Explaining the Gender Gap in Juvenile Shoplifting: A Power-Control Theoretical Analysis

Abstract

Shoplifting is one of the crimes with the smallest gender gap among all offense types. Contrary to common stereotypes, males commit shoplifting more frequently than females. Apart from the insight that the share of female offenders is highest for crimes that are well compatible with the prevailing female role, the roots of the gendered distribution of shoplifting have not been extensively studied. Its focus on gender roles and gendered socialization processes makes power-control theory an obvious explanatory approach, yet it has never specifically been utilized to examine the gender gap in shoplifting. This study attempts to close this research gap. Based on a large-scale student survey from Austria, this study scrutinizes the theory's potential to account for the gender divide in juvenile shoplifting. Results provide somewhat more support for the control and risk-related parts of the theory than for its power component.

Notes

¹Acknowledging the comparatively small gender gap in common property offenses, however, Hagan and colleagues (1990) based their reply to the criticisms levied by Jensen and Thompson (1990) on a special analysis of data on adolescent theft delinquency in general.

²Initial identification of family types specified the ideal types as patriarchal and egalitarian. However, criticisms highlighting the fact that even women in authority positions hold lower status in the greater patriarchal society and that women who leave the workforce to rear children may not lose household power (see, e.g., Morash and Chesney-Lind 1991) yielded a shift in references to discuss these types as more or less patriarchal.

³Leiber and his colleagues (Leiber and Wacker 1997; Mack and Leiber 2005) have questioned, for example, where single-parent families fit into the theoretical specification. Early descriptions of PCT included these families in the category of egalitarian, or less patriarchal, structures because there is no imbalance of power within the household (Hagan 1989; Hagan et al. 1987, 1990, 1993; McCarthy et al. 1999). Other researchers have noted that while an imbalance of power may not be present at the family level, the elements of patriarchy that influence decision-making at the societal level may have a more profound impact for single-parent families, specifically mother-headed households, than for two-parent families, and that there are differences across single-parent households in the degree to which they must rely on the larger social system for assistance. In other words, there is variation in the experience of “power” among single-parent households (Blackwell 2000; Leiber and Wacker 1997; Mack and Leiber 2005; Morash and Chesney-Lind 1991).

⁴An exception is Hagan and colleagues (1990) article where they focus on general theft criminality.

⁵Since the ISRD-2 draws solely on juveniles from urban regions, the addressed 1-year-prevalence rates might be a bit higher than for the overall population.

⁶Calling shoplifting a common crime is also warranted by the comparatively weak inverse relationship between shoplifting activity and social class. Shoplifting rates to not vary dramatically across classes (Klemke 1992).

⁷Even these low amounts will be overestimated since the risk of being reported increases with the value of the stolen goods (Wittenberg 2009).

⁸To realize a net sample of 92 schools, 115 schools had to be contacted. This corresponds to a participation rate of 80%.

⁹In Austria, for the age groups surveyed here, three types of schools are available: grammar schools, comprehensive schools, and schools for children with special needs. Due to their small number, grammar schools and schools for children with special needs were oversampled.

¹⁰The online research tool EFS Survey was used to implement the questionnaire (www.unipark.com).

¹¹With r = .62 (p = .000), maternal and paternal control are correlated substantially, but not closely enough to cause problems with multicollinearity.

¹²“Your [mother/father] is—currently gainfully employed / currently unemployed / a housewife / a pensioner.”

¹³“Does your [mother/father] have co-workers, whom she/he can order what to do? no / yes.”

¹⁴The artifact hypothesis is challenged by evidence that the correlation between subjects' delinquency and friends' delinquency also emerges when different data sources are used to assess both concepts (Matsueda and Anderson 1998).

¹⁵A “differential vulnerability hypothesis” also has been raised, positing that boys are more susceptible to peer influence than girls, even when the actual level of peer exposure is held constant (Moffitt et al. 2001; Warr 2002). Empirical research provides more support for the exposure thesis than the vulnerability thesis (Weerman and Hoeve 2012).

¹⁶The preponderance of co-offending applies both to boys (74%) and girls (70%) (p > .05).

¹⁷According to a Lagrange multiplier test, fitting a Poisson regression is thwarted by a significant overdispersion of the incidence measure (p ≤ .001).

¹⁸Such an asymmetric distribution of maternal control can be demonstrated both for instrumental (p = .000) and relational (p = .000) control.

¹⁹A detailed analysis shows that fathers direct both instrumental (p = .022) and relational control (p = .003) specifically to sons.

²⁰With p = .076, significance is barely missed by the corresponding effect parameter (B = −0.59).

²¹In the reduced power-control model, the significance threshold is missed only by a hair's breadth by the gender-shoplifting-effect (B = +0.76; p = .054).

²²Confidence in this theory-conforming finding is attenuated by the fact that the difference in the family type specific gender–delinquency relationships is not large enough to meet significance criteria. To test for cross-group variation of the gender gradients, negative binomial regression models were first fitted for both types of households that solely contained gender as predictor of shoplifting frequency (see Table 6), then the statistical test for the equality of regression coefficients proposed by Paternoster and colleagues (1998: 862) was conducted. In terms of the resulting test statistic (Z = 0.58; p = .562), no significant difference between the conditional regression coefficients could be established. The lack of evidence for systematic cross-group variation of the gender–shoplifting effects certainly questions a core assumption of PCT, but also should not be overestimated. According to more recent methodological works (Allison 1999; Mood 2010), applicability of this coefficient difference test in non-linear models appears to be problematic. It is based on the assumption that the true residual variance is the same in all compared groups, which is equal to demanding that the influence of omitted predictor variables is identical for all subgroups—a requirement that is difficult to fulfill.

²³To calculate a Z-statistic that can be used to determine whether two regression slopes differ significantly, Paternoster and colleagues (1998:862) recommend dividing the difference between the two conditional regression coefficients by the square root of the sum of the squared standard errors of these coefficients.

²⁴The bigger gender divide for maternal control in egalitarian households observed in our sample results from an increased control of daughters in these families. The reason for a comparatively intense control of girls in egalitarian families may be found in the fact that mothers, who themselves are successful in job and career, also attach great importance to their daughters being vocationally successful. Attention and supervision may be employed as tools to promote education and ambition.

²⁵Calculating for each respondent the difference between received maternal and paternal control and comparing the mean difference scores across boys and girls with a T-test demonstrates a significantly greater parental control differential for daughters (T = 8.67; p = .000).

B … unstandardized regression coefficient Z … B/S.E. p … error probability.

²⁶Once again, comparison of the explanatory power of the predictor variables is based on Z-statistics.

²⁷In egalitarian households for gender differences in paternal control the significance threshold is missed only by a hair's breadth (p = .068).

²⁸This is true also for alternative classifications of the two family types. For example: Construing egalitarian families analogs to Hagan himself (1989:223 ff.) by combining “single mother households” with “both parents command class households” and contrasting this group with patriarchal families constructed as before does not change the findings.

Additional information

Notes on contributors

Helmut Hirtenlehner

HELMUT HIRTENLEHNER is Associate Professor of Criminology and head of the Centre for Criminology at the University of Linz (Austria). His current research interests include institutional-anomie theory, power-control theory, and fear of crime. Recent publications were “Anxieties About Modernization, Concerns About Community, and Fear of Crime: Testing Two Related Models” (with Stephen Farrall) in the International Criminal Justice Review (2013) and “Culture, Institutions, and Morally Dubious Behaviors: Testing Some Core Propositions of the Institutional-Anomie Theory” (with Stephen Farrall and Johann Bacher) in Deviant Behavior (2013).

Brenda Sims Blackwell

BRENDA SIMS BLACKWELL is an Associate Professor in the Department of Criminal Justice and Criminology, and an Affiliate Faculty member of the Women's Studies Institute, the Gerontology Institute, and the Institute of Public Health at Georgia State University. Her research reflects interests in women, crime, and criminal justice and criminological theory, as well as sentencing strategies and outcomes. She has published numerous journal articles and book chapters on power-control theory and additional articles on system responses to female offenders, and treatment of domestic violence cases. Recently, she has been engaged in projects examining responses to the commercial sexual exploitation of children.

Heinz Leitgoeb

HEINZ LEITGOEB is a Research Associate at the University of Linz, Austria. His main areas of research include deviant behavior, quantitative methods, and sociology of education. His most recent publications are “Educational Participation of Youth with Migration Background in Austria” (in German, 2012, with Norbert Lachmayr) and “Modeling Sexual Recidivism. A Comparison of Semi-parametric and Parametric Survival Models” (in German, in print, with Johann Bacher and Helmut Hirtenlehner).

Johann Bacher

JOHANN BACHER is Professor of Sociology and Research Methods at the University of Linz (Austria). His main research areas are quantitative methods, especially complex sampling and data analysis, social inequality and sociology of childhood, and deviant behavior. He has published more than 150 articles and books in his research fields. Recent publications include “Cluster Analysis” (in German, 2010, with Andreas Poege and Knut Wenzig) and “Culture, Institutions, and Morally Dubious Behaviors: Testing Some Core Propositions of the Institutional-Anomie Theory” (published 2013 in Deviant Behavior, with Helmut Hirtenlehner and Stephen Farrall).