Does Residuals-on-Residuals Regression Produce Representative Estimates of Causal Effects?

원문 초록

Double Machine Learning is commonly used to estimate causal effects in large observational datasets. The "residuals-on-residuals" regression estimator (RORR) is especially popular for its simplicity and computational tractability. However, when treatment effects are heterogeneous, the proper interpretation of RORR may not be well understood. We show that, for many-valued treatments with continuous dose-response functions, RORR converges to a conditional variance-weighted average of derivatives evaluated at points not in the observed dataset, which generally differs from the Average Causal Derivative (ACD). Hence, even if all units share the same dose-response function, RORR does not in general converge to an average treatment effect in the population represented by the sample. We propose an alternative estimator suitable for large datasets. We demonstrate the pitfalls of RORR and the favorable properties of the proposed estimator in both an illustrative numerical example and an application to real-world data from Netflix.