Swift, certain, and fair (SCF) sanction programs typically target probationers and parolees whose substance use is viewed as a driver of their criminal behavior. For these individuals, sobriety is often a condition of community supervision. However, as noted above, detection and punishment, p, for violating these conditions can be inconsistent in practice. Often, offenders fail drug tests but are not punished consistently, and when they are (occasionally) punished, the penalty, s, is severe (e.g. revocation of parole). SCF programs offer a new model, focused on swift, certain, and fair (modest) sanctions in response to substance abuse. Programs typically involve frequent, random drug tests, where a failed test is met with an immediate, short sanction (e.g., a night or two in jail). The goal is to induce behavioral change through clear expectations and consistent responses to breaking the rules–a focus on increasing p while also dramatically reducing the associated penalty, s (which results in an increase in Uc2). Proponents argue that reducing s has little effect in practice because targeted offenders typically have high discount rates. This would imply that the increase in p is the most important element of these programs, but the net effect is an empirical question. This model assumes that those who abuse drugs or alcohol still respond to changes in p in a rational way; critics of these programs point out that learning to manage addiction likely requires meaningful treatment, not a simple change in incentives. It is therefore unclear how many people would change their substance use in response to a change in p alone.
One of the first studies evaluating this model in the context of reducing substance abuse was a randomized controlled trial (RCT) of HOPE in Hawaii. Hawken and Kleiman (2009) compared individuals randomly assigned to HOPE with those who received probation as usual. Eligible probationers included individuals with a substance abuse problem. Twelve months after assignment, probationers in the treatment group had spent significantly less time incarcerated than those in the control group. The success of this program prompted many other jurisdictions to implement HOPE-style models. Unfortunately, results of subsequent replication studies in other jurisdictions have been mixed (Hawken and Kleiman, 2011; Grommon et al., 2013; Lattimore et al., 2016).
Kilmer et al. (2013) evaluated another SCF program in South Dakota called 24/7 Sobriety. The program requires individuals arrested for alcohol-related offenses to take a breathalyzer test twice per day or wear an alcohol-monitoring bracelet that continuously checks whether the person has been drinking. This dramatically increases p. If someone tests positive for alcohol consumption, they receive swift, certain, and modest sanctions. This program was gradually phased in across counties in South Dakota, allowing a difference-in-differences analysis. Trends in places that adopted 24/7 Sobriety were compared with trends in places that had not yet adopted the program. The researchers found that adoption of the program caused a 12% reduction in repeat DUI arrests and a 9% reduction in domestic violence arrests. Both effects were statistically significant. A follow-up study found that 24/7 Sobriety also caused a significant reduction in deaths (Nicosia, Kilmer and Heaton, 2016).
An RCT of various frequencies of drug testing for high-risk, young parolees in California found that being randomly assigned to more frequent testing had no significant impact on re-arrests; those assigned to high-testing groups had higher rates of violent arrests on average. An additional study considering a similar model, O’Connell, Brent and Visher (2016), found no effect on recidivism. However, the analysis controls for participants’ employment, which itself appears to be an outcome of treatment. This will bias the estimate toward zero. (Haapanen and Britton, 2002). However, implementation of the assigned frequency of drug testing was poor, so it is unclear whether participants perceived the probability of punishment as differing across the groups. A subsequent study by Kilmer (2008) aggregated the subjects into two groups: those assigned to no drug testing or to some drug testing. He also used random assignment as an instrument for whether drug tests were actually administered, to measure the TOT effect. He found that drug testing dramatically increased the likelihood of being employed or in school during the first 30 days of parole. Effects on recidivism were not measured. There was substantial heterogeneity by race: drug testing had no effect for black parolees, but had very large effects for Hispanic parolees.
All told, the literature provides strong support for the hypothesis that increasing p can encourage desistance from crime. However, many programs that increase p also change other parameters, and these could counteract any beneficial effects. In addition, changing p is likely to be more effective for some groups than others, and we do not yet understand heterogeneity by offender or crime type.