Atlantic, The Brain on Trial:

In the past, researchers have asked psychiatrists and parole-board members how likely specific sex offenders were to relapse when let out of prison. Both groups had experience with sex offenders, so predicting who was going straight and who was coming back seemed simple. But surprisingly, the expert guesses showed almost no correlation with the actual outcomes. The psychiatrists and parole-board members had only slightly better predictive accuracy than coin-flippers. This astounded the legal community.

So researchers tried a more actuarial approach. They set about recording dozens of characteristics of some 23,000 released sex offenders: whether the offender had unstable employment, had been sexually abused as a child, was addicted to drugs, showed remorse, had deviant sexual interests, and so on. Researchers then tracked the offenders for an average of five years after release to see who wound up back in prison. At the end of the study, they computed which factors best explained the reoffense rates, and from these and later data they were able to build actuarial tables to be used in sentencing.

Which factors mattered? Take, for instance, low remorse, denial of the crime, and sexual abuse as a child. You might guess that these factors would correlate with sex offenders’ recidivism. But you would be wrong: those factors offer no predictive power. How about antisocial personality disorder and failure to complete treatment? These offer somewhat more predictive power. But among the strongest predictors of recidivism are prior sexual offenses and sexual interest in children. When you compare the predictive power of the actuarial approach with that of the parole boards and psychiatrists, there is no contest: numbers beat intuition. In courtrooms across the nation, these actuarial tests are now used in presentencing to modulate the length of prison terms.