The use of a Marakov dependent Bernoulli process to model the relationship between employment status and drug use

This study uses the 1984,1988,1992, and 1994 samples of the National Longitudinal Survey of Youth (NLSY) to estimate the effect of illicit drug use on the probability of employment of those in the labor force. The effects of the frequency of drug use, and duration of such use, are investigated using longitudinal logit models. The weekly employment status of the respondents is modeled as a Markov dependent Bernoulli process. This model assumes that the probability of employment is influenced by the personal and socio-economic characteristics of the respondents as well as by the employment status of the previous week. Separate analyses are carried out for male and female samples. For both male and female samples, the results are consistent with the hypotheses that illicit drug use, in particular past and long-term use, have a significant negative effect on employment. The analyses show that current and short-term drug use have no significant impact on employment. The odds ratio is used to draw quantitative conclusions. The ratio of the odds of being employed under various drug use scenarios vs. nonuse is computed by gender, marital status, type of job, and race. A consistent picture of reduced odds of employment among drug users emerge from the analyses. Within each female or male sample, the odds ratios are very similar across the categories. However, the odds ratios for women are considerably smaller than the odds ratios for men. Thus, drug use appears to have a greater impact on the employment status of women than that of men --Abstract, page iii