Reconceptualising toddler aggression : a thesis presented in partial fulfilment of the requirements for the degree of Master of Education at Massey University, Palmerston North, New Zealand

Abstract

This study used sorting techniques and Multidimensional Scaling analyses (MDS) to interrogate the comprehensiveness of a toddler aggression construct - Aggressive Behavior (AGG) - as empirically defined by the Child Behavior Checklist for Two and Three-year-olds (CBCL/2-3). MDS provided a way of unravelling the latent dimensional structures of the checklist item set and creating an inter-item semantic-like map. This map acted as a tool in the summary of patterns in sets of archival CBCL toddler data to reveal latent coherence or dimensional consistency across toddlerhood with specific regard to the aggression or aggressive-like behaviour construct – the main research goal of the present study. First New Zealand participants (N=70, x=30 years) completed four independent semantic-like matching tasks by sorting CBCL/2-3 checklist items according to their "face value". MDS algorithms transformed individual data into a map and hierarchical trees (h-trees) showing inter-item proximities. After validation of the map clusters sets of archival CBCL data were represented and interpreted in the 5-dimensional MDS solution (P=5) as vectors using PROFIT analysis. A measure of the stability of the vector components in terms of the amount of common variance captured across 24 to 42 months - of – age. demonstrated better fit than CBCL subscale stability for the first three dimensions. Candidates for dimensional stability across toddlerhood indicated by the MDS analyses and map were suggested. Replication of the toddler map. the second objective, involved creating an expanded item set that included items from the CBCL/4-18. The new Combined item set was then sorted following identical procedures by a different group (N=49, x=30). PROFIT analyses of archival 4-18 data on the rotated Combined configuration was compared with stability of Achenbach's CBCL/4-18 subscales between 60-months- and 72-months-of-age but correlations were no better than chance. Additional analyses were undertaken that revealed the archival CBCL/4-18 items had little variance when interpreted in the MDS solution. This study succeeded in identifying alternative candidates for continuity of aggressive - like behaviour across toddlerhood in patterns in raw CBCL data that may contribute to the reported CBCL/2-3 Aggressive Behavior construct stability. Three alternative constructs are suggested: a construct which features high frustration, anger and resistance to control - believed to interact with punitive or restrictive parenting practices, and central in theories of the development of coercive parent-child relationships; a construct which appears to index insecure attachment styles; and a construct reflecting toddlers' developing ability to control their attention and behaviour.