No. 06 Errors of Generalization
Hasty Generalization
In Turkish: Genelleştirme Safsatası
You saw two and now you know them all. Congratulations, statistician.
Every minibus driver in this town drives like a lunatic!
I took two today and both were flying. All of them are like that!
You took two. There are forty. Try the other thirty-eight first.
What is it?
Judging a whole group from one or two examples. The examples are few, and usually the most memorable ones; the group is large, and you never met its boring majority.
The pattern
- The few A’s I saw were B.
- Therefore all A’s are B.
Why it fails
Generalising isn’t bad; it’s how anyone learns anything. The problem is the sample: two or three cases, especially the ones you remember because you were angry, don’t represent a group’s average. The worst driver sticks in the mind; the forty calm ones vanish. Most stereotypes are born this way: few examples, high confidence.
How to spot it
- Words like “always”, “all of them”, “none”, backed by two or three stories.
- The examples are always personal and always dramatic: something that happened to someone.
- Ask “how many did you see?” and the answer is “I saw it, that’s enough.”
What to say
“How many did you see?” Then: “And how many didn’t you see?” The second question usually does it. Offer a counterexample straight away: “The three I took were perfectly calm.”
When it’s not a fallacy
If the sample is large and representative, generalising is legitimate; surveys and clinical trials work exactly that way. Some single cases are strong too: one poisonous mushroom is enough to generalise about that species. The difference lies in how many examples there are and whether they’re typical.
Examples
- Family“That phone brand is junk; your cousin’s second one broke too.”
- Work“New graduates don’t work; both the ones we hired last year left.”
- Health“That medicine does nothing, the neighbour took it and didn’t get better.”
- Football“This referee is always against us; two matches, two penalties denied.”