Redesigning NYC’s Pretrial Release Assessment


  • Pretrial assessments can be useful in aiding judges on release decisions and reducing racial disparities.
  • Behavioral barriers can impede effective use of the assessment in court.
  • Redesigning the assessment report can increase use of its recommendation and thereby decrease pretrial detention.

The Challenge

Within 24 hours of being arrested and held in custody in New York City, a person in detention is brought before a judge for arraignment. At the arraignment, a judge determines the likelihood that the individual will return for their subsequent court appearances and, accordingly, whether they should be released on their own recognizance (ROR), released with non-monetary conditions, required to post bail in order to be released, or remanded (detained until trial).

Many factors contribute to a judge’s determination, and the consequences of this early decision can be severe. Cash bail contributes to harmful pretrial incarceration that disproportionately impacts lower-income individuals, as many people in detention are unable to pay the required sum. Restrictive non-monetary conditions (e.g. electronic monitoring) can also be harmful if overused, as they contribute to the “net-widening” problem—the unnecessary expansion of social control over people and communities.


Our Approach

One of the key functions of the New York City Criminal Justice Agency (CJA) is to provide judges and the court with evidence-based information to improve decision-making and increase the likelihood that individuals return for all their court dates.

From 2003 to 2019, the previous CJA pretrial release assessment aided judges, prosecutors, and defense attorneys in their arraignment arguments and decisions. The assessment combined information about a person in detention’s community ties (gathered from pretrial interviews) and information about the person’s history with the criminal justice system (drawn from court databases) to generate a recommendation for or against ROR. At arraignment, the CJA assessment was included as a physical sheet of paper in the lawyers’ and judges’ packets of case materials.

Between 2017 and 2019, ideas42 participated in a collaborative effort to redesign the CJA assessment. We partnered with the NYC CJA, NYC Mayor’s Office of Criminal Justice (MOCJ), Crime Lab New York, and the research firm Luminosity to create a new release assessment based on three goals:

  1. Maintain the current high court appearance rates in NYC for people released pretrial.
  2. Reduce the use of pretrial detention when possible.
  3. Reduce racial and other disparities in pretrial settings.

Crime Lab New York and Luminosity analyzed over 1.6 million cases in NYC to find the best predictors of court appearance, and then incorporated the most predictive factors into a new release assessment algorithm. Reducing racial disparities was a fundamental aim of this work. We applied a behavioral lens to ensure that judges would adopt the new assessment in practice and use it effectively to support their arraignment decisions.

For more on our research and approach, read the project brief.



The updated CJA release assessment has been used in NYC arraignments since January 2020, and a CJA brief highlights the performance of the updated release assessment in its first several months. The updated release assessment recommended Black and White individuals for ROR at very similar rates (83.9% and 83.5% respectively) and recommended for ROR slightly more Latino individuals (85.8%). The updated release assessment also performed well in terms of predicting court appearance rates, according to the data collected prior to the suspension of court appearances due to the COVID-19 pandemic. Judges’ release decisions were generally consistent with the assessment’s recommendations, however, smaller disparities exist. While remand rates were very similar (0.4-0.5%) amongst individuals recommended for ROR, Latino individuals were the most likely to be released on recognizance (74.5%) followed by White individuals (72.0%) and Black individuals (69.4%), who were the most likely to receive bail (by up to 3.5 percentage points) of all three groups.

To learn more about these results, read the project brief.



There is still work to be done to fully eliminate racial/ethnic disparities and improve judicial decision-making overall. Organizations must pursue additional ways to counter implicit bias in decision-making throughout the justice process, and create systems that more effectively support communities of color.

If we build on this significant progress and implement new tools informed by behavioral science, we can continue to advance towards the critical mission of creating a more equitable arraignment process and reducing pretrial detention in New York City.

Interested in learning more about this work? Reach out to us: