About the Public Safety Assessment

The PSA uses nine factors related to age, criminal history, current offense, and past court appearance. These factors suggest the likelihood of court appearance or arrest-free behavior based on how others with the same or similar factors behaved while on pretrial release. The information used by the PSA is the same information judges normally consider when making pretrial decisions.

The factors and methods used to calculate PSA scores are publicly available. Jurisdictions should provide individual PSA scores to the person charged, judicial officer, defense counsel, and prosecution.

The PSA is free. Jurisdictions agree to the PSA Terms of Use and Core Requirements (PDF).

Judges may consider the information provided by the PSA, but their decisions must be based on the law and consideration of all the evidence.

Alongside the PSA, judges typically have access to more detailed information about the facts of the case, the strength of the evidence, the person’s criminal history, ties to the community, such as family relationships, employment history, housing, mental health and substance use history, and other critical information that may inform their decisions. They may also hear arguments from the defense and prosecution, review the arresting officer’s report, and hear input from victims. The final decision about release, conditions of release, and detention always lies with the judge.

Video Transcript

Local jurisdictions must develop their own policies to use the information provided by the PSA. While the PSA provides a score that estimates a person’s likelihood of appearing in court and remaining arrest-free during the pretrial period, the assessment itself does not recommend release, conditions of release, or detention. Pretrial system stakeholders create a Release Conditions Matrix for their local system. State law and court rules typically set forth legal standards and criteria for detention. For those who are released, the Release Conditions Matrix suggests levels of release conditions that will be most effective in improving court attendance and law-abiding behavior based on community safety standards and available government and community resources.

Arizona, Kentucky, New Jersey, and Utah implemented the PSA statewide. Many counties and cities throughout the country also use the PSA. Hundreds of thousands of assessments are conducted every year, providing pretrial decision makers with additional information to inform their decisions, which are then weighed alongside other factors to make more deliberate and informed decisions.

PSA research falls into two types: validation studies and implementation studies. Validation studies use historical or post-implementation data after a community begins using the PSA to test its performance in predicting pretrial outcomes. They also evaluate how accurately the PSA estimates outcomes across race and gender. Implementation studies examine how jurisdictions decide to use the PSA, how it is implemented, and how its use in decision-making may influence pretrial outcomes.

Implementing the PSA requires careful planning, dedicated resources, and collaboration across the pretrial system. Successful implementation depends on a collaborative team supported by key stakeholders, including judicial officers, prosecutors, defense counsel, and pretrial services. Jurisdictions also need access to local, state, and national criminal history and court data to score the PSA accurately, technology to automate the assessment, and sufficient staff to manage implementation and train all system and community stakeholders.

The Arnold Ventures philanthropy developed the PSA. APPR provides jurisdictions with information about the PSA, including implementation guides, training, and technical assistance.

The PSA was created using the largest, most diverse set of pretrial records ever assembled: approximately 750,000 cases from roughly 300 jurisdictions across the United States. Researchers analyzed the data to determine which factors were most predictive of failure to appear pretrial, new criminal arrest while on pretrial release, and new violent criminal arrest while on pretrial release. Immediately after its development, the PSA was validated using a dataset of over 500,000 cases from multiple different jurisdictions. Since then, it has been revalidated in many additional jurisdictions, showing it to be predictive of these key pretrial outcomes, and to predict well across all racial and ethnic groups.

Yes, it uses a fixed, publicly available point system to calculate scores. The factors and methods used to calculate PSA scores are publicly available.