Behavioral Prediction and the Problem of Incapacitation*

Abstract

Traditional views of incapacitation as a crime control strategy rely fundamentally on behavioral prediction. Most attempts at predicting offenders' behaviors have relied on simple dichotomous dependent variables. Recent attention to components of the "criminal career," such as the rate of offending and the nature and potential patterning of offending behavior, could provide significant advantage to the prediction problem while informing debates about incapacitation strategies. W e examine this possibility in the context of a study of some 6,000 offenders followed for more than 25 years. Little predictive advantage is realized by attention to more sophisticated behavioral outcome criteria. In agreement with other studies, we observe little support for hypotheses concerning the pattern of careers with respect to offense behaviors. The utility of incapacitation as a crime control strategy seems to be constrained by the limits imposed by predictive validity and perhaps by the nature of the criminal career.

Concepts of Risk in Relation to Organizational Structure and Functioning within the Personal Social Services and Probation

“…(The impact of the policy shift upon current Probation practice is reviewed in Kemshall later in this number.) However, since the 1970s the prison population both in the United States and in Britain has increased substantially (Gottfredson and Gottfredson 1994), creating a situation in which the demand for incarceration has continually outstripped resources. The pragmatic, and to a large extent fiscally driven, response to this "penal crisis" has been the pursuit of selective incarceration and "ways to reduce the populations consistent with the public safety".…”

Sentencing Using Statistical Treatment Rules: What We Don’t Know Can Hurt Us

“…Our interpretation of a risk prediction exercise depends critically on our understanding of the goals of the actors (in this case state troopers) in the current system. Similarly, in the sentencing context, the interpretation of a risk prediction exercise like the one in Gottfredson and Gottfredson (1994) depends critically on what the researcher thinks the judges and other actors (and the sentencing rules and guidelines) seek to accomplish. Kuziemko (2006) analyzes the behavior of parole boards in Georgia and provides two pieces of evidence consistent with our main point.…”

“…For example, prior criminal history increases risk, but it also induces treatment (incarceration) that could alter behavior and our ability to observe that behavior. This clouds the meaning of variables measuring the effects of past criminal history on recidivism in exercises like that in Gottfredson and Gottfredson (1994). For the same reason, the current risk assessment tools reviewed in Andrews et al (2006) do not provide clear answers to the question they address because they fail to take account of this simultaneity.…”

“…The general consensus in the criminological literature -see e.g. Gottfredson and Gottfredson (1994) and Auerhahn (1999) -views these efforts to identify high risk offenders as largely ineffective.…”

“…They claim not that risk prediction cannot be done, but rather that the methods employed in recent analyses have so many methodological problems as to make the exercises largely uninformative. They hold up earlier work done in criminology, such as the oft-cited paper by Gottfredson and Gottfredson (1994) that uses 25 years of follow-up data on California prisoners, as exemplifying the proper way to create and validate these risk prediction scales.…”