Survivor’s Bias and the Nintendo Generation

Military service in the United States is not exactly a low impact, safe, employment option. There are rigorous physical demands at times, and often inconsistent levels of appropriate medical care for injury prevention and recovery.

So when I see clickbait headlines like this:

Well, it pisses me off because that is a damn good example of “survivor bias.”

What is the “normal rate” for a duty limiting injury per Soldier per Year? Well, according to the Army, across the whole military, is right at 50%. Every year you serve, you are a coin toss from being so broke you can’t fully do your job.


So if half the military is getting sprains and stress fractures, at a rate of 50% per year, what does that tell us about initial entry training? By law, all initial military training must be no less than 12 weeks of actual training in order to be deployable, and if we add in one week for “admin” (such as showing up for reception) we can safely say that Initial Military Training is at least 25% of a year. So if 50% of Soldiers get an injury year over year, then we expect the “normal injury rate” for Initial Military Training to be right at 12.5% of Trainees get sprains, tears, stress fractures to the point where they cannot continue training.

Now…here’s the interesting part. The 50% of injuries are only the ones we know about, reported through medical channels. The actual injury rate is much higher because the military medical system isn’t designed to keep you healthy, it’s design to patch you back up only when you are so broke you can’t do your job. The good news is that it’s hard to hide a duty limiting injury in basic training, so we generally have “good data” for that. The table below was published in 2016, so relatively “Nintendo Generation” data.

Ignoring the “overuse” category for now, the acute category looks only slightly elevated over the Army normal rate of 12.5% expected, until you get to the Cavalry numbers which are actually concerning. But, if you know anything about the combat arms, you know those poor 19Ds are doing all the Infantry and most of the Armor training tasks, so it makes sense they would have more injuries than both of those groups.

Now…these injury numbers are right in line with the Australians.

Results: Injury prevalence for ASC recruits was 17.8% and for ARC recruits 13.9%. Injury incidence for the ASC and ARC were 17.8 injuries/100 soldiers/100 days and 17.4 injuries/100 soldiers/100 days respectively. The majority of injuries for both courses were sprains and strains.

This is a pretty good indicator that US military training is in line with expectations, as far as producing injury in our volunteers.

So where does this “survivor bias” come in? Well it comes in because by the time you get to the Pentagon, you are the very definition of a “survivor” in military terms. The Pentagon is where the Majors make the coffee and the Lieutenant Colonel is the power point slide monkey… So when they see an injury rate in initial military training, it’s super easy to think, “damn kids these days with their computer games and pale skin” while forgetting that by the time they made E8 or O5, that more than HALF their cohort has left the service. Less than 1 in 5 Soldiers serves to 20 years or beyond, so we expect that as your length of service increases, your time spent “utterly broken” to decrease as well (not that it will disappear mind you, as there are old Paratroopers and young office clerks).

So without hard data, looking at actual number over actual years, a Pentagon spokesperson saying that they are concerned about the bone density of American teenagers is about as tone deaf as you can get.

This entry was posted in defense, history. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s