Last weekend the world learned of the U.S. air attack on the MSF (Médecins Sans Frontières or Doctors Without Borders) hospital in Kunduz. Unfortunately, research that I’ve been part of, which I’ll detail here, suggests that we should be shocked but not surprised. Air attacks compete with mortars to be the most indiscriminate weapon in recent wars. And  there is a long and sad history of Afghan civilians killed in NATO airstrikes including this one, this one and this one.

Which weapons systems are the most indiscriminate? See below.

In 2011 I was part of a group that published a paper using Iraq Body Count (IBC) data to classify the range of lethal effects on Iraqi civilians caused by the various weapons systems used there (see also this). The table below shows some pertinent comparisons drawn from this work.

As you see, women and children account for a high percentage of all civilians killed by air attacks and mortars (at least according to the data we have). Their civilian victims almost resemble a random draw from the whole population, with women nearly as numerous as men, and children outnumbering both men and women. From this perspective, air and mortar attacks seem remarkably indiscriminate.

Guns, in contrast, appear to be far more discriminating, or at least their bearers are. In a war, shooters seem reluctant to pull the trigger with women or children in their sights, overwhelmingly shooting adult males.

I can’t think of any reason why people carrying out airstrikes would intentionally target women and children more than shooters do. Rather, I think that airstrikes are relatively indiscriminate, taking down a fairly random cross-section of victims regardless of intentions.

The data suggest that when a military chooses to carry out an airstrike, it implicitly expresses a relatively high tolerance for killing women and children. After all, killing women and children is a predictable consequence of an airstrike.

That’s one of the main reasons my colleague Madelyn Hsiao-Rei Hicks and I created the Dirty War Index (from which these tables are drawn): to focus attention on predictable consequences of choosing a particular weapon, rather than on the intentions — whether benign or otherwise — of the militaries that operate the weapons.

Our IBC paper is loaded with caveats. IBC does not have a comprehensive list of all civilian deaths. More important, it only has information on gender and age for around one-third of the deaths we analyzed. The true percentages could differ noticeably from the percentages given in the table above.

Nevertheless, the differences between air attacks and mortars, on the one hand, and small arms gunfire, on the other, are large. It is very unlikely that better information would substantively overturn the sharp contrasts you see.

Last week a similar paper came out using the data of the Center for the Documentation of Violations in Syria (VDC). The table below is the Syrian analog to our chart on Iraqi civilian deaths. The pattern is strikingly similar to what we found for Iraq, although less pronounced.

Again airstrikes and mortars appear to be the least discriminating weapons — or to put it differently, the ones that predictably kill women and children.

Some observers have, I believe, misguidedly emphasized U.S. intentions in the MSF attack. But there may be no explicit evidence proving that military decisionmakers knew about the MSF hospital and ordered airstrikes anyway. If so, the focus on intention could make it easier for U.S. officials to return to business as usual.

Focusing on the predictably indiscriminate nature of airstrikes and mortars would make it easier to call those ordering the use of such weapons to account.

If airstrikes continue to be used, we can safely predict another devastating attack on civilians sometime soon.

Michael Spagat is head of the economics department at Royal Holloway, University of London. He blogs at War, Numbers and Human Losses.