It’s a travesty when people are stuck behind bars for months or even years before they have even been convicted of a crime. However, there are sometimes good reasons to deny a person bail, such as when they are an obvious flight risk or a clear danger to the community.
Many jurisdictions around the U.S. are using pretrial assessment tools to help decide whether people are safe to release before trial. Ideally, these tools would accurately gauge a person’s likelihood of flight or of committing a new offense if released. Ideally, they would be less biased than judges have proven to be. Ideally, they would reduce the jurisdiction’s reliance on cash bail and pretrial incarceration.
Are these tools working, though? A new report and database indicate that there is little evidence they are doing the job advertised.
The organizations behind the report and database are the Media Mobilizing Project and MediaJustice. They collected data on pretrial risk assessment tools being used in 46 states and the District of Columbia. They performed in-depth interviews with 38 pretrial services agencies and did email interviews with officials from 18 others.
Jurisdictions not even tracking the tools’ effectiveness
Unfortunately, these interviews revealed that few jurisdictions are keeping track of whether their tools serve their intended purposes. Moreover, when the groups themselves looked into whether the tools are effective, they found:
- The tools are ineffective at reducing money bail and pretrial detention
- The tools have little impact on racial disparities in pretrial release rates
For example, when Spokane County, Washington, expanded its use of pretrial assessment tools in 2017, it hoped to reduce the number of defendants in pretrial detention. Almost three years later, however, the county’s jail population remains at the same level it had been at in 2016.
Or consider the experience of Kentucky. When it mandated the use of pretrial assessment tools in 2011, it did see an initial increase in release rates. Research found, however, that judges soon began overriding the tool. While the tool recommended release for about 90% of defendants, judges only followed that recommendation in 29% of cases.
Furthermore, the algorithms used in these pretrial assessment tools are not race-neutral. They rely heavily on criminal history and even prior arrests, which means that African-American and Latinx defendants are more likely to be considered high risk simply because they generally have more contact with law enforcement than their white peers.
Multiple studies have found that very few people — between 1% and 2% — commit violent crimes after being released before trial, even when they are labeled high risk. Yet the system seems geared toward locking people up before trial, especially poor and minority defendants.