(May 23, 2017) After a spate of controversial police shootings, the Washington Post and Guardian newspapers in mid-2015 launched public databases to track and report on law enforcement-involved homicides. Meanwhile, the Treatment Advocacy Center released the only study to report the role of serious mental illness in such fatalities. The study found that individuals with severe psychiatric disease were 16 times more likely to die in an encounter with law enforcement than the general public.
In the course of unearthing relevant statistics for Overlooked in the Undercounted, we determined that only one federal agency - the Bureau of Justice Statistics (BJS) in the Department of Justice - systematically collected mental health data on individuals involved in deadly law enforcement encounters. Unfortunately for efforts to quantify the role of mental illness in such incidents, BJS had determined its methods of collecting accurate data on what the agency terms law enforcement homicides were so flawed that only about half of these fatalities were being identified.
"Despite their limitations, independent databases have proven to be so much more complete and accurate than any government source that a bizarre feedback loop of incomplete information has emerged," we wrote in Overlooked. "Because government agencies themselves lack complete and accurate data on fatal law enforcement encounters, they collect their data from media and other online datasets. But because media and online sources don't have definitive data from government, they generate incomplete data."
BJS announced it "suspended data collection and publication of the ARD data until further notice" and set out to overhaul its methods and improve its reporting.
New Methods, Better Data
The results are in based on the new approach, along with preliminary state reports, and they are illuminating. As previously projected by BJS, using previous methods, arrest-related deaths (ARDs)* were underreported by approximately 50%.
Identifying arrest-related deaths now begins with automated surveys of Google Alerts, media and open-source websites to identify potentially eligible cases and follows up with manual confirmation of all deaths with individual police departments and medical examiners/coroners, whose reporting of such deaths is voluntary.
Applied to the months from June 1, 2015, through March 31, 2016, the survey of unofficial sources identified 1,348 potential arrest-related deaths, many reported in multiple media reports. To validate and improve on those sources, approximately 150,000 news articles were manually reviewed and coded to train automated classification programs, and then 250,000 media articles were manually coded to run through the classifiers to determine their accuracy in identifying potential arrest-related deaths.
Findings developed with this approach included:
- The largest state, California, had the largest number of potential ARDs during the period
- The District of Columbia, Wyoming and New Mexico had the highest rate of potential ARDs per million residents (influenced in cases like Wyoming by the very low number of deaths); all exceeded 6.52 ARDs per million residents
- New Jersey, New York and Rhode Islands had the lowest rates of potential ARDs per million; all had less than 2.64 ARDs per million population.
State-by-State Data
The findings from the media survey were then distributed to the chief executive in each state and local law enforcement agency involved with the 379 potential ARDs from June through August 2015, along with the chief medical examiner/coroner in the jurisdiction of the agency. The agencies were asked to complete incident reporting forms with information associated with the decedents and the incidents, including whether the death occurred during "mental health assistance."
A total of 75% of the agencies responded with some information, and 72% of those provided complete responses. Open investigations may have depressed the reporting rates, the BJS technical report said.
Of the 379 potential ARDs, four out of five were confirmed by either law enforcement or the medical/coroner offices, and slightly more than half were confirmed by both. Sixty deaths were confirmed by one agency but denied by the other. Slightly more than one in 10 of the deaths were reported by the agencies but not initially found in media reports.
More notable than the disparity between law enforcement and medical officials on whether a death was arrest-related was the disparity between how many such deaths the media reported and the relevant agencies had previously volunteered.
A total of 29 states reported no ARDs at all from June-August 2015, while the media reported deaths in all of them but Alaska. Florida, for example, reported no ARDs, but media reported 25 in the state.
Seven additional states reported only one death where media reported multiple deaths, and official underreporting was common. California officially reported four ARDs, for example, but the media reported 60, while Texas reported 15 ARDs while the media reported 43.
All the state statistics are published online in the technical report "Arrest-related deaths program redesign study, 2015-16: Preliminary findings."
The role of mental illness in the ARDs is not included in the technical report. However, the 21st Century Cures Act enacted in 2016 included provisions for collecting and reporting such data, a recommendation that was made in our 2015 study. Now that the BJS has developed a thoroughgoing approach to collecting ARD data overall, data-driven information about the role of mental illness in law enforcement homicides becomes attainable.
Doris A. Fuller
Chief of Research and Public Affairs
*Arrest-related deaths are defined by BJS as all deaths attributed to any use of force by on-duty law enforcement personnel and any death that occurs while a decedent's freedom to leave is restricted by law enforcement before, during or following an arrest.
References:
- Banks, D., et al. (December 2016). Arrest-related deaths program redesign study, 2015-16: Preliminary findings. US Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
- Fuller, D.A., et al. (December 2015). Overlooked in the undercounted: The role of mental illness in fatal law enforcement encounters. Treatment Advocacy Center.
Follow ORPA on Twitter at @TACResearch.
Research Weekly is a summary published as a public service of the Treatment Advocacy Center and does not necessarily reflect the findings or positions of the organization or its staff. Full access to research summarized may require a fee or paid subscription to the publications.
To comment, please visit our Facebook page.