That delusional and delusion-prone individuals gather less evidence before reaching a decision (“jumping to conclusions”) is arguably the most influential finding in the literature on cognitive theories of delusions. However, the cognitive basis of this data-gathering tendency remains unclear. Suggested theories include that delusion-prone individuals gather less data because they 1) misjudge the information value of evidence; 2) find data gathering more taxing than controls; and/or 3) make noisier decisions than controls. In the present study we developed a novel, incentivized, non-serial data-gathering task to tease apart these alternatives. Higher delusion-proneness was associated with gathering less information on this task, even when accounting for gender, risk aversion, and intelligence. Our findings suggest that misjudging the information value of evidence contributes substantially to the “jumping to conclusions” bias and that neither higher subjective costs nor noisy decision-making can fully account for it.