The effectiveness of service work: An analysis of recidivism

The Israeli "service work" law of 1987 enables a court to commute prison sentences of up to 6 months to service work in the community. This paper examines the correctional effectiveness of this new sanction by comparing the rate of recidivism (over a period of 14 months) among 407 offenders sentenced to service work to that of 950 comparable offenders sentenced to imprisonment. As the research design is quasi-experimental, an adjustment for confounders is carried out using the propensity score (PS) methodology. The estimation of the odds ratio of recidivism with respect to sanction comprises two steps: (a) the PS, which is the conditional probability of assignment to a particular sanction given a set of confounders, is estimated by a logistic model; and (b) the conditional probability of recidivism, given the PS and other covariates, is estimated by a second model. The findings indicate that before an adjustment for the systematic differences between the two sanctions was carried out, the odds for recidivism among prisoners were 2.4 times higher than the odds for service workers. After the adjustment, the odds ratio was reduced to 1.7. This estimate indicates that the service work sanction has a considerable correctional effect. The need to address additional criteria for the effectiveness of service work (e.g., net-widening) is emphasized.