Disentangling two causes of biased probability judgment: Cognitive skills and perception of randomness

This experimental study investigates the interaction of two influential factors of biased probability judgments. Results provide new insights on the preconditions for an application of either the gambler's fallacy or its exact opponent, the hot hand fallacy. The first factor is cognitive ability, measured in a cognitive reflection test. The second one is the level of perceived randomness in the observed outcomes. Probability judgments are found to vary significantly across salience of randomness treatments as well as across subgroups with high or low cognitive abilities. Like in previous research, subjects with higher cognitive skills are more likely to engage the gambler's fallacy, yet only if perception of sequential randomness is low. In a setting where randomness is very salient the exact opposite can be observed. Similarly surprising insights are revealed when controlling for cognitive abilities in the analysis of salience treatments. Past results are only confirmed for a subgroup with lower cognitive skills, while their peers' beliefs are completely opposite.