| Public | Private | |||||
|---|---|---|---|---|---|---|
| Coeff. | SE | #Obs. | Coeff. | SE | #Obs. | |
| Male | −0.036 | 0.052 | 3,572 | −0.005 | 0.030 | 8,592 |
| Correct school grade | 0.016 | 0.029 | 2,548 | −0.01 | 0.012 | 5,943 |
| White | 0.010 | 0.044 | 2,957 | 0.008 | 0.026 | 7,314 |
| Father with college degree | 0.002 | 0.010 | 3,572 | 0.015 | 0.025 | 7,705 |
| Mother with college degree | 0.010 | 0.016 | 2,548 | 0.007 | 0.024 | 7,705 |
| Family income < 10 · min. wage | −0.004 | 0.007 | 2,957 | −0.031 | 0.033 | 6,601 |
Notes: Source is ENEM microdata. *p<0.1, **p<0.05, ***p<0.01. We report the results of our regression discontinuity design estimates based on local linear regressions with triangular kernels and optimal bandwidth as suggested by Calonico et al. (2016). Metropolitan area fixed effects in all columns. Covariates: gender, age, race, parental schooling, family income, teacher schooling, presence of a science lab, and ratios of teachers, computers, and staff to students. Dependent variable not included as a covariate. Standard errors are clustered at the schools level.