(Evidence-based, Actuarial Pretrial) Risk Assessment


A lot of folks are talking about “risk assessment” these days. And that’s good. But it’s important that everyone is clear about what it means (and doesn’t mean) when we use this term. At PJI, risk assessment is almost always accompanied by a handful of other words: pretrial, evidence-based, and actuarial. Here’s why:

Pretrial—The criminal justice system uses risk assessment at different times for a variety of purposes—including decisions about sentencing, incarceration security levels, and conditions for post-adjudication supervision. Needs assessments are a completely separate evaluation done to determine whether an individual might benefit from supports such as substance use or behavioral health treatment—which is important when crafting a good support structure for adjudicated people who may benefit from treatment. Among all of the points at which risk is being assessed, pretrial risk stands out because it focuses on measuring only two specific behaviors—court appearance and rearrest—during a specific time frame—between arrest and trial. As such, the factors used to determine pretrial risk level (e.g., prior convictions) and the comparison cases (similar people released before trial in the recent past) are far more limited than those used in other risk assessments. This makes pretrial tools and their results much more accurate.

Evidence-based—The fact is, some kind of risk assessment happens every time a person appears before a court officer charged with making a bail decision. Their assessment can range from a purely subjective, visual analysis—does the person look risky?—to something more objective and, one hopes, fairer. Many early objective tools looked at the strength of the person’s ties to the community—things like home ownership and local family members—and required an interview with the arrested person. Subsequent statistical analysis showed, however, that those things not only don’t matter, but they can unfairly favor wealthier people. Today’s evidence-based tools—the product of ongoing data analysis—have not been shown to exacerbate or institutionalize any bias; instead they are and can be used to narrow the likelihood of bias. And this is about capturing data about decisions that heretofore were opaque.

Actuarial—When we talk about actuarial pretrial risk assessment, we are referring to tools that have been developed using statistical analysis of carefully gathered data; in this case, as many relevant cases of previous pretrial behavior as possible. The Public Safety Assessment (PSA) developed by the Laura and John Arnold Foundation, for example, is based upon statistical analysis of 1.5 million cases across 300 jurisdictions. This is no different than auto insurers measuring your likelihood of getting into an accident based on the previous accidents of other people—except, as discussed above, it is much more narrowly focused.

Compared to purely subjective decision making, or tools that measure a broad range of behavior over an extended period of time, evidence-based, actuarial pretrial risk assessments are a compelling improvement. They deliver better results for accused people and for the systems that use them. And as the process of collecting and analyzing past case data continues, evidence-based, actuarial pretrial risk assessment promises to deliver even better results going forward.

There’s a lot more to talk about here and we will. Share your two cents in UP. And stay tuned.

Comments are closed.