Abstract
This study, using data from a large sample of prison releasees, examined the similarities and differences in men's and women's risk factors for recidivism involving rearrest for any crime and rearrest for a violent crime during an average 3.4-year follow-up period. Logistic regressions revealed several gender differences. Prior incarceration, time served, and specific types of prior arrest histories had differential associations with women's and men's overall and violent recidivism. Age, race, and conviction offense were gender-specific risk factors for overall recidivism, and education level and marital status were gender-specific risk factors for violent recidivism. Implications for risk assessment are discussed.
Notes
Across most prison systems in the United States, the measure of recidivism usually used is return to prison, as this is information readily available to prison systems. However, whether or not someone is returned to prison usually represents the culmination of decisions made by multiple practitioners in the criminal justice system, including parole agents, the police, prosecutors, and judges, depending on the reason for the return to prison (i.e., a technical violation vs. a new sentence).
Arrests for possession of marijuana were excluded from the prior drug possession arrests because almost all marijuana possession offenses were classified under Illinois law as misdemeanor offenses and because of shifts in the degree to which local law enforcement agencies formally arrest those found in possession of marijuana. Possession of all controlled substances other than marijuana is classified as felony offenses and thus is considered much more serious. The measure of drug sale offenses included marijuana and other drugs.
In cases in which an offender was sentenced to prison for multiple offenses, the most serious offense was used to identify the current offense. The most serious offense was identified as that which carried the longest prison sentence.
Logistic regression models were run that excluded and included missing cases. The models presented here are those with the missing cases included. Some differences in results emerged between those models run with and without the missing cases and primarily had an impact on the influence of current offense felony class, months served in prison, and prior drug sale arrests in the general recidivism models and on the influence of disciplinary incidents in the violent offense recidivism models.