Last week, Michael Fumento had a fascinating, skeptical piece about the potential for armed civilians to stop mass shootings.

He makes a number of important points. Sometimes civilians do stop mass shootings, but this is pretty rare—about 3 percent of mass shootings in a study Fumento cites. Even when a civilian is armed, we can’t expect him to charge into danger rather than escaping with his own life. I’d add that civilians normally pack pistols, often “compact” or “subcompact” pistols at that, which have neither the range nor the power of the long guns often used by modern mass shooters. I’m slightly optimistic about increasing the percentage of these shootings that are stopped—the population of concealed-carry holders is growing at an amazing rate, and we could try to cut back on “gun-free zones” (though bars and nightclubs are a special case). But we are starting from an incredibly low baseline.

What especially got me thinking, though, was Fumento’s claim that many people will be psychologically unable to pull the trigger on a fellow human being. To support this point, he cites Dave Grossman’s controversial On Killing. Grossman argued that through World War II, most soldiers in combat couldn’t even aim at the enemy; it was only the advent of modern military training that changed this. Fumento notes his experience killing animals with his bare hands during his own training.

The debate over Grossman’s work is often really a debate over the work of S.L.A. Marshall, a World War II historian. While Marshall isn’t Grossman’s only source, he’s mentioned by name more than 50 times. Marshall claimed that only a small minority of World War II soldiers actually shot at the enemy, citing interviews he conducted with American troops. Critics allege that Marshall didn’t even collect the data he would need to make that claim, and the Canadian historian Robert Engen said he found, if anything, the opposite when he consulted interviews of World War II soldiers from his own country.

But my goal here isn’t to assess Grossman’s case as presented. What occurred to me is that we have a new source of data that can shed light on this question: the numbers collected by the Washington Post on killings by American police. Is there any sign that officers are refusing to shoot when they are threatened? Grossman’s theory predicts that there should be. Police are trained, obviously, but many say the training is poor, and at the very least I doubt many cops are required to kill animals with their bare hands. And Grossman claims that “throughout history the majority of men on the battlefield would not attempt to kill the enemy, even to save their own lives or the lives of their friends.”

In total, American police killed 990 civilians in 2015, according to the Post. 783 of those people were armed with a deadly weapon, not including vehicles or toy guns that may have appeared real. (I should note that a suspect’s being unarmed doesn’t prove that there was no threat to the officer’s life—I’m just trying to err on the conservative side.) Some of those people, in turn, might have been committing “suicide-by-cop,” meaning they were acting threateningly but were not actually homicidal. More than one-quarter were known to have been mentally ill; this is a rough proxy, but it’s in line with the other estimates we have of suicide-by-cop. Among those who had deadly weapons and were not known to be mentally ill, 491 were killed with an “Attack in Progress.” (The other options are “Other” and “Undetermined.”)

Compare that with 52: that’s the number of officers who were killed feloniously in 2015. Even with a cautious estimate of true threats against police, cops killed more often than they got killed by more than nine-to-one. Maybe you can exclude even more cases where civilians were killed than I did, but that’s a lot of cushion before we get to Grossman’s strong majority in the opposite direction—and we could also remove some cases from the other side of the ledger, such as those where an officer did not have time to fire (e.g., six officers died in ambushes in 2015) or where the officer actually fired but didn’t eliminate the threat (one admittedly dated study found that 15 percent of slain officers were able to discharge their weapons first).

Concealed carry won’t solve mass shootings. But humanity’s innate reticence to kill might play a pretty small role in that.

Robert VerBruggen is managing editor of The American Conservative. Twitter: @RAVerBruggen