A Multigroup, Longitudinal Study of Truant Youths, Marijuana Use, Depression, and STD-Associated Sexual Risk Behavior

ABSTRACT

Truant youths are likely to engage in a number of problem behaviors, including sexual risky behaviors. Previous research involving non-truant youths has found sexual risk behaviors to be related to marijuana use and depression, with differential effects for male and female youths. Using data collected in a National Institute on Drug Abuse (NIDA)-funded, prospective intervention project, results are reported of a male-female, multi-group, longitudinal analysis of the relationships among truant youth baseline sexual risk behavior, marijuana use, and depression, and their sexual risk behavior over four follow-up time points. Results indicated support for the longitudinal model, with female truants having higher depression scores, and showing stronger relationships between baseline depression and future engagement in sexual risk behavior, than male truants. Findings suggest that incorporating strategies to reduce depression and marijuana use may decrease youth sexual risk behavior.

KEYWORDS:

• depression
• marijuana use
• sexual risk
• STDs

Funding

The research reported in this article was funded by National Institute on Drug Abuse (NIDA) Grant No. DA021561.

Notes

1. Following the National Institute on Drug Abuse (NIDA)-approved protocol for the study, the length of follow-up was premised on year of entry into the project. Since some youths had missing follow-up data due to this sampling design, as well as typical attrition, the multigroup analysis of the just-identified model was replicated involving only cases with complete data on all variables in the model. The analysis involved 136 male and 74 female youths. With one exception, the same pattern of results was found for those including missing data. For the one exception, the original analysis involving estimation of missing data found a significant relationship between sexual risk behavior at six-month follow-up and baseline marijuana use among females; however, this relationship was nonsignificant in the analysis involving cases with complete data on all variables. Therefore, missing data due to sampling design did not substantively affect the model.

2. As with the just-identified model, the over-identified model was replicated using only cases with complete data on all variables in the model to gain an idea of model fit by eliminating the correlation between depression and marijuana use at baseline. Consistent with the version of the model containing missing data, the results indicated an excellent fit of the model to the data: chi-square = 2.19, df = 2, p = 0.33; RMSEA = 0.030; CFI = 1.000; TLI = 0.996. In particular, all residual correlations were at or near zero. Therefore, missing data due to the sampling design did not substantively affect the model.