Table 2. Response variable descriptions. See appendix for response variable correlations calculated using restricted maximum likelihood. Note: a NeerNikam is responsible for maintaining the irrigation channels, overseeing water distribution, and quelling conflict between farmers. Variable Type Description workdays days / yr poisson Number of village-wide communal workdays attended in the last year for work on ponds, channels, paths, and emergency repairs. (Mean = 3.1). Only eight farmers reported above 10 workdays, and two outliers (reporting 49 and 64 days worked) were additionally given low overall reliability rankings by research assistants, and excluded from the analysis. adequacy 0 - 5 index binomial Sum of four binary response variables on the relative adequacy of the irrigation system. Questions 36, 37, 38, 39, 43 from survey appendix, paraphrased: Is there enough water for all? Is there enough water for you? Same or better water availability than other villages? Same or better water availability than 10yr ago? Is water distribution reliable here? Questions 38 and 39 were recoded as binary variables. Item reliability: Cronbach’s a = 0.699 fairness 0 - 5 index binomial Sum of five binary response variables about procedural fairness in the irrigation system. Questions 56, 57, 58, 59, 60, paraphrased: Does NeerNikam distribute water fairly? Do village leaders keep NeerNikam honest? Are you satisfied with irrigation system fairness? How fair are irrigation rules? How fair are penalties for breaking irrigation rules? Questions 59 and 60 recoded such that completely = 1 and mostly = 1, not = 0, and no rules = NA. Item reliability: Cronbach’s a = 0.670