How Much Risk, and Risk of What? A Closer Look at Pretrial Rearrest and Risk Assessment

Abstract

Many studies of pretrial rearrest, including validations of risk assessment instruments, lump multiple types and severities of crimes together. Using a dataset of over 15,000 felony defendants who were released pretrial over a four-year period in New Mexico, we measure rearrest rates for specific types and severity of crime, and compare these with the risk scores provided by the widely-used Public Safety Assessment (PSA) developed by Arnold Ventures. Our data classifies both the original charge and new charges, if any, by severity (1st through 4th degree felony, misdemeanor, and petty misdemeanor) and by type (violent, drug, property, public order, and DWI).

We find that the rates of rearrest for serious crimes during pretrial release are lower than overall rearrest rates suggest. Across all PSA score categories, about 1/3 of rearrests are for misdemeanors or petty misdemeanors. About 2/3 of rearrests are for felonies, most of which are fourth degree. Rearrest for 1st or 2nd degree felonies is very rare—less than 0.1% and 1% respectively—even among defendants whose initial charge is severe. We argue that policymakers who decide how to translate risk scores into recommended conditions of release, and judges who consider PSA scores or other risk assessments as factors in release decisions, should be provided with this richer picture of pretrial crime. We also urge that future validation studies of the PSA be carried out at this level of specificity rather than simply reporting rates of NCA (New Criminal Activity) and NVCA (New Violent Criminal Activity) in each score category.

Keywords: criminal justice, recidivism, pretrial detention, risk assessment, validation