Table 4.

Results of the best-fitting GEE models for the Sharing Game and Taking Game (n = 290, dyads = 1312). Higher log-odds estimates in the Sharing Game indicate an increased likelihood of ego giving to an alter of that category, while higher coefficients for the Taking Game are associated with an increased propensity for ego to take from alter. These models contain controls for the amount each individual gave/took (not displayed here). Dashes indicate that the variable was not included in the final model. 95% CIs are displayed in brackets.

variable	Sharing Game log-odds coefficient	Taking Game log-odds coefficient
intercept	2.43 (2; –2.87)	−3.55 (−4.23; −2.87)
reciprocity (1 = reciprocal giving/taking)	0.39 (0.1; 0.68)	—
primary kin (ref. unrelated individuals)	1.33 (0.76; 1.91)	—
distant kin (ref. unrelated individuals)	0.59 (0.04; 1.14)	—
primary kin's spouse/spouse's primary kin (ref. unrelated individuals)	0.31 (−0.25; 0.87)	—
spouse's distant kin/other affines (ref. unrelated individuals)	0.08 (−0.32; 0.48)	—
spouse (ref. unrelated individuals)	−0.16 (−0.83; 0.52)	—
resource quantity (1 = high)	n.a.	2.38 (1.87; 2.89)
proximity	—	−0.15 (−0.3; 0.00)