Advocates of pretrial RATs promote them as more objective, and therefore less racially biased, than human judges or magistrates.
However, the use of RATs does not take bias out of the equation, and we cannot assume that using pretrial RATs will reduce incarceration or racial and economic disparities.
Several jurisdictions using RATs have not successfully reduced the racial disparities in their jails, and many more do not track any racial changes.
Many of the jurisdictions we interviewed did not explicitly connect using a risk assessment with reducing racial disparities in incarceration and supervision demographics in their jurisdictions. While some did say that they were aware of disproportionate racial populations in their jails and wanted to tackle this issue, others did not view their program or the RAT as tools to change jail demographics.
For example, our interviewee from Buncombe, North Carolina, acknowledged the racial disparities in their jail system and is working to address these disparities and implicit bias through trainings for judges and other decision-makers.1Interview with Media Mobilizing Project, 4/4/2017. See Interview Summary for more information Our interviewee from Douglas County, Kansas, recognized that though they tried to keep the tool they developed race-neutral, even prior convictions can include racial bias. They are making efforts to address racial disparities in issues like traffic stops.2Interview with Media Mobilizing Project, 7/18/2019. See Interview Summary for more information
Several other interviewed jurisdictions were trying to address racial disparities, account for bias, and study their own practices and tools as they relate to race, including Connecticut;3Interview with Media Mobilizing Project, 8/30/2017. See Interview Summary for more information Hennepin4Interview with Media Mobilizing Project, 11/3/2017. See Interview Summary for more information and Cass Counties, Minnesota;5Interview with Media Mobilizing Project, 7/18/2019. See Interview Summary for more information Santa Clara County, California;6Interview with Media Mobilizing Project, 9/15/2017. See Interview Summary for more information Nashville, Tennessee;7Interview with Media Mobilizing Project, 7/18/2019. See Interview Summary for more information and Saline County, Kansas.8Interview with Media Mobilizing Project, 7/16/2019. See Interview Summary for more information
Some interviewees noted the complexity of understanding a tool’s full impact on different racial populations, and pointed out that the racial impact of a particular tool may vary between different jurisdictions using the same tool due to the way a tool is implemented on the ground.
However, interviewees from New Orleans, Louisiana;9Interview with Media Mobilizing Project, 5/1/2017. See Interview Summary for more information San Antonio, Texas;10Interview with Media Mobilizing Project, 5/15/2017. See Interview Summary for more information Shasta County, California;11Interview with Media Mobilizing Project, 11/13/2018. See Interview Summary for more information Yakima County, Washington;12Interview with Media Mobilizing Project, 4/12/2017. See Interview Summary for more information Mesa County, Colorado;13Interview with Media Mobilizing Project, 5/1/2017. See Interview Summary for more information and Armstrong County, Pennsylvania,14Interview with Media Mobilizing Project, 7/24/2019. See Interview Summary for more information stated that the pretrial assessment tool was not implemented to explicitly reduce racial disparity in pretrial incarceration. Some felt that race had nothing to do with assessment, or that because race was not an explicit input, the tool had no racial impact.
None of our interviewees could explicitly link the use of their RAT to a reduction in racial disparities in their jail or pretrial release rates. Many said they do not collect or could not tell us about the racial demographics of their jails at all.
Megan Stevenson’s report evaluating the impacts of risk assessment in Kentucky, an early adopter of a statewide pretrial RAT, found that using a risk assessment tool did not alter racial disparities in the pretrial population.15Megan Stevenson: Assessing Risk Assessment in Action, George Mason University Stevenson stated that the 2011 bail reform law mandating risk assessment “benefited white defendants more than Blacks” due to the way judges implemented the changes.16Megan Stevenson: Assessing Risk Assessment in Action, George Mason University
In addition, before statewide implementation of risk assessment in Kentucky, the pretrial release rate without bail was roughly equal between Black and white defendants; after RATs became mandated in 2011, the release rate for white defendants was 10% higher than for Black defendants.17Tom Simonite: Algorithms Should’ve Made Courts More Fair. What Went Wrong? Wired
In 2017, New Jersey began using the PSA statewide and completely reformed their money bail and pretrial systems. Although the overall jail population has decreased, “the racial and ethnic makeup within New Jersey’s jail population has remained largely the same.”18Glenn A. Grant: Jan 1-Dec 31 2018 Report to the Governor and the Legislature, New Jersey Courts, Criminal Justice Reform
Researchers studying the impact of the PSA in Mecklenburg, North Carolina, found that the PSA had no impact on racial disparities in their systems, and that Black individuals were more likely than others to be identified through the PSA as being high-risk.19Cindy Redcross, Brit Henderson, Luke Miratrix, and Erin Valentine: Evaluation of Pretrial Justice System Reforms that use the Public Safety Assessment: Effects in Mecklenburg County, North Carolina, MDRC Center for Criminal Justice Research
The Center for Court Innovation tested their tool for accuracy using New York City-based data in 2019.20Heather M. Harris, Justin Goss, and Alexandria Gumbs: Pretrial Risk Assessment in California, Technical Appendices, Public Policy Institute of California The researchers found that 37% of Black and 29% of Latinx individuals were classified as moderate-high or high risk, compared to 18% of white individuals, and Black and Latinx individuals were more often incorrectly classified as higher risk than white individuals.
These researchers concluded that “pretrial risk assessment tools are likely to exhibit inequity – especially racial inequity,” often because using input data such as prior arrests is racially biased.21Heather M. Harris, Justin Goss, and Alexandria Gumbs: Pretrial Risk Assessment in California, Technical Appendices, Public Policy Institute of California