This month has been a harrowing month, even for those numbed by the steady drumbeat of violence in America. The public was exposed to video evidence of two police killings in successive days, with both victims—Alton Sterling and Philando Castile—being African American men. And then to video of a sniper attack on police officers in Dallas that left five of them dead. The former added to a long list of black victims shot out of apparent fear, even though they posed no objective threat. And the latter served as a reminder that policing in America can be dangerous, and that fear of injury or death is often warranted.
In an unusually frank statement, the Governor of Minnesota conceded that Philando Castile would probably still be alive if he were white. This is an admission that at least for some officers, fear of black male violence is out of proportion to the risks that they actually face. However, while videos can provide compelling evidence of unwarranted fear in individual instances, it cannot tell us if the situation would have been resolved peacefully if Castile were white. For this we need national data on encounters between police and citizens, including not only those in which deadly force was used, but also those in which suspects were successfully pacified and disarmed.
Unfortunately, such data remain unavailable. Federal statistics, compiled by the FBI and the Bureau of Justice Statistics,under-count the number of people killed by police by more than half. Data collection efforts by independent organizations such as The Washington Post and The Guardian provide more accurate information, by using information from news outlets, research groups, and open-source reporting projects. This has brought the enormous scale of police brutality and racial disparities into the spotlight. However, no database keeps track of nonfatal encounters at the national level, making it impossible to determine whether encounters with the police tend to unfold differently for blacks than for whites if circumstances are otherwise the same. FBI Director James Comey has called the lack of accurate data “ridiculous.”
Without a comprehensive national database, we can’t even determine the extent to which eliminating racial bias would reduce police violence against African Americans. It is uncontroversial that racial disparities exist. Of those killed by police in 2015, for instance, 27% were black. This is more than twice the share of the black population nationwide, but it is roughly comparable to the rate at which black citizens are arrested. This has led some to argue that it is the encounter rate rather than bias during encounters that is largely responsible for the disparity. But this argument is based on the questionable hypothesis that police encounters with blacks are as objectively threatening to officers on average than encounters with whites. The available evidence suggests otherwise: among those killed, black citizens are disproportionately likely to have been unarmed.
The existing data also do not allow us to assess the efficacy of different police practices and policies. For example, there is some evidence that residency requirements—which require police officers to live in the communities they serve—make the police force less rather than more diverse. However, that does not necessarily mean that eliminating residency requirements would reduce police violence. Residency requirements may improve community relations and regular contact between communities and the police may reduce stress and perceived threat on both sides.
Likewise, while body and dashboard cameras have the potential to increase transparency, provide evidence, and improve awareness, their effect on police violence is ambiguous. While the use of police cameras is associated with a reduction in police violence in some jurisdictions, it is not clear whether introducing body cameras nationwide will have the desired effect: perhaps police departments that chose to participate in a study have more positive attitudes towards body cameras than those who do not. Moreover, existing studies cannot tell us how police cameras affect bias and differential treatment.
To address these and other questions, comprehensive and accurate data is needed not only on police violence but also on its possible determinants. This includes local data on crime rates, poverty, and neighborhood conditions, and detailed data on police practices and incident reports, collected and distributed in a uniform manner. Studies released this past week show that using data on nonfatal encounters can provide valuable new insights. A report by the Center for Policing Equity shows that crime rates do not drive racial disparities in police violence, and an analysis by Harvard economist Roland Fryer finds an absence of racial bias in shootings by Houston police. However, both studies rely on the voluntary participation of law enforcement agencies, which could bias findings if participating agencies follow different practices and have a different culture than other departments.
Without reliable and accurate statistics, it is impossible to uncover the drivers of police brutality. And if we do not have a clear understanding of the problem, it will be much harder to find the right remedies, and to build consensus around implementing them.
Willemien Kets is an assistant professor of managerial economics and decision sciences at the Kellogg School of Management at Northwestern University and a fellow in The OpEd Project’s NU Public Voices Fellowship. Rajiv Sethi is an economics professor at Barnard College, Columbia, and a faculty member at the Santa Fe Institute.