Drawing a broad conclusion from a sample that is too small, unrepresentative, or cherry-picked.
"I've met three people from [country X] and they were all rude — people from [country X] are rude." More insidiously: "Three companies in our sector tried remote work and failed — remote work doesn't work." The sample is small, selection bias is likely (we hear about the failures), and the conclusion is stated as universal.
Anecdotes feel real in a way that statistics don't. "I know someone who..." bypasses the abstract and makes a claim vivid and personal. The vividness creates an illusion of representativeness — we treat the salient example as typical even when we have no reason to believe it is.
Universal claims ("always," "never," "all," "none") based on a handful of examples. Ask: "How many cases is this based on, and how were they selected?"