The blog of the Urban Institute
October 17, 2014

There's more than one way to measure recidivism

October 17, 2014

As long as Californians are footing the bill for public safety, it is critical that state corrections leaders track the success of public safety programs in the most precise and accurate way.

The state is currently engaged in an important conversation about how it measures rates of reoffending for people who have been in the criminal justice system. For better or worse, recidivism is the most commonly used performance measure of the correctional system—but it frequently lacks the precision necessary to effectively inform policy decisions.

In California, decisions about the future of realignment and the allocation of billions of state and local budget dollars will be greatly impacted by recidivism rates in the coming years. This is why, in a letter to state law enforcement officials, Attorney General Kamala Harris recently wrote that “[u]niversally defining recidivism is a fundamentally important issue if we are to be smart on crime.”

Collecting the right data, establishing metrics for success, and assessing the results are crucial steps toward understanding the impact of the massive changes in correctional policy in recent years. Currently, there are two competing proposals about what event should trigger recidivism: arrest or reconviction. Harris, who has proposed using arrest as the statewide indicator, hopes to find compromise with the Board of State and Community Corrections, which has chosen reconviction.

I have an alternate proposal. They’re both right and California should not only choose both, but add a few others as well.

The most important step that California can take is to move beyond the traditional idea of a single statewide rate of success or failure. There is no “right” measure of recidivism, and California should adopt multiple measures that reflect different outcomes. Recidivism reduction is the responsibility of multiple agencies at the state, county, and city level, and reoffending measures should reflect that complexity. Different individuals receive different interventions, from state prison sentences to local supervision in the community. No single measure can possibly capture this complexity and yield meaningful information.

So, yes, California should track rearrest, reconviction, and return to custody for new crimes—they all measure different points in the criminal justice system. But that is just the beginning. California should take this opportunity to consider capturing additional measures of success. These include:

Desistance: How many people released from prison never return?

New research indicates this number is as high as 6 in 10—this other side of the recidivism coin raises different but no less important policy questions. This reframing results in very different policy questions about the prison experience, and shifting focus to those who desist demands we confront the question of length of stay. If someone never returns to prison, did the imprisonment “work” or would s/he have stopped committing crimes even in the absence of imprisonment? Was the duration of their imprisonment optimal or could they have served a shorter term?

Time to failure: How long do people remain crime free?

Keeping people out of prison for good is the ideal outcome, but not the only measure of correctional success. Increasing the duration between criminal offending is a positive outcome.

Severity: Are the crimes of those who reoffend less or more severe than prior crimes?

Again, while the ultimate goal is desistance, preventing more serious crimes is a positive public safety impact that should be measured.

Behavior change: What interventions are most effective at reducing recidivism?

The population of all people released from prison is so large and diverse that it is unwise to draw much in the way of conclusions. Recidivism of subpopulations—people who received treatment; people who were impacted by a policy change—is likely to elicit more informative results.

The first step to improve recidivism measurement is to make sure you are correctly defining what you intend to measure. And in this case, more is better. A single measure is insufficient to draw meaningful conclusions about the performance of an enormous criminal justice system that touches a diverse population exposed to a variety of interventions. Failing to make these improvements can result in recidivism data that do not convey a true picture of reoffending.

So, rather than worry about reconciling their two definitions, the Attorney General and Board of State and Community Corrections should start with both of their definitions and build from there.

Photo of Pacer County, California Courthouse by Flickr user Charles Nadeau (CC BY 2.0)

Follow Ryan King on Twitter.


As an organization, the Urban Institute does not take positions on issues. Experts are independent and empowered to share their evidence-based views and recommendations shaped by research.


Amen! S If something isn't measured, it doesn't exist within organizations or even in politics. And sadly, as the New Jim Crow, by Michelle Alexander points out, one of the objectives of the laws is to create an underclass of African Americans whose liberty and lives are diminished by the Justice system. I recommend her book to everyone trying to understand why the prison system in the USA has gotten so out of control. Maybe it is just the way some people want it to be.