Although recidivism is denoted by a return to crime, criminologists may not have a valid way of measuring whether a crime
has occurred. Officially recorded criminal justice events such as arrest conviction are imperfect measures for assessing criminal
activity because many crimes are committed without detection.
Another way of assessing criminal activity is to interview study participants and ask them to recall the crimes they have
committed. Because of memory decay and other methodological issues, this, too, is an imperfect measurement. As a compromise,
recidivism is measured in different ways to see whether the different assessments correspond.
Key considerations involved with measuring recidivism are:
- How the study determines that a re-offense has occurred.
- When the offender recidivates.
- How risk is factored into the research design.
How Recidivism Rates Are Determined
Both the practitioner and the theorist are most interested in whether an intervention or sanction has an effect on criminality.
Determining the rate of recidivism is one way to measure this effect.
Recidivism cannot accurately be measured just by using arrest data because not all crime is discovered. It can, however, be
measured in different ways, such as:
- Interviewing offenders to determine whether they have committed crimes since entering or exiting a program or sanction.
- Analyzing officially recorded criminal justice events such as arrests, convictions, supervision violations and commitments
to jail or prison.
- Charting a new offense over an elapsed time frame (e.g., Has the person been arrested since entering community-based drug
treatment? Has the person been arrested within three years of his or her release from prison?).
- Measuring time elapsed until the next crime (e.g., number of days passed until someone was rearrested after his or her release
The last measure is the most sensitive; however, statistical methods to measure time to an event are more complicated. When an offender recidivates is key for measurement, as discussed below.
For example, one might compare the difference between the effects of different drug treatments on post-program relapse and
criminality. Comparing recidivism rates across programs or jurisdictions, however, can be quite difficult because the programs
are likely to have significant differences in measurement definitions, between the types of supervision offered, and among
the offenders being studied.
Timing Is Key to Measuring Recidivism
The timing of recidivism is key not only to its measurement but also to understanding the processes underlying the effects
of sanctions and interventions with respect to the propensity of the individual to commit crime. Recidivism is delineated
by starting and stopping events. The starting event can be the entry into a program or the release from prison. Other criminal
justice events such as starting probation or the beginning of parole also qualify as starting events. The stopping event is
typically a criminal justice action such as an arrest or revocation of supervision.
Recidivism refers to both the type of stopping event (such as the arrest) and the amount of time between the starting and stopping criminal justice events (such as between entering a program and re-arrest). Sometimes researchers
report only statistics on the stopping event, such as the percentage of people arrested. Other times, researchers report the
average amount of time from starting to stopping event(s).
How Risk Affects Recidivism Research
The "at-risk environment" must be considered when measuring recidivism. The level of risk for someone released from prison
may depend on the level of post-release supervision. For example, it may depend on whether drug testing is conducted. Studies
have shown that the higher the at-risk environment, the more likely someone will recidivate.
One of the many difficulties with measuring recidivism is that analysts tend to assume that the risk environment of re-arrest
is the same for everyone who is being studied. This factor is important for practitioners and criminologists to be able to
study programmatic differences in an intervention. If someone released from one program is put in a "riskier" environment
than someone from a different intervention or a control group, any observed differences in re-arrest or recidivism rates may
be due to the risk environment rather than the intervention. For example, one can think of a program that releases offenders
to higher levels of supervision than a contrasting program. Observed differences in recidivism may be due to the different
levels of supervision rather than the effect of the program.
These types of problems are inherent in the study of most social indicators. Although the problems are formidable, techniques
have been developed that allow analysts to make credible comparisons. 
 See the seminal work, Recidivism, by M.D. Maltz, 1984, NCJ 146886.
Date Created: February 20, 2008