Low self-control, gender, race, and offending in late life

Abstract

Self-control theory has been one of the most scrutinized general frameworks of crime for over 20 years. A majority of evidence pertaining to the theory, however, is derived from samples of teenagers and young adults. Relatively little information exists regarding whether self-control explains offending among people in late adulthood. As such, the generality of the framework has yet to be fully examined. This study uses a representative sample of people aged 60 years and older from interviews conducted in Arizona and Florida. The current study tests two key propositions regarding the generality of the theory: (1) the extent to which self-control accounts for the relationship between demographic variables and criminal offending, and (2) the invariance thesis which stipulates that self-control will have a uniform effect on offending across social groups. The analyses reveal two findings regarding theoretical generality: (1) low self-control explains late-life criminal behavior but does not account for the relationship between offending and gender, and (2) low self-control has an invariant effect on offending across gender and race when measured behaviorally. Taken together, the analyses address important elements of the supposed generality of self-control theory and extend the framework's scope to the explanation of offending in late life.

Keywords:

• low self-control
• gender and race gaps in offending
• invariance thesis
• criminal behavior
• late adulthood

Acknowledgements

The opinions, findings, and conclusions or recommendations expressed in this publication are those of the author and do not necessarily reflect those of the Department of Justice. The author thanks Michael Reisig, Kristy Holtfreter, and Travis Pratt for their comments on earlier drafts of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The American Association for Public Opinion Research’s (2011) protocol was used for the response rate calculation. Specifically, the response rate excluded cases of unknown eligibility (e.g., answering machines or no answer), known ineligibility (e.g., disconnected numbers, business/fax lines, or no individual in household 60 years of age or older), and individuals who failed the cognitive screener. A 10 callback rule was used in an attempt to contact unknown eligibility cases (e.g., no answer on initial call attempts).

2. Table 1 also reveals that all correlations fall well below an absolute value of 0.70 which is a threshold used to indicate potentially harmful collinearity (Tabachnick & Fidell, 2007). Additional diagnostic tests were conducted to determine whether problematic collinearity existed in the multivariate regression models presented in Tables 2 and 3. These tests provide confidence that the observed correlations between the independent variables do not produce multicollinearity because all variance inflation factors (VIF) fell below the 4.0 threshold (Tabachnick & Fidell, 2007) and all condition indices were below the threshold of 30 (Belsley, Kuh, & Welsch, 1980; Mason & Perreault, 1991).

3. Behavioral measures of low self-control can sometimes have limited discriminate validity with offending scales. In the case of the present study, taking a ‘higher dosage of medicine’ (a behavioral low self-control item) is conceptually similar to one item used in the offending scale (‘Took medication that was not prescribed to you’). As would be expected, these items were correlated with one another. Given the importance of examining the relative influence of attitudinal and behavioral indicators of low self-control, it was necessary to examine whether this overlap was responsible for any of the results. As a sensitivity analysis, the ‘medication’ item was removed from the offending frequency scale and all models reported below were re-estimated. All substantive results were unchanged using the 6-item offending scale. Therefore, one can have confidence that the behavioral low self-control results are not attributed to tautology. All analyses reported in the main text use the seven-item offending scale.

4. The author thanks the Editor and anonymous reviewers for suggesting this supplemental analysis.

Additional information

Funding

This research was supported in part by Award No. 2010-IJ-CX-0008, awarded by the National Institute of Justice, Office of Justice Programs, US Department of Justice.