بررسی طولی عوامل پیش بینی کننده بزهکاری: تجزیه و تحلیل داده ها از سازمان جوانان موبایل

Abstract This study analyzed the relationships among adolescent delinquency, self-worth, peer influence, and family cohesion (i.e., maternal and paternal warmth). The longitudinal analysis identified how these relationships develop and change through adolescence. Using data from the Mobile Youth Survey, a 14-year longitudinal study of high-poverty, primarily Black American youths living in Alabama (N = 5400), delinquency, self-worth, and peer influence were analyzed in linear growth models. Results from these three linear growth models are presented. Findings include a significant increase in delinquency over time for the adolescents in the study and significantly lower rates of delinquency overall for females than males. Delinquency was also found to have negative relationships to both parental warmth and self-worth, with higher levels leading to decreased delinquency. Peer influence was found to have a gender effect, with males exhibiting steady rates, while females exhibit an increase in peer influence over time. Furthermore, maternal warmth and self-worth are also found to increase the rates of peer influence as well as significantly increasing self-worth

Introduction Despite decades of research and intervention, juvenile delinquency remains a social problem in modern American culture. Studies have shown that deviant behavior in adolescence leads to an increased likelihood of adult criminal behavior (Haynie, 2001 and Patterson et al., 1989). The literature abounds with empirical support for several causal factors. Although early research was often limited by its cross-sectional nature and attention to few variables, more recent research has been more complex, with studies based on longitudinal data and examining variables such as gender, ethnicity, and socioeconomic status. Still, such research is in its relative infancy and additional research is needed to increase knowledge regarding the various pathways through which certain juveniles become classified as delinquent. Although relatively little research has investigated predictors of delinquency among youths living in extreme poverty, some factors, such as inconsistent or low levels of parental monitoring and association with deviant peers have emerged as promising for further investigation for these youths as well as youths living in other situations (Brody et al., 2006; Hoeve et al., 2009, Murphy et al., 2012, O'Donnell et al., 2012 and Simons and Burt, 2011). In contrast, the literature on the causes and consequences of delinquent behavior among Black American adolescents is extensive. For example, peer relationships (O'Donnell et al., 2012), family structure and relationships (Farrington, Jolliffe, Loeber, Stouthamer-Loeber, & Kalb, 2001), and self-image (McMahon & Watts, 2002) have been cited as factors associated with deviant behavior in Black American adolescents, although few models have been developed to explore the strength of these associations. It is likely that age, gender, and ethnicity influence how various factors, both independently and interactively, influence delinquent behaviors among adolescents (Daigle et al., 2007, Ge et al., 2002, O'Donnell et al., 2012 and Whaley et al., 2010). Association with delinquent peers, for example, has been found to be a better predictor of delinquency for males than for females (Piquero, Gover, MacDonald, & Piquero, 2005). Across gender and ethnicity, the effects of relationships with parents and problem behaviors have been shown to be partially mediated by school success and time spent with friends (Pilgrim, Schulenberg, O'Malley, Bachman, & Johnston, 2006). In addition, gender differences in rates of juvenile delinquency and in factors that predict juvenile delinquency have been reported (Fagan and Wright, 2012, Jennings et al., 2009, Neumann et al., 2010 and Snyder et al., 2008). In this study, we use longitudinal data from the Mobile Youth Survey (Bolland, 2004) to explore the effects of self-worth, peer influence, and family cohesion on delinquency in a sample of black adolescent males and females living in extreme poverty. By focusing on strengths of associations rather than just existence of associations, the study fills a gap in the literature. Studies of delinquency often have been framed in one or more of three criminological theories: differential association/social learning; social control/social bonding, and general strain (Cullen, Wright, & Blevins, 2006; Lilly, Cullen, & Ball, 2007). Research that is framed within a particular theory, however, focuses on the variables important to that theory; such research may, therefore, fail to examine the influence of other variables important to the phenomena of interest (Daigle et al., 2007). We have avoided this limitation by selecting variables that have been demonstrated to be important, rather than by attempting to test one or more theories. We call upon these theories; however, to explain the findings, building a framework that links individual choice and environmental influence.

. Results 4.1. Model 1: delinquency A linear growth model was estimated with delinquency of the individual as the dependent variable, peer influence on delinquency, self-worth, warmth toward the mother, and warmth toward the father as time-varying covariates, and gender as a time-invariant covariate. Non-significant terms were removed from the final model. Estimates for the unconditional growth model and the final conditional model are shown in Table 1. Table 1. Linear growth model of delinquency of the adolescent. Parameter Unconditional growth model Full model Estimate SE Estimate SE Intercept 2.327⁎⁎⁎ 0.095 8.732⁎⁎⁎ 0.509 Peer inf. − 0.115⁎⁎⁎ 0.015 Self-worth − 0.390⁎⁎⁎ 0.053 Mom − 0.358⁎⁎⁎ 0.059 Gender − 2.885⁎⁎⁎ 0.522 Gender ∗ self-worth 0.104⁎⁎ 0.047 Gender ∗ mom 0.260⁎⁎ 0.079 Age 0.497⁎⁎⁎ 0.026 0.987⁎⁎⁎ 0.101 Age ∗ gender − 0.207⁎⁎⁎ 0.053 Age ∗ self-worth − 0.039⁎⁎ 0.013 Age ∗ dad − 0.021⁎⁎ 0.008 Variance (intercept) 7.407⁎⁎⁎ 0.915 4.026⁎⁎⁎ 0.893 Variance (age) 0.479⁎⁎⁎ 0.065 0.238⁎⁎⁎ 0.066 Error variance 15.775⁎⁎⁎ 0.275 14.787⁎⁎⁎ 0.304 Deviance 90,881.0 71,179.6 ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options The unconditional growth model demonstrates a significant increase in delinquency over time for the adolescents in the study, γ = 0.50, t(4264) = 374.84, p < .001. The conditional growth model shows a statistically significant improvement in fit over the unconditional growth model, χ2(9) = 19,701.4, p < .001. This significant growth in delinquency is also present in the final model, γ = 0.99, t(3402) = 96.38, p < .001. We find this growth rate to be moderated by gender, as we find a significant interaction between age and gender, γ = − 0.21, t(3265) = 15, p < .001. Males have a higher growth in delinquency than do females. Gender also appears as a main effect, γ = − 2.89, t(3265) = 30.52, p < .001, indicating that males have higher delinquency rates at age 11 than do females. From these two effects, we see that males exhibit higher delinquency at age 11 and then exhibit a higher growth in delinquency throughout their adolescent years than do females. The amount of maternal warmth had a significant effect on delinquency, γ = − 0.36, t(3265) = 37.54, p < .001. Those adolescents with higher maternal warmth exhibited lower overall delinquency. Gender moderated the effect of maternal warmth at age 11, as seen by the significant interaction effect, γ = 0.10, t(3265) = 10.7, p = .001. The impact of maternal warmth on the delinquency of the adolescent was greater for males, with greater maternal warmth predicting lower delinquency at age 11. While this trend is similar in direction for females, the effect is smaller. Paternal warmth had a significant effect on the change in delinquency over time, γ = − 0.02, t(3265) = 7.57, p = .006. Higher paternal warmth results in less of an increase in delinquency across adolescence. This effect was found to be identical for both genders. Additionally, self-worth was found to have a significant effect on delinquency at age 11, γ = − 0.39, t(3265) = 53.3, p < .001, as well as having a significant impact on delinquency over time, γ = − 0.04, t(3265) = 8.74, p = .003. Adolescents having higher levels of self-worth will not only have lower delinquency at age 11, but they will also have significant slower rates of increase in their delinquency rates as they age. The rate of increase in delinquency as they age is proportional to their amount of self-worth, meaning the higher the self-worth, the less the increase. Peer influence also had an effect on the delinquency of the adolescents, γ = − 0.12, t(3265) = 56.69, p < .001. Surprisingly, the coefficient was negative, indicating that adolescents not engaged in delinquent behaviors receive pressure from their peers to participate in delinquent behavior. 4.2. Model 2: peer influence A linear growth analysis was performed using peer influence as a dependent variable, self-worth, maternal warmth, and paternal warmth as time-varying covariates, and gender as a time-invariant covariate. Non-significant terms were removed from the final model. Estimates for the unconditional growth model and the final conditional model are shown in Table 2. Table 2. Linear growth model of peer influence. Parameter Unconditional growth model Full model Estimate SE Estimate SE Intercept 15.659⁎⁎⁎ 0.059 13.039⁎⁎⁎ 0.192 Self-worth 0.257⁎⁎⁎ 0.018 Gender 1.830⁎⁎⁎ 0.266 Mom 0.129⁎⁎⁎ 0.028 Gender ∗ mom − 0.123⁎⁎ 0.040 Gender ∗ self-worth − 0.063⁎ 0.024 Age 0.058⁎⁎⁎ 0.014 0.001 0.020 Gender ∗ age 0.070⁎ 0.028 Variance (intercept) 3.285⁎⁎⁎ 0.343 2.918⁎⁎⁎ 0.327 Variance (age) 0.081⁎⁎⁎ 0.021 0.073⁎⁎⁎ 0.020 Error variance 6.884⁎⁎⁎ 0.021 6.516⁎⁎⁎ 0.106 Deviance 78,057.7 73,273.3 ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options The unconditional growth model shows a significant increase in reported levels of peer influence over time, γ = 0.06, t(4455) = 16.06, p < .001. The conditional growth model shows a statistically significant improvement in fit over the unconditional growth model, χ2 (6) = 4784.4, p < .001. Interestingly, in the final model, we find that age itself is not significant, γ = 0.001, t(4283) = 0.00, p = .96, but rather is moderated by gender, γ = 0.07, t(4925) = 6.18, p = .013. These effects show that females see an increase in the amount of peer influence, while the peer influence for males remains relatively stable. At age 11, females report a higher level of peer influence than their male counterparts, γ = 1.83, t(4925) = 47.20, p < .001. From these two results we see that not only are females receiving higher peer pressure at age 11, but this peer influence will increase steadily as they get older. Males are receiving less pressure from peers, with no significant increase over time. In this model, self-worth also has a significant effect on peer influence, γ = 0.26, t(4925) = 203.88, p < .001. This effect is also moderated by gender, γ = − 0.06, t(4925) = 6.62, p = .01. Males show an increase in peer influence as their self-worth increases. Interestingly, the opposite effect is shown for females, with increases in self-worth resulting in decreasing amounts of peer influence on delinquency. Maternal warmth also has a significant effect on the peer influence measure, γ = 0.13, t(4925) = 20.49, p < .001. This effect is also moderated by gender, γ = − 0.12, t(4925) = 9.57, p = .002. The males in the sample exhibit a positive relationship between maternal warmth and peer influence, with increases in maternal warmth leading to an increase in peer influence. For females, however, the level of peer influence on delinquency remained steady irrespective of the level of maternal warmth. Paternal warmth was not significant in this model. 4.3. Model 3: self-worth A linear growth model was estimated with self-worth as the dependent variable, maternal warmth and paternal warmth as time-varying covariates, and gender as a time-invariant covariate. Non-significant terms were removed from the final model. Estimates for the unconditional growth model and the final conditional model are shown in Table 3. Table 3. Linear growth model of self-worth of the adolescent. Parameter Unconditional growth model Full model Estimate SE Estimate SE Intercept 6.119⁎⁎⁎ 0.038 5.004⁎⁎⁎ 0.090 Dad 0.088⁎⁎⁎ 0.015 Mom 0.125⁎⁎⁎ 0.018 Gender ∗ mom 0.087⁎⁎⁎ 0.019 Gender ∗ dad − 0.059⁎⁎ 0.020 Age 0.107⁎⁎⁎ 0.009 0.109⁎⁎⁎ 0.010 Variance (intercept) 2.267⁎⁎⁎ 0.143 2.918⁎⁎⁎ 0.153 Variance (age) 0.061⁎⁎⁎ 0.008 0.073⁎⁎⁎ 0.009 Error variance 2.302⁎⁎⁎ 0.037 6.516⁎⁎⁎ 0.041 Deviance 65,707.6 53,263.9 ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options The unconditional growth model shows a significant increase in self-worth over time for the adolescents in the study, γ = 0.11, t(4541) = 136.11, p < .001. The conditional growth model shows a statistically significant improvement in fit over the unconditional growth model, χ2 (4) = 12,443.7, p < .001. In the final model, we do see this significant increase in self-worth over time for adolescents, γ = 0.11, t(3740) = 122.23, p < .001. Unlike the previous two models, these growth rates are not moderated by gender, rather, they are identical for the two genders. There are no moderating effects of the increase in self-worth over time. What does change in this model is the self-worth at age 11. We find a significant effect of maternal warmth, γ = 0.13, t(3978) = 46.15, p < .001, as well as for paternal warmth, γ = 0.09, t(3978) = 34.27, p < .001. Both of these effects are also moderated by gender, as seen by the interactions with gender and maternal warmth, γ = 0.09, t(3978) = 20.01, p < .001, as well as with gender and paternal warmth, γ = − 0.06, t(3978) = 8.82, p = .003. Adolescents with greater maternal warmth exhibit higher self-worth ratings. Although maternal warmth is positively related to self-worth for both males and females, the relationship is stronger for females. Paternal warmth is also positively related to self-worth for both males and females, with a stronger relationship for males. The same gender parent has greater influence on the self-worth of the adolescent. Self-worth of the adolescent is strongly predicted by their beginning self-worth at age 11, as all groups experience identical rates of increase