Sayers et al look in Some descriptive aspects of two-person non-zero-sum games at variations of the prisoner dilemma. They had two participants (α and β) pressing press either a black (1) or a red (2) button. In one example they used the matrix:
In the discussion they write:
In none of these games does the collaborative strategy, (α:1 and β:1) or black-black, occur more frequently than it would if subjects pressed red and black at random with equal probability.
Is this finding replicated in other studies or is it only a non-replicateable one-of-finding? In case there was a replication and it was better then random, did still a significant number of participants chose against picking the collaborative option?