Table 3

Main Results—Treatment Effect on Test Scores at Post-Intervention (T2) and in the One-Year Follow-up (T3)

Post-intervention (T2)Follow-up (T3)
Summary ScoreMathematicsExecutive FunctioningLanguageSummary ScoreMathematicsExecutive FunctioningLanguage
Model 1
Wild P0.0790.1450.0570.8700.0850.0040.2960.680
Adj. R20.6120.4400.4920.5320.5220.3640.3820.492
Model 2
Wild P0.1220.2080.0560.8670.1880.0430.3480.744
Adj. R20.1470.1140.1140.1010.1230.0830.0840.176
Model 3
Wild P0.0950.0790.1150.9870.1080.0040.1860.912
Adj. R20.6130.4330.4920.5210.5170.3490.3790.456
Model 4
Wild P0.3870.1650.2630.5880.3770.0350.4000.632
Adj. R20.0170.0040.0220.0150.0150.0260.0070.031
  • Notes: *p < 0.1, **p < 0.05, ***p < 0.01. Each column in each panel presents regression coefficient of treated (standard error) using ordinary least squares. For both assessment period, Model 1 regresses outcome on the treatment indicator, controlling for baseline test scores, gender, birth month, parental characteristics (mother and father’s education level, earnings, an indicator for non-Western country of birth), and indicators for late consent and not having participated in the T1 assessment. In Model 2, we exclude baseline test scores from the model. In Model 3, we restrict controls to baseline test scores and an indicator for not having participated in the T1 assessment. Model 4 has no controls. All regressions are clustered on and control for randomization block. We have utilized the boottest package in Stata to do a few-cluster-correction of the p-value, reported in the table as Wild P.