The reason a friend’s recommendation outranks a study

A study covering ten thousand cases is set against one friend’s experience, and the friend wins. Described as a reasoning error this is puzzling, since the sample sizes are not comparable. Described in terms of what each source supplies, it is less puzzling: only one of the two can be asked follow-up questions and held responsible for the answer.

What everyone sees

The preference is read as innumeracy — people cannot handle statistics and default to anecdote. Statistical reasoning is genuinely difficult. But the account leaves out that a personal source provides things a study does not, and that some of those things are legitimately evidential rather than merely vivid.

What is actually happening

Brown and Reingen examined word-of-mouth referral and found the strength of the tie affected which recommendations were acted on, with strong ties exerting more influence at the decision stage, indicating that the relationship is doing work independent of the content. Nisbett and Ross documented the wider pattern that concrete, vivid single cases outweigh pallid statistical information in judgement. Combining the two: the friend is trusted partly through a mechanism that is not evidential at all, and partly through one that is — they can be interrogated, their circumstances are known, and they bear a reputational cost if the recommendation fails.

Why it stays hidden

The reasonable half of the preference hides behind the unreasonable half. Because the vividness effect is well documented, the whole behaviour gets classified as bias, and the accountability the friend provides goes unexamined. A study cannot be asked what it would do differently, and that difference is real even where the sample argument favours the study.