Study 1
To examine the impact of the Washington State Earned Release Law, Drake, Barnoski, and Aos (2009) conducted a quasi-experimental design study with a matched comparison group, to examine the recidivism of individuals who were released from incarceration between July 2003 and July 2010 in Washington State. During this time, when an individual entered prison, the criminal history in the Department of Corrections (DOC) operational database was automatically searched to determine if the individual was eligible for 50 percent release time. When an individual was found to be potentially eligible, the individual’s criminal history was further investigated through a comprehensive interstate exchange of criminal history maintained by the Federal Bureau of Investigation, to ensure eligibility for early release. There were 2,297 individuals eligible for earned release time who formed the intervention group for the study.
A comparison group was selected by the study authors, who developed an algorithm to determine who would have been eligible for 50 percent release time prior to July 2003. There were 4,840 individuals who were eligible for earned release before 2003, and these individuals formed the comparison group.
The intervention group was 78 percent male. In terms of race/ethnicity, 74 percent were white, 19 percent were Black, and 19 percent were Hispanic. The comparison group was 80 percent male. In terms of race/ethnicity, 71 percent were white, 23 percent were Black, and 17 percent were Hispanic. There were significant differences between the intervention and comparison groups at baseline. The comparison group had higher risk factors associated with recidivism, including criminal history, offense seriousness, sentence length, and demographics. Additionally, the comparison group had higher risk factors, including non-drug risk score, violent-felony risk score, and Sentencing Reform Act (SRA) seriousness level. To account for these differences, the study authors created three additional comparison groups to test the effect of the earned release law: 1) risk-variable matched groups; 2) risk-variable matched groups in which the study authors and DOC agreed on eligibility; and 3) matched groups based on individual scores, offense severity level, and number of days at the midpoint of the sentencing grid. For the CrimeSolutions review, the outcomes of the overall intervention group, compared with the overall comparison group, were considered (the results of the three additional comparison groups are provided under Other Information).
Regression analyses were used to adjust for the observed differences between the intervention and comparison groups within the 3 years of the follow up.