جنسیت، پیوندهای اجتماعی و بزهکاری: مقایسه مدل های پسران و دختران

Abstract Past research has assessed gender differences in delinquency due to differential social controls, yet important questions remain regarding gender and social bonding. As much of this work was premised on Hirschi's measurement of the social bond, we examine whether gender moderates two parts of the social bond: the measurement of the social bond and structural differences between the social bond and delinquency. Using multiple-group structural equation modeling, we find that neither the measures of the social bond nor their relationships with property crime are gender-specific. The structural relationship between the elements of the social bond and violent delinquency differs slightly for boys and girls. We discuss implications of this research for social control theory, measuring the social bond and for gender-specific theories of social bonding and control.

. Introduction Feminist criminologists have suggested that the causal process of criminal offending differs by gender and, as such, the field of criminology has seen a growth of gender-specific theories of criminal offending (Hagan et al., 1987; Heimer and DeCoster, 1999; McCarthy et al., 1999; Ogle et al., 1995; Steffensmeier and Allan, 1996). In particular, several authors have suggested a gendered process of social control stemming from differences in familial bonds and control (Hagan et al., 1987; McCarthy et al., 1999; Heimer and DeCoster, 1999; Ogle et al., 1995). It was suggested that girls and boys experienced different familial processes, developed different types of social bonds or were differentially controlled. Such theoretical developments are based on two lines of theory and research: feminist research on gender as structure (Risman, 1998; West and Zimmerman, 1987) and Hirschi's social bonding theory (Hirschi, 1969). Researchers are increasingly suggesting that gender as structure is created, maintained, and differentially experienced within families leading to gender differences in boys' and girls' delinquency. Underscoring this line of theoretical development is the assumption that social controls operate differently or have differential importance for boys and girls, thus necessitating gender-specific theories of social control. For example, Hagan et al.'s power-control theory suggests that Marxist-patriarchal arrangements within the family translate into differential control and social bonds for boys and girls. Whereas, Heimer and DeCoster's (1999) gendered theory of violent delinquency suggests that girls' relational socialization and gendered social bonds impact their involvement in violence (see also, Steffensmeier and Allan, 1996). Much of the argument concerning gender differences in relational control rests squarely in a gendered “ethic of care” (Gilligan, 1982; Mears et al., 1998; Steffensmeier and Allan, 1996), which implies a gendered process of affective bonding. The assumption of gendered social bonding is not entirely in keeping with Hirschi's version of the social bond in which he suggests that social controls are gender neutral. Yet his proposition that boys and girls bond similarly to conventional others has never been tested. If gender moderates how the social bond is measured, then models using a “general” social control model for male and female offending would be mis-specified. More importantly, previous findings of gender differences in social bonding may have incorrectly attributed differences in bonding to structural, rather than measurement, differences. Two critical measurement and modeling questions remain concerning gender, social control, and social bonding. First, does gender moderate the measurement of social bonds? Second, does gender moderate the structural relationship between the social bond and delinquency for boys and girls? Before we address these two research questions, we review the relevant literature on gender, social bonds, and social control.

Conclusion Researchers interested in explaining gender differences in delinquency often use theories that assume gendered social bonding and control. Much of this work is premised on Hirschi's measurement of the social bond and its inhibitory effect on delinquency. Implicit in this idea of a gendered social bond is the assumption that either the social bond is measured differently for boys and girls or its effects on delinquency are moderated by gender. Our research was designed to test this assumption. As we demonstrate, the elements of the social bond are measured similarly for boys and girls. Gender does not moderate the measurement of the social bond or the structural relationships between the elements of the social bond and theft. We have evidence, however, that the structural relationship between the social bond and violent delinquency differs for boys and girls. We have evidence that suggests that constraining the structural paths to be equal in the violence model does not make the model fit much worse. The constrained model is also the most parsimonious model for boys' and girls' violent delinquency. As with much research in this area, our conclusions are mixed regarding whether gender moderates the relationship between social bonds and violent delinquency. Other researchers have suggested that violent crime is indeed gendered in its prediction (Heimer and DeCoster, 1999; Ogle et al., 1995) and our findings tentatively affirm this. Multiple-groups structural equation modeling provides a way to specify Hirschi's (1969) theoretical model and to assess whether gender moderates the measurement and modeling of social bonds. Thus, we tested for overall interaction effects and for gender-specific relationships nested within the full model. This type of nuanced analysis is critical to address the debates in contemporary literature regarding gender and criminological theory, specifically theories that utilized concepts of gendered social bonding and control. We generally affirm Hirschi's original hypotheses that increased conventionality is associated with lowered delinquency (theft and violent delinquency), and that the associations between the elements of the social bond and delinquency are similar for boys and girls. One exception exists however, in the unique effect that peer attachment has on boys' violent delinquency. We find that boys who report higher levels of peer attachment, report lower levels of violent delinquency. This finding is echoed in Agnew and Brezina's (1997) work in which they find poor peer relations with peers was positively associated with delinquency for boys. Other research on peer attachment and delinquency in general mirror our findings for boys. Haynie (2001) finds that highly popular respondents, measured as the number of times respondents listed an individual as their “best friend,” located in non-delinquent peer groups reported lower levels of delinquency than less popular, or less central youths. Few studies, however, look at gender differences in the impact of the quality of peer ties on delinquency, particularly for violent delinquency. This is an area of fruitful inquiry. Previous tests of gender and social controls or social bonding were limited by incomplete specifications that omitted elements of the social bond, used only single item indicators of more complex concepts or failed to model the elements of the social bond as latent variables. Additionally, previous tests of the importance of social controls or social bonding for boys' and girls' offending that did not account for possible measurement differences may have attributed gender difference or similarity in social bonds to structural factors rather than measurement factors. Our research suggests that no gendered measurements differences exist for this sample but that structural differences between the elements of the social bond and violent delinquency may need to be addressed. We would like to see the findings of our measurement model replicated in other samples that include boys and girls more highly involved in serious delinquency, as the seriousness of delinquency may impact the gender invariance of the social bond as our results tentatively suggest. This is particularly relevant for samples testing the theory for gender differences in violence as girls typically engage in less violence, resulting in girls' models with less variation in the dependent variable which attenuates associations. We would also like to see our measurement model and our gender invariance finding tested in longitudinal research as the delinquency and social bond relationship may be reflexive (Liska and Reed, 1985) and somewhat weaker in longitudinal research (Agnew, 1991). With this reflexive relationship of delinquency and social bonds in mind, it would be interesting to see if deviance is more costly for girls' rather than boys' social bonds during adolescence, suggesting gender variance in the social bond process or equivalent over time, suggesting gender invariance in social control over time—questions only addressed via longitudinal research. Our findings do not mean that gender is irrelevant for understanding delinquency, but quite the opposite. Gender matters for rates of offending and for the degree of attachment boys and girls experience, as is evident in the means for this sample. Clearly, gender matters in boys and girls rates of offending for many crimes, yet theories of gendered control or bonding may not adequately explain the sex gap in delinquency. Boys and girls are far more similar regarding their social bonds and sensitivity to social control than they are different. There are many ways in which gender matters in the process leading to delinquency, however. Girls are more likely to suffer from sexual and physical abuse in their families of origin precipitating running away from home and increased exposure to criminal opportunities and particularly, sexual victimization (Chesney-Lind and Shelden, 1998; Miller, 1998; Tyler et al., 2001). The experience of sexual abuse may uniquely affect girls' ability to form attachment bonds needed to deter delinquency, or may precipitate involvement as offenders in violence. Additionally, social control via direct control may be gendered, as is suggested by power-control theory (Hagan et al., 1987). Omitted from our research is an analysis of parental monitoring and supervision. Gender may moderate the relationship between direct social controls and delinquency as found in past research (Seydlitz, 1991) and needs to be assessed more critically. The differential social location of girls and boys creates different social worlds in childhood and early adolescence. However, it appears that they react similarly to social controls in later adolescence. It is possible that gender-specific theories of deviance are needed when considering adolescent processes outside of the social bond. What is clear is that social bonds are measured similarly for girls and boys and for a given change in social bonds, boys and girls exhibit a similar change in offending. Our findings raise serious questions regarding the gendered nature of social bonding in adolescence and theories of gendered bonding and delinquency.