10
Sex differences in the causes of self-control: An examination of mediation, moderation, and gendered etiologies Constance L. Chapple a, , Jamie Vaske b , Trina L. Hope c a University of Oklahoma, United States b Western Carolina University, United States c University of Oklahoma, United States abstract Sex is one of the most robust predictors of self-control, with a consistent nding that girls score higher on a variety of measures of self-control. In this research, we investigate three possible reasons for why this is true: rst, we examine whether current predictors of self-control mediate the effect of sex on self-control, second, we examine whether sex moderates the effect of current predictors on self-control and third, we examine the possibility that the causes of self-control are gendered, necessitating different causal models for boys and girls. Using data from the Children of the National Longitudinal Survey of Youth79, we assess three, related questions: Is the sex effect on self-control mediated by current predictors of self-control? Does sex moderate the effects of current predictors of self-control? Does the causal model predicting self-control differ for boys and girls? We nd that the sex effect on self-control is robust; does not moderate the etiology of self-control; and although partially mediated by etiological variables, remains a signicant predictor of self-control. We also nd that current predictors do a poor job of explaining girlsacquisition of self-control, suggesting a gendered etiology of self-control. © 2010 Published by Elsevier Ltd. Introduction Researchers have long known that sex is a signicant predictor of self-control. While this construct has alternatively been termed low self-control (Burton, Cullen, Evans, Alarid, & Dunaway, 1998; Hayslett- McCall & Bernard, 2002; LaGrange & Silverman, 1999; Tittle, Ward, & Grasmick, 2003), impulsivity (Chapple & Johnson, 2007; Moftt, Caspi, Rutter, & Silva, 2001) or preference for risk (Hagan, Simpson, & Gillis, 1987, 1988), it is clear that across samples, location, and time, girls exhibit greater self-control than boys. Gottfredson and Hirschi acknowledge that boys and girls differ signicantly on self-control but maintain that this is simply a difference in levels rather than a difference in causal processes (1990, p. 149). Their argument implies a mediation hypothesisas differing levels of etiological variables are taken into account, the sex effect should diminish, if not disappear. A second explanation is that while the same variables are predictive of self-control for boys and girls, sex interacts with, or moderates, the strength of the effects that these variables have on self- control. Hagan and associates suggest that sex interacts with the causes of preference for risk due to gendered socialization practices in the home; differences that reect a gender stratied social system (Hagan et al., 1987, 1988). As a result of gendered socialization processes, boys are taught to be risk-taking while girls are taught to be risk-averse. Hagan and associateshypothesis regarding gender, socialization and preference for risk implies that sex moderates the effect from parental socialization to preference for risk. In addition to a focus on boysweakened controls rather than girlsgreater controls, Hayslett-McCall and Bernard suggest that boysgreater exposure to attachment disruptions consistent with the development of tradi- tional masculinity uniquely and negatively impacts boysself-control (2002). Hagan and associateswork, in addition to Hayslett-McCall and Bernard, open the door for suggestions that the causal processes and perhaps even models of self-control differ by sex. To assess the extent to which sex conditions the etiology of self-control, modera- tion analyses are needed. Finally, within feminist criminology, theorists are increasingly calling for examinations of gendered pathwaysto delinquency (Chesney-Lind, 2006; Daly, 1992; Hagan & Foster, 2003). In particular, feminist criminology suggests that girls and boys differ in how they arrive at delinquency, allowing for different causal models of boysand girlsdelinquency (Belknap & Holsinger, 2006; Mullins and Miller, 2008). Of increased importance in this line of inquiry are examina- tions of the etiological processes leading to delinquency that may differ for boys and girls. In particular, it is suggested that girlsunique exposure to institutionalized sexism and family adversity differen- tially inuence their delinquent involvement. As the causes of crime, according to the gendered pathways perspective, are diverse and complex for girls and women, researchers should use a wider lens to Journal of Criminal Justice 38 (2010) 11221131 Corresponding author. E-mail address: [email protected] (C.L. Chapple). 0047-2352/$ see front matter © 2010 Published by Elsevier Ltd. doi:10.1016/j.jcrimjus.2010.08.004 Contents lists available at ScienceDirect Journal of Criminal Justice

Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

Embed Size (px)

DESCRIPTION

Gender and Self-Control

Citation preview

Page 1: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

Journal of Criminal Justice 38 (2010) 1122–1131

Contents lists available at ScienceDirect

Journal of Criminal Justice

Sex differences in the causes of self-control: An examination of mediation,moderation, and gendered etiologies

Constance L. Chapple a,⁎, Jamie Vaske b, Trina L. Hope c

a University of Oklahoma, United Statesb Western Carolina University, United Statesc University of Oklahoma, United States

⁎ Corresponding author.E-mail address: [email protected] (C.L. Chapple).

0047-2352/$ – see front matter © 2010 Published by Edoi:10.1016/j.jcrimjus.2010.08.004

a b s t r a c t

Sex is one of the most robust predictors of self-control, with a consistent finding that girls score higher on avariety of measures of self-control. In this research, we investigate three possible reasons for why this is true:first, we examine whether current predictors of self-control mediate the effect of sex on self-control, second,we examine whether sex moderates the effect of current predictors on self-control and third, we examinethe possibility that the causes of self-control are gendered, necessitating different causal models for boys andgirls. Using data from the Children of the National Longitudinal Survey of Youth79, we assess three, relatedquestions: Is the sex effect on self-control mediated by current predictors of self-control? Does sex moderatethe effects of current predictors of self-control? Does the causal model predicting self-control differ for boysand girls? We find that the sex effect on self-control is robust; does not moderate the etiology of self-control;and although partially mediated by etiological variables, remains a significant predictor of self-control. Wealso find that current predictors do a poor job of explaining girls’ acquisition of self-control, suggesting agendered etiology of self-control.

lsevier Ltd.

© 2010 Published by Elsevier Ltd.

Introduction

Researchers have long known that sex is a significant predictor ofself-control. While this construct has alternatively been termed lowself-control (Burton, Cullen, Evans, Alarid, &Dunaway, 1998; Hayslett-McCall & Bernard, 2002; LaGrange & Silverman, 1999; Tittle, Ward, &Grasmick, 2003), impulsivity (Chapple & Johnson, 2007;Moffitt, Caspi,Rutter, & Silva, 2001) or preference for risk (Hagan, Simpson, & Gillis,1987, 1988), it is clear that across samples, location, and time, girlsexhibit greater self-control than boys. Gottfredson and Hirschiacknowledge that boys and girls differ significantly on self-controlbut maintain that this is simply a difference in levels rather than adifference in causal processes (1990, p. 149). Their argument implies amediation hypothesis—as differing levels of etiological variables aretaken into account, the sex effect should diminish, if not disappear.

A second explanation is that while the same variables arepredictive of self-control for boys and girls, sex interacts with, ormoderates, the strength of the effects that these variables have on self-control. Hagan and associates suggest that sex interacts with thecauses of preference for risk due to gendered socialization practices inthe home; differences that reflect a gender stratified social system(Hagan et al., 1987, 1988). As a result of gendered socialization

processes, boys are taught to be risk-takingwhile girls are taught to berisk-averse. Hagan and associates’ hypothesis regarding gender,socialization and preference for risk implies that sex moderates theeffect from parental socialization to preference for risk. In addition toa focus on boys’weakened controls rather than girls’ greater controls,Hayslett-McCall and Bernard suggest that boys’ greater exposure toattachment disruptions consistent with the development of tradi-tional masculinity uniquely and negatively impacts boys’ self-control(2002). Hagan and associates’ work, in addition to Hayslett-McCalland Bernard, open the door for suggestions that the causal processesand perhaps even models of self-control differ by sex. To assess theextent to which sex conditions the etiology of self-control, modera-tion analyses are needed.

Finally, within feminist criminology, theorists are increasinglycalling for examinations of “gendered pathways” to delinquency(Chesney-Lind, 2006; Daly, 1992; Hagan & Foster, 2003). In particular,feminist criminology suggests that girls and boys differ in how theyarrive at delinquency, allowing for different causal models of boys’and girls’ delinquency (Belknap & Holsinger, 2006; Mullins andMiller,2008). Of increased importance in this line of inquiry are examina-tions of the etiological processes leading to delinquency that maydiffer for boys and girls. In particular, it is suggested that girls’ uniqueexposure to institutionalized sexism and family adversity differen-tially influence their delinquent involvement. As the causes of crime,according to the gendered pathways perspective, are diverse andcomplex for girls and women, researchers should use a wider lens to

Page 2: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1123C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

capture the interplay of factors that affect female offending. Suchinquiries often necessitate an examination of the “causes of thecauses” of delinquency and how they may differ for boys and girlsgiven their different social structural locations. Essentially, oursuggestion here is that the causes of delinquency may be genderedif the causes of self-control are gendered.

In our research,we tackle three possible explanations forwhy sex isa significant predictor of self-control. First, girls report higher levels ofself-control because they report higher levels of the variables thatcause self-control. Second, sex differences in self-control are explainedby sex differences in the predictive power of the traditional variablesthat produce self-control. Third, sex differences in self-control are theresult of gendered etiological processes. While all three are probableexplanations, each taps into different empirical mechanisms forexplaining the sex effect on self-control. The argument that differinglevels of etiological variables produce the sex effect is best addressedas amediation effect; the argument for an interaction between sex andthe traditional predictors implies a moderation effect; while genderedetiological processes suggest that the causal model itself may differsignificantly for boys versus girls. Using prospective data frommothersand children in the National Longitudinal Survey of Youth79, weinvestigate the robustness of the sex effect on self-control throughanalyses of mediation, moderation, and differing causal models. Wethen addresswhat etiological divergencesmaymean for empirical andtheoretical work on gender and self-control.

Literature review

The etiology of self-control

Recent research on the causes of self-control has highlightedfamilial, extra-familial, and biological sources. Self-control has beenpredicted by family variables such as attachment and supervision(Beaver, DeLisi, Vaughn, Wright, & Boutwell, 2008; Beaver, Wright, &DeLisi, 2007; Boutwell & Beaver, 2010; Gibson, Sullivan, Jones, &Piquero, 2010; Vaughn, DeLisi, Beaver, & Wright, 2009; Vazsonyi &Huang, 2010; Wright & Beaver, 2005), family structure (Hope,Grasmick, & Pointon, 2003), and maternal deviance (Chapple, Hope,& Whiteford, 2005; Hope & Chapple, 2005). Research has alsohighlighted the importance of extra-familial sources of self-controlthrough collective socialization at the school and community levels(Pratt, Turner, & Piquero, 2004; Turner, Piquero, & Pratt, 2005).Finally, emerging research (Beaver, DeLisi, Vaughn, & Wright, 2010;Beaver et al., 2007; Wright & Beaver, 2005) has emphasized theimportance of biological and genetic influences on self-control.

Familial sources of self-control

According to Gottfredson and Hirschi, the primary source ofvariation in levels of self-control can be traced back to early childhoodsocialization processes. The socialization process that Gottfredson andHirschi describe begins with parental attachment. Indeed, researchexploring the role of parental attachment/parental involvement in theacquisition of self-control finds support for Gottfredson and Hirschi'sassertions (Beaver et al., 2007; Beaver, et al., 2008a; Boutwell &Beaver, 2010; Chapple et al., 2005; Gibson et al., 2010; Hope et al.,2003; Hope & Chapple, 2005; Vaughn et al., 2009; Vazsonyi & Huang,2010; Wright & Beaver, 2005).

However, love and affection alone do not insure the acquisition ofself-control; parentsmust also “monitor and correct” errant behavior intheir children (Gottfredson & Hirschi, 1990, p. 97). Researchersexploring the role of supervision or monitoring on self-control havefound consistent support for the idea that effective parental supervisionis related tohigher levels of self-control (Chapple et al., 2005;Hopeet al.,2003; Hope & Chapple, 2005; Gibson et al., 2010; Pratt et al., 2004;Unnever, Cullen, & Pratt, 2003). In terms of punishment, researchers

exploring the role of effective but non-corporal discipline have foundsupport for the assertion that effective discipline is associated withhigher levels of self-control (Hay, 2001; Unnever et al., 2003). Theconverse is also true; use of physical punishment is predictive of lowerlevels of self-control (Beaver et al., 2007; Hay, 2001; Vaughn et al.,2009). Finally, when elements of parenting are combined into globalparental management/parental socialization measures (Feldman &Weinberger, 1994; Gibbs, Giever, & Martin, 1998; Hay & Forrest, 2006;Perrone, Sullivan, Pratt, &Margaryan, 2004), consistent support is foundfor Gottfredson and Hirschi's emphasis on the importance of parents inthe acquisition of self-control.

Other family variables found to be important predictors of self-control include family structure (Chapple et al., 2005;Winfree, Taylor,He, & Esbensen, 2006), parental education (Boutwell & Beaver, 2010;Hope et al., 2003), mother's age at child's birth (Boutwell & Beaver,2010), mother's age at first intercourse (Hope & Chapple, 2005),parental stress (Vaughn et al., 2009), maternal drug use (Chappleet al., 2005), and level of parental self-control (Boutwell & Beaver,2010; Nofziger, 2008).

Community and school sources of self-control

Despite a strong theoretical focus on familial correlates of self-control, individuals also acquire self-control from sources beyond thefamily. Although not suggested by Gottfredson and Hirschi as a sourceof self-control, community context and organization has long beenthought to be an important source of extra-familial socialization(Shaw & McKay, 1942). Along these lines, Pratt et al. (2004) foundthat community context (measured by maternal reports) is predictiveof self-control independent of parental socialization. Specifically,youth living in neighborhoods characterized by adverse conditionshad lower self-control. This relationship remained significant net ofdemographic and family socialization variables in both cross-sectionaland longitudinal analyses. This suggests that as much as socialorganization can be a source of social control whichmay translate intoself-control, community disorganization may exacerbate alreadydysfunctional family influences on the acquisition of self-control. Incontrast to the findings of Pratt et al., Boutwell and Beaver (2010),using independent measures of neighborhood conditions, did not findthem to be significant predictors of self-control once individual-levelfamily socialization variables were included in the model. Boutwelland Beaver, however, utilize a single-item measure of neighborhoodsafety that does not measure the larger concept of communityorganization and/or collective socialization.

Although they do not specifically mention the possible role ofcommunities in levels of self-control, Gottfredson and Hirschi doacknowledge that outside of family sources, school can provideimportant self-control lessons (1990, p.105). Schooling regularizeschildren's lives and teaches them self-restraint and cooperation. To theextent that schooling provides important lessons in timeliness, order,thinking of others and delaying gratification, school attachment—and toa lesser degree, school organization—may influence the acquisition ofself-control. On the one hand, schools may act as a “safety net” in whichpoorly parented children are given a chance to acquire self-control. Onthe other hand, well-socialized children may increase their self-controlwhen rewarded and reinforced by schooling. In their recent study,Turner et al. (2005) expanded their previous researchwith the additionof school socialization as an explanatory variable. Results of this studysuggest that school socialization is indeed important in the explanationof self-control, and this relationship remained significant after control-ling for demographic and family process variables.

Biological sources of self-control

While Gottfredson and Hirschi (1990) emphasized the familialsources of self-control, recent research has shown that genetic and

Page 3: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1124 C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

biological factors are significantly related to levels of self-control aswell (Beaver et al., 2008b). For instance, Beaver, Wright, DeLisi, andVaughn's (2008) analysis of twins from the National LongitudinalStudy of Adolescent Health revealed that genetic factors explained 52percent of the variance in low self-control at wave I and 64 percent ofthe variance in low self-control at wave II. The authors also reportedthat genetic factors were significantly related to the stability andchange in low self-control. More specifically, 82 percent of thestability in low self-control was accounted for by genetic factors, and20 percent of the change in low self-control was attributed to geneticfactors. Other studies have also found that genetic factors explain asubstantial proportion of variance in self-control (Wright & Beaver,2005; Wright, Beaver, DeLisi, & Vaughn, 2008).

In addition to genetic contributions to low self-control, research hasfound that self-control is a function of other biological factors such asbirth complications (Elgen, Sommerfelt, & Markesad, 2003), ADHD(Unnever et al., 2003), prenatal substance use (Brown et al., 1991;Cornelius, Ryan, Day, Goldschmidt, &Willford, 2001; Leech, Richardson,Goldschmidt, & Day, 1999) or neuropsychological deficits (Boutwell &Beaver, 2010). A study of 155 twin pairs revealed that anoxia (oxygenstarvation during birth) was significantly related to parent and teacherreports of low self-control, controlling for youths’ demographics andparental involvement (Beaver & Wright, 2005). More recently,Ratchford and Beaver (2010) found that birth complications and lowbirth weight were indirectly related to youths’ low self-control viagreater neuropsychological deficits in the prefrontal cortex. Additionalstudies have also reported that individuals with low self-controlperform poorly on assessments of neuropsychological functioning,particularly those associated with the prefrontal cortex (Beaver et al.,2007, 2008b; Vaughn et al., 2009). The prefrontal cortex is the brainregion that is activated during tasks that involve behavioral inhibition,concentration, abstract reasoning, decision-making, working memory,and regulation of emotions and cognition. Given the role of theprefrontal cortex in these tasks, it is not surprising that deficits in thisregion have been linked to lower levels of self-control (Knoch & Fehr,2007).

While research on the causes of self-control is less common thanresearch on the consequences of self-control, there is mountingevidence that self-control emerges from multiple sources. The family,the community, and the school emerge as key social agents of self-control while neuropsychological deficits and heredity impact self-control outside of environmental sources. For this research, however,we are interested in moving beyond the determinants of self-controlto questions concerning how these determinants relate to sexdifferences in self-control. Do sex differences in the levels of thepredictors of self-control account for the effect of sex on self-control(mediation hypothesis)? Additionally, do the known determinates ofself-control predict equally well for boys and girls (moderationhypothesis)? Or, do a different set of variables predict self-control forboys vs. girls, suggesting that separate causal processes are at play?While these issues have been grappledwith in the literature with self-control as an independent variable and crime/delinquency as adependent variable, they have not been explicitly addressed whenself-control is the outcome variable.

Sex and self-control

Gottfredson and Hirschi note that “gender differences for all typesof crime are established early in life and that they persist throughoutlife. This fact implies a substantial self-control difference between thesexes” (1990, p. 147). They maintain that girls have higher levels ofself-control because parents are more likely to recognize and punishdeviant behavior in girls compared to boys (p. 149), suggesting thatthe sex effect on self-control is simply a matter of mediation—i.e., girlsscore higher on positive predictors of self-control, such as attachment,supervision, and effective discipline, while boys score lower. This

suggests that when familial variables are included in modelspredicting self-control that the effect of sex should be partially, ifnot fully, mediated.

While most of the research on the determinants of self-control hasnot explicitly investigated sex differences, sex is often used as acontrol variable in the models; so although we can determine if itremains a significant predictor when other independent variables areincluded in the models, the analyses are not presented with an eyetoward the mediation question. We found only two studies where sexwas not a significant predictor of self-control. Pratt et al. (2004) foundthat sex was a significant predictor of supervision (boys experiencinglower levels of supervision), but not monitoring/discipline. In modelspredicting self-control with sex, supervision, and monitoring/disci-pline among the independent variables, supervision was significant,but sex was not, suggesting that the effect of sex on self-control mayhave been mediated by supervision. Hay's (2001) models predictingself-control included sex simultaneously with monitoring anddiscipline, and while monitoring and discipline were significant, sexwas not. Because questions of mediation were not the focus of Hay'sanalyses, however, he did not present the models with sex and thefamily variables included separately, so we cannot know if the effectof sex on self-control was mediated by the family variables.

The findings of Pratt et al. and Hay are more the exception than therule, however, as most research finds that sex remains a significantpredictor of self-control when the primary independent variables areincluded in the models (Beaver et al., 2007; Boutwell & Beaver, 2010;Chapple, Tyler, & Bersani, 2005; Gibson et al., 2010; Lynskey, Winfree,Esbensen, & Clason, 2000; Perrone et al., 2004; Unnever et al., 2003;Wright & Beaver, 2005).

Mediation, moderation, and differing causal models

While sex is always included as a control variable in researchexploring the causes of self-control, it is rarely the focus of theresearch. This means that sex is almost always included simulta-neously with the other variables, so we cannot know if its effect wasmediated by the other independent variables. In addition to the Prattet al. (2004) study discussed above, we found two other studies thatallowed us to indirectly address the mediation hypothesis. Lynskeyet al. (2000) first present a model with just demographic variablespredicting self-control, and then add both parental attachment andparentalmonitoring in the secondmodel. The standardized coefficientfor “male” decreases from (-.129) to (-.081), a reduction of 37%,suggesting that a portion of the sex effect on self-control is due toboys’ lower levels of parental attachment and/or monitoring. Beaveret al., however, find less evidence of mediation effects. They presentan initial model including sex and other controls and then addmaternal characteristics one at a time in subsequent models. Thestandardized coefficient for sex remains significant at (-.07) whenneuropsychological defects, age at child's birth, and mothers’ highestlevel of education are added to the model, and changes only slightlywhen maternal involvement (-.06) and maternal smoking (-.08) areadded to the models. Beaver et al. reproduce the same analyses usingpaternal measures and find similar results, with one exception. Whilethe standardized coefficient for sex changed little when most of thepaternal characteristics are added in subsequent models, whenpaternal involvement is added to the model the coefficient for sexbecomes considerably stronger (from -.08 to -.71), suggesting asuppression rather than a mediation effect.

To test the moderation hypotheses, we searched for analyses thateither included interaction terms between sex and the key indepen-dent variables, or presented the analyses separately by sex. Lynskeyet al. (2000) and Beaver et al. (2007) presented their modelsseparately by sex, although a gendered analysis of self-control wasnot the primary focus of their research. In our version of themoderation hypothesis, we would expect to see the same variables

Page 4: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

Table 1Means and t-tests for dependent and independent variables by sex

Boys Girls Range

Self-control*** 26.15 27.39 10-30Supervision* 1.78 1.86 0-2Maternal attachment 11.44 11.74 5-14Non-physical punishment*** 0.74 0.86 0-1Maternal education 0.84 0.83 0-1Neighborhood socialization 7.58 7.65 3-9School socialization 8.70 8.95 2-10Normal birth weight 0.05 0.07 0-1No prenatal substance use 0.59 0.60 0-1Non-white 0.40 0.43 0-1

* p≤ .05, *** p≤ .001 (indicate significant mean differences by sex).

1125C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

significantly predicting self-control for boys and girls, but that thestrength of the coefficients may differ by sex. Lynskey et al. found littleevidence of a moderation effect. Both parental attachment andparental monitoring were significant predictors of self-control forboys and girls, and the coefficients were quite similar (for attachment,.222 for boys and .228 for girls; for monitoring .381 for boys and .338for girls). Beaver et al. (2007) looked at predictors of low self-controlin early childhood (kindergarten and first grade) and presented dataon male and female samples. For both boys and girls, the sameindependent variables were significant predictors of self-control,including fine motor skills, gross motor skills, parental involvement,parental withdrawal, parental affection, and physical punishment.Additionally, the coefficients were very similar. The coefficients forparental withdrawal were the most divergent, at .24 for boys versus.17 for girls, suggesting a slight moderating effect of sex on parentalwithdrawal (with the relationship being somewhatmore powerful forboys than girls). Finally, Chapple and Johnson (2007) examined theeffect of sex on impulsivity (a component of our measure self-control)using multiple group path analysis and found that girls’ impulsivitywas not significantly predicted by any of the etiological variablesassessed. Whereas, boys’ level of impulsivity was significantlypredicted by poor motor skills and reading ability in childhood, aswell as weak attachment to mothers, poor supervision, and receipt ofcorporal punishment. They found that the effect of maternalattachment and corporal punishment were significantly stronger forboys, suggesting a moderation effect. Overall, the existing research onboth mediating and moderating effects is limited and not particularlyconclusive.

With regards to the idea of different causal processes producingself-control for boys versus girls, we did not find any existing researchthat explicitly tests whether a different set of variables are predictiveof self-control for boys than for girls, although the analyses by Lynskeyet al. and Beaver et al. described above suggest that the causalprocesses are similar.

Finally, research on how sex may relate to community or schooldeterminants of self-control is virtually non-existent. Other research,however, suggests an interactive relationship between communitycontrols and self-control that has potentially important implicationsfor studies of sex and self-control. Wikström and Sampson (2003)suggest that the impact of the community on self-control ismoderated by current levels of self-control: “We specifically hypoth-esize that the community strength of impact on motivation to offendvaries inversely by the individual propensity to offend. That is, thecommunity context has the strongest influence on individualmotivation to offend for those with the weakest propensity to offend”(p. 139). Those with strong propensities to offend will not rely to thesame degree on “situational inducements” so will therefore be lessaffected by the community context. This theory suggests possible sexdifferences in the role of community in the development of self-control. If we assume girls to have weaker propensities to offend, thenthey should be more affected by the “community-context supply ofcriminogenic behavioral settings” (p. 139) (see also Wikström &Loeber, 2000). In essence, girls should respond more favorably toincreased criminal opportunities via disorganized neighborhoods.

As our discussion of the prior research on sex and self-controlillustrates, there are still a number of unanswered questionsconcerning the nature of sex differences in levels of self-control. Weseek to clarify these questions by investigating mediating, moderat-ing, and divergent causal processes in the role of sex in the acquisitionof self-control.

Data and sample

Data for this study were taken from the Children of the NationalLongitudinal Survey of Youth (NLSY79-Child data). The originalNLSY79 is a nationally representative, longitudinal sample of youth

aged 14 to 21 in 1979. The NLSY79 contains “extensive informationabout the employment, education, training, and family experiences ofthe respondents,” and beginning in 1982, the female respondents ofthe NLSY79 were surveyed about “pregnancy, postnatal fertility, andchild care experiences” (Center for Human Resource Research, 2000,p. 1). In 1986, these data were further enriched through the initiationof the Children of the NLSY79 survey in which mothers and theirchildren were assessed. The biannual assessments include measuresof “cognitive ability, temperament, motor and social development,behavior problems, and self-competence of the children as well as thequality of their home environment” (Center for Human ResourseResearch, 2004, p. 4). Starting in 1988, personal interviews wereconducted with the children themselves after they reached the age of10. The children were asked about their “schooling, family, peer-related and other attitudes and behaviors” (Ibid, p. 4).

The NLSY79-Child sample “approximately represent[s] a cross-section of children born to a nationally representative sample ofwomen who were between the ages of 31 and 38 on January 1, 1998”(Center for Human Resource Research, 2000, p. 3).1 For the analyses,we used mother and child assessments taken from twowaves of data;the children were aged 10-11 in 2000 and 12-13 in 2002. All of theindependent variables were measured in 2000, while self-control, ouroutcome variable, was measured in 2002. As differences in self-control are considered to be stable after age 10, we chose to measureself-control in late childhood/early adolescence. Because siblings areincluded in the original sample, we randomly selected one respondentout of each sibling group for inclusion in the sample, bringing the finalsample size to 492 respondents (262 boys, 230 girls).2 We includethree samples in our analyses: the full sample, the boys’ sample andthe girls’ sample (see Table 1).

Measures

In terms of the demographic profile of the samples, 40% of the boysample and 43% of the girl sample are non-white. Girls and boys wereboth approximately 10.5 years old in 2000 and 47% of the sample isfemale.

Dependent variable

We use items from the Behavioral Problems Index (BPI) tomeasure self-control. Other researchers using the NLSY79 Child data(Chapple, 2005; Chapple & Johnson, 2007) have used similarmeasures. Mothers were administered the Behavior Problems Index(BPI), which was designed to measure the incidence and severity oftheir child's behavior problems (Peterson & Zill, 1986). Mothers wereasked to answer how true certain statements were of their child suchas: your child cheats or lies, is cruel to others, is impulsive, is not sorryafter misbehaving, and so on. The items were scored so that 1=oftentrue, 2=sometimes true, and 3=not true. Ten of the 32 itemsthought to specifically capture self-control were selected for inclusion

Page 5: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1126 C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

in the scale (see Appendix A for a list of items in this scale). When theitems were tested for reliability, the scale had an alpha of .84.

Independent variables

Family predictors of self-controlOur review of the familial and environmental predictors of self-

control highlighted the importance of family function, maternaleducation, and school and neighborhood socialization as predictors ofself-control. We describe our measurement of familial, extra-familialand biological predictors below. Supervision, measured in 2000 whenthe youth were 10-11 years old, was assessed with a single item inwhich youth were asked how often their mothers knew who theywere with when they were not home (0=hardly ever, 1=some-times, 2=often). Overall, approximately 81% of boys and 89% of girlsreported that their mothers often knew who they were with whenthey were not home. Girls reported significantly higher levels ofsupervision.

Following the lead of prior research with the NLSY79-Child andYoung Adult data (Hope & Chapple, 2005; Tyler, Stone, & Bersani,2006), maternal attachment was assessed via a four item, child self-report measure. Children were asked four questions regarding howclose they felt to their mother, how well their mother shared ideaswith them, how often their mother listened to their side of anargument, and how often their mother talked over importantdecisions with them. Two of the items (mother listens to child'sside and talks over important decisions) were reverse coded so thathigher scores indicated greater attachment on all items. The fourordinal items were summed together to form a composite measure ofmaternal attachment (alpha=.54).

A measure of non-physical punishment was also included in theanalyses, and was assessed through maternal reports. Mothers wereasked how theywould respond if their child threw a tantrum. Subjectsreceived a score of “0” if they said they would spank their child, andthey received a score of “1” if they responded that they would notspank their child. Eighty-six percent of mothers of girls respondedthat they would not spank their child for throwing a tantrum and 74%of mothers of boys in the sample responded the same.3

To measure maternal education, mothers were asked to report thehighest grade they had completed as of 2000. The responses werecoded so that 0=less than a high school diploma and 1=high schooldiploma or more. The prevalence of maternal high school diploma isapproximately 84 percent for boys and 83 percent for girls. Because ofmissing data problems, income could not be used in this analysis andwe use maternal education both as a substantive predictor of self-control and as a (rough) demographic control for social status.

Extra-familial predictorsSimilar to past research on environmental predictors of self-

control we use items derived from Pratt et al. (2004) and Turner et al.(2005) to measure neighborhood and school socialization. As bothprior works used the same data as we use, we simply replicated theirmeasures. Mothers were asked to rate their neighborhoods on variousdimensions. Neighborhood socializationwas measured by three items:“People don't respect rules,” “People don't supervise their kids,” and“The extent to which people keep to themselves.” Responses werecoded 1=big problem, 2=somewhat, 3=not a problem. The scaleranged in value from 3-9 and boys reported a mean of 7.58 and girlshad a mean of 7.65. The sex difference in means on neighborhoodsocialization was not significant. Mothers were also asked to evaluatethe school their child attends on several dimensions. Schoolsocialization was measured with two items: Grade you give yourchild's school for “ability to maintain discipline” and “teach right fromwrong.” The items were coded so that A=5 and F=1, or that higherscores reflected greater school socialization. The scale has a mean of8.71 for boys and 8.95 for girls. The sex difference in means for school

socialization is significant andmothers of girls reported greater schoolsocialization.

Biological predictorsTwo biological risk factors related to low self-control were

included in the current analyses: birth weight and prenatal substanceuse. Beginning in 1983, mothers were asked questions concerningtheir prenatal and postnatal practices with the target child. Thisincluded questions about birth weight, and prenatal substance use.Subjects were considered to have normal birth weight if they weighedover 5.5 pounds at birth (0=low birth weight, 1=normal birthweight). Prenatal substance use is assessed by asking mothers fourquestions on whether they used alcohol, tobacco, marijuana, orcocaine any time in the 12 months prior to the subject's birth. Thesefour items were summed together to form an index of prenatalsubstance use. The index was then dichotomized so that 0=motherused drugs or alcohol during pregnancy and 1=mother did not usedrugs or alcohol within 12 months of the subject's birth. Approxi-mately 59 percent of mothers stated they had used drugs or alcoholwithin the 12 months prior to the subject's birth (i.e., score of “0” onthe scale). Unfortunately, we cannot disentangle substance use whilepregnant from substance use in the two months prior to becomingpregnant, and our measure is a conservative estimate of in-uteroexposure to drugs or alcohol.4

Control variables

Finally, we control for race, with non-white coded as 1 and whitecoded as 0.

Analyses

As we propose differing mechanisms for explaining the sex effecton self-control, we performed two different types of analyses. First, toassess whether current predictors of self-control mediate the effect ofsex on self-control, we first established that boys and girls differedsignificantly in self-control and which etiological variables differedsignificantly by sex (Baron & Kenny, 1986). Those variables thatdiffered significantly by sex (supervision and physical punishment)were assessed as potential mediators of the sex and self-controlrelationship.5 To the extent that the coefficient for sex declines insignificance or becomes non-significant, mediation, either partial orfull, is implied. In addition, we further examine the mediation processby estimating the path models in MPLUS using the mean adjustedweighted least squares estimator. Mean adjusted weighted leastsquares estimator is used because this estimator is appropriate forcategorical and dichotomous outcomes (i.e., physical punishment andsupervision). The models are represented as:

Ypunish=α+γ11Xsex+ζ1Ysuper=α+γ21Xsex+ζ2Yselfcon=α+γ31Xsex+β31Ypunish+β32Ysuper+ζ3

Where γ11 describes the regression of punishment on sex, γ21

describes the regression of supervision on sex, and γ31 describes theregression of self-control on sex (once punishment and supervisionare taken into account). β31 refers to the regression of self-control onphysical punishment (controlling for sex and supervision), and β32

refers to the regression of self-control on supervision (controllingfor sex and physical punishment). ζ refers to the residuals for eachequation.

Second, to assess whether sex moderates the acquisition of self-control, or whether different variables altogether predict self-control,we split the sample into boys’ and girls’ samples and used hierarchicalmultiple regressions with the substantive variables entered in blocks.We then include a series of coefficients tests to determine if any

Page 6: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1127C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

variables differed significantly for boys and girls (Brame, Paternoster,Mazerolle, & Piquero, 1998). Rather than examining whetherindividual predictors of self-control are moderated by sex (e.g. asindividual interaction terms), we chose to assess sex moderation onthe entire causal model, by using split samples and coefficient tests.This allows us to test both the moderation and the differing causalmodel hypotheses.

Results

Before we discuss the multivariate models, we first discuss thefindings of our bivariate analyses. Chi square and t-test analysesreveal that boys are significantly more likely to receive physicalpunishment than girls, score lower on the self-control scale, andreport significantly lower levels of supervision. Boys and girls,however, do not differ in their level of maternal attachment (t=-1.702, p=.089), neighborhood socialization (t=-.468, p=.640), orschool socialization (t=-1.800, p=.073). Further, boys and girls donot differ on the prevalence of low birth weight (χ2=.603, p=.437),maternal substance use during pregnancy (χ2=.077, p=.781), ormaternal education level (χ2=.073, p=.787). Contrary to feministtheorizing in the literature, girls in this sample did not report higherlevels of extra-familial control but do experience higher levels offamilial control via supervision and physical punishment.

Multivariate linear regression models are estimated in STATA 9.1.Test statistics are considered statistically significant if p≤ .05. Adifference of coefficient test is used on to examine whether thestrength of the coefficients differs between girls and boys (Clogg,Petkova, & Haritou, 1995). A z value of ±1.96 or greater indicates thatthe strength of the coefficient varies significantly by sex.

Investigating mediators of the effect of sex effect on self-control

The bivariate analyses suggest that the average level of self-controlsignificantly differs for boys and girls. Further, t-test analyses revealedthat there are significant sex differences in levels of physicalpunishment and supervision. The next step in the analyses is toexamine whether sex differences in physical punishment andsupervision mediate the sex effect in self-control. It may behypothesized that girls have higher levels of self-control becausethey receive less physical punishment than boys and are subject tomore supervision than boys. This hypothesis is tested in two ways.First, we regress self-control on sex (0=male, 1=female) as thebaseline model, and then enter physical punishment and supervisioninto the model to examine whether the effect of sex on self-controlbecomes non-significant. The effect of sex should become non-significant if physical punishment and supervision mediate the effectsof sex on self-control. Second, we verify the mediation effects usingpath analyses in MPLUS.

As shown in Table 2, sex has a significant effect on self-control(b=1.238, pb .001), and this effect remains significant (b=.814,p=.013) after controlling for the effects of physical punishment andsupervision. While the effects of sex on self-control remain statisti-cally significant, the coefficient of sex is reduced by 34% once physicalpunishment and supervision levels are taken into account. This result

Table 2Non-physical punishment and supervision as partial mediators of sex on self-control

b b

Female 1.238*** .814*Non-physical punishment 1.520***Supervision 1.621***F 15.86 16.69R2 .03 .10Adj. R2 .02 .09

* p≤ .05, *** p≤ .001.

suggests that sex differences in physical punishment and supervisionpartially mediate the sex gap in self-control, but that a largepercentage of the gap (66%) is not explained. Individual regressionmodels show that introducing supervision into the model reduces thesex coefficient by 18% [(1.238 – 1.016) / 1.238=.18), and physicalpunishment reduces the sex coefficient by 15.6% [(1.238 – 1.044) /1.238=.156). As an additional check on our etiological model, we rana series of regressions to assess the robustness of sex in light of our fulletiological model. As can be seen in Appendix B, sex remains a robustpredictor of self-control net of all our etiological variables.

The path analyses results further suggest that physical punishmentand supervision mediate the effects of sex on self-control (χ2=2.774,df=1, p=.09, CFI=.971, TLI=.825, RMSEA=.060). Sex has apositive, significant effect on physical punishment (γ11=.448,Standardized γ11=.218, p=.001) and supervision (γ21=.321,Standardized γ21=.158, p=.028) indicating that girls are less likelyto be physically punished and they receive higher levels ofsupervision. In turn, physical punishment (β31=.791, Standardizedβ31= .232, pb .001) and supervision (β32= .850, Standardizedβ32=.247, pb .001) have a significant effect on self-control. Thedirect effect of sex on self-control is non-significant (γ31=.612,Standardized γ31=.088, p=.091). However, the indirect effect of sexon self-control via physical punishment is statistically significant(.354, p=.012) and the indirect effect via supervision is marginallynon-significant (.273, p=.051). The path analysis results suggest thatsupervision and physical punishmentmediate the effect of sex on self-control; our results imply that girls develop higher levels of self-control because they experience more supervision and less physicalpunishment. However, when you examine the model fit statistics, themodel proposed is a poor fit to the data. We will follow up on thispoint at the end of the next section.

Investigating sex as a moderator and differing causal models by sex

Table 3 displays the multivariate linear regression results of ourmoderation and differing causal models analysis. As shown in thetable, only one of the predictors of self-control is significant for girls inthe full model (last column), maternal supervision (b=1.137,β=.153, p=.040). As found in the earlier path analysis, theindependent variables do a poor job of predicting self-control forgirls (F=2.02, p=.039), and the model only explains 4 percent of thevariation in girls’ self-control.

In contrast to the findings in the girls’ model, the boys’ modelshows that a number of variables have a significant effect on self-control. Boys’ levels of self-control are significantly related to theirparents’ use of physical punishment (b=2.058, β=.229, p=.001),mothers’ educational level (b=1.916, β=.175, p=.008), andmaternal attachment (b=.575, β=.254, pb .001). For boys, higherlevels of self-control are associated with non-physical punishment(i.e., not getting spanked) and greater maternal attachment—or lowerlevels of self-control are associated with higher levels of physicalpunishment and less maternal attachment among boys. Boys withhigher levels of self-control also reported that their mothers weremore likely to graduate from high school. It appears that the familialvariables largely contribute to the proportion of explained variation inself-control for boys. The percentage of explained variation in themodel that includes familial risk factors (R2=.21) is significantlylarger than the model that includes only the demographic andbiological variables (R2=.07, F=12.40, df=3, 210, pb .001). Thisfinding indicates that parenting factors significantly contribute to thedevelopment of self-control among boys. Overall, the full modelexplains approximately 22 percent of the variation in self-control forboys (F=7.65, pb .001). In addition, the last column of Table 3 showsthat the effect of maternal attachment is significantly strongerfor boys (z=2.614, p=.009). These results suggest that maternal

Page 7: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

Table 3Unstandardized regression coefficients of self-control on demographic, biological, familial, and extra-familial variables

Boys

Demographics Biological Family Environmental

Non-White -.915 -.940 -.315 -.100Maternal education 2.499*** 2.462*** 2.283*** 1.916**Normal birth weight -.941 -.503 -1.331No prenatal substance use .186 .142 -.079Non-physical punishment 2.258*** 2.058***Maternal attachment .607*** .575***Maternal supervision 1.022 1.021Neighborhood socialization .289School socialization .319F 10.76 4.75 8.15 7.65R2 .07 .07 .21 .25Adj. R2 .06 .05 .18 .22

Girls

Demographics Biological Family Environmental z test

Non-White -.416 -.443 -.517 -.437 .474Maternal education 1.084* 1.076* 1.068 .977 .999Normal birth weight .930 1.148 .970 -1.449No prenatal substance use .046 .142 .089 -.251Non-physical punishment .609 .434 1.861Maternal attachment .088 .092 2.614Maternal supervision 1.253* 1.137* -.144Neighborhood socialization .088 .920School socialization .161 .726F 3.22 2.03 2.44 2.02R2 .03 .04 .08 .09Adj. R2 .02 .02 .05 .05

* p≤ .05, *** p≤ .001.

1128 C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

attachment has a significantly stronger effect on self-control for boysthan for girls.6

Based on our findings with these data and the poor model fit forthe girls’ sample, it appears that sex differences in self-control are nota function of moderation effects, but of different etiological processesproducing self-control for girls vs. boys.

Discussion and conclusion

We posed three research questions in this paper. First, do sexdifferences in the levels of the predictors of self-control account for theeffect of sex on self-control (mediation hypothesis)? Second, do theknown determinates of self-control predict equally well for boys andgirls (moderation hypothesis)? Or, do a different set of variablesaltogether predict self-control for boys vs. girls, suggesting thatseparate causal processes are at play?

As expected based on the findings from previous research, wefound that compared to boys, girls scored significantly higher on ourmeasure of self-control. We also found that girls had significantlyhigher mean levels of maternal monitoring and that boys weresignificantly more likely to receive physical punishment when theymisbehaved. Because monitoring and discipline are consistentpredictors of self-control, these significant mean differences shouldostensibly translate into higher levels of self-control for girls, sug-gesting a mediating effect. Similar to the findings of Lynskey et al.(2000), we did find evidence that the effects of sex on self-controlwere partially mediated by monitoring and discipline. The coefficientfor sex was reduced by 34% when maternal supervision and physicalpunishment were included in the regression models presented onTable 2.

The analyses presented in Table 3 tested our moderation andseparate causal model hypotheses. Recall that we viewed the mod-eration hypotheses through a somewhat narrow lens—i.e., whetherthe same variables are predictive of self-control for boys and girls, butthe strength of those relationships differs by sex. Surprisingly, we

found that only one of the variables, maternal supervision, thatpredicted self-control for boys predicted self-control for girls. Thisfinding is not without precedent, however, as Chapple and Johnson(2007) using similar data for the NLSY79-Child data, found only onesignificant predictor of girls’ impulsivity as compared to six significantpredictors of boys’ impulsivity. Although Chapple and Johnson (2007)measure their outcome of impulsivity slightly differently than we do,and they used both family variables (supervision, maternal attach-ment and discipline) and developmental variables (motor skills andreading ability) as predictors of self-control, they also found anabsence of significant predictors for girls’ impulsivity.

Additionally, none of the “new” predictors (i.e., those not theo-rized by Gottfredson & Hirschi, 1990) that we examined weresignificant predictors of girls’ self-control. The striking finding fromour research is the absence of significant predictors of girls’ self-control coupled with the poor model fit for our self-control model inthe girls’ sample. Unfortunately, in our research, the origins of girls’self-control and the sources of girls’ higher self-control, remain amystery. Our findings are in contrast to prior research on the cor-relates of self-control by sex from Lynskey et al. (2000) and Beaveret al. (2007) in which the correlates of self-control and model fit weresimilar for boys and girls. We can speculate on methodologicalreasons why our findings diverge, such as small sample sizes anddifferent sets of predictors, but on a more substantive note, otherresearch on the acquisition of self-control (Higgins, 2004) findssimilar model misspecification and raises the possibility that perhapswe are inappropriately measuring and modeling girls’ self-control.Considering the small number of studies to date that present analysesseparately by sex, it seems to us that there is good reason to encouragefuture researchers to investigate more carefully potential sexdifferences in the causal models for self-control.

Our results for the boys were farmore encouraging. Despite havingsignificantly lower mean levels of self-control, boys’ self-control wassignificantly predicted by greater school and neighborhood socializa-tion (opposite from what Wikström and Sampson suggested), greater

Page 8: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1129C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

maternal monitoring and attachment, and negatively, by mother's useof physical discipline. The origins of boys’ self-control appear to bediverse and include both familial and extra-familial sources, as well aspositive socialization and control experiences (monitoring, attach-ment, and neighborhood organization) and negative experiences(physical punishment). Although our research was intended as anexploration of what made girls’ self-control greater than boys, weuncovered that the predictors of boys’ self-control are robust, andemerge from familial, school, and community sources. Finally, theeffect of maternal attachment on self-control was significantlystronger for boys than for girls, which suggests a potentially genderedetiological path through boys’ attachment to their mothers. Again, thisis an interesting finding worthy of replication for its implications ontheory and policy.

We are left, however with two unanswered questions: first, whydo girls have higher mean levels self-control? And, why d traditionalpredictors of self-control do such a poor job of predicting girls’ self-control? We believe that the two issues are related—that girls havehigher self-control, due in part, to unmeasured variables. When poormodel fit is found it is primarily to either or both measurement errorin the predictors or outcomes or model misspecification. To assessmeasurement error, we re-ran our analyses with a latent variable forself-control and found results identical to the ones we reported. Wealso ran exploratory factor analysis in MPLUS to assess genderdifferences in the measurement of self-control. With the exception ofthe poor loading of “restless and over-active” for girls, self-control wasmeasured similarly for boys and girls and as a scale, has similarreliabilities for boys and girls (.84 for boys; .80 for girls).

This subsequent analysis leaves us with the possibility of modelmisspecification. Although a few of our individual items are not ideal(parental supervision prenatal substance use), our models are acomprehensive representation of the extant criminological literatureon the predictors of self-control. So, we turned to the literaturepredicting similar constructs such as impulsivity, ADD/ADHD (lowself-control), social cognition and empathy (high self-control). Whilenone of these alternate constructs is as comprehensive as self-control,each shares an element of self-control that could impact sexdifferences in the levels of self-control. Our review suggests thatgender identity and peer relations are important predictors ofimpulsivity, social cognition and empathy. Research indicates thatrespondents with higher levels of peer functioning (low rejection andhigh cohesion) report lower rates of ADD/ADHD (Blachman &Hinshaw, 2002; Thurber, Heller & Hinshaw, 2002) and lower ratesof impulsivity (Snyder, Prichard, Schrepferman, Patrick, & Stoolmiller,2004). Peersmay be an important, untapped, developmental source ofself-control for boys and girls. Similarly, research has found thatrespondents who report more feminine gender characteristicsreported higher rates of empathy (Skoe, Cumberland, Eisenberg,Hansen, Perry, 2002), social cognition (Wood, Heitmiller, Andreasen,& Nopoulos, 2007;Wood,Murko &Nopoulos, 2008) and lower rates ofrisky behaviors (Murnen & Smolak, 1997; Williams & Ricciardelli,2003). This research also opens the possibility that some domains ofself-control may be more “gendered” than others—particularly thedomain of empathy or on its converse, disregard for the pain andsuffering of others.We believe that additional etiological paths to self-control exist and that researchers should assess domain specificgendered etiologies.

The current study has several strengths. Because longitudinal NLSYdata were employed, we were able to assess the etiology of self-controlamong a diverse sample of children from across the United States withan over-sample of children born to youngmothers, minorities, and poorwhites. We were also able to follow children longitudinally from ages10-11 until they were 12-13 using multiple reporters, thus allowing usto establish causal order and increase the validity of ourfindings. Finally,we included a wider variety of possible predictors of self-control thanmuch of the past research, including familial, neighborhood and school,

and biological variables. Despite these strengths, there are weaknesses.As mentioned above, the use of secondary data frequently posesunavoidable challenges. The measure of maternal monitoring was notideal; however, it was the best measure available given the restrictionsof the data. Also, we usedmother reports of child's self-control. Thoughthis may not be the ideal measure of low self-control, it is likely aconservative estimate. Finally, our sample sizes at 262 and 230 for boysand girls, respectively, are quite small for criminological research.

We believe that our work opens the door for more theorizing onthe causes of self-control for girls—as it may differ from boys. Ourresearch on boys’ self-control echoes the traditional literature andfinds that boys who come from more functional and supportivefamily, school, and neighborhood settings report greater self-control.However, what makes girls exhibit greater self-control remainsunclear—in these data at least. We believe this theoretical groundremains ripe for theory development and extension.

Appendix A

Items Comprising the Self-Control ScaleHow often are these statements true about your child…?Child cheats or tells lies.Child has difficulty concentrating.Child bullies or is cruel to others.Child is not sorry after misbehaving.Child has trouble getting along with other children.Child is impulsive, acts without thinking.Child is restless or over-active.Child is stubborn, sullen or irritable.Child has a very strong temper.Child has trouble getting along with teachers.Alpha=.85

Appendix B

OLS Regression of Sex and All Etiological Variables on Self-Control (Full Sample)

Demographics

Biological Family Environmental

b

b b b b

Female

1.238*** 1.261*** 1.247*** .845** .768* Non-White -.670* -.689* -.384 -.264 Maternaleducation

1.818***

1.714*** 1.634*** 1.434**

Normal birthweight

.212

.758 .254

No prenatalsubstanceuse

.112

.110 -.011

Non-physicalpunishment

1.645***

1.466***

Maternalattachment

.341***

.328***

Maternalsupervision

1.237***

1.169**

Neighborhoodsocialization

.180

Schoolsocialization

.249*

F

15.86 14.44 7.81 10.06 9.14 R2 .03 .08 .07 .16 .19 Adj. R2 .02 .07 .06 .14 .17

* p≤ .05, ** p≤ .01, *** p≤ .001.

Notes

1. The youngest women of the original NLSY79 sample are now over 40 years old.When these women complete their childbearing, the child and young adult sub-samples will be a nationally representative sample of children born to women whowere born between 1960-1965.

Page 9: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1130 C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

2. Our sample size is constrained and a bit small for multiple group analysis. This isdue to several limitations in the data. First, the family variables used as predictors areonly asked of children (not young adults) and children must be at least 10 years old toreceive these questions. Also, although there is measure of impulsivity/self-control inthe young adult sample (see Turner and Piquero, 2002), it is less reliable than themeasures of impulsivity/self-control in the child data. Finally, because the NLSY79switches over a proportion of 14 year olds to the young adult questionnaire who arecloser to their 15th birthday at survey administration, when 14 year old respondentsare included in a sample using child data, approximately 30% of the sample hasmissing data. It is for all these reasons we confined our sample to the child data onlyand compromised on sample size.

3. For the ease of interpretation in the multivariate analyses, physical punishmentis reverse coded so that high values indicate non-corporal punishment.

4. Of the women who reported substance use prior to the birth of the target child,the majority reported infrequent alcohol use. Therefore, our prenatal drug use variableis a conservative estimate of substance exposure in utero.

5. We also include our full model of predictors of self-control, entered in blocks, toshow the robustness of the sex effect net of all the causal predictors.

6. Due to the limited sample size, we also conducted a series of regressions using amultiplicative term to test the interactions between sex and the exogenous variables.We used the interaction command in STATA 9.1 to create the individual interactionterms. The only individual interaction term that was statistically significant was thesex X maternal attachment interaction (b=-.453, p=.011). The sex X physicalpunishment interaction was marginally significant (b=-1.493, p=.079).

References

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction insocial psychological research: Conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychology, 51, 1173−1182.

Beaver, K. M., DeLisi, M., Vaughn, M. G., Wright, J. P., & Boutwell, B. B. (2008a). Therelationship between self-control and language: Evidence of a shared etiologicalpathway. Criminology, 46, 393−970.

Beaver, K. M., DeLisi, M., Vaughn, M. G., & Wright, J. P. (2010). Intersection of genes andneuropsychological deficits in the prediction of adolescent delinquency and lowself-control. International Journal of Offender Therapy and Comparative Criminology,54, 22−42.

Beaver, K. M., & Wright, J. P. (2005). Evaluating the effects of birth complications lowself-control in sample of twins. International Journal of Offender Therapy andComparative Criminology, 49, 450−471.

Beaver, K. M., Wright, J. P., & DeLisi, M. (2007). Self-control as an executive function:Reformulating Gottfredson and Hirschi's parental socialization thesis. CriminalJustice and Behavior, 34, 1345−1361.

Beaver, K. M., Wright, J. P., DeLisi, M., & Vaughn, M. G. (2008b). Genetic influences onthe stability of low self-control: Results from a longitudinal sample of twins. Journalof Criminal Justice, 36, 478−485.

Belknap, J., & Holsinger, K. (2006). The Gendered Nature of Risk Factors forDelinquency. Feminist Criminology, 1, 48−71.

Blachman, D. R., & Hinshaw, S. P. (2002). Patterns of Friendship Among Girls With andWithout Attention-Deficit/Hyperactivity Disorder. Journal of Abnormal ChildPsychology, 6, 625−640.

Brame, R., Paternoster, R., Mazerolle, P., & Piquero, A. (1998). Testing for equality ofmaximum-likelihood regression coefficients between two independent equations.Journal of Quantitative Criminology, 14, 245−261.

Brown, R. T., Coles, C. D., Smith, I. E., Platzman, K. A., Silverstein, J., Erickson, S., & Falek, A.(1991). Effects of prenatal alcohol exposure at school age II. Attention and behavior.Neurotoxicology and Teratology, 13, 369−376.

Boutwell, B. B., & Beaver, K. M. (2010). The intergenerational transmission of low self-control. Journal of Research in Crime and Delinquency, 47, 174−209.

Burton, V. S., Cullen, F. T., Evans, T. D., Alarid, L. F., & Dunaway, R. G. (1998). Gender, self-control, and crime. Journal of Research in Crime and Delinquency, 35, 123−147.

Center for Human Resource Research. (2000). 1998 Child & young adult data: A user'sguide. Columbus, OH: The Ohio State University.

Center for Human Resourse Research. (2004). 2002 NLSY79 User's guide: A user's guide.Columbus, OH: The Ohio State University.

Chapple, C. L. (2005). Self-control, peer relations, and delinquency. Justice Quarterly, 22,89−106.

Chapple, C. L., Hope, T. L., & Whiteford, S. W. (2005a). The direct and indirect effects ofparental bonds, parental drug use, and self-control on adolescent substance use.Journal of Child and Adolescent Substance Use, 14, 17−38.

Chapple, C. L., & Johnson, K. A. (2007). Gender differences in impulsivity. Youth Violenceand Juvenile Justice, 5, 221−234.

Chapple, C. L., Tyler, K. A., & Bersani, B. E. (2005b). Child neglect and adolescentviolence: Examining the effects of self-control and peer rejection. Violence andVictims, 20, 39−53.

Chesney-Lind, M. (2006). Patriarchy, Crime and Justice: Feminist Criminology in an Eraof Backlash. Feminist Criminology, 1, 6−26.

Clogg, C. C., Petkova, E., & Haritou, A. (1995). Statistical methods for comparingregression coefficients between models. American Journal of Sociology, 100,1261−1293.

Cornelius, M. D., Ryan, C. M., Day, N. L., Goldschmidt, L., & Willford, J. A. (2001). Prenataltobacco effects on neuropsychological outcomes among preadolescents. Journal ofDevelopmental and Behavioral Pediatrics, 22, 217−225.

Daly, K. (1992). Women's Pathways to Felony Court: Feminist Theories of Lawbreakingand Problems of Representation. Southern California Review of Law & Women'sStudies, 2, 11−52.

Elgen, I., Sommerfelt, K., & Markestad, T. (2003). Population based, controlled study ofbehavioural problemsandpsychiatricdisorders in lowbirthweightchildrenat 11 yearsof age. Archives of Disease in Childhood Fetal and Neonatal Edition, 87, F128−F132.

Feldman, S. S., & Weinberger, D. A. (1994). Self-restraint as a mediator of familyinfluences on boys’ delinquent behavior: A longitudinal study. Child Development,65, 195−211.

Gibbs, J. J., Giever, D., & Martin, J. S. (1998). Parental management and self-control: Anempirical test of Gottfredson and Hirschi's general theory. Journal of Research inCrime and Delinquency, 35, 40−70.

Gibson, C. L., Sullivan, C. J., Jones, S., & Piquero, A. R. (2010). Does it take a village?Assessing neighborhood influences on children's self-control. Journal of Research inCrime and Delinquency, 47, 31−62.

Gottfredson, M. R., & Hirschi, T. (1990). The General Theory of Crime. Stanford, CA:Stanford University Press.

Hagan, J., & Foster, H. (2003). S/He's a Rebel: Towards a Sequential Stress Theory ofDelinquency and Gendered Pathways to Disadvantage in Emerging Adulthood.Social Forces, 82, 53−86.

Hagan, J., Simpson, J., & Gillis, A. R. (1987). Class in the household: A power-controltheory of gender and delinquency. American Journal of Sociology, 92, 788−816.

Hagan, J., Simpson, J., & Gillis, A. R. (1988). Feminist scholarship, relational andinstrumental control, and a power-control theory of gender and delinquency. TheBritish Journal of Sociology, 39, 301−336.

Hay, C. (2001). Parenting, self-control, and delinquency: A test of self-control theory.Criminology, 39, 707−736.

Hay, C., & Forrest, W. (2006). The development of self-control: Examining self-controltheory's stability thesis. Criminology, 44, 739−774.

Hayslett-McCall, K. L., & Bernard, T. J. (2002). Attachment, masculinity, and self-control:A theory of male crime rates. Theoretical Criminology, 6, 5−33.

Higgins, G. E. (2004). Gender and self-control theory: Are there differences in themeasures and the theory's causal model? Criminal Justice Studies, 17, 33−55.

Hope, T. L., & Chapple, C. L. (2005). Maternal characteristics, parenting, and adolescentsexual behavior: The role of self-control. Deviant Behavior, 26, 25−45.

Hope, T. L., Grasmick, H. G., & Pointon, L. J. (2003). The family in Gottfreson and Hirschi'sgeneral theory of crime: Structure, parenting and self-control. Sociological Focus, 36,291−311.

Knoch, D., & Fehr, E. (2007). Resisting the power of temptations: The right prefrontalcortex and self-control. Annals of the New York Academy of Sciences, 1104, 123−134.

LaGrange, T. C., & Silverman, R. A. (1999). Low self-control and opportunity: Testing thegeneral theory of crime as an explanation for gender differences in delinquency.Criminology, 37, 41−72.

Leech, S. L., Richardson, G. A., Goldschmidt, L., & Day, N. L. (1999). Prenatal substanceexposure. Effects on attention and impulsivity of 6-year-olds. Neurotoxicology andTeratology, 21, 109−118.

Lynskey, D. P., Winfree, T. L., Jr., Esbensen, F., & Clason, D. L. (2000). Linking gender,minority group status and family matters to self-control theory: A multivariateanalysis of key self-control concepts in a youth-gang setting. Juvenile and FamilyCourt Journal, 51, 1−19.

Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. A. (2001). Sex differences in antisocialbehavior: Conduct disorder, delinquency and violence in the Dunedin longitudinalstudy. Cambridge, UK: Cambridge University Press.

Mullins, C. W., & Miller, J. (2008). Temporal, situational, and interactional features ofwomen's violent conflicts. The Australian and New Zealand Journal of Criminology,41, 36−62.

Murnen, S. K., & Smolak, L. (1997). Femininity, masculinity, and disordered eating: AMeta-Analytic review. International Journal of Eating Disorders, 3, 231−242.

Nofziger, S. (2008). The ‘cause’ of low self-control: The influence of maternal self-control. Journal of Research in Crime and Delinquency, 45, 191−224.

Perrone, D., Sullivan, C. J., Pratt, T. C., & Margaryan, S. (2004). Parental efficacy, self-control, and delinquency: A test of a general theory of crime on a nationallyrepresentative sample of youth. International Journal of Offender Therapy andComparative Criminology, 48, 298−312.

Peterson, J. L., & Zill, N. (1986). Marital disruption, parent-child relationships, andbehavior problems in children. Journal of Marriage and the Family, 48, 295−307.

Pratt, T. C., Turner, M. G., & Piquero, A. R. (2004). Parental socialization and communitycontext: A longitudinal analysis of the structural sources of low self-control. Journalof Research in Crime and Delinquency, 41, 219−243.

Ratchford, M., & Beaver, K. M. (2009). Neuropsychological deficits, low self-control, anddelinquent involvement: Toward a biosocial explanation of delinquency. CriminalJustice and Behavior, 36, 162–147.

Shaw, C. R., & McKay, H. D. (1942). Juvenile Delinquency in Urban Areas. Chicago:University of Chicago Press.

Skoe, E. E. A., Cumberland, A., Eisenberg, N., Hansen, K., & Perry, J. (2002). The influencesof sex and gender-role identity on moral cognition and prosocial personality traits.Sex Roles, 46, 295−309.

Snyder, J., Prichard, J., Schrepferman, L., Patrick, M. R., & Stoolmiller, M. (2004). Childimpulsiveness-inattention, early peer experiences and the development of earlyonset conduct problems. Journal of Abnormal Child Psychology, 6, 579−594.

Tittle, C. R., Ward, D. A., & Grasmik, H. G. (2003). Gender, age, and crime/deviance: Achallenge to self-control theory. Journal of Research in Crime and Delinquency, 40,426−453.

Thurber, J. R., Heller, T. L., & Hinshaw, S. P. (2002). The social behaviors and peerexpectations of girls with attention deficit hyperactivity disorder and comparisongirls. Journal of Clinical Child and Adolescent Psychology, 4, 443−452.

Page 10: Journal of Criminal Justice Volume 38 Issue 6 2010 [Doi 10.1016%2Fj.jcrimjus.2010.08.004] Constance L. Chapple; Jamie Vaske; Trina L. Hope -- Sex Differences in the Causes of Self-control-

1131C.L. Chapple et al. / Journal of Criminal Justice 38 (2010) 1122–1131

Turner,M.G., & Piquero, A. R. (2002). The stability of self-control. Journal of Criminal Justice,30, 457−471.

Turner, M. G., Piquero, A. R., & Pratt, T. C. (2005). The school context as a source of self-control. Journal of Criminal Justice, 33, 327−339.

Tyler, K. A., Stone, R. T., & Bersani, B. (2006). Examining the changing influence ofpredictors of adolescent alcohol misuse. Journal of Child & Adolescent SubstanceAbuse, 16, 95−114.

Unnever, J. D., Cullen, F. T., & Pratt, T. C. (2003). Parental management, ADHD, anddelinquency involvement: Reassessing Gottfredson and Hirschi's general theory ofcrime. Justice Quarterly, 20, 471−500.

Vaughn, M. G., DeLisi, M., Beaver, K. M., & Wright, J. P. (2009). Identifying latent classesof behavioral risk based on early childhood: Manifestations of self-control. YouthViolence and Juvenile Justice, 7, 16−31.

Vazsonyi, A. T., & Huang, L. (2010). Where self-control comes from: On the de-velopment of self-control and its relationship to deviance over time. DevelopmentalPsychology, 46, 245−257.

Wikström, P. H., & Loeber, R. (2000). Do disadvantaged neighborhoods cause well-adjusted children to become adolescent delinquents? A study of male juvenileserious offending, individual risk and protective factors, and neighborhood context.Criminology, 38, 1109−1142.

Wikström, P. H., & Sampson, R. J. (2003). Social mechanisms of community influenceson crime and pathways in criminality. In B. B. Lahey, T. E. Moffitt, & A. Caspi (Eds.),

The Causes of Conduct Disorder and Serious Juvenile Delinquency (pp. 118−148).New York: Guilford Press.

Williams, R. J., & Ricciardelli, L. A. (2003). Negative perceptions about self-control andidentification with gender-role stereotypes related to binge eating, problem drinkingand co-morbidity among adolescents. Journal of Adolescent Health, 32, 66−72.

Winfree, L. T., Jr., Taylor, T. J., He, N., & Esbensen, F. (2006). Self-control and variabilityover time: Multivariate results using a 5-Year, multisite panel of youths. Crime andDelinquency, 52, 253−286.

Wood, J. L., Heitmiller, D., Andreasen, N. C., & Nopoulos, P. (2007). Morphology of theventral cortex: Relationship to femininity and social cognition. Cerebral Cortex, 18,534−540.

Wood, J. L., Murko, V., & Nopoulos, P. (2008). Ventral frontal cortex in children:Morphology, social cognition and femininity/masculinity. Social Cognitive andAffective Neuroscience, 3, 168−176.

Wright, D., & Beaver, K. (2005). Do parents matter in creating self-control in theirchildren? A genetically informed test of Gottfredson and Hirschi's theory of lowself-control. Criminology, 43, 1169−1202.

Wright, J. P., Beaver, K. M., DeLisi, M., & Vaughn, M. (2008). Evidence of negligibleparenting influences on self-control, delinquent peers, and delinquency in a sampleof twins. Justice Quarterly, 25, 544−569.