The sociospatial ecology of deaths during police interactions in the United States, 2016–2020

Abstract

Who dies in police custody? Where? To answer these questions, we use spatially disaggregated georeferenced data that measure 9098 deaths occurring by multiple causes during interactions with police throughout the U.S. from 2016 to 2020. We use a Sociospatial Ecology framework and Bayesian statistics over U.S. counties that establishes the relationship between social contexts – regional poverty, White/non-White population, violent crime rates, and political identity – and the risk of dying during police interactions. In addition, we evaluate the effects of police force Whiteness on deaths during police interactions. Controlling for alternative explanations, we show heterogeneous distributions of fatality risk, with large clusters in the Southwest and isolated high-probability pockets in other states. Risk maps allowing for visualization of these patterns are provided. We arrive at five main results. 1) There is a general trend of higher death during police interaction in areas of high poverty, fewer White people, higher violent crime rates, and higher populations with conservative values. 2) A great risk of deadly encounters for Black people exists throughout most of the U.S., while regional patterns of high risk exist for all other people of color. 3) White deaths during police interactions are most sensitive to ecological factors. 4) The risk of Blacks getting killed by police increases in White areas regardless of violent crime rates. 5) Higher proportions of White police within U.S. counties leads to higher interactive death risk for all races/ethnicities except Asian/Pacific Islander. Ultimately, our findings identify widespread racial/ethnic biases in situations of power and control.