Use of weighted path analysis in testing the influence of self-regulation, risk proneness, peer pressure, and substance use on adolescent sexual risk behavior

Crocket, Rafaelli and Shen (2006) explored the relationship among self-regulation (behavioral problems) in early childhood, risk proneness (sensation seeking), peer pressure and substance use in early adolescence for their effect on sexual risk taking in later adolescence, using three waves of the National Longitudinal Survey on Youth-child data (1990, 1994 and 1998 respectively). Their structural equation model (SEM) revealed behavioral problems in early childhood predispose youth in mid-adolescence to perceive themselves as engaging in higher sensation seeking (assessed in 1994). This successive combination leads to alcohol use and sexual risk taking in mid-adolescence (outcomes in 1998). Though their research substantiated the relationship among these underlying mechanisms longitudinally, their computations conducted without application of sampling weights, did not yield significant pathways between self-regulation in mid-childhood and peer pressure in early adolescence. Further, differences between racial/ethnic groups were not detected. In order to control for oversampling of underrepresented minorities, their study is replicated in this paper by applying the transformed raw weights to the covariance matrix calculated in SPSS and analyzed in AMOS. The weighted path analysis (i.e. an algebraic formula employed in calculation of the covariance matrix to adjust for post-study design effect) findings demonstrate both ethnic and gender variation in the link among self-regulation, risk proneness and consequential sexual risk taking. Mathematical weighting technique thereby yields results supporting the need for targeted culturally sensitive mental health interventions, tailored to adolescents based on their race and gender.