Confidence in Knowledge
The authors propose that confidence depends too much on the apparent strength of evidence and too little on its reliability or weight. This can produce overconfidence from striking but weakly grounded impressions and underconfidence when extensive evidence provides a consistent, modest signal.
THE IDEA
CONFIDENCE IN KNOWLEDGE
The authors propose that people are more sensitive to the strength of evidence than to its weight. On this account, people tend toward overconfidence when strength is high and weight is low, and toward underconfidence when strength is low and weight is high. This conclusion, we propose, applies not only to judgments about chance processes such as coin spinning, but also to judgments about uncertain events such as who will win an upcoming election, or whether a given book will make the best-seller list. When people assess the probability of such events, we suggest that they evaluate their impression of the candidate or the book. These impressions may be based on a casual observation or on extensive knowledge of the preferences of voters and readers. In an analogy to a chance setup, the extremeness of an impression may be compared to sample proportion, and the credence of an impression may correspond to the size of the sample, or to the discriminability of the competing hypotheses. If people focus on the strength of the impression with insufficient appreciation of its weight, then the pattern of overconfidence and underconfidence observed in the evaluation of chance processes should also be present in evaluations of nonstatistical evidence.
The account extends to judgments where strength and weight are difficult to define separately. The authors also connect it to the “difficulty effect.”
When to use it
Evaluating the likelihood of a project's success based on initial feedback versus comprehensive data.
Assessing the probability of a candidate winning an election by considering both the strength of public impressions and the statistical weight of polling data.
Making forecasts about product adoption by balancing anecdotal evidence with market research sample sizes.
FROM THEORY TO EVERYDAY LIFE
In practice
Election prediction
A person may have a very strong impression that a particular candidate will win based on a few compelling anecdotes (high strength), but ignore the fact that the sample of information is small and unrepresentative (low weight), leading to overconfidence.
Book best-seller list
A publisher sees a modest but consistent advantage for a book across extensive, reliable market research. Focusing on the modest size of the advantage while overlooking how consistently it appears could leave the publisher less confident than the combined evidence warrants.
A MOMENT TO REFLECT
Put it to work.
Think of a recent decision you made where you had strong feelings about an outcome. Did you consider the weight (reliability, sample size) of the evidence, or only its strength?
How might the strength-weight distinction explain why experts are sometimes overconfident in their predictions?
Can you identify a situation where you were underconfident despite having substantial evidence? What factors led to that?
Often confused with
CONTINUE THE CONVERSATION
Source texts
- Heuristics and Biases The Psychology of Intuitive Judgment (Thomas Gilovich, Dale Griffin, Daniel Kahneman)