Steelman
Survivorship bias
Also called survivor bias, selection bias (broader).
Drawing conclusions only from the winners, because the losers aren’t visible.
What it is
Reasoning from the cases that made it through some selection process while overlooking the ones that didn’t, which leads to overly optimistic or simply wrong conclusions.
The pattern
Look only at the survivors. Notice what they have in common. Conclude that this trait causes survival.
An example
“Bill Gates, Mark Zuckerberg and Steve Jobs all dropped out of college. Dropping out is a smart path to success.”
We hear about the famous dropouts, not the far larger number who dropped out and didn’t become billionaires.
Why it fails
The failures may have had the same trait. Without them in the data, you can’t tell whether the trait helps, hurts or makes no difference.
When it’s not a fallacy
Most lists skip this part. Calling a fair point a fallacy is a mistake too.
It can be fair when
- The question really is about survivors (what do buildings that lasted have in common?), or the failures are included in the analysis.
Try one
A question from the app. The wrong answers are its usual look-alikes.
Family argument about a 12-year-old’s first phone. Mom: “She’s 12. I want her off social media for a few more years.” Lily: “But every creator I watch started posting at my age, and now they’ve got millions of followers and real careers!” Aunt Rosa: “She’s right. Starting young is how you learn the platforms. Keep her off and you’re keeping her behind.” Mom: “We’re not deciding this over dessert.” Lily: “That sounds like a maybe.” Mom: “It’s a no for tonight.”
What’s the flaw?
The creators Lily watches are the few who made it. The thousands of kids who started posting at 12 and got nowhere aren’t in her feed, so the winners’ early starts don’t show that starting young helps.
A fair reply
“You’re only seeing the kids who made it, so how many started at 12 and got nowhere?”
Often confused with
- Cherry-picking
Cherry-picking selects evidence on purpose. Survivorship bias happens because the failures have already dropped out of view.
- Hasty generalization
Hasty generalization uses too small a sample. Survivorship bias uses a sample filtered by the very outcome being studied.
Where the name comes from
The name says it: you only see the survivors, the ones that made it through. A bias is a lean that tilts your conclusions one way. “Survivor bias” is the same name, shortened. “Selection bias” is the wider family: any skew that comes from how the examples were picked.
The classic illustration comes from statistician Abraham Wald’s WWII work at Columbia University’s Statistical Research Group, estimating aircraft vulnerability from the damage on planes that returned. The damage you can see is on the planes that survived it.
Sources
- Survivorship bias, Wikipedia
In the app
Practice spotting it on arguments that sound real.
Steelman has more examples of survivorship bias, how to answer it in a real conversation, and practice rounds that mix it with its look-alikes, so you learn the difference, not just the name. Coming soon to iPhone and Android.
You’re on the list.
We’ll email you the day Steelman is out.
Thanks for offering to test. We’ll write before the beta with how to join.