Instrumented Quintile Interaction Effects of the Impact of Acceleration into Algebra I in Eighth Grade
| Independent Variable | Algebra I Test Score | Pass Algebra I by Tenth Grade 2SLS | Pass Geometry by Eleventh Grade 2SLS | Pass Algebra II by Twelfth Grade 2SLS | |
|---|---|---|---|---|---|
| 2SLS | RFQR with Imputation | ||||
| Quintile 1 student * enrolled in Algebra I by eighth grade | −0.425*** (0.101) | −0.167*** (0.039) | 0.196*** (0.030) | −0.128*** (0.038) | −0.0527 (0.089) |
| Quintile 2 student * enrolled in Algebra I by eighth grade | −0.428*** (0.033) | −0.436*** (0.014) | 0.0192 (0.029) | −0.105*** (0.012) | −0.0198 (0.028) |
| Quintile 3 student * enrolled in Algebra I by eighth grade | −0.343*** (0.040) | −0.382*** (0.011) | 0.00780 (0.011) | −0.123*** (0.013) | −0.0388 (0.036) |
| Quintile 4 student * enrolled in Algebra I by eighth grade | −0.282*** (0.041) | −0.305*** (0.011) | 0.0400*** (0.012) | −0.0774*** (0.014) | −0.0331 (0.021) |
| N | 121,572 | 135,752 | 135,752 | 135,752 | 135,752 |
| Adjusted R2 | 0.696 | 0.354 | 0.429 | 0.530 | 0.440 |
Notes: Standard errors, corrected for clustering at the decile-cohort-district level, in parentheses. Algebra I test score is taken from the student’s first test administration. Course passage for Algebra I and Algebra II is defined as obtaining a standardized test score at or above the twentieth percentile of the statewide distribution. Course passage for Geometry is defined as obtaining an achievement level at or above 3 on the test. Grade-retained students are kept with their original cohort. All models control for average sixth and seventh grade math test score decile, cohort and district fixed effects, and instrument for Algebra I enrollment by eighth grade using an indicator representing the probability of taking Algebra I by eighth grade within your decile-cohort-district cell. Columns headed “2SLS” are estimated by two-stage least squares. Column headed “RFQR with Imputation” applies the Neal and Johnson (1996) method of imputing poor performance for 14,180 non-Algebra I-takers and estimating using the Chernozhukov and Hansen (2005) method.
↵*** denotes a coefficient significant at the 0.1 percent level,
** the 1 percent level,
* the 5 percent level.