Estimates of the attendance equations for each year and each gender are in Table 4,
while estimates of the completion equations are in Table 5. The independent variables
in the attendance and completion models are the same, with one exception. Since the
attendance equation is for 6-11 year-olds, it includes a set of age dummies and, in the
sample that pools genders, this is interacted with a dummy for whether the child is a
girl (1) or a boy (0). The completion equation, which is for 12 year-olds, simply
includes a gender dummy.
To take account of the survey design, all estimates are weighted using weights
available in the datafile. Reported standard errors are robust to arbitrary forms of
heteroskedasticity that we expect are likely given clustering in the data structure.
Tests of the joint significance of subsets of variables (e.g. village infrastructure, state
fixed effects) are reported in the Tables. The contribution of the state fixed-effects to
the total explained variation as measured by the Pseudo R-squared is also reported in
the Tables.