Revisiting the volunteer's dilemma: group size and public good provision in the presence of some ambiguity aversion

Conventional game theory dictates that in a volunteer's dilemma, the probability of public good provision decrease in group size. However, experimental evidence does not support this: the probability of public good provision approaches one in large groups. I build a theoretical model addressing this puzzle, where a fraction of the group has maximin preferences, while the rest are expected utility maximizers. In small groups, the probability of public good provision may decrease with group size, but it reaches one in larger groups. While randomization is common in small groups, larger groups have some individuals who always volunteer, and others who never volunteer (another experimentally observed phenomenon).