Experiments 1–3: Summary statistics.

z-scores for binomial tests versus chance (50%) the number of participants predicting a changed dice outcome (i.e., gambler’s fallacy) versus continuation of the dice outcome on the post-run critical trial. Positive values indicate that more participants predicted a change in the dice outcome (e.g., predicting an even sum after an odd sum), while negative values indicate that more participants predicted that the previous dice outcome would continue (e.g., predicting an even sum after an even sum). Asterisks indicate significance at p<0.001. Where asterisks do not appear, the binomial test was not significant at p<0.05.