As the argument goes, we put guards with guns in banks to stop thieves, so why not use that same strategy to protect our children? After all, we cherish these young lives far more than we do our bank deposits, don’t we?

There is a gut appeal to such an argument to be sure—and if only the facts agreed with it, it might be persuasive.

But they don’t. Indeed, the facts (and by “facts,” I mean the real facts) counter this line of reasoning in three clear-cut ways.

First off—and here’s a shocker—banks hardly use armed guards any more. Despite virtually every bank robbery movie you’ve ever seen, most banks and bank branches are guardless nearly all of the time. We can see that from looking at one key subset of banks—um…those that have been robbed.

In 2016, there were 3,733 commercial bank robberies across the U.S, according to the FBI’s own bank crime statistics. (There were also an additional 400-plus robberies at credit unions, S&Ls, and mutual savings banks.) Lest you think this subset is a carve-out of just the wimpy banks, or the careless banks, or the robbery-prone banks—and that’s why they were marked…it isn’t. Rather, this huge sample is fairly representative of the U.S. commercial bank population on the whole—which totaled under 4,900 in the latest quarter, according to the Federal Reserve Bank of St. Louis.

So what can we learn from these incidents? Start with this: In only 224 of them did the “victim institution” even have a policy of hiring security guards and in just 177 of the cases was a guard actually on duty at the time of crime (see the bottom of page 2 in the FBI stats).

Why? Well, that brings up the second surprise: The presence of armed guards tends to dramatically increase the likelihood of violence and injury during a robbery, as the Center for Investigative Reporting found when it analyzed detailed FBI crime incident reports from 2007 to 2011—which the group got through a Freedom of Information Act request. In robberies where there were no guards present, there were violent events (such as the firing of a weapon or injury of a customer or employee) in 4% of cases. When an armed guard was on duty, that rate more than tripled, to 12.8%.

So what’s surprise No. 3? Armed guards aren’t nearly the kind of deterrent that many think they are. For evidence, consider the most armed and armored kind of bank repository there is: the armored car. These same FBI statistics paint a fairly frightening picture here too. In 2016, no fewer than 36 of these tank-like vehicles were robbed at gunpoint (see pages 9 and 10).

In these assaults, three people were killed: Two of them, sadly, were guards.

We can’t afford those kinds of odds at our schools.

This essay appears in today’s edition of the Fortune Brainstorm Health Daily. Get it delivered straight to your inbox.