Program Profile: Law Enforcement Assisted Diversion (LEAD) Program (Seattle, Washington)

Study Title: Seattle’s Law Enforcement Assisted Diversion (LEAD): Program Effects on Recidivism (which is associated with Outcome 1 and Outcome 2)

Research Design

Collins and colleagues (2017) conducted a nonrandomized, two-group longitudinal controlled evaluation of Seattle’s Law Enforcement Assisted Diversion (LEAD) program. The study took place in Seattle, Washington—a large urban city—between October 2011 and July 2014. Adults suspected of low-level drug and prostitution offenses were eligible for inclusion in the study. Adults were ineligible for participation if they possessed more than 3 grams of drugs, were suspected of dealing drugs for profit or dealing above a subsistence income, were not amenable to diversion, exploited minors or others in dealing, promoted prostitution, or had serious criminal histories (e.g., murder, arson, robbery 1, assault 1, kidnapping, firearm or sex offenses). Additional exclusions applied to those with certain convictions within the past 10 years (domestic violence, robbery 2, assault 2 or 3, burglary 1 or 2, firearm violations) or those already in King County Drug Diversion or Mental Health Court.

Participants entered the LEAD treatment group or the system-as-usual control group through two methods. They  were 1) arrested during designated officer shifts, or 2) referred as social contacts based on known involvement in recent criminal activity. First, police officer shifts, not individual arrestees, were randomly assigned as LEAD or control shifts. Eligible people arrested during those shifts were placed into the corresponding group. A second pathway allowed “social contacts” (people suspected of recent drug or prostitution activity who were encountered in the geographic zone from which participants are drawn original catchment area) to enter LEAD without arrest. Because the original catchment area had few eligible individuals over time, most individuals were approached about participating in LEAD. As a result, an insufficient number of individuals were available for inclusion in the control group. To meet the need for an adequately sized, comparable control group, control areas (in addition to control shifts) were added to the evaluation. The control areas included neighborhoods patrolled by the same squads, using identical inclusion criteria. 

Sample

A total of 318 participants were included in the study: 203 individuals were assigned to the LEAD treatment group (146 as arrestees, 57 as social contacts), and 115 individuals were assigned to the control group. The control group received standard criminal justice processing, including arrest, prosecution, and potential incarceration, without access to LEAD services.

Of the total study sample, 34 percent were female. As indicated by police records, 60 percent identified as African American, 26 percent as European American, 4 percent as American Indian/Alaska Native, 4 percent as Multiracial, 3 percent as Hispanic/Latino, 2 percent as Asian American, and 1 percent as Other. The mean age was 40.2 years. Notably, 82 percent of participants had experienced homelessness.

The study authors used generalized boosted regression modeling to calculate propensity score weights, controlling for differences in age, biological sex, race/ethnicity, and prior arrests. 

Outcome Measures/Data Collection

The outcome of interest was recidivism, measured as arrests and felony charges. Arrests were defined as instances in which individuals were taken into police custody for crimes committed during the program evaluation period. Arrests related to new offenses, and they did not include parole or probation violations or failure-to-comply offenses pursuant to prior violations. Felony charges included criminal charges for felonies that occurred during the evaluation timeframe. Data were collected from Seattle Police Department records, LEAD case management logs, the FBI’s National Crime Information Center (NCIC) [these data were compiled by the King County Prosecuting Attorney’s Office], and participant records.

Statistical Analysis

 The primary analyses were generalized logistic estimating equation models, which tested the effects of time, intervention group, and the time and intervention group interaction adjusting for repeated measures. Outcomes were assessed up to nearly 5 years for recidivism outcomes. Odds ratios were reported for all outcomes. No subgroup analyses were conducted. 

Citation:

Collins, Susan E., Heather S. Lonczak, and Seema L. Clifasefi. 2017. "Seattle’s Law Enforcement Assisted Diversion (LEAD): Program Effects on Recidivism Outcomes." Evaluation and Program Planning 64:49–56.

This study is a high-quality randomized controlled trial.