NEW YORK — Judges in New York City frequently override official pretrial release recommendations, leading to unnecessary pretrial incarceration and widening racial disparities in bail decisions, according to new research from John Jay College’s Data Collaborative for Justice (DCJ).
Between 2021 and 2023, the city’s Pretrial Release Assessment recommended release on recognizance (ROR) — unconditional release without bail — for nearly 80% of people charged with felonies. But judges followed those recommendations only about half the time for nonviolent felony cases and less than a third for violent felonies, the DCJ found.
The assessment is administered by the NYC Criminal Justice Agency (CJA) and incorporates multiple factors statistically associated with court attendance to make one of three recommendations: release on recognizance, consider all options, or not recommended for ROR.
The tool aligns with New York State law, which focuses on ensuring court appearance. Validation studies using data from 2014, 2022 and 2023 found the assessment to be accurate and effective in predicting court attendance.
Analyzing 251,917 arraignments from 2021 to 2023, DCJ researchers found that the assessment recommended ROR in 88% of cases overall — 79% for violent felonies, 77% for nonviolent felonies and 92% for misdemeanors. However, judges granted ROR in only 25% of violent felonies, 42% of nonviolent felonies and 78% of misdemeanors.
Judicial alignment was especially low in felony cases. Judges followed ROR recommendations 30% of the time for violent felonies and 51% for nonviolent felonies, compared with 83% for misdemeanors. Even in violent felony cases where bail was legally allowed, judges imposed bail in 41% of cases where the assessment recommended release.
The study also uncovered racial disparities. The assessment recommended release at similar rates for Black, Hispanic and white defendants (87%, 88% and 85%, respectively). Yet judges granted ROR in violent felony cases only 26% of the time for Black defendants, compared with 32% for Hispanic and 43% for white defendants. Judges imposed bail or remand on Black defendants 44% of the time, compared with 39% for Hispanic and 29% for white defendants, despite equivalent likelihoods of returning to court.
Borough differences were also pronounced. Judges in the Bronx and Brooklyn were most likely to follow the tool’s guidance, while those in Staten Island were least likely. In nonviolent felony cases, alignment ranged from 39% to 43% in Manhattan, Queens and Staten Island to 58% in Brooklyn and 72% in the Bronx.
Judges also appeared to overemphasize factors such as housing instability and prior warrants or convictions — even though the assessment already accounted for them — often overriding ROR recommendations when these factors were present.
The DCJ acknowledged the study’s limitations, noting that researchers did not observe proceedings directly or survey judges about their decision-making. The organization also recognized that judges may exercise discretion based on case-specific details and that no predictive tool is infallible.
Still, the researchers warned that frequent judicial deviations could undermine fairness. “The pretrial release assessment provides empirically grounded guidance to judges making release decisions, while still allowing for discretion for case-specific circumstances,” said Joanna Weill, associate director of research at DCJ. “More consistent use of these recommendations would foster fairness by limiting the extent to which outcomes depend on the judge assigned.”
Michael Rempel, DCJ’s executive director of data, called for greater judicial training and feedback. “While New York City has a validated assessment shown to be effective in predicting return to court, we found limited reliance on it in felony cases,” he said. “We hope our study presents an affirmative opportunity for city officials and reform-minded state court leadership to provide judges with greater training and information, and a chance to have their questions answered, in how this tool could help them in their work.”
Follow the Vanguard on Social Media – X, Instagram and Facebook. Subscribe the Vanguard News letters. To make a tax-deductible donation, please visit davisvanguard.org/donate or give directly through ActBlue. Your support will ensure that the vital work of the Vanguard continues.
