Table 3.
Tests for the random assignment rules.
| Variables | OLS | School Fixed Effects | Grade-by-School Fixed Effects |
|---|---|---|---|
| (1) | (2) | (3) | |
| Age | 0.081 | 0.094 | 0.101 |
| (0.528) | (0.141) | (0.165) | |
| Ethnicity | −0.430 *** | 0.024 | 0.051 |
| (0.153) | (0.036) | (0.041) | |
| Only child in family | −0.892 *** | −0.095 | −0.130 |
| (0.075) | (0.082) | (0.086) | |
| Gender | 0.005 | −0.034 | −0.016 |
| (0.032) | (0.088) | (0.070) | |
| Hukou type | 0.737 *** | 0.182 | 0.224 * |
| (0.085) | (0.122) | (0.125) | |
| Mother with a college education | −0.432 *** | −0.068 | 0.015 |
| (0.055) | (0.055) | (0.053) | |
| Father with a college education | −0.501 *** | −0.209 *** | −0.105 |
| (0.061) | (0.073) | (0.068) |
Notes: The data source is China Education Panel Survey (CEPS) 2013–2014. Robust standard errors reported in parentheses are clustered at the class level. Estimate in each cell is obtained from a separate regression in which the dependent variable is the student’s predetermined variable, as listed above. An ordinary least squares (OLS) estimation without any fixed effects is applied for regressions in Column (1). Regressions in Column (2) include a set of grade and school fixed effects. Specifications in Column (3) include grade fixed effects, school fixed effects, and grade-by-school fixed effects. *** p < 0.01, * p < 0.1.