Work Effects on Substance Use Among Formerly Detained Emerging Adults: Findings from Panel Data

Background/Purpose: Substance use among employed adults is well documented, reporting a positive association between the two factors. But previous studies were mostly based on a cross-sectional design; little is known whether the same directionality can be confirmed by a longitudinal design. Moreover, the association between employment and substance use focusing on at-risk young adults who were formerly incarcerated has received relatively little analysis. To fill the gaps in literature, guided by social control theory, we examine to what extent employment is associated with substance use over time among formerly incarcerated young people through panel analysis.

Methods: We used the most recent three waves of survey data from the Pathways to Desistance study (N = 1,354; 2000-2010). Dependent variables were binary substance use, including alcohol, marijuana, and illegal drugs (no/yes). A main independent variable was the work measured by the number of weeks worked across all community and under-the-table jobs. Time-variant and -invariant covariates included socio-demographics. After conducting descriptive and bivariate analyses, we used a hybrid approach combining within- and between-cluster effects.

Results: Overall, the participants worked less than half a year; the mean use of alcohol, marijuana, and illegal drugs were 52%, 31.2%, 11.5%, respectively, across waves. A Friedman test indicated there was a significant change over time in employment (p < .001). Cochran’s Q test revealed there was a significant change in the proportion of alcohol use across waves (p < .05). Our hybrid model with between-effects confirmed employment was associated with alcohol (b = .07, CSE = .01, p < .001) and illegal drug use (b = -.02, CSE = .01, p < .05), but not with marijuana. Regarding the time-variant covariates, friendship quality was negatively associated with all substance use; gang involvement was positively associated with all substance use; adverse neighborhood conditions and combination of mental health and IPV were positively associated with marijuana and drug use. For the models with within-effects, only age was positively associated with alcohol and illegal drug use. The random effects estimates of time-invariant predictors confirmed that females and those who enrolled in school before detention were less likely to use marijuana than counterparts; African-Americans were less likely to use illegal drugs than Whites.

Conclusions and Implications: Our results indicate that with respect to inter-individual processes, people who had more weeks worked are more likely to engage in alcohol use, but less likely to engage in illegal drug use. This implies that employment may allow them to socially drink alcohol, thus playing a role in the socialization of participants’ drinking behavior, given the small magnitude of the coefficient. However, employment may also have a deterrent effect on illegal drug use. These findings are different from previous cross-sectional research, reporting positive associations between employment and all types of substance use. Interestingly, with respect to intra-individual processes, the association remains insignificant in all models, highlighting that employment may not contribute to changing an individual’s proclivity for substance use. The causal dynamics underlying the association between employment and substance use remain to be further investigated.