Black lives matter. And it turns out the police agree.

There’s a big footnote here, though.

In officer-involved shootings in … 10 cities, officers were more likely to fire their weapons without having first been attacked when the suspects were white. Black and white civilians involved in police shootings were equally likely to have been carrying a weapon. Both of these results undercut the idea that the police wield lethal force with racial bias…

[I]n the arena of “shoot” or “don’t shoot,” Mr. Fryer found that, in tense situations, officers in Houston were about 20 percent less likely to shoot suspects if the suspect were black. This estimate was not very precise, and firmer conclusions would require more data. But, in a variety of models that controlled for different factors and used different definitions of tense situations, Mr. Fryer found that blacks were either less likely to be shot or there was no difference between blacks and whites…

And intriguingly, he found that the rise of mobile video did not substantially change the results in Houston. Racial gaps were about the same in years when iPhones and Facebook were prevalent and in years when they weren’t.

The author, an economics professor at Harvard, calls it “the most surprising result of my career.” The fact that cops kill more white suspects than black ones (twice as many, in fact) was already well known but the significance of that is disputed. On the one hand, it’s no surprise that they’d kill more whites: There are many, many more white Americans than black ones. Much more than twice as many, which means that police are killing black suspects at a higher rate than you’d expect given the percentage of blacks that compose the total population. In fact, when it comes to unarmed suspects, a WaPo analysis conducted last year found the number of whites and blacks killed by cops to be identical, another disproportionate result. On the other hand, blacks are charged with a much higher percentage of robberies, murders, and assaults than you’d expect per their share of the population. If cops responding to reports of violent felonies find themselves confronting a disproportionately high number of black suspects, the disproportion in shootings of black suspects might be explained.

The wrinkle added by the Harvard study is the possibility that blacks might actually be less likely to get shot in a situation where lethal force could be used than white suspects are. Why would that be? Before you answer, consider the big footnote: When it comes to nonlethal force, cops are indeed more likely to get rough with black suspects than white ones.

Why would a cop be more willing to push a black suspect around more than a white one but not more likely — maybe even less likely — to shoot him? One theory, per the data above on violent felonies, is that cops respond more aggressively because they fear, justly or unjustly, that black suspects will react more aggressively. The professor’s theory is that cops know that killing a black suspect is apt to mean lawsuits, protests, and possibly criminal charges a la the Freddie Gray case in Baltimore but pushing a black suspect into a wall is nothing anyone will get too excited about. That may also help explain why black and Latino cops were more likely to fire at black suspects, per a different study, than white cops are. The racial dimension of a minority officer killing a minority suspect is far less fraught than it is if a white cop does it. (Of note: The Minnesota cop who shot Philando Castile was Latino, not white.) “Black Lives Matter” and other civil rights groups may have succeeded in raising awareness of white-cop-black-suspect killings to the point where white cops are willing to cut black suspects a break in lethal situations that white ones don’t get.

There’s another possibility. If cops are more likely to categorize encounters with black suspects as potentially requiring lethal force than encounters with white suspects, that might help explain the lower rate of shooting blacks even though it suggests some anti-black bias. For instance, if an officer is more likely to judge a black suspect’s movements as suspicious, an encounter that might be routine with a white suspect might seem dangerous with a black one, e.g., if a black driver made a sudden move to grab something from under his seat and it turned out to be his driver’s license. An officer might reach for his gun in alarm in that case but not fire; that would push the stats on shooting of black suspects in “dangerous” situations down further, but it would obscure the question of whether cops are unfairly more likely to see black suspects as dangerous in the first place. (Which might also explain why they’re more likely to use nonlethal force against them.) In any case, file this study away as a narrative-challenger for the next BLM protest.