Base Rate Fallacy

How common something actually is can get completely ignored in favor of a single striking detail. That detail drives a judgment about likelihood that's badly out of step with reality.

This bias underweights the prior probability when evaluating a piece of evidence, distorting the resulting judgment. It arises when diagnostic information gets overemphasized relative to the background frequency, producing a suboptimal probabilistic inference.

A person tests positive on a medical screening test for a rare disease. Even though the test is fairly accurate, they immediately assume they almost certainly have the disease — ignoring that the disease affects only 1 in 10,000 people. Because the disease is so rare, most positive tests are actually false positives, so the real probability of having it is still quite low.

A fraud detection model flags a transaction as suspicious with a reported sensitivity of 95% and a false-positive rate of 1%. An analyst, focusing on that high sensitivity, concludes the flagged transaction is almost certainly fraudulent. But if the true base rate of fraud in the transaction population is only 0.1%, a straightforward application of the actual math yields a positive predictive value of just under 9% — meaning over 91% of flagged transactions are false positives. The high sensitivity dominated the analyst's judgment while the low base rate got effectively discarded, producing a badly miscalibrated estimate.

A striking clue draws more attention than how common the outcome actually is. That extra attention is exactly what inflates the perceived likelihood past what it really deserves.

The prior frequency gets underweighted relative to the incoming diagnostic cue, producing an asymmetry in how the evidence gets integrated. The system applies far stronger weight to the salient likelihood than to the slow-moving base rate behind it.

Remembering to check how common the event actually is, before deciding, is the direct fix. Comparing the clue against that base rate keeps the judgment calibrated.

Explicitly combining the observed likelihood with the prior frequency, via proper Bayesian updating, corrects the estimate directly. Decision rules that penalize excessive weight on a salient cue keep the base rate from being discarded.

Overattribution to salient evidence; Underweighting of prior frequency; Systematic probability overestimation

An adversarial actor can weaponize the base rate fallacy by flooding a decision context with vivid, emotionally salient diagnostic cues—such as anecdotal testimonies, dramatic case studies, or alarming statistics—that systematically crowd out background frequency information, causing targets to wildly overestimate the probability of a desired conclusion. In security, intelligence, and medical contexts, this can be exploited to manufacture false threat assessments or inflate perceived risk by cherry-picking high-likelihood signals while suppressing base rate anchors. Propagandists and marketers routinely exploit this by presenting exceptional cases as representative, causing audiences to treat rare events as common and act accordingly.

Explicitly elicit and document the base rate (prior frequency) as a mandatory first step before evaluating any diagnostic evidence, institutionalizing it as a checklist item in high-stakes decision protocols. Train evaluators in Bayesian updating procedures or use structured analytic templates that force numerical combination of prior probabilities with likelihood ratios before forming a posterior judgment. Implement red-team analyses by assigning a devil's advocate role to surface ignored base rates whenever salient evidence dominates group deliberation.