Try (not) again Sam! Framing effect and control policies on the almost-winning bias: An experiment

We propose the results from a lab experiment aimed at reducing the almost-winning bias effect (Reid, 1986) by comparing two informative mechanisms. The almost-winning (or near-miss) cognitive bias induces individuals to wrongly perceive the probability to succeed in a chance game. A near-miss event encourages further attempts because it increases the perceived probability of future success: this is actually true in all skill tasks where nearly reaching a specific goal should induce further attempts, though in a chance game near-misses should be meaningless. In this experiment we propose two (chance) frames: an investment game and a slot machine game; our results show the significant effect of the almost-winning bias to future bets, in particular when playing the investment game. Finally, we introduce two potential mechanisms to correct such bias: participants were warned about the independence across periods (nudging) or were informed about the probability of winning (awareness). We find that the average bet decreases through time, though the nudge intervention was not as effective as the awereness.