| Half Optimal BW =5.6 | Optimal BW= 11.2 | Twice Optimal BW =22.4 | |
|---|---|---|---|
| Private · treatment | −0.027 | −0.005 | −0.023 |
| (0.190) | (0.139) | (0.125) | |
| Private · below median · treatment | 0.359* | 0.389*** | 0.365** |
| (0.188) | (0.145) | (0.142) | |
| Observations | 6,963 | 14,201 | 26,077 |
| R-squared | 0.402 | 0.410 | 0.424 |
↵Notes: Source is ENEM microdata. *p<0.1,**p<0.05, ***p<0.01. We report the results of our benchmark specification (Equation 1) adding the terms below_mediansm, below_mediansm ⋅ privs, and below_mediansm ⋅ privs ⋅ ds with triangular kernels and optimal bandwidth as suggested by Calonico et al. (2016). Covariates: gender, age, race, parental schooling, family income, teacher schooling, presence of a science lab, and ratios of teachers, computers, and staff to students. Metropolitan area fixed effects in all columns. Standard errors are clustered at the school level.