Evidence shows that non-custodial measures may be more effective to reduce recidivism compared to incarceration [1].
An examination of two datasets with 222 comparisons of offender groups (n = 68,248) indicated that offenders who were incarcerated longer had a slight 3% increase in recidivism compared to offenders who were incarcerated for a shorter period. Furthermore, comparing an additional sample of 267,804 incarcerated individuals versus others who received a non-custodial measure in the community identified that those who had been incarcerated showed a 7% increase in recidivism [2].
Another study compared the effectiveness of non-custodial measures with incarceration using different methodological approaches and a population sample [3]. The results revealed that incarceration increases recidivism in the one-, two-, and three-year follow-up periods, even after controlling for different variables, such as age, gender, and race/ethnicity.
A systematic review by the Campbell Collaboration included 14 high-quality studies, including three Randomized Controlled Trials (RCTs) and two natural experiments, and found conflicting evidence on the effectiveness of non-custodial sanctions.
The review included two meta-analyses. The first examined only RCTs and natural experiments, and found no significant differences between custodial and non-custodial sanctions in terms of their impact on recidivism. Differences in recidivism are modest at best, although slightly in favor of non-custodial sanctions. The second meta-analysis included only quasi-experimental studies, and its results showed a significant “criminogenic” effect of custodial sentences.
Overall, the meta-analyses lead to contradictory conclusions, namely: no effect if the analysis is limited to RCTs and natural experiments, and a statistically significant effect in favor of non-custodial sanctions if the analysis is conducted with quasi-experimental studies. The authors speculate that this difference may be explained by the fact that RCTs offer better conditions for controlling for potentially influential variables, such as offenders’ personal characteristics, which may affect sentencing decisions as well as the risk of recidivism. Nevertheless, the authors conclude that, although several studies have reached the level required for inclusion in the systematic review, further studies of high methodological quality are still needed.