the other problem with risk

I mentioned once that there was something important to say about risk and perception at the individual level versus the population level. Let’s try saying it this way:

Every time I put my shoes on, my dog gets excited.

“No, Mr Pants, I’m just going to work. What’s the matter with you?” I say.

Why does he get excited when I put my shoes on? I mean, he only gets to go out, oh, definitely less than half the time that I put my shoes on.


Me putting my shoes on happens EVERY TIME he goes out. Only my touching the leash is a better predictor of getting to go out.

So from one perspective, my putting my shoes on only predicts his going out less than half the time. But from another perspective, going out is predicted 100% of the time by my putting my shoes on.

When you think about this concept in terms of sexual health, it’s kind of inside out. Like, suppose you use a condom maybe half the time, and yet you still don’t get an STI. What that teaches you is that NOT using a condom is just as predictive of getting an STI as using a condom is. Dig? You may know intellectually that condoms prevent lots of STIs, but as far as your direct experience goes, condom use isn’t associated with prevention.

Maybe the dog isn’t a helpful example. Taking drinking. Now, on my campus 80% of the drinking that happens is beer and wine, but 100% of the medical emergencies related to alcohol are related to liquor. So at the population level, you vastly increase your chances of having things go seriously wrong if you drink liquor.

But look at it from the individual perspective. If something like 500 students each weekend consume liquor (these are imaginary but realistic numbers). And about two students per month end up at the hospital (fictional but realistic). So you’ve got 8 nights, 500 students each night and maybe two emergencies. That’s a VERY low correlation, right, so from the individual point of view it’s easy to think that the risk is very low. Which it is – except in comparison to not drinking or drinking only beer or wine. This other group is much larger and has zero emergencies.

It’s about perspective, and we humans are deeply irrational when we’re deciding how to assess risk. Somehow we believe we will never be struck by lightening but, you never know, we might win the lottery, even though the odds are similar.

The odds of getting an STI from any individual sexual event aren’t terrifically high – something like .004% for HIV, if I remember correctly (which I might not). But when you add a condom, the odds go way, way, way, way down. And the thing is, once you have HIV, you, like, have HIV. Which changes your life.

But it’s not compelling for me to stand in front of students and say, “You’ve got a minute fraction of a percent of chance of getting HIV, if you have intercourse, so use a condom every time in order to make it a vastly smaller minute fraction of a percent.”

It’s a problem, friends, this deep inability we have to understand the nature of risk.