The gambler's fallacy in problem and non-problem gamblers

Although numerous correlational studies have shown an association between cognitive distortions and problem gambling, only a few behavioral studies have investigated this topic by comparing problem (PGs) and non-problem gamblers (N-PGs). This quasi-experiment investigated the occurrence in both groups of a widespread cognitive distortion, the gambler's fallacy (GF), using a fictitious roulette game. Moreover, it investigated whether the GF increased the bet amount and whether impulsivity and sensation seeking were associated with the GF.

Two indices of the GF were used: a cognitive index, the probability estimate of each outcome (black/red) after manipulating the final run length (the same outcome occurring four times/once), and a behavioral index, the choice of the outcome on which to bet. A total of 320 (160 PGs and 160 N-PGs) unpaid male volunteers, aged between 18 and 68, participated in this study.

Erroneous probability estimates should mediate the effect of longer runs on the alternation choice (i.e., the choice of an outcome different from the previous one) to support the occurrence of GF. The GF should increase betting. PGs should be more prone than N-PGs to GF.

The choice of the outcome depended on both cognitive (erroneous probability estimates) and affective (preference for red) factors. PGs bet more than N-PGs but they were not more prone than N-PGs to incurring GF. Although impulsivity and sensation seeking were more intense in PGs than in N-PGs, they scarcely affected GF.

Overall, our results corroborate the tested model of the GF that links mistaken probability estimates, choice of the outcome on which to bet, and bet amount. However, they are similar to PGs and N-PGs and fail to corroborate the hypothesis that the GF is more evident in PGs.