The number of available sample observations modulates gambler’s fallacy in betting behaviors

Abstract The gambler’s fallacy is a commonly observed cognitive bias in betting behaviors. It refers to the erroneous belief that an independent and identically distributed random process exhibits negative serial correlation. This belief is often triggered when individuals observe a sample of historical outcomes of the random process. Our study examines the impact of providing more historical information or a larger sample of recent outcomes on the propensity for individuals to commit the gambler’s fallacy in repeated betting scenarios with varying probabilities of winning. Experimentally, we find strong evidence of both the gambler’s fallacy and the hot-outcome fallacy, which predicts the opposite pattern. These fallacies correlate respectively with the frequency and length of consecutive occurrences in the observed sample. Providing a larger sample of observations increases the tendency for the gambler’s fallacy when the probability of winning is low but significantly reduces the fallacy when probabilities are moderate or high. Our results indicate that gambling fallacies are highly sensitive to contextual factors in decision-making, and that in real-life gambling situations, which often involve low chances of winning, providing more sample observations could amplify the gambler’s fallacy.