Three examples do not a universal truth make
Somebody says they won’t hire anyone under 25 because “young people don’t want to work.”
Asked how many young people they’ve actually interviewed, the answer is three.
Three people out of roughly fifty million Americans in that age range.
This is hasty generalization.
Drawing a broad conclusion from insufficient evidence. Taking a small, unrepresentative sample and treating it as though it speaks for an entire category. It’s how stereotypes form. It’s how prejudice gets rationalized. It’s how people end up making serious decisions based on a handful of anecdotes and a confident voice.
The brain is built to do this.
Essay 3 covered how heuristics are mental shortcuts that save cognitive energy. Generalizing from limited experience is one of the most basic of those shortcuts, and one of the most error-prone. A few experiences with members of a group, and the brain constructs a rule. The rule then gets applied to millions of people nobody has ever met. The shortcut feels like wisdom because it happened fast.
In everyday life it looks harmless.
A friend got food poisoning at a Thai restaurant, so Thai food is suspect. Somebody met one rude person from a particular city, so everyone from that city is rude. A relative tried a new medication and had a bad reaction, so the medication is dangerous. Each of these might feel like a reasonable conclusion from a specific experience. None of them actually is.
In political life, the stakes get bigger fast.
“Crime statistics in that neighborhood prove those people are dangerous.” “My experience with government programs shows they never work.” “Every politician I’ve encountered has been corrupt, so they’re all corrupt.” Each statement might be rooted in a real experience. None of them is adequate evidence for the sweeping conclusion being drawn from it. The jump from the specific thing to the universal rule is where the fallacy lives.
Sample size matters.
If somebody has had five interactions with members of a group of millions, they know something about those five people. Full stop. The five-person data set doesn’t automatically extend to the fifty million they’ve never met. It can’t, because five people can’t be a representative sample of fifty million. The math doesn’t work.
Representativeness matters too.
Even a larger sample can mislead if it isn’t drawn from across the full range of the population. A hundred interviews conducted in one city will tell you something about that city. It won’t tell you about a country. A thousand responses collected from people who already agreed to be interviewed will tell you about volunteers, not about the general public. The number isn’t enough by itself. Who got counted matters just as much.
This is closely related to the anecdotal evidence fallacy in 15.17.
The difference is the direction of the mistake. Anecdotal evidence treats one story as proof of a claim. Hasty generalization takes a handful of stories and generalizes them to a whole category. Both are flavors of “not enough data, too confident a conclusion.” Both feel like arguments because they use real examples. Real examples don’t automatically count as proof.
The right question when a generalization starts forming – how many cases is this based on, and are those cases representative of the full range?
If the answer is “three” and the conclusion is about millions, the reasoning hasn’t reached the conclusion. Something else has.
Usually prejudice, dressed in the language of evidence.