Estimates of the impact of formal childcare availability on school enrollment in children over 6 are summarized in Table 2 . Of the original sample of 8366 children, information on school enrollment was available for 8157 at midline (4508 enrolled, 3649 unenrolled). 5134 of these children (2628 treated, 2506 control) were over 6 years old at midline and were therefore included in our models. Of the 3649 children specifically coded as unenrolled at midline (1974 treated, 1675 control), 1523 were over 6 years old (810 treated, 713 control). Our analyses of non-enrollment due to domestic responsibilities were restricted to this subgroup of 1523 children.

Our findings in the overall sample suggested that, by the time of the midline survey, the childcare intervention did not result in a meaningful change in either the probability of siblings over 6 being enrolled in school (ITT RR: .98 (95% CI: .94, 1.02)), or being unenrolled specifically due to domestic responsibilities (ITT RR: 1.10 (95% CI: .91, 1.33)). Sex-stratified analyses did not reveal any meaningful differences between boys and girls. The per-protocol analysis produced similar estimates, although the sample size decreased substantially and estimates were correspondingly imprecise.