Eugene has kindly invited me to present findings from my new book, “The War on Cops.” The book argues that the Black Lives Matter narrative about a racist criminal-justice system is false and dangerous — dangerous to innocent black lives that are being lost to rising crime and, it would increasingly appear, to law and order itself as cops become the target of assassination.

In future posts, I will discuss the most controversial aspect of “The War on Cops”: my claim that the current rise in violent crime in many American cities is the result of officers backing off of proactive policing.

Today, however, I want to address the question of police shootings and race, in light of the recent fatal police shootings of Alton Sterling in Baton Rouge and Philando Castile in Falcon Heights, Minn. Those shootings have amplified the charge that the U.S. is experiencing an epidemic of racially driven police shootings. A New York Times editorial, for example, asked, “When Will the Killing Stop?” The chair of the Congressional Black Caucus, Rep. G.K. Butterfield (D-N.C.), asserted on the House floor that “most” of the people fatally shot by the police this year have been African American.

In fact, as of July 9, whites were 54 percent of the 440 police shooting victims this year whose race was known, blacks were 28 percent and Hispanics were 18 percent, according to The Washington Post’s ongoing database of fatal police shootings. Those ratios are similar to last year’s tally, in which whites made up 50 percent of the 987 fatal police shootings, and blacks, 26 percent. (The vast majority of those police homicide victims were armed or otherwise threatening the officer.) But Butterfield could be forgiven his error, given the virtually exclusive media focus on black victims of police officers.

Does the actual distribution of police victims confirm the Black Lives Matter allegation that policing is lethally biased? That depends on the benchmark chosen for assessing police actions.

Typically, activists and the media measure police actions against population ratios. Given that blacks are 13 percent of the nation’s population, a 26 to 28 percent black share of police gun fatalities looks disproportionate. But policing should be measured against crime rates, not population percentages, because law enforcement today is data-driven. Officers are deployed to where people are most being victimized, and that is primarily in minority neighborhoods.

In America’s 75 largest counties, comprising most of the nation’s population, blacks constituted 62 percent of all robbery defendants in 2009, 57 percent of all murder defendants, and 45 percent of all assault defendants — but roughly 15 percent of the population in those counties. In New York, where blacks make up 23 percent of the city’s population, blacks commit three-quarters of all shootings and 70 percent of all robberies, according to victims and witnesses. (Whites, by contrast, commit less than 2 percent of all shootings in New York City and 4 percent of all robberies, though they are nearly 34 percent of the population.)

New York City’s crime disparities are repeated in virtually all American metropolises. They will determine where officers are most often called to a drive-by shooting or an armed robbery, and where officers are most likely to face violent and resisting criminals — encounters which can lead to officers’ own use of deadly force. Police critics have never answered the question of what they think non-biased policing data should look like, in light of the vast differences in rates of criminal offending. Blacks commit homicide at nearly eight times the rate of whites and Hispanics combined. Black males between the ages of 14 and 17 commit gun homicide at nearly 10 times the rate of white and Hispanic male teens combined. Should police stops, arrests and those rare police shootings nevertheless mirror population ratios, rather than crime ratios? The answer is not forthcoming from Black Lives Matter activists.

In 2015, the police fatally shot 36 unarmed black males, according to The Washington Post’s typology, and 31 unarmed white males. The Post’s classification of victims as “unarmed” is literally accurate but sometimes misleading. The label can fail to convey the charged situation facing the officer who used deadly force.

At least five “unarmed” black victims had tried to grab the officer’s gun, or had been beating the cop with his own equipment. Some were shot from an accidental discharge triggered by their own assault on the officer. One had the officer on the ground and was beating him on the head so violently, breaking bones and causing other injuries, as to risk the officer’s loss of consciousness. And one individual included in the Post’s “unarmed black male victim” category was a bystander unintentionally struck by an officer’s bullet after an illegal-gun trafficker opened fire at the officer and the officer shot back. If a victim was not the intended target of a police shooting, race could have had no possible role in his death.

Police critics may reject the claim that a suspect was grabbing the officer’s gun, citing the discredited police narratives in the shootings of Walter Scott in North Charleston, S.C., in April 2015 and of Laquan McDonald in Chicago in October 2014. In the case of Walter Scott, ballistics and autopsy evidence would eventually have undermined Officer Michael Slager’s exculpatory story, even without the cellphone video. But if skepticism toward police narratives has now become routine, equal skepticism is warranted toward witness accounts of allegedly unjustified officer shootings, as demonstrated by the bystander hoax that Michael Brown was gunned down in cold blood by Ferguson, Mo., Officer Darren Wilson in August 2014. Whether one trusts officer accounts more than bystander accounts or vice versa will depend on one’s prior assumptions about the police and the community, unless and until there is a critical mass of such conflicting narratives resolved in one direction or the other.

The Post’s “unarmed” cases do not support the idea that the police have a more demanding standard for using lethal force when confronting unarmed white suspects than unarmed black suspects. According to the press accounts, only one unarmed white victim attempted to grab the officer’s gun. In Tuscaloosa, Ala., a 50-year-old white suspect in a domestic assault call ran at the officer with a spoon; he was Tased and then shot. A 28-year-old driver in Des Moines led police on a chase, then got out of his car and walked quickly toward the officer, and was shot. In Akron, Ohio, a 21-year-old suspect in a grocery store robbery who had escaped on a bike did not remove his hand from his waistband when ordered to do so. Had any of these victims been black, they would have stood a good chance of becoming household names; instead, they are unknown.

Of course officers do make lethal and sometimes criminal errors in judgment. The Castile and Sterling homicides may prove to be as horribly unjustified as the Scott and McDonald shootings; we do not yet have enough information to know. But those grotesque miscarriages of tactics and justice are a minute fraction of the tens of thousands of encounters with armed suspects and the nearly half-million arrests for violent felonies that officers make each year.

It goes without saying that every unjustified police shooting of a law-abiding civilian is a stomach-churning tragedy. Given the appalling history of racism in this country and the complicity of the police in that history, police shootings of black men are particularly and understandably fraught. Police training must work incessantly to eliminate all unjustified uses of police force and to make sure that officers treat everyone they encounter with courtesy and respect and within the confines of the law.

But contrary to the Black Lives Matter narrative, there is no government agency more dedicated to the proposition that black lives matter than the police. The data-driven, proactive policing revolution that began in the mid-1990s has saved tens of thousands of black lives that would have otherwise been lost to urban gun violence had crime remained at its early 1990s rate. Unfortunately, those crime gains are now at risk, thanks to the false narrative that police officers are infected with homicidal bias.

[Heather Mac Donald is the Thomas W. Smith Fellow at the Manhattan Institute and a contributing editor of City Journal.]