ticipants who graduate from the program with drug court participants who drop out. Comparisons between drug court graduates and drug court dropouts introduce self-selection bias, as graduates may fare better than dropouts because of their own commitment to abstaining from drugs rather than program effects. The more appropriate way to measure effectiveness is to compare all drug court participants—graduates and dropouts—to a control group that did not participate in a drug court program (Belenko, 1998; U.S. General Accounting Office, 1997).
Even in the handful of studies that make appropriate comparisons between all drug court participants (dropouts and graduates) and a control group, no mention is typically made of what percentage of those eligible for these programs were willing to participate. Yet if very few persons brought before the courts are willing to volunteer to participate in these programs, they will ultimately do little to relieve overburdened courts. Furthermore, since the willingness to volunteer for a drug court program may vary from court to court, depending on what other options are available with standard adjudication, it may be difficult to generalize any findings on the willingness to volunteer from one court to another.
In addition, very few studies make use of experimental designs in which drug court participants are compared with a no-treatment control group. However, in one study of the Maricopa County drug court, study subjects were randomly assigned to either a drug court or traditional probation and followed over a 3-year period (Turner et al. 1999). In the 3-year follow-up, drug court participants were less likely to be rearrested than those on traditional probation. Indeed, the Maricopa County study compared drug court participants with those on traditional probation in terms of a number of outcome measures, including rearrests for drug crimes and non-drug crimes as well as for convictions or reincarceration in jail or prison during the 3-year follow-up. As noted previously, it is doubtful that outcome measures such as rearrests, convictions and reincarcerations can provide accurate counts of drug use levels or criminal behavior in a follow-up period. After all, many of the subjects in this study could have easily managed some episodes of drug use or criminal behavior in the follow-up period without being rearrested or reincarcerated. Hence it is difficult to know what to make of the drug court participants’ lower overall rearrest rates for all crimes.
Not only did the Maricopa County study rely on some of the same questionable outcome measures used in other evaluation studies, but it likewise confined its follow-up period to the 3 years that the subjects were subjected to criminal justice supervision (Turner et al., 1999). Yet follow-ups that occur after criminal justice supervision is over are important because they make it possible to determine if drug court participants are