Table 2.

Results of the model-averaging procedure, pooled across five imputed datasets, for the Sharing Game using the sample for which camp stability data were available (n = 183, camps = 11). Positive parameter estimates indicate an increase in rice kept for self (%). Relative importance denotes proportion of top models used in model-averaging which contain each variable (a value of 1 means all models used for model-averaging contain this variable, while lower values indicate that less models contained this variable).

variable	level	parameter estimate	s.e.	relative importance	significance level
intercept	—	119.73	16.15	—	<0.001
camp stability	camp	−67.73	8.99	1.00	<0.001
harvesting rice (1 = no)	camp	20.15	5.73	1.00	<0.001
affinal closeness	individual	−13.59	4.48	1.00	0.003
stored rice (1 = no)	individual	9.54	3.58	1.00	0.008
no. of dependent offspring	individual	1.84	0.78	1.00	0.02
cash labour involvement	individual	−7.55	5.29	0.78	0.215
no. of primary kin in camp	individual	0.52	0.93	0.39	0.581