Pseudodiagnosticity
Also known as: Pseudodiagnostic reasoning
Gathering evidence that fits a hypothesis without checking the alternative.
What it means
The tendency, when testing competing hypotheses, to seek and use information about how likely the evidence is under the favored hypothesis while neglecting how likely the same evidence is under the alternative — even though only that comparison is diagnostic. A datum that strongly fits your hypothesis tells you nothing if it fits the rival hypothesis equally well, yet people treat high consistency with the preferred explanation as confirmation. The error reflects a failure to think in terms of likelihood ratios, the quantity that actually distinguishes hypotheses in Bayesian inference, and it leads to overconfidence in conclusions that the evidence does not warrant. It is conceptually distinct from confirmation bias in pinpointing the specific neglect of the alternative's likelihood. It matters in medical diagnosis, intelligence analysis, and scientific reasoning, where compelling-looking but non-diagnostic evidence routinely props up the wrong conclusion.
Examples
A doctor noting that a patient's symptom is common in the suspected disease, without asking whether that symptom is just as common in the diseases being ruled out.
Sold on a candidate, the manager notes she once ran a marathon — proof of grit — without asking how many people who lack grit have also run marathons.
The mechanic hears the rattle that fits a worn wheel bearing, orders the part, and never checks that a loose heat shield rattles in exactly the same way.
First described in Doherty, Mynatt, Tweney & Schiavo (1979).