Skip to content

Evaluation and Validation of the PSA in San Francisco

San Francisco adopted a pretrial risk assessment tool to help the county reduce the use of cash bail and to address racial and ethnic disparities in release decisions.

Background

San Francisco began implementing the Public Safety Assessment (PSA) in May 2016 to inform pretrial release decisions and supervision levels. The PSA is a static risk tool that uses criminal history, current offense, and age to generate estimates of the likelihood that a person will fail to appear, be arrested for a new offense, or be arrested for a new violent offense. 

Under SB 36 (2019), jurisdictions utilizing a pretrial risk tool are required to validate the accuracy and reliability of the tool in (a) predicting failure to appear and commission of new criminal activity while on pretrial release and (b) assessing any disparate effect or bias of the tool based on sex, race, or ethnicity.

Research Project

Validation: CPL will assess overall validity of the PSA in San Francisco. To be considered “valid”, the PSA must predict the likelihood of failure to appear or new (violent) criminal activity in San Francisco at statistically significant rates and rates acceptable to the PSA Workgroup. We will also measure any disparate effect and differential predictivity of the risk assessment tool based on sex, race, or ethnicity.

Evaluation of the PSA: CPL is also assessing the extent to which the PSA is reducing the length of time people spend in custody pretrial and its overall impact on annual jail bed days.  This analysis will yield important findings about the effect the PSA has had on racial/ethnic disparities in the population that is detained pretrial.

Research Team

 Johanna Lacoe (Co-Principal Investigator), Steven Raphael (Co-Principal Investigator), Elsa Augustine, Alissa Skog

Results

Read the report

Stay Informed