Study 1
Collins, Lonczak, and Clifasefi (2015) conducted a quasi-experimental study to examine the shorter- and longer-term recidivism outcomes for participants who were involved in the Law Enforcement Assisted Diversion (LEAD) program, as compared with those in the “system-as-usual” control group.
For the evaluation, Seattle Police Department (SPD) officer shifts for squads making referrals to LEAD were randomly divided into red- and greenlight shifts. The LEAD group (n = 203) included individuals who were encountered by SPD officers during greenlight shifts in the LEAD catchment area (the Belltown neighborhood in Seattle). They were screened for eligibility and diverted to the LEAD program (if they were eligible and completed the intake process). The LEAD group also included social contacts, which were individuals eligible for the program due to known recent criminal activity, but were recruited by officers outside of a criminal incident during a greenlight shift within the catchment area. The control group (n = 115) included individuals who were arrested during redlight shifts, or in non-LEAD neighborhoods (i.e., areas adjacent to Belltown that were not part of the LEAD program, but were patrolled by the same officers). Control group participants were processed through the criminal justice system as usual (i.e., jail booking, criminal charges).
All participants were recruited by the same officers using the same criteria. Propensity score weighting was used to balance and match the LEAD group and control group. The study authors used two weighting variables: one for estimating average treatment effect (ATE) and one for estimating the average treatment effect for treated participants (ATT). The CrimeSolutions review of this study focused on the analysis related to the ATT.
The LEAD group was 39 percent female, with an average age of 41. The race/ethnicity of LEAD participants included African American/Black (55 percent), white (27 percent), American Indian/Alaska Native/Pacific Islander (6 percent), Hispanic/Latino (5 percent), Asian American (less than 1 percent), more than one race (4 percent), and other (1 percent). The control group was 26 percent female, with an average age of 37. The race/ethnicity of participants included African American/Black (68 percent), white (25 percent), Hispanic/Latino (1 percent), Asian American (3 percent), and more than one race (3 percent). The groups differed on age, race/ethnicity and gender. Therefore, propensity score weighting was used including these covariates.
The primary outcome of interest was recidivism, measured as new arrests. Data was collected from the King County Prosecuting Attorney’s office, from the FBI’s National Crime Information Center. New arrests referred to having been taken into police custody for a crime committed during the LEAD program evaluation timeframe. New arrests did not include parole or probation violations, or arrests due to violations prior to entering the program. Arrests for new law violations were examined separately from warrant arrests pursuant to incidents occurring after participants entered the study.
The evaluation examined the impact of the LEAD program on recidivism outcomes over the shorter term (i.e., 6 months prior and 6 months subsequent to program involvement) and the longer term (i.e., encompassing 2 years prior to the LEAD start date–October 1, 2009 through July 31, 2014). The longer-term analyses involved unequal windows of time for participants that started and ended the program at different times. This was statistically controlled for in the longer-term models. The data was analyzed using generalized estimating equations.