Table S4.

Results of six logistic GLMs on the probability of understanding the game depending on behavioral type

Classification schemeNumber of typesχ2(df)Significance
Expansive (Fig. 1)615.3(5)0.009
F01 + negative514.8(4)0.005
F01412.9(3)0.005
Three-way312.9(2)0.002
Binary (cooperator/non)211.9(1)0.001
Binary (constant/inconstant)212.1(1)0.001
  • Each model uses a different scheme to classify types according to their contribution schedule in the strategy method. Expansive (Fig. 1) uses the classification scheme presented in Fig. 1, which is five categories plus unclassified. F01 + negative uses the four-way classification scheme published by ref. 1 plus the addition of negative cooperators that show a negatively conditional strategy. F01 uses the four-way classification scheme published by ref. 1. Three-way is the same as F01 but combines conditional and humped cooperators. Binary (cooperator/non) is a binary scheme comparing all players that contributed something to those that contributed 0 MU. Binary (constant/inconstant) is a binary scheme comparing those that contributed a constant amount (including 0 MU) or not.