Title: 

Covariate adjustment in randomized experiments with missing outcomes and covariates

Abstract:

Covariate adjustment can improve precision in analyzing randomized experiments. With fully observed data, regression adjustment and inverse probability weighting are asymptotically equivalent in improving efficiency over unadjusted analysis. When some outcomes are missing, we consider combining these two adjustment methods with inverse probability of observation weighting for handling missing outcomes, and show that the equivalence between the two methods breaks down. Regression adjustment no longer ensures efficiency gain over unadjusted analysis unless the true outcome model is linear in covariates or the outcomes are missing completely at random. Propensity score weighting, in contrast, still guarantees efficiency over unadjusted analysis, and including more covariates in adjustment never harms asymptotic efficiency.  Moreover, we establish the value of using partially observed covariates to secure additional efficiency by the missingness indicator method, which imputes all missing covariates by zero and uses the union of the completed covariates and corresponding missingness indicators as the new, fully observed covariates.  Based on these findings, we recommend using regression adjustment in combination with the missingness indicator method if the linear outcome model or missing complete at random assumption is plausible and using propensity score weighting with the missingness indicator method otherwise.

Bio: 

Fan Li is a professor in the Department of Statistical Science at Duke University, with a secondary appointment at Biostatistics and Bioinformatics. Her primary research interest in causal inference, and have been working at the intersection between causal inference, machine learning, and population health. She also works on Bayesian analysis and missing data problems. She is the editor for Social Science, Biostatistics and Policy of the Annals of Applied Statistics, and an elected fellow of the American Statistical Association and the Institute of Mathematical Statistics (IMS).