We sat down with Jeremy Travis, executive vice president of criminal justice, Kristin Bechtel, director of criminal justice research, Virginia Bersch, director of criminal justice, and Stuart Buck, vice president of research, to ask them a few questions about pretrial risk assessment, since the team is celebrating the launch of the first-ever Public Safety Assessment (PSA) website.
Jeremy Travis: There’s a tremendous amount of momentum in the pretrial justice movement. The calls for change have spurred jurisdictions across the country to ask important questions such as, “Who is in our jails? Who is being released or detained before trial, and at what cost to individuals, families, and community safety? How should judges make decisions about pretrial liberty, and using what information?” These questions are being asked against the backdrop of strong evidence showing that many individuals charged with crimes, but presumed innocent, are spending time in jail awaiting trial because they can’t afford to pay money bail, often low amounts. The evidence also shows that many of them could be released with little risk that they wouldn’t return to court or would pose a threat to public safety. Finally, we have learned that pretrial detention can have serious negative consequences for their lives.
In short, this fundamental question of who should be detained and who should be released before trial has become the subject of a robust and unprecedented national movement. State legislatures are crafting legislation to reduce the reliance on money bail, provide for pretrial services, and create speedy trial provisions. Police, prosecutors, and public defenders are asking tough questions about how lower-level crimes are enforced and charged, and are weighing potential alternatives. One component of these reform initiatives—supported by the Laura and John Arnold Foundation—has been the introduction of risk assessments to guide judges in making the critical decisions on pretrial detention. This is an exciting time for the pretrial reform movement. We’re making great progress, but there’s a lot more work to be done.
2. Is that why LJAF decided to expand access to the PSA through the website?
Virginia Bersch: Pretrial risk assessment is only one part of the broader reform efforts that Jeremy described, but it has played an important role. There’s a hunger for pretrial risk assessment because people recognize risk-based decision making as fundamentally fairer than the status quo in most jurisdictions. We’ve been contacted by more than 600 people working in the criminal justice system who have expressed interest in implementing the PSA. We’ve been able to establish partnerships with approximately 40 of them. We are excited that by launching the PSA website and partnering with a new National Provider organization, we will be able to expand the reach of the PSA to all jurisdictions considering risk assessment. The National Provider will help provide a range of technical assistance to jurisdictions around the country and will eventually operate the PSA website. It houses a robust suite of resources on risk assessment implementation, all of which have been created in collaboration with technical assistance providers, current PSA jurisdictions, and other leaders in pretrial reform. It reflects lessons learned, best practices, and up-to-date research, which will be refreshed continuously. We hope—and think—the field will find these resources helpful.
3. What are the results from the research you’ve seen so far?
Kristin Bechtel: Early research findings on the PSA’s impact and the way judges and other pretrial stakeholders are using it are promising, but we want more data and more results. We recently issued an RFP calling for researchers to undertake a very robust research agenda and received a lot of great proposals. We formed a Pretrial Research Advisory Board to review those research proposals and the findings as they roll out. We are very excited about the Board—it’s an extraordinary group of practitioners, scholars, and policymakers who have a range of perspectives on pretrial risk assessment. They will add an objectivity and depth of experience to the research.
The agenda as we’ve laid it out looks at jail populations, release rates, pretrial recidivism, and court appearances before and after PSA implementation. We’re asking researchers to conduct validation studies that test for predictive accuracy and predictive bias, and we’re also going to be conducting at least one randomized controlled trial to find out to what extent use of the PSA leads to reductions in rates of failure to appear in court, new criminal activity, and new violent criminal activity.
We are committed to building the evidence base so jurisdictions are able to make the most informed decisions possible. As this evidence base is developed, we will share the findings to the national network of jurisdictions working with the National Provider. We are committed to using our findings to improve practice across the country.
4. There have been concerns that risk assessments perpetuate racial bias. Could the expansion of the PSA exacerbate this problem?
Kristin Bechtel: One of our foundational principles as a team and as individuals is a commitment to racial justice. We want to do everything possible to ensure the PSA does not perpetuate pre-existing biases that exist within the criminal justice system. That’s why a key component of the research RFP calls for thorough testing in that area. The research we have so far is promising. For example, independent researchers did a preliminary study of the PSA’s impact in Yakima County, Washington, and found that after PSA was put into place, Yakima County actually saw a reduction in racial disparities in its pretrial release rates. The release rate for white defendants remained the same, but increased for defendants who are Latino/Hispanic, African American, Asian, and Native American.
In a separate study of the PSA in Kentucky, researchers from RTI found that the PSA is predicting equally well for black and white defendants—in other words, the PSA is exhibiting strong parity across racial lines. Researchers found a slight disparity in the area of Failure to Appear, in that the PSA was underestimating the rate at which black defendants would fail to appear at court. Our research agenda will look deeply into this issue in Kentucky and other jurisdictions, and explore ways to improve the PSA using data from local jurisdictions.
5. What is the value of pretrial risk assessment?
Jeremy Travis: In most jurisdictions, under the existing system, defendants are released from jail to await trial if they can afford bail, while others who can’t afford to pay are detained—regardless of the defendants’ levels of risk to community safety or risk of skipping court. This has profoundly negative consequences for individuals, families, and communities, and it disproportionately impacts people of color and those with low incomes. In analyzing the system, we found that oftentimes, judges don’t have access to basic information about defendants at the time they are considering their release decision—such as a defendant’s criminal history—and these decisions are consequently often made without objective data or with the use of predetermined, offense-based fixed bail schedules.
We believe that moving towards a system that incorporates data-driven assessments is the best way to help judges make better decisions as to who is released and who is detained before trial. This type of data-informed approach can help to support judicial decision making, increase public safety, promote the fair treatment of all individuals, and ensure the responsible use of taxpayer funds. At the same time, we recognize that any risk assessment should be used as a guide; judges ultimately should make decisions about pretrial detention. Likewise, the simple adoption of the PSA should not be viewed as synonymous with comprehensive bail reform. A jurisdiction should examine all of its practices that contribute to unnecessary pretrial detention in designing a system of pretrial justice.
6. You said you’re open to tweaking the algorithm—why?
Virginia Bersch: We are constantly looking for new information and new ways to improve the PSA—we recognize that we can always learn more. We firmly believe that the PSA represents a dramatic improvement over the money bail system that has been in place for decades. We also point out that only ten percent of all jurisdictions in the country use any form of pretrial risk assessment, so we think the launch of the PSA website will lead to significant uptake of the PSA. But if, through research, we discover that the PSA can be improved, then we will embrace the opportunity to improve it. Through the Pretrial Research Advisory Board, we’re bringing in outside experts who have a history of thinking critically about risk assessment and who understand jurisdictions’ needs on the ground to help us think through these very questions. Those are the steps that lead to progress. If you get entrenched in your beliefs, you run the risk of tolerating a system like money bail, where blatant unfairness is excused simply because “that’s the way we’ve always done it.”
Stuart Buck: As Virginia said, we are committed to evaluating the PSA, validating it in local jurisdictions using local data, and seeking constant improvement over time. We agree with colleagues who say that algorithms cannot capture everything relevant about a defendant. No one assessment should ever substitute for informed judgment by a human being, in the case of pretrial decision making, a judge. The point of the PSA is to help human judgment be better informed.
7. Is the PSA intended to decrease pretrial detention populations?
Kristin Bechtel: The PSA was developed to give judges access to objective data they could consider when making release decisions. We hope and expect that the incorporation of better data into the decision-making process will lead to decreased pretrial jail populations. That’s because we know, based on numerous studies, that the vast majority of the nearly 750,000 people sitting in our country’s jails are awaiting trial and pose little risk to community safety. They’re in jail because they can’t afford their bail amount.
There are sometimes local forces at play that can heavily influence pretrial release decisions and detention numbers. For example, some jurisdictions that use the PSA must also abide by federally imposed caps on their jail populations or respond to local capacity issues. In other jurisdictions, research has found that judges may follow the PSA’s recommendations less closely over time, leading to increases in pretrial detention. We are very interested in understanding these local factors and how they impact jail populations, and plan to support researchers to study how jurisdictions implement the PSA and use it over time.
In general, research has found that use of the PSA leads to decreases in pretrial jail populations, without upticks in crime or missed court dates. New Jersey, for example, which implemented the PSA statewide as part of a larger reform effort, has seen its pretrial jail population decline by 20 percent with no increase in crime.
8. What are you hearing from judges about the PSA?
Virginia Bersch: We’re hearing very encouraging things. Researchers at RTI recently published a qualitative report that asked criminal justice practitioners across the country—judges, prosecutors, public defenders, and pretrial staff—about how they use the PSA and what they think of it. The study found, among other things, that 98 percent of the judges surveyed indicated that the PSA informs their decision, at least some of the time. Almost 80% of the judges surveyed reported that the PSA “always” or “often” informs their decision. That tells us that judges are using the information provided by the PSA, while still maintaining judicial discretion. Out of all the respondents, 61 percent report they “often” agree with the PSA’s recommendation, and 69 percent report they consider it a strength that the PSA is research-based.
It’s great to have this feedback from the field. We hope the new PSA website will provide practitioners with even more resources that they can use to help support pretrial decision making.