Bad Watchdog Season 2 launches June 20.

Policy Letter

POGO Urges Improvements to Justice Department Death Data Collection

(Illustration: Renzo Velez / POGO; Photo: ajmexico / Flickr, CC BY 2.0)

Laura Wyckoff
Bureau of Justice Assistance
810 Seventh Street NW
Washington, DC 20531

Re: Agency Information Collection Activities; Proposed eCollection eComments Requested; Death in Custody Reporting Act Collection

The Project On Government Oversight (POGO) is submitting this comment regarding the Death in Custody Reporting Act (DCRA) collection notice published on June 28, 2021.1

POGO is a nonpartisan independent watchdog that investigates and exposes waste, corruption, abuse of power, and when the government fails to serve the public or silences those who report wrongdoing. We champion reforms to achieve a more effective, ethical, and accountable federal government that safeguards constitutional principles.

As we outlined in a letter to Attorney General Merrick Garland in July,2 we are concerned that the Bureau of Justice Assistance (BJA)’s program for collecting state and local data on deaths in custody has several shortcomings that will likely prevent it from providing an accurate account of deaths in custody or insight into how to prevent them. Accordingly, we urge the bureau to use a more detailed and standardized collection form and to take additional steps to ensure the completeness of collected data. We also urge the bureau to proactively develop and publicize its plan to make DCRA data, and analysis of that data, available to the public.

A More Detailed and Standardized Collection Form Is Necessary

The collection notice indicates that BJA has not improved Form DCR-1A.3 As we wrote in our letter, the most recent version of the form is inadequate to ensure that the bureau will receive sufficient details about death in custody incidents, and it lacks the standardization necessary to allow for meaningful analysis.

In addition to providing the decedent’s name and time and date of death, agencies using the form must complete nine data fields: Four relate to basic demographic information about the decedent; four relate to the facility and agency with custody; and one relates to the cause and circumstances of the death. While all of these fields are important for discerning patterns in deaths in custody, the last is likely to elicit some of the most important information. Unfortunately, it is the worst-designed. The rest of the fields use either checkboxes or fill-in-the-blank lines for specific information. The final field has checkboxes to indicate one of seven general causes of death (including “unavailable, investigation pending” and “other”). It then contains a prompt to enter a “brief description” of the circumstances in plain text.

At best, the design of this final field may lead to confusion. The prompt for a text description is presented on the form as an eighth checkbox, giving the impression that checking one of the seven preceding boxes eliminates the need to provide any further description. This would seriously reduce the amount of information the department receives.

Moreover, relying solely on an open-ended text-entry field for information about the circumstances of a death leaves too much discretion to the agency responsible for filling out the form and introduces another source of potential bias and inaccuracy into the analysis. Different people may have different interpretations of what details are important, and agencies that may wish to hide or gloss over information can use the discretion inherent in the text field to do so. Even if the information provided is complete, text-based data is much harder to analyze; in order to compare different events and determine trends, bureau staff will have to sort the text information into categories that can be analyzed mathematically. In addition, a method using an expansive text field is also likely more time consuming for those filling out the forms.

Improving the collection form does not need to be an onerous process: The form used by the Bureau of Justice Statistics for its federal DCRA collection provides a good template.4 It utilizes significantly more checkbox data fields, which should result in more granular data being submitted in a standardized and readily analyzed format.

The Collection Program Should Include Data Improvement Measures

The current collection program lacks data quality-control measures, which we have noted in multiple communications to the department.5 In 2016, BJA’s proposed collection methodology would have required reporting agencies to submit plans for improving the quality of their data, and it would have checked submitted data against public sources, drawing from a methodology used by the Bureau of Justice Statistics (BJS).6 However, that methodology was withdrawn in 2018 and replaced by the current program, which lacks both safeguards.

We are concerned that the current methodology will yield significant undercounts, an issue that is not hypothetical. In 2014 BJS had to suspend a similar program, the Arrest-Related Deaths Program, due to data problems. A subsequent analysis suggested the program had only captured half of all homicides by law enforcement.7 Since monitoring and preventing homicides by law enforcement were Congress’s primary motivations for passing DCRA,8 and remain two of the most pressing policy issues facing the country today, continuing with a methodology that has been proven inadequate in the past is unacceptable. BJA should restore the safeguards from its 2016 collection proposal.

Ensure Public Access to Selected Data

For the information collected under DCRA to truly foster accountability and inform meaningful reforms to reduce deaths, the public must have timely access to the data. However, the department’s commitment to facilitating this access has been uneven at best. During the prior administration, guidance from BJA indicated that data collected under DCRA would only be released in response to Freedom of  Information Act requests.9 We were heartened this spring to see that guidance replaced by a statement that federal privacy law allows the department to release non-personally identifying information and that the Office of Justice Programs does publish “general data related to deaths in custody.”10 However, neither of those statements represents an affirmative commitment from the Office of Justice Programs or BJA to release data from the state and local collection program in the future. We urge the department to make that affirmative commitment.

Crucially, public access must be timely. It took until the end of 2020 for BJS to release its report on DCRA data for fiscal years 2016 and 2017. When it did release the data, it was in the form of a highly aggregated analytical report.11 Delays of this magnitude render the data nearly useless for informing public consideration of reforms. To speed public access, non-identifying data could and should be made available to the public before the department’s analysis is complete. Notably, using a more standardized collection form, as discussed above, would likely make it easier to make certain types of data available more quickly.

DCRA can be an important tool to shed light on a crucial issue. However, BJA’s implementation has prevented it from serving this purpose to date. Over a year since the bureau started collecting data under the act, it is time to take these straightforward steps to ensure the data it receives is of the highest possible quality, is in the most readily useable form, and is made available to the public as quickly as possible. If you have any questions or wish to discuss these recommendations further, please do not hesitate to contact Brandon Brockmyer ([email protected]) or David Janovsky ([email protected]).


Brandon Brockmyer
Director of Research

David Janovsky