A Note on Postrandomization Adjustment of Covariates

Abstract

The adjustment for postrandomization variables has been widely used in practice to obtain additional information in randomized experiments despite its conceptual and statistical difficulties. To enhance the chance of more appropriate postrandomization adjustment analyses and interpretation, this paper systematically summarizes the potential problems of the conventional regression-based postrandomization adjustment method with its reliability, precision, and causality. Some available alternative methods that could possibly provide either more powerful, less biased evaluation or more appropriate assessment for causality are outlined. A working example is used to illustrate the application of the different postrandomization adjustment methods.