The killings of young, often unarmed black men by white police officers have been heatedly discussed in recent years. Since Michael Brown’s killing in Ferguson, Mo., in 2014, which amplified the Black Lives Matter movement that began after Trayvon Martin’s killing in Florida, killings by police of young black men have generated more attention — and have increased tension between police departments and citizens.

One difficulty in understanding these killings — and thus figuring out how to prevent them — is that we lack data on their occurrences. Until August 2016, police departments did not have to report officer-involved deaths. While the Justice Department is now collecting this data, that effort is too new to have yielded much insight.

To offer more insight, our recent research draws on a new data set of every confirmed, line-of-duty killing by police of a civilian in 2014 and 2015.

How we collected our data

We started by drawing from, a Web-based repository of police killings that is maintained by a citizen. KilledByPolice provides links to media stories that report killings. We then independently verified all the reported killings by cross-checking them with other data sets that included Fatal Encounters, the (now defunct), police department websites and local media reports. Through this process, we identified 860 verified, nonaccidental police killings in 2014, and 1,092 in 2015.

We started with 2014 so that we could see whether Michael Brown’s killing increased attention to the issue, which is why we did not draw data from The Washington Post’s database, which starts with 2015. But we are currently bringing our data set to the present and are drawing on The Post’s extensively reported data in that effort.

For each killing, we gathered data on the person killed, the police officers involved and the circumstances. Because we rely on media reports for most details of these killings, the data are incomplete. Many of the reports lacked the details we sought. Sometimes the reports did not include the officer’s name, thus making it impossible to ascertain race, years of job experience and the like. Sometimes they did not include whether the suspect was armed.

Nonetheless, we believe our data are the most complete and accurate that exist.

What we found

Of the people killed by police in 2014 and 2015, 51 percent were white, 28.1 percent were black, 19.3 percent were Latino, and 1.7 percent were Asian. The remainder came from other ethnic backgrounds.

Most people killed by police were white. In 2014 and 2015, white people made up about 62 percent of the U.S. population and are underrepresented in this group. Meanwhile, blacks made up 17.9 percent of the country and are dramatically overrepresented. In other words, African Americans are disproportionately more likely to be killed by police than white people. Latinos also are overrepresented in data on killings by police, making up 17.6 percent of the population but 19.3 percent of these deaths.

What’s more, African Americans and Latinos killed by police were, on average, younger than white people who were killed by police. African Americans killed by police were an average of 30 years old, Latinos were 31, and whites were 39.

Understandably, national discussion has focused on the killings by police of unarmed civilians, but fewer than 1 percent of the killings we found were of people who were unarmed. We found that 65 percent possessed a firearm during an encounter with police. The rest were armed with other weapons, such as knives, bats and so on.

We should note that we couldn’t always tell from news accounts and public records whether a person was armed, so these figures come from the 83 percent of officer-involved deaths in which we could.

We collected information about the race of people who were killed by police, but also the race of the officers. We were able to identify the race of officers in 68 percent of the cases. We found that 59.2 percent of people killed by white police officers were white; 28.2 percent of people killed by white officers were African American. Nonwhite officers were significantly more likely to kill nonwhite citizens, especially Latinos: 33.7 percent of the people killed by nonwhite officers were African American, and 32.6 percent were Latino. This disparity is probably driven by urban police departments’ efforts to deploy nonwhite officers in nonwhite neighborhoods.

Law enforcement officers of all races disproportionately killed African Americans.

Among the unarmed people for whom we have full data, 50 percent were African American, 25 percent were white and 25 percent were Latino.

Many law enforcement strategies over-police African American neighborhoods

Many public policies, especially criminal justice policies such as the war on drugs, are designed to over-police predominantly poor African American communities.

Our findings join other scholarship, including research on stop-and-frisk and “driving while black,” that suggest that racially disproportionate killings by police are the result of institutional approaches to policing rather than individual racist police officers.

In other words, a major reason that young black men are disproportionately likely to be killed by police is because police are disproportionately likely to come into contact with young black men. Reducing these deaths may require different approaches to policing in the United States.

Logan Strother is assistant professor of political science at Purdue University.

Charles Menifield is dean of the School of Public Affairs and Administration (SPAA) at Rutgers University at Newark.

Geiguen Shin is a postdoctoral research associate at SPAA Rutgers University at Newark.