When the numbers lie, policy suffers
In 1999, Washington State passed the Offender Accountability Act (OAA) to make changes to how the state classifies and supervises individuals released from prison. The Washington State Institute for Public Policy (WSIPP) was charged with assessing whether the policy was having its intended effect of reducing recidivism by targeting supervision resources to those most likely to reoffend.
A 2010 report by WSIPP found that reconviction rates had increased prior to the reform and continued to rise after the passage of the 1999 reform. It would seem that the OAA failed in its goal to reduce recidivism. Or did it?
Isolating the effect of policy
Recidivism is traditionally measured by the rates of arrest, reconviction, or return to custody of all people released from prison in a year. Naturally, the characteristics of those released from prison will vary from year to year due to sentencing and release policies. Differences in characteristics like offense type, admission type (new crime versus technical violation of supervision), length of stay, and age at first arrest matter because they are linked to risk of reoffending.
But without accounting for these differences, it’s impossible to know whether recidivism rates are affected by interventions such as the OAA—or if they’re simply due to the characteristics of those who were released from prison that year.
Fortunately, there are ways for analysts to control for the underlying characteristics of the recently released and better isolate the impact of policy interventions.
What happened when WSIPP adjusted for the characteristics of the released population? The story changed completely. WSIPP concluded that “actual recidivism rates are lower today than they would have been without the policy (and other) changes.”
It was the “increasing underlying risk of the offender population” that had driven the increase in recidivism during the study period. In fact, the OAA had been meeting its objective.
How to improve recidivism data
Washington was actually experiencing a decline in rates of reoffending, yet common measures of recidivism were telling the opposite story. What would have happened had WSIPP not conducted the analysis that controlled for this risk? Would policymakers have chosen to explore a different strategy? Maybe change how they managed supervision in their state?
The Washington example is significant because it illustrates the importance of carefully designing analyses that focus on the impact of a specific policy or program. Unfortunately, this example is the exception, not the rule.
Recidivism is most frequently reported as a single statewide rate, which is too imprecise to draw meaningful conclusions and insufficient for assessing the impact of changes to policy and practice. But too often, that is all policymakers and practitioners have to base important decisions about where to direct precious state resources.
Our new brief outlines four steps that states can take to improve how they measure, collect, analyze, and disseminate recidivism data:
- States need to define recidivism using multiple measures of success, not a single state-level rate.
- States should develop protocols that ensure data are accurate, timely, and consistently collected.
- Any analysis must account for changes in the underlying composition of the population.
- Recidivism data and analysis must be packaged for a policymaker audience to impact decisionmaking.
The brief offers a blueprint for improving recidivism data, a critical first step to understanding the effectiveness of sentencing and corrections interventions. When the numbers tell a misleading story, as illustrated above, policy could suffer.
For more information on measuring recidivism, join us for our October 15 webinar.
Photo from Shutterstock