Responses were coded as ��1�� selleck chemicals llc for yes and ��0�� for no. Results Race differences in mean levels of measured indicators of risk, protection, and smoking were tested using multivariate analysis of variance for continuously measured variables and chi-square for dichotomous variables. Correlated measures of parental influences (monitoring, discipline, attachment, and guidelines) and deviant peer behaviors (marijuana and alcohol use, delinquent behavior, and trouble at school) were modeled as latent constructs (J?reskog, 1971). Multiple-group confirmatory factor analysis (Bontempo & Hofer, 2006; Muth��n, 1989) was used to establish measurement invariance and to test for race differences in levels of latent constructs. Blacks were less likely than Whites to report smoking in the 10th grade (see Table 1); however, the difference was not significant.

Blacks were significantly more likely than Whites to have a parent who currently smoked (��2 = 31.4, df = 3, p < .0001). Black families reported significantly lower per capita income than White families (t = 9.67, p < .0001). Black parents reported significantly higher guidelines against substance use (t = 3.49, p = .0006). Black teens reported slightly higher association with peers who get in serious trouble at school (��2 = 3.86, df = 1, p = .05). Multiple-group confirmatory factor analysis was conducted on the two latent constructs, parenting and deviant peers. The intercept and loading for guidelines were significantly higher in the Black than White sample and were freely estimated, whereas all other intercepts and loadings were constrained to be equal across groups.

These results suggest an acceptable level of equality of the measures across race. No differences in mean levels of parenting and deviant peers between Black and White families were detected (t = ?.43 and ?1.36, respectively). Similar analyses confirmed no differences in measurement parameters by gender. Multiple-group structural equation modeling (SEM; Muth��n & Muth��n, 2004) was used to compare the magnitude of the impact of risk and protective factors by race (see Figure 1). We compared the fit of multiple-group SEM with all path coefficients allowed to differ between race groups to multiple-group SEM with all path coefficients constrained to be equal across AV-951 race using a derivatives differences test. This comparison revealed no significant different in fit between the constrained and unconstrained models, indicating the magnitude of the impact of risk and protective factors was not significantly different for Black and White families. The constrained model fit the data adequately (Figure 1; ��2 = 63.41, df = 9, p = .08, comparative fit index = 0.98, Tucker-Lewis index = 0.98).