Working Paper: NBER ID: w15200
Authors: Weili Ding; Steven F. Lehrer
Abstract: This paper introduces an empirical strategy to estimate dynamic treatment effects in randomized trials that provide treatment in multiple stages and in which various noncompliance problems arise such as attrition and selective transitions between treatment and control groups. Our approach is applied to the highly influential four year randomized class size study, Project STAR. We find benefits from attending small class in all cognitive subject areas in kindergarten and the first grade. We do not find any statistically significant dynamic benefits from continuous treatment versus never attending small classes following grade one. Finally, statistical tests support accounting for both selective attrition and noncompliance with treatment assignment.
Keywords: Class Size; Education Production; Dynamic Treatment Effects
JEL Codes: C31; I21
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Attendance in small classes during kindergarten and first grade (I21) | Significant benefits in cognitive achievement across all subject areas (I24) |
| Continuous treatment compared to never attending small classes (D29) | No statistically significant dynamic benefits by the end of first grade (I21) |
| Selective attrition and noncompliance with treatment assignment (C90) | Affects the estimation of treatment effects (C22) |
| Noncompliance (H26) | Bias in traditional intention-to-treat (ITT) estimators (C51) |
| Empirical strategy (C90) | Estimation of dynamic average treatment effects (C22) |