Probabilistic reasoning

Categorical Perception of p-Values

Statistically trained graduate students are more likely to judge two p-values as being different (vs. similar) when they cross the traditional .05 significance boundary (e.g., .046 vs. .052), compared to when they did not (.026 vs. .032), after controlling for known numerical cognition effects. This is consistent with a categorical perception effect for p-values at the .05 boundary.