Can self-control change substantially over time?: Rethinking the nature and role of self-control in Gottfredson and Hirschi's general theory of crime

Abstract

The primary goal of this study is to verify if the changing level of structural and situational `sensitivity' to costs and benefits associated with deviant behaviors (e.g., Hirschi (2004) and Gottfredson (2006)'s redefined self-control, Tittle, Ward, and Grasmick's (2004) "desire to exercise self-control," Wikström and Treiber's (2007) "situationally-based" self-control) is associated with the changing level of more general `ability' to measure costs and benefits within individuals (e.g., Gottfredson and Hirschi's (1990) trait-like self-control, Tittle, Ward, and Grasmick's (2004) "capacity for self-control," Wikström and Treiber's (2007) "executive capability"). More importantly, to better disentangle the causal mechanisms underlying stability and change in offending behaviors over time, This study examines how low self-control as one of the constituent elements of offending propensity changes over time in the general population and across different study groups using both a hierarchical linear model (HLM) and a second-order latent growth model (LGM). Then, structural equation modeling (SEM) is employed to examine the on-going processes of cumulative advantage and disadvantage by more explicitly testing the bidirectional relationship of key theoretical constructs (e.g., self-control vs. social control/bond) over time.

In contrast to the Gottfredson and Hirschi's prediction, this study found meaningful differences in the growth pattern of self-control among individuals in the population in general and especially across different study groups. Interestingly, the changing level of social control/bond triggered by experimental conditions accounted for the between-group difference observed. The same pattern persists when different analytic techniques and model specifications are applied to test the same research hypotheses, which suggests that the results are not an artifact of measurement error, model specification, or statistical methods. Most of all, this study was able to better disentangle the `long-term' relationship between self- and social control variables, which is found to be more dynamic and bidirectional than previously hypothesized.