Pretrial Release

The Problem: Prison overcrowding and associated issues with state and federal sentencing policy receive ample attention among criminal law scholars and practitioners. Just as important, however, are the problems created by ineffective detention policies before trial. A number of reform efforts are underway to better predict an arrestee’s actual risk of failure to appear (FTA) and recidivism. The most prominent new tool is the actuarial risk assessment mechanism, one of many evidence-based practices now used in criminal and other legal spheres. The theoretical use of such mechanisms has come under fire from various quarters. Yet we do not know in practice how effectively these risk scores mitigate FTA and recidivism.

May 25, 2016

Pretrial Release

These instruments, despite their promise, give rise to some concerns worthy of study. Many rely on subjective factors, which require the exercise of judgment to score; their predictive power has not been assessed using data from the jurisdictions in which they are used; most instruments depend on information that, realistically, can only be obtained from an arrestee interview; and, crucially, only two instruments have undergone randomized evaluation in over six decades of use in the United States.

The Study

In partnership with Arnold Ventures, the A2J Lab will evaluate the effectiveness of the Public Safety Assessment (PSA) in several jurisdictions across the country. The purpose of the study is to test whether an actuarial risk assessment tool that avoids the need for arrestee interviews produces better pretrial incarceration decisions and associated decreases in FTA and repeat offenses. (Visit our Open Science Framework page for more information.)


In partnership with leading criminal justice researchers, Arnold Ventures developed the PSA to help judges evaluate the risks mentioned above. To date, it has been fully implemented, and is fully operational, in a variety of U.S. jurisdictions.

The perceived benefits of the PSA include:

  • Reliance only on objective inputs, i.e., those that can be observed without the exercise of significant subjective decision-making.
  • Calculation using only administrative data, obviating the need for an arrestee interview.
  • Avoidance of factors that are highly correlated with problematic characteristics, such as race, national origin, or economic class.
  • Amenability for use in jurisdictions throughout the United States, regardless of a community’s demographics or its criminal activity levels.

Thflowchart2e PSA’s inputs are, according to Luminosity research, correlated with the risk of FTA or engaging in new criminal activity (e.g., arrestee age and whether he has failed to appear within the last two years). An algorithm transforms the various inputs into two scores on a 1-6 scales, one each for FTA and new criminal activity (NCA), as well as a “Violence Flag,” signifying an elevated risk of committing new violent criminal activity (NVCA). A Decision Making Framework (DMF) is then applied and consists of a set of instructions to transform the two FTA and NCA scores, the Violence Flag, and other factors (such as whether the current charge is for certain domestic violence-related offenses) into a recommendation for a judge making an initial release/detention decision. The translation of scores into recommendations involves fundamental questions of a jurisdiction’s budget, tolerance for risk of arrestee misbehavior, attitudes toward incarceration, current and anticipated jail population, and other jurisdiction-specific factors.  As a result, the DMF’s recommendations are the result of a conversation between the LJAF and an implementing jurisdiction, with the LJAF insisting that the jurisdiction adhere to the principle that the DMF’s recommendation be reached based on a consideration of the PSA scores, not solely the current charge against the arrestee.

Field operation

The A2J Lab is currently engaged to evaluate the PSA’s effectiveness in Dane County, Wisconsin.

Cases will be randomized either to the “treatment” group, in which instance the Clerk’s Office will produce the PSA printout, append it to the case file, and make it available to the prosecuting and defense attorneys as well as to the Commissioner in time for the initial appearance; or to the “control” group, in which instance, the Clerk’s Office will not produce the PSA printout. Randomization will be by case number.


The principal outcomes of interest for this study include the following:

  • Duration between initial appearance and case disposition;
  • Whether the arrestee was convicted of any charge;
  • Whether the arrestee was sentenced to any term of incarceration and the number of months ordered to serve, if any;
  • Failure to appear, both whether the arrestee missed any court date in the case as well as how many FTAs the arrestee exhibited.
  • Whether the arrestee engaged in any new criminal activity and the number of separate incidents (both between initial appearance and case disposition and up to two years after initial appearance);
  • Whether the arrestee engaged in any new violent criminal activity and the number of separate incidents (both between initial appearance and case disposition and up to two years after initial appearance); and
  • Whether the arrestee was released at any point between initial appearance and case disposition, the number of days the arrestee spent incarcerated during that period , and the share of days from initial appearance to case disposition that the arrestee spent incarcerated.

The Research Team

Jim Greiner, Faculty Director, Access to Justice Lab; Professor of Law, Harvard Law School

Matthew Stubenberg, Associate Director of Legal Technology, Access to Justice Lab

Ryan Halen, Data Analyst, Access to Justice Lab

Heidi Liu, J.D./Ph.D. candidate, Harvard Law School and Harvard Kennedy School

Chris Griffin, Visiting Professor and Research Scholar, James E. Rogers College of Law, University of Arizona