مسیرها و عوامل پیش بینی کننده رفتارهای ضد اجتماعی در نوجوانان آمریکایی آفریقایی تبار از محله های فقیر

Abstract Antisocial behavior among youth remains a serious personal and social problem in the United States. The purposes of this study were to (1) identify the shape and number of developmental trajectories of antisocial behavior in a sample of poor, inner-city African American youth, and (2) test predictors of group membership and the developmental course of antisocial behaviors. Using growth mixture modeling, we examined predictors of antisocial behavior pathways and the likelihood of arrest in a sample of 566 poor, urban African American adolescents (ages 11 to 16). Three distinct trajectory classes of antisocial behavior were identified over a period of six years: one low-risk group (low steady) and two high-risk groups (incremental and high starter). The conditional probabilities for being arrested during ages 14–16 were 0.18 for the low steady class, 0.68 for the incremental class, and 0.31 for the high starter class. Prevention strategies for adolescents at high risk are discussed.

Introduction Antisocial behavior, often interchangeably used with delinquency, violent behavior, conduct problem, and deviant behavior, is one of the most serious behavioral problems in the United States and incurs costs to individuals, families, and the society as a whole. Antisocial behavior is defined in this paper as behavior that violates social norms or the rights of fellow human beings. Because of its significance, a number of efforts exist to understand how antisocial behavior develops over the life course and to comprehend the predictors and consequences (e.g., Schaeffer, Petras, Ialongo, Poduska, & Kellam, 2003). Although a general tendency of decreasing rates of antisocial behavior is reported as youth reach adulthood, research also indicates a strong link between childhood antisocial behavior and subsequent chronic offenses during adulthood ( Bongers et al., 2008 and Huesmann et al., 2002). In particular, minority youth in urban, poor neighborhoods have high risks of being exposed early to violence and of developing antisocial behavior ( Spano et al., 2006, Tolan et al., 2003 and Walker et al., 2007). Thus, it seems paramount to develop effective prevention and intervention efforts for minority youth to reduce the chances of subsequent criminal behaviors. Research indicates that biology and personal attributes such as social maladjustment and value orientation play an important role in predicting antisocial behavior of youth (Moffitt, 1993), yet attention has also been placed on contextual characteristics that shape the environment for the youth. An ecological perspective indicates that an environment surrounding individuals (e.g., family, school) can have a profound influence on adolescents' antisocial behavior (Gorman-Smith et al., 2000 and Seidman et al., 1998). Nevertheless, neither developmental perspectives nor environmental characteristics alone may account for trajectories of antisocial behavior. The developmental–ecological perspective argues that individual development is influenced by the ongoing qualities of the social settings in which the child lives and interacts with outer world (Gorman-Smith et al., 1998, Gorman-Smith et al., 2000 and Le Blanc and Kaspy, 1998). That is, personal characteristics and environmental settings may collectively contribute to the formation and development of the trajectories of antisocial behavior.

6. Results Fig. 2 shows the observed pooled means of antisocial behavior over time. The figure suggests that over the six years, antisocial behavior remained steady; this assumes that everyone follows one population distribution. Since this approach does not allow for subgroup differences in the population, we classified the entire group into multiple classes using GMM. Observed means of antisocial behavior (N=566). Fig. 2. Observed means of antisocial behavior (N = 566). Figure options To decide the optimal number of classes, the linear models of one class through five classes were estimated. Table 1 presents the results based on the model-selection criteria. The three-class model had the lowest BIC value whereas the four-class model provided the highest entropy (0.91) followed by the three-class model (0.88). BLRT favored the three-class model over the two-class model and the four-class model over the three-class model. However, the four-class model included a class with a very small number of members (21), which does not provide distinct class characteristics. In consideration of the parsimony of modeling and overall model quality based on evaluation criteria, the three-class model was selected as the final model. Also, the matrix of average conditional probabilities for cases to be in their respective classes in the three-class model had high diagonal values of 0.93, 0.90, and 0.96, indicating good classification quality by this solution. Table 1. Model-selection criteria. Model BIC Entropy BLRT (H0 = k − 1 classes) 1 class 11,235.87 N/A N/A 2 classes 11,087.34 0.85 p < 0.0001 3 classes 11,044.40 0.88 p < 0.0001 4 classes 11,053.24 0.91 p < 0.0001 5 classes 11,102.21 0.87 p = 0.43 Table options Fig. 3 presents the means weighted by estimated class probabilities of three classes over six years. Three distinct groups were identified considering their trajectories of antisocial behavior. The majority of the adolescents fell into the low steady group (n = 435, 77%), which maintained antisocial behavior at low levels over six years. The incremental group (n = 86, 15%) consisted of adolescents whose antisocial behavior was low at age 11 but kept increasing thereafter. Finally, the high starter group (n = 45, 8%) followed the trajectory that started high at age 11 but decreased to a similar level of the low steady group by age 16. Antisocial behavior means weighted by estimated class probabilities (N=566). Fig. 3. Antisocial behavior means weighted by estimated class probabilities (N = 566). Figure options To further understand the profiles of each class obtained, logistic regressions were completed to examine the effects by each predictor on class membership. Considering the high amount of collinearity among predictors, a separate model for each predictor was tested, where the low steady group was used as the reference group. Table 2 presents odds ratios and confidence intervals. Membership in the incremental group over the low steady group was associated with being suspended from school, smoking, and using alcohol/drugs at age 11, even though involvement of antisocial behavior was quite comparable between the two groups. The membership of the high starter group compared to the low steady group was related to poorer parental behavioral control, school suspensions, smoking, use of alcohol and drugs, and greater feelings of hopelessness at baseline. Table 2. Association between predictors and class membership. Predictor Incremental High starter Odds ratio 95% C.I. Odds ratio 95% C.I. Family factor Parental behavioral control 1.11 0.91–1.34 1.30⁎ 1.00–1.68 School factors Suspension 2.37⁎⁎ 1.48–3.80 3.56⁎⁎ 1.89–6.69 Expulsion 1.99 0.93–4.28 1.08 0.31–3.71 Individual factors Smoke 2.25⁎⁎ 1.32–3.84 2.23⁎ 1.11–4.47 Alcohol/drug 7.33⁎⁎ 2.48–21.72 89.37⁎⁎ 32.96–242.32 Self-worth 0.93 0.82–1.04 0.86 0.74–1.01 Hopelessness 0.97 0.85–1.12 1.54⁎⁎ 1.28–1.84 Note. Odds ratios are relative to the low steady group. ⁎ p < 0.05. ⁎⁎ p < 0.01. Table options To check the predictability of the identified trajectories for later antisocial behavior, the odds ratios of being arrested were obtained using the low steady group as the reference group. The odds ratios of being arrested was 9.77 (p = 0.001) for the incremental group and 2.14 (p = 0.015) for the high starter group. That is, the incremental group had the highest chance of being arrested later, followed by the high starter group, and the low steady group had the lowest chance of being arrested.