There is a natural tendency to believe that the offenders who have committed the most serious crimes are the most dangerous. We assume that the commission of a heinous crime reveals intrinsic character flaws that will inevitably manifest themselves in future offenses. But is this true? Empirical research casts doubt on the assumption. For instance, the well-known study by the Bureau of Justice Statistics on recidivism among prisoners released in 1994 found lower rearrest rates among violent offenders than property offenders (61.7 percent versus 73.8 percent), with homicide offenders having the lowest rearrest rates among all of the categories studied (40.7 percent).
A new study of parole violations by Ryken Grattet, Jeffrey Lin, and Joan Petersilia reaches a similar conclusion. Their dataset included the records of more than 250,000 parolees in California in 2003 and 2004. Here is what they found regarding the effect of offense category:
The commitment offense variables indicate that parolees who had last been incarcerated for property offenses pose the greatest risk to violate, followed by parolees committed for drug (the omitted category), violent, and sexual offenses. Parolees committed to prison for violent offenses have a 19.1 percent lower hazard of violation than drug offenders, and parolees committed for sexual offenses have a one-third lower risk than drug offenders. Individuals with greater numbers of prior violent convictions also have a lower hazard of violation. For each additional violent conviction, a parolee has a 2.0 percent lower hazard of violation. The number of serious convictions also lowers the hazard of violation by 3.4 percent per prior serious offense. . . .
[P]olicy makers and the public, who often assume that the seriousness of a parolee’s past behavior is positively correlated with risk, might be surprised to learn that markers of the seriousness of the offender’s criminal history actually lower the risk of violation. In other words, the type of crime a parolee has been convicted of is indeed predictive of future bad behavior; however, it is drug and property—so-called low-level offenders—that pose heightened risks of violations. (385-87)
What is new and particularly helpful about this study is that it attempted to hold supervision intensity constant.
It might be argued that more serious offenders recidivate at lower rates because they are more effectively deterred by more intense supervision. But what Grattet et al. actually found was that the more serious offenders remain relatively less risky even when adjusting for supervision intensity. Indeed, factoring intensity in actually made the more serious offenders appear even less risky. The data demonstrated that, as a general matter, more intense supervision results in more, not fewer, findings of violations — when it comes to parole violations, the rule seems to be the harder you look, the more you will find. In California, the parolees with more serious convictions are as a matter of course given the most intense supervision. Taking this into account, it becomes all the more impressive that those convicted of violent and sexual offenses are tagged with violations less frequently than the lower-level offenders.
To be sure, there is something incomplete in this analysis, which treats riskiness as purely a matter of likelihood of violations. As a matter of efficient risk-management, however, it is important to distinguish among different types of violations, which might range from minor technicalities to serious new instances of criminal offending. For instance, we might regard the “low-risk” violent offenders as riskier than the “high-risk” property offenders if the violent offenders’ violations are much more frequently violent than those of the property offenders.
As far as I can tell, the new study does not address this issue. The earlier BJS study does provide some support for the intuition that it matters more when violent offenders recidivate than when property offenders do so, although the differences may not be as great as you would think. Of the violent offenders who were rearrested within three years of release from prison, 27.5 percent were rearrested for a new violent offense, which is not a lot greater than the 21.9 percent of rearrested property offenders who were picked up for a violent offense. Clearly, there is no strict divide between the ex-cons who are violent offenders versus those who are property offenders, and there remains considerable doubt (from a risk-management perspective) whether those convicted of the most serious crimes should categorically receive the most intense parole supervision.
Here are some other interesting tidbits from the new study.
First, those whose criminal careers start the earliest are not necessarily the biggest risks:
Age at first commitment to the California prison system, which we included as a measure of an offender’s onset of criminality, has a positive effect, indicating that each additional year increases the hazard of violation by 1.7 percent. This runs counter to conventional wisdom about onset—that offenders who begin their institutional careers earlier are more prone to deviant behavior as they age. (387)
Second, race matters:
Controlling for their offense history, Black parolees have an 11.1 percent higher risk of violation than parolees of other races. Whether this disparity results from Black parolees being more likely to exhibit unmeasured risk factors, such as substance abuse, employability deficits or criminogenic cognitive orientations, or from bias by parole agents, is not knowable with these data. (387)
Finally, who your parole agent is also matters:
Consistent with our hypotheses, controlling for offense history, personal characteristics, supervision intensity, and supervision capacity, parolees with Black agents and agents with more than 3 years of experience all have lower risks of violations. Although significant, the effects of parole agent race and job tenure are not large. Parolees with Black agents have a 3.6 percent lower hazard of violation. Parolees supervised by agents with less than 3 years experience have a 7.4 percent higher risk of incurring a violation compared with parolees with agents who have between 3 and 10 years of tenure, and a 4.3 percent higher risk than parolees with agents that have more than 10 years on the job. However, contrary to expectations, we do not find that the gender of the parole agent or having a parole agent with prior prison employment experience increases the risk of violation. (390)
The citation for this study is “Supervision Regimes, Risk, and Official Reactions to Parolee Deviance,” 49 Criminology 371 (2011).