This report presents the findings of a local validation of the Public Safety Assessment (PSA) as implemented in Pennington County, South Dakota. Policy Research began this validation study in 2023 with funding from the John D. and Catherine T. MacArthur Foundation’s Pretrial Risk Management Project.

Policy Research researchers examined the performance of the PSA scores in a sample of 4,570 cases that were processed in the Pennington County court, that required an initial appearance hearing, and that were PSA eligible between June 6, 2018, and December 31, 2021. Researchers examined the PSA’s ability to predict failure to appear and new criminal arrest scores during the pretrial period, as well as differences in PSA performance across demographic characteristics.

Key Findings

  • About two-thirds of people who had been released from jail on personal recognizance bonds in Pennington County appeared in court and were not arrested for a new criminal charge during the pretrial period. In other words, most people were successful while awaiting their trial.
  • Overall, the failure-to-appear subscale scores demonstrated poor validity in predicting failure to appear, and the new criminal arrest subscale scores demonstrated fair validity in predicting new criminal arrests. These findings indicate that the PSA subscale scores do not distinguish between people at lesser and greater risk of failing to appear in court but do distinguish between people at lesser and greater risk of new criminal arrest.
  • There were differences in the performance of the PSA subscale scores across groups defined by race. Both the failure-to-appear and new criminal arrest subscales were better at predicting pretrial outcomes for white people than Native American people. Moreover, the relationship between failure to appear subscale scores and rates of failure to appear differed significantly for white and Native American people.
  • There were also differences in the performance of the PSA subscale scores across groups defined by gender. Both subscales were consistently better
    at predicting pretrial outcomes for men than women. Again, the relationship between failure-to-appear subscale scores and rates of failure to appear differed significantly for men and women.

Recommendations

Based on the results of the current validation, researchers offered the following five recommendations:

  1. Limit the use of the failure-to-appear subscale scores to inform decisions about service delivery.
  2. Consider adopting an instrument that assesses dynamic factors.
  3. Clarify override procedures and monitor their use and impact on decisionmaking.
  4. Implement continuous quality improvement measures.
  5. Conduct a validation with a more representative sample and additional data.

This report was first released in 2025. For a 508-compliant version of this report, please email communications@prainc.com.

Suggested Citation

Zottola, S. A., Desmarais, S. L., & Chen, L. (2025, January). Local validation of the Public Safety Assessment in Pennington County, South Dakota. Policy Research Associates. https://www.prainc.com/resource-library