What Went Wrong With This Foundation-Backed Effort to Lower Jail Populations?



We have written often about the John D. and Catherine T. MacArthur Foundation’s work to reform criminal justice systems. Through its Safety and Justice Challenge, launched in 2015, MacArthur has awarded $148 million to 52 cities and counties committed to criminal justice reform.

By many indications, MacArthur’s investments are starting to show positive results, and the foundation was encouraged enough to expand its effort last fall. A 2018 report from the City University of New York (CUNY) examined average daily jail populations (ADP) and racial/ethnic disparities in jails among Year 1 grant recipients. It showed ADP down by double digits in Toledo, Ohio, and in Philadelphia, as well as more modest decreases between 7 and 10 percent in Milwaukee and New Orleans.

However, Spokane, Washington, was conspicuous for its increase. According to the CUNY report, the average daily population in Spokane County rose 10.3 percent.

Recently, Spokane authorities announced they were abandoning an ambitious algorithm designed to help courts determine which inmates should remain in jail pending trial in favor of a free alternative assessment program from another funder. Spokane received $1.75 million from MacArthur in 2016 to support its efforts to reduce jail populations. Since then, a series of technological and personnel problems hampered implementation of the program. Meanwhile, jail populations did not decrease. In December, Spokane abandoned the effort.

At a moment when there’s a lot of funding action around criminal justice reform, and—inevitably—some degree of trial and error in this space, it’s worth taking a closer look at what went wrong in Spokane.

The idea seemed simple enough: Using data on more than two dozen factors ranging from criminal records to employment history, the proposed algorithm would aid judges in deciding which defendants should be held in jail and which could be released as their cases moved through the justice system. Advocates of the proposed Spokane Assessment for Evaluation of Risk (SAFER) said the tool would reduce jail overcrowding and racial/ethnic disparities.

In 2016, just before being awarded the MacArthur grant, Spokane County awarded Zachary Hamilton, a Washington State University criminologist, a $70,000 contract to develop the SAFER software from scratch. The tool Hamilton developed would produce a score indicating the probability that a particular defendant would miss a court date or reoffend. Applying this tool, Spokane hoped to reduce its jail population by holding low-risk offenders.

Unfortunately, the richness of the data—encompassing 30 factors—contributed to the complexity of its implementation. The system never worked as intended. Problems first surfaced during development and initial testing. Staff turnover at the firm contracted by Hamilton to write the program delayed its development. Technical issues ranging from the display of risk scores on monitors to syncing SAFER with state criminal history data hampered testing in late 2016. Hamilton told the Spokesman-Review newspaper that the state courts office was reluctant to share data on adult offenders with individual counties.

This problem was especially critical, as it forced staff at Spokane County’s pretrial services office to search databases manually. About $600,000 from the MacArthur grant was allocated to additional staffing in pretrial services, but even with the additional personnel, the manual searches proved a heavy burden.

Cheryl Tofsrud, Spokane’s manager of pretrial services, identified this issue as a key factor in the project’s failure. Having to conduct manual lookups added time to an already lengthy process.

“What was thought to be a time saver [actually] wasn’t,” she told The Inlander, a weekly publication in Spokane.

Hamilton then produced a scaled-down version of the tool dubbed “SAFER Lite” to buy time to sort out the technical issues and resolve access to the offender data from the state. Turnover in the pretrial services office, however, posed a new obstacle. Further, judges indicated that the risk scores were not as helpful as the evaluations that the pretrial services office had provided previously, according to the Spokesman-Review. Tofsrud opted to suspend use of SAFER, in part because of lack of sufficient staff.

Eventually, the state courts administrative office did approve access to the requested adult offender data, but even with that approval, there would still be months of programming necessary to bring the project to fruition. By then, Spokane officials were ready to give up. In December, they did just that, halting the SAFER project altogether. But Spokane went a step further, choosing a replacement tool: the Public Safety Assessment (PSA).

The PSA was developed by the Laura and John Arnold Foundation (LJAF), another high-profile funder we’ve written about often. LJAF has been a leading backer of criminal justice reform efforts for years, with a major focus on “supporting innovation in the pretrial justice arena, specifically the bail reform movement.” On its website, the foundation describes what’s at stake in this work: “The decision whether to detain a defendant during the pretrial phase has far-reaching impact. Individuals held in jail stand to lose their jobs, suffer negative health effects, and experience disruption in their family lives.” Low-income people who can’t afford bail suffer the most in the current system.

LJAF has poured huge resources into developing better, evidence-based approaches to pretrial justice—including its extensive work to develop, disseminate and evaluate the PSA. The funder offers the tool for free to any jurisdiction interested in using it. Like SAFER, the PSA is intended to improve decision-making in the criminal justice system in a way that increases public safety, makes the system more efficient, and treats all defendants fairly. However, the PSA uses only nine factors, far fewer than the 30 used by the SAFER assessment.

More than 40 jurisdictions across the country, including three states—Kentucky, New Jersey and Arizona—use the PSA. Yakima County, Washington, is one of the jurisdictions using the PSA, and officials there report that the tool has helped reduce racial disparities in the pretrial jail population. Spokane officials are optimistic the tool will produce similar results for them.

The PSA is not without its critics, who contend that because of the limited range of factors examined, the tool may exacerbate racial disparities rather than reduce them.

In hindsight, some in Spokane argue that the county should not have poured so much grant money into a single project. Some say the expectation that a single tool could reduce the county’s jail population may have been unrealistic. However, it appears that the issues surrounding data sharing were another significant issue marring the project. Data sharing among governmental entities is an area fraught with technological and legal complexities. Federal laws such as the Health Insurance Portability and Accountability Act (HIPAA) and the Family Educational Rights and Privacy Act (FERPA) strive to protect privacy. Government agencies are often reluctant to share data with other entities for fear of inadvertently violating these or other laws governing data privacy.

But access to data from multiple sources—when properly analyzed and securely stored—has enormous potential to improve decision-making in criminal justice, education and other areas. Helping to navigate and untangle the complexities in this area could be valuable for some ambitious funder to explore.