Table 4. Regression results and diagnostics. Coefficients with standard errors in parantheses. Bold estimates have 95% confidence or greater. Max variance inflation factors (VIF) is the largest variance inflation factor for any predictor in the given model. A VIF of 10 or greater is considered an indication of multicolinearity (Fox and Monette 1992). A random-effects poisson was over-dispersed (dev/df = 2.17; Lindsey 1999), while a negative binomial model without village was not (dev/df = 1.29). Village-level variance were calculated with a mixed effects poisson. The diversity predictor variable used for workdays was number of castes with 10 or more households.

Workdays Adequacy Fairness
Family negative binomial binomial binomial
Intercept -1.58 (0.83) -3.43 (1.03) -2.45 (1.31)
Village-level
LNPOP 0.77 (0.15) -0.25 (0.13) 0.57 (0.19)
DISTANCE -0.01 (0.01) 0.00 (0.01) 0.02 (0.01)
CASTES -0.26 (0.08) 0.16 (0.06) -0.11 (0.09)
WEALTHGINI -0.03 (0.01) 0.05 (0.02) -0.11 (0.02)
Household-level
AGE 0.00 (0.00) 0.00 (0.00) 0.00 (0.01)
EDUC -0.01 (0.02) -0.01 (0.02) -0.02 (0.02)
HHSIZE 0.06 (0.04) 0.06 (0.05) 0.15 (0.06)
LNWEALTH -0.01 (0.04) -0.01 (0.04) 0.22 (0.06)
FRAC 1.14 (0.44) 0.27 (0.44) 2.56 (0.64)
DALIT -0.03 (0.23) -0.15 (0.25) -1.24 (0.58)
CHANNEL 0.38 (0.17) 0.96 (0.18) 3.36 (0.28)
DALIT*CHANNEL 0.12 (0.31) 0.36 (0.33) 1.95 (0.67)
Village variance
Full Model 0 0.001 0
Model Fit & Diagnostics
ML pseudo-R² 0.22 0.28 0.94
DF 241 243 243
Deviance 310 143 473
Max VIF 7.9 7.2 7.2