Self-Control and Intimate Partner Violence: Does Gender Matter?

ABSTRACT

The present study focuses on the interrelationships among gender, self-control, and intimate partner violence (IPV). The extant bodies of literature suggest that both IPV and low self-control are gendered phenomena. That is, males tend to display lower mean levels of self-control than females, and there are notable differences in IPV between males and females. Since both phenomena are gendered, it is plausible that the relationship between self-control and IPV may also be gendered. To that end, the purpose of the present study is to explore the role played by gender in the relationship between self-control and IPV. To accomplish this, a sample of currently dating undergraduate and graduate college students (n = 960) was analyzed. The first objective is to determine whether self-control and IPV vary across gender. The second objective is to examine whether self-control predicts IPV differently by gender. Results indicated partial support for Gottfredson and Hirschi’s general theory of crime. Self-control operated similarly on IPV for both males and females, supporting the gender-neutrality of their theory. However, the interaction between gender and self-control did not significantly impact IPV offending. Moreover, the gender gap in crime as it relates to self-control remains in question as females were more likely than males to commit more types of IPV. Implications and directions for future research are discussed.

Acknowledgments

The authors would like to thank the anonymous reviewers at Deviant Behavior for commenting on earlier versions of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The Revised Conflict Tactics Scale (CTS-2; Straus et al. 1996) addresses some of the limitations of the CTS by including more types of behaviors, and injury and physical outcome measures (DeKeseredy and Schwartz 1998).

² Researchers contacted instructors to secure permission to administer the survey in each randomly selected class. Students had the option to decline participation, and no extra credit was offered for participation.

³ The authors acknowledge that data collected in 1995 is unlikely to reflect current trends in IPV. However, the focus of the current study is on the theoretical and empirical relationship between self-control and IPV, which should not be period specific, that is, limited to data collected at some but not other points in time (Cochran et al. 2016).

⁴ Although 985 students reported being involved in a current relationship at the time of data collection, 25 respondents did not indicate their sex on the survey. As the current study includes gendered analyses, these 25 respondents were excluded from the final sample.

⁵ Current data were collected before the Revised Conflict Tactics Scale was developed (CTS-2; Straus et al. 1996).

⁶ Five of the IPV items had strong factor loadings between .74 and .83. The three remaining IPV items with weaker factor loadings were the less frequently reported IPV items: beating up (.34), threatening with a gun/knife (.43), and using a gun/knife (.10). Additional factor analyses were conducted without these three items; however, removal from the scale did not substantially increase the Cronbach’s alpha reliability coefficient (.786). Therefore, all eight IPV items were retained in the final offending scale to maintain consistency with the CTS.

⁷ Although the unidimensionality of self-control has been questioned, the cognitive scale developed by Grasmick et al. (1993) has been found to maintain construct validity even in the prediction of more serious forms of crime (Longshore and Turner 1998; Longshore et al. 1996; Piquero and Rosay 1998). Analyses using alternative measures of self-control (see DeCamp 2015; Pechorro et al. 2022) revealed no substantively or statistically different final results.

⁸ A principal components factor analysis was performed on the 24 self-control items intended to measure the concept of self-control, which produced six factors with eigenvalues greater than one. However, the scree discontinuity test revealed a single factor solution. Twenty of the self-control items had strong factor loadings between .40 – .62. The four remaining self-control items with weaker factor loadings were items measuring physical activities, a result comparable to other studies utilizing a similar scale (Grasmick et al. 1993; Piquero and Tibbetts 1996). Additional factor analyses were conducted without these items; however, removal from the scale did not substantially increase the Cronbach’s alpha reliability coefficient (.842). Therefore, all twenty-four self-control items were retained in the final self-control scale.

⁹ In a similar fashion, Sellers (1999) also constructed one opportunity variable from these two items (e.g., frequency of interaction with one’s partner and whether the partners were cohabitating).

¹⁰ Seven of the prior IPV offending items had strong factor loadings between .50 – .90, with one moderate factor loading (.44) for the least reported physical aggression item (e.g., used a knife or gun). An additional factor analysis was conducted without this item; however, removal from the scale did not substantially increase the Cronbach’s alpha reliability coefficient (.835). Therefore, all eight prior IPV offending items were retained.

¹¹ All eight prior IPV victimization items had strong factor loadings between .50 – .90 (Cronbach’s alpha = .893). Therefore, all eight prior IPV victimization items were retained in the final offending scale to maintain consistency with the CTS.

¹² We acknowledge that a measure of sex (male/female) is not synonymous with the theoretical construct of gender, which is far more complex. However, to the extent that gender identity is often consistent with sex (Meerwijk and Sevelius 2017), the current study uses sex as a proxy measure of gender. Gender is used throughout the manuscript as a theoretical concept while sex refers to the measured variable.

¹³ The authors chose to utilize SEM as it has a few advantages over other multivariate analysis techniques. For instance, in multiple regression, paths within the model are analyzed iteratively. This process is not able to accurately account for the variance in all of the measures simultaneously, resulting in possible bias in estimates (Kline 2004). SEM, however, allows for the examination of multiple paths at the same time, and thus, allows one to correct for these measurement issues (Bollen 1989; Kline 2004).

¹⁴ Tables of results are available upon request.

¹⁵ Those who had ever used resistance in a dating relationship had a weak, yet positive association with committing more types of IPV (r = .147). Those who had higher levels of opportunity had a positive association with committing more types of IPV (r = .121). Those who perpetrated IPV on more previous partners had a moderately positive association with committing more types of IPV (r = .406). Those who were victimized by more previous partners had a modest positive association with committing more types of IPV (r = .182).

¹⁶ The Revised Conflict Tactics Scale (CTS-2; Straus et al. 1996) addresses some of the limitations of the original CTS by including more types of behaviors, and injury and physical outcome measures (DeKeseredy and Schwartz 1998).

Additional information

Notes on contributors

Laura M. Gulledge

Laura M. Gulledge is an Assistant Professor in the School of Criminal Justice, Forensic Science, & Security at The University of Southern Mississippi. Her research interests include intimate partner violence, intervention strategies for juveniles, and correlates of juvenile delinquency and recidivism. She has recently published in Journal of Experimental Criminology and Journal of Personality Assessment.

Christine S. Sellers

Christine S. Sellers is a Professor in the School of Criminal Justice at Texas State University. She is co-author with Ronald Akers and Wesley Jennings on Criminological Theories: Introduction, Evaluation, and Application, now in its seventh edition. Her research interests include criminological theories and the role of gender in the explanation of criminal and delinquent behavior.

John K. Cochran

John K. Cochran is a Professor of Criminology at the University of South Florida. His research interests involve tests of micro-social theories of criminal behavior, the social control functions of religion, and examinations of issues in macro-social deterrence and rational choice perspectives. He has recently published in the Journal of Research in Crime and Delinquency and Criminology.