New Investigation:

How Lax EPA Oversight Enabled Jackson's Water Crisis.

Protecting Civil and Human Rights

DOJ Can and Must Improve Data on Deaths in Custody

(Photos: Getty Images; Photo Illustration: Leslie Garvey / POGO)

There’s no question there is a crisis of people dying while in government custody across the country. For years, police killings, predominately of Black people, have dominated news and led to mass demonstrations. Countless investigations have exposed abuse and medical neglect in prisons and jails.

We know there’s a problem, but we don’t know the full scope of it. There is no comprehensive data on how many people die while in government custody — while the government is responsible for keeping them safe.

In 2014, Congress reauthorized the Death in Custody Reporting Act (DCRA). The law had a vital and seemingly straightforward goal: Improve the Department of Justice’s efforts to track how many people die in government custody around the country and the circumstances of their deaths, and study the data to identify ways to reduce the number of deaths going forward.

Eight years later, there’s very little to show for the Justice Department’s effort to comply with the DCRA. In a Senate hearing September 20, the head of the Justice Department office responsible for overseeing death in custody data had a new explanation for why its efforts have failed: It’s actually Congress’s fault. The department now says a handful of seemingly minor differences in wording between the 2014 version of the DCRA and an earlier version have tied its hands, preventing it from designing a program that actually works.

There is no comprehensive data on how many people die while in government custody.

There’s no denying that the program in place now isn’t working. The Government Accountability Office released an audit September 20 of the department’s data for fiscal year 2021. It found that the Justice Department numbers were substantially inaccurate. The department missed 990 “potentially reportable” deaths in custody that had been reported on elsewhere, to say nothing of deaths that have gone wholly unreported. The audit also found that 70% of the records the department did have were missing at least one piece of information. Even if all deaths were reported and the records were complete, the data they produce is so imprecise and jumbled, it would still be useless — a point we’ve been making to the department for over a year.

Having spent a lot of time analyzing the law and the department’s past efforts to implement it, we were taken aback by the explanation for these problems. For some reason, the U.S. Justice Department has convinced itself that a law designed to give it more tools to collect data on the number of people who die in custody in this country and share that data with policymakers has taken away its power to do just that.

It doesn’t take a deep dive to see that the department has more flexibility than it claims.

According to the Justice Department, the wording of the current version of the law prevents it from doing three key things it had done in the past: Collect data directly from state and local agencies, assign the work of collection to the department’s own statistics experts, and release the data to the public. It used to do all of these things, and while the data was far from perfect (indeed, in some ways it was deeply flawed), it was better and more transparent than anything that’s been produced under the newer law.

Preventing these practices certainly wasn’t the intent of Congress when it reauthorized the law in 2014. A bit of history is helpful to understand why. When the first DCRA became law in 2000, the Justice Department’s Bureau of Justice Statistics (the name tells you about its expertise) became responsible for collecting death in custody data. It did so by reaching out directly to the thousands of local jails, state prisons, and law enforcement agencies across the country. The DCRA law lapsed in 2006, but the statistics bureau kept collecting data anyway.

It was still collecting that data when Congress passed the updated version of the law in 2014, and there’s a paper trail that shows Congress thought it was simply making sure the data collection had the weight of a statute behind it. The House Judiciary Committee noted in its report endorsing the bill that collection was ongoing. The purposes of the bill were to incentivize states to report data, extend the collection program to deaths in federal custody, and order the Justice Department to make policy recommendations based on what the data revealed.

There’s simply no suggestion that Congress thought reauthorizing the DCRA would mean anything about the existing collection would change. As former Representative Robert Goodlatte (R-VA), then-chair of the judiciary committee, noted, “Because the Bureau of Justice Statistics has continued to collect the information … this bill will not impose any new cost on the agency.”

Even though Congress thought it was ensuring the continuation of an ongoing program, the Justice Department is right in noting that there are some wording differences between the old and updated statutes. But these differences aren’t as significant as the Justice Department says.

For instance, the department says the new law prevents it from directly contacting local agencies (which account for the overwhelming majority of police departments and jails, and which come in contact with a mind-boggling number of people). The basis of this prohibition, it says, is the text of the statute saying that a “state shall report to the Attorney General … pursuant to guidelines established by the Attorney General, information regarding” deaths in custody.

It’s true that the language doesn’t explicitly mention local governments as sources of information. But neither did the old law, which required a state to provide “assurances that it will follow guidelines established by the Attorney General in reporting … information regarding” deaths in custody.

Neither version of the law has anything to say about local governments, except that deaths in their custody must be reported.

It’s certainly plausible to read the new text and conclude that states should report data on behalf of localities. But it’s hard to compare the statutes and conclude that the new one requires or prohibits anything different from the old.

The department says another source of trouble is a provision in the 2014 law that allows it to withhold some grant funding from states that don’t report. Specifically, the law gives the attorney general the discretion to apply a 10% cut to a state’s award under the Byrne Justice Assistance Grant program, one of the most prominent federal law enforcement grants, if it fails to comply with data reporting requirements.

The department has taken the dubious position that a provision mandating a single report to Congress on how to reduce deaths in custody prevents it from publishing any other data.

The Justice Department says that this provision has actually impeded data collection. The reasoning gets bureaucratic and jargony fast, but the gist is that, by introducing this tool for compliance, Congress forced the department to assign data collection responsibilities to a bureau with less experience with this kind of work. Managing states’ grant funding is a policy issue. Since the Bureau of Justice Statistics is only supposed to conduct research, not do policy work, the Justice Department decided it could no longer remain in charge of state-level data collection.

But the first version of the DCRA was also tied to funding: States had to certify that they would follow data reporting guidelines to be eligible for so-called “truth-in-sentencing incentive grants.” And this data was reported to the experts at BJS.

Finally, the department has taken the dubious position that a provision mandating a single report to Congress on how to reduce deaths in custody prevents it from publishing any other data. It points to another section of law that says it can only release statistical information for “the purpose for which it was obtained.”

This restriction didn’t keep the department from publishing DCRA data under the old law, or in the interlude when there was no DCRA law. But now, the department has decided that the entire purpose of the DCRA is to facilitate the report to Congress — and since making data public wouldn’t help the report, it can’t release the data. This interpretation is easily belied by the fact that DCRA data collection is supposed to go on in perpetuity, even after that single report is done. If the whole point of the law was to produce a report, wouldn’t the collection end once the report is out?

Collecting data on deaths in custody is just a small step toward actually reducing those deaths.

We think the explanation isn’t that the 2014 law prevents the Justice Department from accurately collecting and reporting on the number of people who die in government custody. The more likely explanation is something the department itself admitted in a recent status report: It has chosen to read the law as narrowly as possible, and to not do anything it doesn’t absolutely have to.

According to the report, the department pared back its more ambitious plans in favor of “provisions specifically required by the statute.” This scaling back meant collecting only the “[data] fields explicitly described in the DCRA of 2013 and exclud[ing] any efforts” to check the quality of the data it received. The department wants to blame Congress, but it made these choices.

Indeed, as those passages suggest, the department hasn’t always been so insistent on refusing to collect this data. In fact, it once proposed a far more comprehensive plan for DCRA data collection — one signed off on by then-Attorney General Loretta Lynch herself. Nothing about the law has changed since then, so there’s clearly room for the department to make better plans.

Credit where it’s due: The department recently committed to a few overdue improvements to its DCRA collection. But as long as it maintains its position that the law imposes these imagined limits on its efforts, the department is committing itself to failure.

Keep in mind, collecting data on deaths in custody is just a small step toward actually reducing those deaths. Without that data, though, policymakers will be deprived of the best possible information for identifying the most meaningful reforms.

It doesn’t have to be this way. This is an eminently solvable problem. We’re currently partnering with The Leadership Conference on Civil and Human Rights on an in-depth report on the sorry state of the DCRA and the path forward, so we’ll have much more to say in the coming months. In the meantime, we hope the Justice Department stops talking itself out of sensible solutions. It can make meaningful progress toward understanding and preventing deaths in custody — if it wants.