Q and A With Howard Snyder: June 2010

Main Story


Hands on prison bars

Howard Snyder is chief of Recidivism, Reentry and Special Projects for the Bureau of Justice Statistics. Donna Lyons, who heads NCSL’s Criminal Justice Program, talked with Snyder for State Legislatures magazine about the challenges and significance of recidivism data in sentencing and corrections policy making.

State Legislatures: We’ve known for a long time that up to half of all prisoners released return to prison within a few years. Why the new attention to recidivism numbers?

Howard Snyder: There are two driving forces. The first is the high cost of incarceration. State budgets are straining under its burden and state legislators are seeking ways to reduce the costs without sacrificing public safety. The second driving force is the general belief that we can do a better job rehabilitating prisoners. In the past many believed that “nothing works” and the best we could do to protect the public was to lock offenders away for a long time. In response, the courts sentenced many offenders to prison and legislators passed truth-in-sentencing and 3-strikes-and-your-out laws to keep them there. Prison population grew to historic levels and state budgets absorbed the costs. However, State budgets are reaching (or have
reached) their breaking points and many believe there are alternatives to incarceration for some would-be prisoners that can both save money and protect the public. If we are to reduce prison populations, we must have a better understanding of which offenders are most likely to reoffend or violate their parole conditions. We must understand which intervention strategies used while in prison or after release are most effective in protecting the public. The measure used to quantify reoffending and program success is recidivism and that is why the public and the decision makers need (and want) to understand recidivism patterns.

SL: While recidivism data are desirable, why are reliable, comparable data so hard to come by?

Snyder: This question can be answered at many levels. Recidivism is generally measured using official records. Official records are not the best measure of recidivism because a large portion of crime never comes to the attention of the criminal justice system. However in the real world, official records are the best information on reoffending available to us. So let’s consider the reliably of official records, specifically rap sheets. Rap sheets summarize
information housed in state criminal history repositories and are the most common source of recidivism information. Rap sheets are reliable when they contain a complete list of the offender’s interactions with the criminal justice system. The rap sheets are unreliable when pieces of information are missing. Sadly, this is often the case, especially with older rap sheet information. Most rap sheets generally are very reliable in reflecting recent arrests
for felonies and serious misdemeanors, because most of these incidents result in the offender being fingerprinted and the fingerprints along with the arrest information being sent to the criminal history repository. The same is not true, for example, with the courts’ responses to these arrests. To be included on the rap sheet, the court’s disposition of the case must be sent to the repository and linked through fingerprints to the offender’s arrest
record. The completeness of the rap sheet’s court (and corrections) information vary from jurisdiction to jurisdiction making the rap sheet an unreliable source of recidivism information, especially when recidivism to defined as a new court conviction or a new imprisonment.

Now let’s talk briefly about the issue of comparable recidivism data. The simple fact is there is no single definition of recidivism. All definitions have three elements. The first is the starting point; this could be, for example, an arrest, a court conviction, or a release from prison. The second element is the failure event. This could be a new arrest, a new conviction, or a new imprisonment. The third element is the time period after the starting event
that you observe offenders to see if they recidivated. So let’s assume we are studying the recidivism of released prisoners. One state could define recidivism as a new conviction within one year of release, while another state defines its recidivism measure as a new arrest within three years of release. If their prisoners had exactly the same behavior after release, the second state would have a much higher recidivism rate than the first due
simply to the difference in how they measured recidivism. This is why, before attempting to compare recidivism rates, you must first determine how recidivism is defined and not compare apples and oranges. Low recidivism rates often come from a measure linked to a short observation period or on a failure event tied to deeper penetration into the criminal justice system, e.g., new imprisonment compared with new arrest.

SL: What can state policy makers do to facilitate more complete and usable recidivism data in their states?

Snyder: First, to improve the completeness of the data in their state, policy makers can work to ensure that their state criminal history repositories contain information on every arrest and then every prosecutor action, court disposition and correctional placement linked to that arrest. If recidivism is measured in the state by a new conviction and the court information is missing, the offender will be mistakenly classified as not recidivating. As I
mentioned earlier, the arrest information is generally well reported; but, as the case moves deeper into the criminal justice system, the completeness of reporting declines. Policy makers should support efforts to insure that their courts and correctional agencies consistently report their information to state repositories. Second, when policymakers asked for recidivism information, they should be clear about the underlying definitions of recidivism
they would prefer be used. If policymakers are funding new programs, they encourage the use of common definitions of recidivism so that comparisons can be made.

SL: You are heading up an effort at the Bureau of Justice Statistics to provide information leadership in areas of recidivism and reentry. Tell me about what you bring to this position what you expect it to contribute to the criminal justice field.

Snyder: As I mentioned, recidivism is generally measured using data housed in criminal history repositories or other administrative data sets. To a researcher, these data sets are gold mines of information. But, like gold mines, their raw contents are difficult to extract and process. I have spent much of my career harvesting data from administrative information systems and transforming these data into useful research databases. At BJS we are
working to extract the criminal history information on individuals efficiently and effectively from state repositories. I have done such work successfully with juvenile court management information systems in the past and hope to establish a system to do the same with the data stored in criminal history repositories. In addition, I have spent my career straddling the fence, with one foot in the research world and the other in the day-to-day world of the
juvenile and criminal justice systems. I know the needs of researchers, practitioners, and policymakers. As a result, the recidivism and reentry data infrastructures we are building at BJS--along with the information and reports that will flow from them--are designed to serve this range of information needs.

Opinions and conclusions provided here are those of Snyder and do not necessarily reflect any official position or policies of the U.S. Department of Justice.