Wednesday, October 8, 2008
Yesterday, we had the pleasure of hosting Dale McNiel from UCSF, who gave a talk about his work with Renée Binder evaluating the effects of the San Francisco Mental Health Court on recidivism. The court offers various programs and close supervision; the defendants' progress is monitored and, at some point, they graduate from the program. Participation in the mental health court is voluntary. Here's a pretty positive take on the program.
McNiel and Binder's study is a great example of an evaluation project on problem-solving courts. One of the problems in measuring the success of sentencing alternatives is that you really need to compare your program to the traditional sentencing and correctional methods. Due to the fact that the mental health court population is self selected (participation in the program is voluntary), the study and control groups are, obviously, not formed through random allocation. Therefore, various pretreatment variables, which might explain why some people might choose to attend mental health court, might also explain differences in recidivism rates. To partially offset this problem, McNiel and Binder use propensity weighting; they select the control group based on criteria that would make them resemble the population that does choose to go to mental health court, as much as possible.
McNiel and Binder's findings clearly indicate a reduction in recidivism for both violent and nonviolent offenses, which grows over time. They control for a variety of demographic and clinical variables. The nuances are important; the court seems to be better for people with certain disorders, and I'm not sure I fully understand its interaction with substance abuse and homelessness, which many people in the study group experience. Also, as they point out in their discussion, this is only one study of one court; recidivism rates may differ across mental health courts, and may be a function not only of the programs themselves, but of the sanctions imposed by the court. Fascinating stuff, and I hope the followup yields more generalizable results.