High-dimensional regression adjustments in randomized experiments

Authors:

Abstract
We study the problem of treatment effect estimation in randomized experiments with high-dimensional regression adjustments. We present an inferential approach that takes any risk-consistent regression adjustment and yields efficient estimates of the average treatment effect. This considerably extends the range of settings where high-dimensional regression adjustments are guaranteed to provide valid inference about the population average treatment effect. We also propose cross-estimation, a simple method for obtaining finite-sample-unbiased treatment effect estimates that leverages high-dimensional regression adjustments. Our method can be used when the regression model is estimated using the lasso, the elastic net, subset selection, or many other methods; we also extend our analysis to allow for adaptive specification search via cross-validation, and flexible non-parametric regression adjustments with, e.g., random forests or neural networks.

This publication has 32 references indexed in Scilit: