Following the fatal shooting of Michael Brown in Ferguson, MO, there was renewed focus on the prevalence of police shootings in the United States. Yet, as was discovered by The Washington Post and scholars everywhere: No federal agency keeps track of this information, and everything the FBI does maintain is limited to raw numbers on “justifiable homicides.” In addition to being self-reported by police (red flag!), this purposely excludes instances when suspects were non-fatally injured or simply shot at and missed. This severely undercounts all instances of police involved shootings (PIS).

Yet, in a highly commendable move, the Dallas Police Department (DPD) released its self-monitored spreadsheets on all PIS incidents from 2003 to 2014 as part of an effort to improve police-community relations. Amazingly, the data includes demographics not only on the victim but also the officer(s) involved. Rightfully so, it was picked up by local and national media with calls that more departments follow the DPD’s lead.

When this release was brought to my attention, the first thing I noticed is that for as rare and rich the DPD’s dataset is, few journalists have actually played with it. The three visuals that exist, which have been copied and shared widely, were made by the DPD themselves — and they’re each pretty similar. Here’s the one that tracks the disposition/outcome of all PIS.

By itself, this is a step forward in that it highlights the deficiencies of monitoring only those instances when the suspect is killed (via “justifiable homicide”). In Dallas, ignoring all cases of injury or shoot and misses would eliminate nearly 2/3 of the data. The fact that this is done elsewhere is egregious and unacceptable.

For as much as I’d like to praise the DPD, though, there was something else that stood out to me. You’ll note in the above that race appears nowhere, even though for researchers interested in this topic, it’s a key component. This is also the case for the DPD’s other visuals (not published here). To simply say that X suspects were involved in police shootings obscures patterns important to policy makers. So, to fix this, I’ve taken the disposition information and broken each down by race. (Note: In my own analyses I’ve removed all females and Asians because they are practically nonexistent in the data).

Identifying the race of the suspects presents a whole different picture. For each year in each disposition, black males nearly constitute a majority. This is despite the fact that, according to the U.S. Census, blacks make up only 25% of the city’s population. Next, because the data also includes whether the suspect had a weapon (and what type), we can pull out the race and dispositions of those who were unarmed.

From 2003 to 2014, nearly half of all unarmed men killed by Dallas police were black. Blacks also formed nearly two-thirds of all unarmed shoot-and-misses and injuries. This clearly is not a coincidence. (And no, I’m not the first person to suggest that police shootings have a disproportionate effect on the black community).

I want to encourage more journalists and scholars to use this data because until the FBI’s reporting system changes and more police departments release PIS data, we have to squeeze as much as possible out of what’s available. As a researcher at the Initiative on Neuroscience and Law, I know there are plenty of interesting questions to be asked and answered: What’s the relationship between the disposition and weapon and race of the suspect? What effect does the officer’s race have on the disposition? And lastly, What crimes were these suspects allegedly committing?

So come on, grad students, get moving.

---

The full DPD dataset can be downloaded from Github (here), though it lacks 2013 data, which will need to be entered manually (or scraped) from the DPD website.