To begin, we have to start with the raw data.
For one, the Washington Post is operating a real-time database to track fatal police shootings.
In 2015, they logged 990 fatal police shootings, of which 494 of the slain were white and 258 were black.
This year, 518 people have been killed by police (a six percent increase from the first 6 months of last year), 239 of those killed being white and 126 being black (a nine percent increase from the first 6 months of 2015).
Are more white people being killed by police than black people? Yes.
However, when you take population into consideration, as any responsible analysis of the data must, black people, making up 13 percent of the population, make up 24 percent of the fatal shootings, meaning, generally speaking, black Americans are more 2.5 times more likely to be killed by police.
When you adjust for unarmed victims "black men were seven times more likely than white men to die by police gunfire while unarmed", when taking into account the 2015 data alone.
After the data for 2015 and 2016 is compiled, "U.S. police officers have shot and killed the exact same number of unarmed white people as they have unarmed black people: 50 each. But because the white population is approximately five times as great as the black population, that means unarmed black Americans were five times as likely as unarmed white Americans to be shot and killed by a police officer."
So while higher numbers of white people (and Hispanics) are killed by police than black people, black people, armed or unarmed, are killed a rate higher than their white counterparts.
There are two prevailing theories for why this is: the racial bias theory and the crime-produced disparity theory (which states that disproportionate levels of crime in the black community drives police use of force).
The racial bias theory makes a lot out of the above disparities and there is indeed evidence on the books showing racial bias against blacks in terms of police dealings.
Most recently, a study published by National Bureau of Economic Research found "[o]n non-lethal uses of force, blacks and Hispanics are more than fifty percent more likely to experience some form of force in interactions with police."
Importantly, the study controls for things like context and civilian behavior finding it "reduces, but cannot fully explain, these disparities."
This makes the data difficult to dismiss by those who hold the crime-produced theory as a explanation for these police conflicts.
Furthermore, mitigating against the crime-produced theory is 2015 study coming from a researcher at the University of California's Anthropology Department.
The study an ,"Analysis of Racial Bias in Police Shootings at the County-Level in the United States", found the the following:
"There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates."
Even when controlling for violent criminality on the part of the person shot, a Center for Policing Equity report released this year found, "[g]iven the rarity of Part I violent crimes [violent and property crimes] and a lack of evidence that arrests for violent crime significantly increase the likelihood of police use of force, these findings suggest that crime rates are an insufficient explanation for disparities in the application of police force."
In other words, the crime-produced disparity theory is whittled down as studies consistently show that arrests for violent crime and local crime rates are not enough to explain the disproportionate killing of black individuals.
This is bad news for people like Heather McDonald, fellow at the Manhattan Institute, who claimed in her popular "The Danger of Black Lives Matter" speech, "[t]he black violent crime rate would actually predict that more than 26 percent of police victims would be black. Officer use of force will occur where the police interact most often with violent criminals, armed suspects, and those resisting arrest, and that is in black neighborhoods."
In other words, while Ms. McDonald is absolutely correct that black people commit violent crime at rates not in keeping with their size relative to the overall population, the conclusion (given without sources) she draws from that directly contradicts years of historical data compiled and evaluated in the CPE report.
Furthermore, her implication that the disparity at which cops are in danger of being killed by black people more than any other racial demographic helps explain the racial disparity in cop killings fares no better without any support from the published data.
This does not mean the racial bias theory wins the day, however.
The theory recently hit a major snag with a new report that shows "on the most extreme use of force – officer-involved shootings – we find no racial differences in either the raw data or when contextual factors are taken into account."
This is the same National Bureau of Economic Research study that found racial bias in police use of non-lethal force.
German Lopez, a staff writer for Vox, criticized the report for using voluntary data (which may be negatively affected by selection bias), which is not an immaterial point.
However, in the same article, he then admittedly used voluntary police data from the FBI to prove the opposite point, justifying his inconsistency by stating the FBI numbers are "more comprehensive".
This is true, but the comprehensiveness of a report does not mitigate against possible selection bias.
Also, the NBER study cannot be invalidated by the FBI statistics Lopex cites because the FBI numbers are not incident-based (see here) and so can say nothing about the controls in the NBER study that show a lack of racial bias.
In the second data set used in the NBER study, the researchers were given access to cases in which lethal force could have been used but was not, to look for potential bias.
And while Lopez claims this data set is "built on police reports of what police claim are arrests in which lethal force was warranted" and that "given the video evidence we’ve seen in the past couple of years, there’s good reason to not take police at their word", he ignores the fact that, as the New York Times reported, it was the lead researcher, Roland G. Fryer Jr., who set the parameters of such cases, not the police.
While some might bandy the idea that the police are lying in these reports, this is intuitively difficult to assent to because these are cases where lethal force was not used, eliminating any apparent reason for dishonesty.
And arguing the officers lied on the reports to make themselves look better in the off-chance someone would care that they could have used lethal force but didn't is just too clever by half.
Perhaps most important, as Mr. Fryer himself explains, "what we actually found was that there were no racial differences in the basic differences analysis. It didn't matter whether we took context--as captured by police reports--into account or not; there was no racial bias in either analysis."
The NBER study is not comprehensive and thus cannot be applied to the nation as a whole, but it is something in the way of rebutting the racial bias narrative.
One the other hand, while some may use the study as proof that minorities are generally handled rougher in the hands of the police, as mentioned earlier, the lack of comprehensiveness cuts this way as well.
Some have suggested that the more frequent stopping of black people by police leads to more violent confrontations.
While this would make sense, unfortunately that just pushes the question back a step and we are left to ask why police stopping are black people more often and the entire debate about racial bias vs crime-produced disparity begins anew (with very sparse data in this case).
Thus, it seems both proponents of the crime-produced disparity theory and the racial bias theory need to go back to the drawing board, and the rest of us should avoid making sweeping judgments apart from what individual cases tell us.
I have already argued elsewhere that the best approach is to not commit crimes, resist arrest, or evade arrest (this solution is, of course, not unique to me).
This is a variation of the crime-produced disparity theory, but argues that in initial police confrontations, not later police violence, of a lethal variety are mostly generated by crime, which all the evidence supports (one can check the WP database).
In addition, it also can accommodate theories of racial bias and thus functions as a way of keeping people safe and keeping officers accountable.
While I now believe this would apply to 99% of all the "hashtag cases" cases of police violence, it does not seem to be the case in the Philando Castile case, where an innocent man was mistaken for a criminal and lost his life at the hands of a scared police officer.
This, however, is not the norm, by any honest reading of the data, and affects the solution negligibly.