Date of Award

Spring 5-15-2015

Author's School

Graduate School of Arts and Sciences

Author's Department

Social Work

Additional Affiliations

Brown School of Social Work

Degree Name

Doctor of Philosophy (PhD)

Degree Type



This project examines the relationship between work and crime among male former prisoners. Criminological theories and observational studies suggest that work reduces crime, but recent studies cast doubt on the ability of employment programs to reduce recidivism among former prisoners. Ongoing weak evaluations may imperil support for employment-focused rehabilitative programming. Using data from the Serious and Violent Offender Reentry Initiative (n = 1,575), this study examines whether selection bias and unobserved heterogeneity contribute to weak evaluation findings.

First, this study tests whether unobserved heterogeneity contributes to jobs programs' weak treatment effects. It uses group-based trajectory modeling and propensity score methods to balance participants and nonparticipants on demographic and criminal risk factors. Lifetime arrest data from administrative records are used to model respondents' prior offending trajectories. Baseline interview data are used to balance respondents on the propensity to receive employment-focused services. After balancing respondents, this study employs duration models to test the effects of educational and employment programming on time to rearrest.

Second, this study tests whether financial problems mediate the work-crime relationship. Longitudinal structural equation modeling is used to model men's labor force attachment, job quality, financial needs, and emotional wellbeing. Models test whether financial problems diminish the crime-reducing effects of employment for men who remain weakly attached to the labor force. Multiple indicators for each latent construct reduce bias due to measurement error.

Results of this study show that education and employment programs in United States prisons have limited effects on the likelihood that participants maintain employment and avoid criminal justice involvement. Male prisoners recruited into the Serious and Violent Offender Reentry Initiative faced multiple barriers to employment before entering prison, due to extensive criminal records, low educational attainment, and limited work experience. Before matching men on the probability of receiving employment-focused services, program participants differed from nonparticipants across an array of demographic and risk factors. The group-based trajectory model derived three latent trajectory groups from the sample that exhibited distinctive demographic characteristics and pre-prison offending trajectories. Due to significant variation at the state-level, a multilevel logit model was used to model the probability of receiving education and employment services. Nearest neighbor matching with caliper resulted in a sample that exhibited balance across multiple demographic, criminal record, employment, and health measures.

After matching, employment program participants were slightly more likely than education participants and nonparticipants to maintain stable employment, and employment program participants exhibited lower rates of rearrest during the first 9 months after release. After that point, there were no significant differences between employment-focused program participants and nonparticipants in labor force and criminal activity.

The longitudinal structural equation model results show that criminal activity has cascading effects on financial and emotional wellbeing, subsequent labor force activity, and ongoing criminal justice involvement. Engagement in crime during the early months of release reduced labor force participation, limited men's ability to obtain higher-quality employment, and increased their financial needs and feelings of psychological distress. In contrast, stable employment led to improved job quality and reduced financial needs over time. Employment did not reduce men's later involvement in criminal activity, however. In fact, employment during the first 9 months of release was associated with increased odds of reporting committing new crimes during the subsequent 6-month period. Overall, the path model results provide no evidence to suggest that stable employment reduces criminal activity among serious and violent former prisoners.

The results of this study cast doubt on theories of crime that presuppose causal associations between work and crime. Observational studies that show associations between stable labor force participation and desistance from crime may be capturing maturation effects that simultaneously directed individuals toward legal work and away from crime. If desistance from crime actually precedes stable labor force attachment for most former prisoners, this may explain the weak empirical evidence for prison-based employment programs. The findings may inform modifications to employment and transitional jobs programs to identify participants on the path to desistance who may be most responsive to these services.


English (en)

Chair and Committee

Carrie Pettus-Davis

Committee Members

Derek Brown, Shanta Pandey, Juan Pantano


Permanent URL:

Included in

Social Work Commons