Cover: Does Parolee Drug Testing Influence Employment and Education Outcomes?

Does Parolee Drug Testing Influence Employment and Education Outcomes?

Evidence from a Randomized Experiment with Noncompliance

Published In: Journal of quantitative Criminology, v. 24, no. 1, Mar. 2008, p. 93-123

by Beau Kilmer

Abstract

Despite the ubiquity of drug testing in criminal justice settings, there is little experimental evidence suggesting that testing reduces drug use or engenders pro-social behavior. This paper estimates the effect of parolee drug testing on labor and education outcomes with data from a randomized experiment involving 1,958 young parolees. It provides the first estimates in the literature suggesting that drug testing with graduated sanctions can improve short-run employment and education outcomes for parolees. After controlling for parole office fixed effects, juvenile criminal history, and a host of other covariates, the analyses suggest that parolees randomly assigned to testing are 6-8 percentage points (~11%) less likely to be unemployed and not in school for the month following release to parole when compared to those assigned to the no-testing condition. Racially- and ethnically-stratified analyses find that Hispanics assigned to testing are 10-13 percentage points (~22%) less likely to be unemployed and not in school, while the estimated coefficient for Blacks is statistically insignificant and hovers around zero. Analyses that use instrumental variable techniques to account for noncompliance by parole officers yield local average treatment effects that are almost twice as large as the intention-to-treat effects.

Research conducted by

This report is part of the RAND Corporation external publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.