonably large variations in VCR rating contributions exist across the variables, 

correction factors could be developed to include VCR ratings in the overnight ratings. 

4.1.3 Measuring Time Shift Viewing 

The last stage of the study examined the measurement of time shift viewing. This section 

investigated issues relating to the current measures of time shift viewing, and then 

explored the possibility of developing other measures. 

This study examined issues relating to the current calculation of VCR ratings, in order to 

confirm and extend previous research. The following objectives were investigated: 

• to compare the average number of programmes recorded per week to the 

number played back. 

• to examine the average delay between recording and playback. 

The first objective investigated how many recorded programmes are played back. 

Previous research has found that not all recordings are played back; the most 

conservative estimate was that only 33 percent of recorded minutes are played back over 

28 days (AGB McNair, 1991). Consequently, ratings that incorporate VCR recordings 

as a surrogate for time shift viewing ( or VCR playback), overstate the size of the 

television audience. To investigate this further, the study compared the average number 

of programmes recorded and played per week by households, and whether the levels 

varied across programme genres. 

The second objective examined how long it takes for VCR recordings to be played back. 

This has an important role in establishing the period of time before actual VCR ratings 



43 

can be calculated. The few studies that have investigated this issue have found that a 

large proportion of video playback occurs within one week of taping; this proportion has 

ranged from 75 to 99 percent, depending on how many days the recordings had to be 

played back (Fiddick, 1984; Hunt, 1987, AGB McNair, 1991). As well as calculating the 

average viewing lag, this study also compared the viewing lags of nine programme 

genres, as previous research had found that some programme genres are played back 

faster than others (Levy, 1981; Levy, 1983; Fiddick, 1984; Mojo, 1994). 

The second part of this section investigated the future measurement of time shift 

viewing. Previous research suggested that the current methods of calculating VCR 

ratings are either inaccurate or can not be included in the overnight ratings (see chapter 

three). To investigate the possibility of improving the accuracy of the overnight ratings, 

the following objective was explored: 

• to propose methods of estimating accurate VCR ratings overnight. 

Research has not previously investigated the potentiality of estimating VCR ratings; this 

study provided an initial investigation. 

The methodology used to investigate the research objectives of this study is explained in 

the following section. 



44 

4.2 Research Methodology 

The data used in this study was obtained from AGB McNair's people meter panel, which 

is able to measure VCR recording and playback through FingerPrinting® technology. In 

practice this is a small device which is attached to the back of household VCRs; when a 

household records a programme the video tape is electronically imprinted ( or "finger 

printed") with the date, time, and channel on a continuous basis. When the programme 

is played back, the "finger print" is recognised, and the people meter records when 

playback occurred, the duration of playback, and which member(s) of the household was 

watching. 

The panel has a sample size of 440 households, which represents approximately 1200 

individuals who are aged five and over1
• Each household's recording activity and each 

individual's viewing behaviour was studied for a six week period beginning August 26, 

1996 and ending October 7, 1996. The study included seven stations: three national 

free-to-air stations and four Sky TV (pay) stations. 

A variety of different measures were used to quantify time shifting levels, including hours 

per week, number of programmes per week, live and VCR ratings, and VCR rating 

contributions. Live correction factors were also calculated to investigate the possibility 

of estimating VCR ratings. The calculation of ratings, VCR rating contributions, and 

live correction factors are detailed in Appendix A. The results of the study are presented 

and discussed in the following chapters. 

1. The sample is stratified by many variables, including region, area type, age of household shopper, household 
size, and weight of television viewing. 



45 

CHAPTER 5. THE EXTENT OF TIME SHIFTING: 
RESULTS AND DISCUSSION 

The first stage of this study measured the extent of time shift viewing in New Zealand, to 

ascertain whether it had the potential to greatly affect programme ratings. This involved 

comparing levels of time shift viewing to pre-recorded viewing levels to monitor the 

extent that VCRs are used for time shifting (section 5.1). The study also compared 

levels of time shift viewing and live viewing to examine the proportion of television 

viewing which is represented by time shift viewing (section 5.2). 

5.1 VCR Use: Home Recorded versus Pre-recorded Viewing 

In the early 1980s, VCRs were predominantly used to play home recorded material, 

however, there is increasing evidence that viewers now play a greater proportion of pre­

recorded material (Levy, 1980; Levy, 1981; Kirkham, 1982; Stoessl, 1982; Fiddick, 

1984; Darkow, 1984; Metzger, 1986; Sims, 1989; Warrens and Thompson, 1992; Mojo, 

1994; Jones, 1996). This trend is causing concern to broadcasters and advertisers alike, 

as pre-recorded tapes contain no advertising and take viewers away from live broadcast 

audiences. The net consequence of this is that programme and commercial audiences 

may be diminishing. 

To investigate the use of VCRs, the total number of hours that all panel members spent 

viewing time shifted and pre-recorded material per week was calculated; the viewing 

hours were averaged over six weeks. The results presented in Table 14 are consistent 

with recent research (Jones, 1996); that is, New Zealanders now spend more time 

viewing pre-recorded material than they spend viewing time shifted material. 



Table 14. 

Material 

Time shifted 

Pre-recorded 

Total 

Playing of time shifted material versus pre-recorded 
material by all panel members per week 

Average hours 
perweek1 

n % 

558 

756 

1314 

42 

58 

100 

I. Includes multiple replays of tapes 

46 

Although the results reported in Table 14 show that viewers watch more pre-recorded 

than time shifted material, previous studies have found that this pattern varies across 

demographic groups (Mojo, 1994; Jones, 1996). Tables 15 and 16 explore this issue 

further. 

Table 15. 

Material 

Time shifted 

Pre-recorded 

Total 

Proportions of time shifted viewing versus pre-recorded viewing by 
gender 

Male Female All 

Hours1 % Hours1 % Hours1
·
2 % 

197 37 346 47 558 42 

333 63 389 53 756 58 

530 100 735 100 1314 100 

I. Average viewing hours per week for all panel members, including multiple replays. 
2. The average hours for all people are not equal to the sum of the average hours for males and females due to 
missing cases. 

Table 15 compares levels of time shifted and pre-recorded viewing by gender. Females 

tended to use their VCRs to view approximately equal proportions of time shifted 

material and pre-recorded material, whereas males watched more pre-recorded material. 

Men and women also differed in the amount of VCR use; that is, compared to males, 

females viewed more time shifted and pre-recorded material. 



47 

VCR usage behaviour also varied across age groups. Table 16 shows that, as viewers' 

got older the proportion of time shift viewing relative to pre-recorded viewing increased. 

Other New Zealand research has found similar results (Mojo, 1994; Jones, 1996). 

Table 16. 

Material 

Time shifted 

Pre-recorded 

Total 

Proportions of time shifted viewing versus pre-recorded viewing by 
age groups 

Age Group 

5-19 20-34 35-50 50+ All 
%1 %1 %1 %1 % 

29 36 40 63 43 

71 64 60 37 58 

100 100 100 100 100 

1. The actual number of hours that all panel members in each age group spent viewing time 
shifted and pre-recorded material per week is presented in Appendix B. 

Further analysis of time shift viewing by gender within the 50 plus age group revealed 

that both males and females viewed more time shifted material than pre-recorded 

material. However, females watched a higher ratio of time shifted material to pre­

recorded material; females also viewed more hours of time shifted and pre-recorded 

material than males. 

The results of this study confirm that the aggregate pattern of VCR use obscures 

variations across demographic groups, providing potentially important information for 

the study of variations in time shift viewing levels. Tables 15 and 16 suggest that 

programmes which are predominantly viewed by females or people aged 50 plus may 

have higher levels of time shift viewing. Thus, it could be hypothesised that one of the 

influencing variables on VCR ratings is the predominant live audience for a programme. 

This hypothesis is explored further in section 6.5. 



48 

Overall, the results of this study are consistent with the patterns reported in earlier 

studies. In particular, these findings confirmed that the extent of time shifting relative to 

pre-recorded viewing is decreasing, and that people now use their VCRs to view pre­

recorded material more than they do to view time shifted material. The data do not 

reveal whether the level of pre-recorded viewing is increasing or decreasing relative to 

live viewing. However, since pre-recorded viewing is equivalent to only four percent of 

total viewing1
, it does not appear to be a significant threat to live viewing. This suggests 

that time shift viewing will also have little influence on live viewing levels. To 

investigate this further, the following section compares levels of time shift viewing to live 

viewing levels. 

5.2 Time Shift Viewing versus Live Viewing 

Recent studies have found that two to three percent of a programme's audience is 

constituted by time shift viewers (Sims, 1989; Robbins, 1996). To explore this issue, 

time shift viewing proportions were calculated using several measures. Table 17 

presents the proportion of television viewing represented by time shift viewing, based on 

average hours per week and average ratings. 

Table 17. Proportion of television viewing constituted by time shift viewing 

Hours per week 
Excluding 

multiple replays 
Viewing n % 

Time shifted 441 2 

Live 17,869 98 

Total 18,310 100 

I. Average rating for all programmes viewed live. 
2. Excluding multiple replays 

Including 
multiple replays 

n % 

558 3 

17,869 97 

18,427 100 

Average rating1 

% 

0.042 

2.00 

2.04 

% 

2 

98 

100 

1. Pre-recorded viewing/ total viewing (excluding multiple replays) = 756 hrs / 19066 hrs= 4% 



49 

Over the sample period of six weeks, panel members viewed, on average, 17,869 live 

hours of television programmes per week across the seven stations; this is approximately 

2.6 hours of viewing daily per individual. This level of viewing appears to be quite high 

and, in fact, is probably at its highest for the year, considering that the study was 

conducted in winter. 

To compare live viewing hours to time shift viewing hours, multiple replays of 

programmes by individuals may be included or excluded. Including multiple replays 

reveals the total amount of time that panel members spent time shift viewing, whereas 

excluding multiple replays is more relevant to media planners as this measure enables 

calculations of the added audience to live programmes2
. Using the latter measure, panel 

members viewed 441 hours of time shift material, on average, per week; this is 

equivalent to four minutes daily per individual. Overall, time shifting accounts for a very 

small proportion of total viewing hours (2%-3%), even when multiple playbacks are 

included. 

The proportion of television viewing constituted by time shift v1ewmg can also be 

measured by comparing live ratings and time shift (VCR) ratings. This study calculated 

average ratings for all programmes viewed live over five weeks3 on seven stations. The 

final column in Table 17 supports the earlier finding, and the results of previous research 

(Sims, 1989; Jones, 1996; Robbins, 1996), that time shift viewing accounts for only a 

very small proportion of total viewing. Analysing the ratings over only the three free-to­

air stations produced a higher average live rating of 3.6 percent and a higher average 

time shift rating of 0.07 percent. However, time shift viewing still only constituted two 

percent of total viewing. 

2. The figures that exclude multiple replays may contain time shifted programmes that were previously viewed by 
the individual while recording. It was not possible to exclude these programmes with the data available. 

3. Analyses of programme ratings are based on five rather than six weeks, because programmes recorded in the 
sixth week had less than one week to be played back. By excluding week six, recordings had at least one week 
to be played back, which is the length of time within which a large majority of programmes are played back 
(Fiddick, 1984; Hunt, 1987; AGB McNair, 1991). 



so 

Considering that popular programmes or episodes typically achieve around 20 rating 

points, the average live rating reported in Table 17 appears to be very low. The majority 

of programmes in the sample had ratings under 0.5 and only one percent of the 

programmes had ratings over 20, this appears to be a representative distribution of live 

ratings for the following reasons. The programmes with ratings around 20 percent 

typically only appear on the free-to-air stations in prime time. Obviously there are many 

programmes that do not fall into this category and have much lower ratings . 

Furthermore, to be included in the analysis, programmes only had to be viewed by one 

panel member, which was the case for many early morning programmes. The 

introduction of 24-hour programming on all of the stations studied, except for Sky 

Movies, is likely to have increased the number of early morning programmes with very 

low ratings. In addition, the Sky TV stations have lower live audiences than the free-to­

air stations. 

The average time shift (VCR) rating, reported in Table 17, also seems very low (0.04). 

In some cases, however, VCR time shifting can increase a programme's live rating by 

one or two rating points. The implication of this is that these programmes have an 

additional 35,000 to 70,000 viewers (1 % to 2% of the viewing population) who are not 

accounted for by live ratings. Table 18 presents the programmes from the sample which 

had VCR ratings greater than one. 



51 

Table 18. Top VCR rating programmes 

Programme Date Start Station Genre VCR Live Total VCR rating 
time rating rating rating con tri bution2 

% % % % 
Montana ST1 15 Sept 8.40pm TVNZl Drama 2. 1 17.5 19.6 10.6 
Free Willy 15 Sept 8.30pm TVNZ2 Movie 2.0 15.2 17.2 11.4 
Montana ST1 22 Sept 8.40pm TVNZl Drama 1.5 17.3 18.7 7.8 
Cybill 10 Sept 8.30pm TVNZ2 Comedy 1.4 12.8 14.2 10.2 
The Topp Twins 5 Sept 7.30pm TV3 Comedy 1.3 13.5 14.8 9.0 
Montana ST1 I Sept 8.35pm TVNZI Drama 1.3 21.1 22.4 5.7 
Montana ST1 8 Sept 8.35pm TVNZI Drama l.3 19.9 21.1 6.0 
Coronation Street 2Oct 7.30pm TVNZI Soap 1.2 17.1 18.3 6.6 
Montana ST1 29 Sept 8.40pm TVNZl Drama 1.1 19. l 20.3 5.6 
Inside N.Z. 12 Sept 8.30pm TV3 Docume I.I 11.9 13.0 8.4 
Indecent Proposal 4 Sept 8.30pm TVNZ2 Movie 1.1 12.5 13.5 7.7 
The 3 Musketeers 22 Sept 8.30pm TV3 Movie 1.1 10.0 11.0 9.5 
Robin Hood 22 Seet 8.30em TVNZ2 Movie 1.0 10.3 11.3 9.1 
I. Montana Sunday Theatre 
2. VCR rating contribution = VCR rating/ total rating. VCR rating contributions were calculated from figures 

with six decimal places. 

The results in Table 18 suggest that some programmes can have very large time shift 

audiences. Montana Sunday Theatre achieved the highest VCR rating (2 rating points) 

when Pride and Prejudice, a six-part drama, screened. Since overnight ratings do not 

include VCR ratings, approximately 73,500 viewers4 were not included in the rating of 

this episode. Previous research has also found that some programmes have much higher 

VCR ratings or time shift audiences than other programmes. For example, in the United 

Kingdom, Robbins (1995) found VCR ratings as high as 3.2 percent; and AGB 

Television (1995) found that 2.3 million time shift viewers watched Wallace and Gromit. 

VCR rating contributions were also computed and presented in Table 18 to examine the 

proportion of the total audience not included in the overnight ratings. All of the VCR 

rating contributions in Table 18 were considerably higher than the average VCR rating 

contribution for all programmes (which was 2%, see Table 17). The highest VCR rating 

contribution was 11 percent for the movie Free Willy. Programmes with low live ratings 

can have much higher VCR rating contributions; the highest contribution over the sample 

period was 92 percent for a comedy, All in the Family, which screened at 5am on 3 

4. Time shift viewers = VCR rating x population => 73,500 = 2.1 % x 3,500,000 



52 

September. Previous research has also found higher VCR rating contributions for 

particular programmes. The top VCR rating programmes in the studies of AGB 

television (1995) and Mojo (1994) had VCR rating contributions between 22 and 24 

percent. These results confirm that time shift viewing has a greater effect on the total 

audience of some programmes than others. 

Further examination of the results in Table 18 revealed that many of the top VCR rating 

programmes were screened simultaneously. Robbins (1996) found similar results for two 

of the top twenty VCR rating programmes in November 1995. There do not seem to be 

any extensive studies on this issue. It seems likely that programme scheduling may have 

an influence on the VCR rating a programme will achieve. 

Another interesting pattern in Table 18 is that the top VCR rating programmes all have 

live ratings of at least ten percent. Simple regression of VCR ratings against live ratings 

produced an R
2 

value of 0.35 (F = 2806, Signif. F < 0.0001), which supports the positive 

relationship between the two variables: higher live rating programmes had higher VCR 

ratings by a factor of 1.79. Figure 7 represents these findings graphically. Programmes 

were sorted by their live rating, and the average live and VCR ratings were calculated for 

every 250 consecutive prograrnmes5
. 

5. Every 250 programmes were chosen to simplify the points on the graph, that is, rather than having 5,310 points 
representing each programme, means were calculated to produce approximately 20 points. 



53 

Figure 7. VCR ratings versus live ratings 

0.35 

0.3 • 
0.25 

en 
C :; 0.2 
a: 
a: 0.15 • 0 
> 

0.1 

0.05 ••• • • 
0 

.,. 
0 2 4 6 8 10 12 14 

Live Rating 

Robbins (1996) also found that "programme types which people choose to record 

automatically follow those programmes which would be most popular for 'live' 

viewing" (p.15). However, earlier studies by Kirkham (1982) and Mojo (1994) reported 

that there was little correlation between live audiences and time shift audiences. 

Kirkham (1982) came to this conclusion after finding that the rank order of the top ten 

recorded programmes was not exactl