Date of Award

2026

Degree Type

Dissertation

Degree Name

Doctor of Philosophy in Behavioral Science

Specialization

Behavioral Science

Department

Psychology

First Advisor

Manshu Yang

Abstract

Randomized controlled trials (RCTs) are commonly used to estimate average treatment effects (ATEs), with baseline covariates often included to improve precision and statistical power for detecting those effects. In applied settings, covariates can be incomplete, and some may moderate treatment effects, leading to heterogeneous treatment effects (HTE). Literature lacks clear direction regarding whether incomplete covariates should be included when estimating the ATE and how different missing data handling methods perform in such conditions. This study aimed to evaluate the performance of four missing data handling methods in RCTs via Monte Carlo simulations, namely listwise deletion (LD), inverse probability weighting (IPW), full information maximum likelihood (FIML), and multiple imputation via chained equations (MI-CE).

The simulation varied sample size, magnitude of ATE, number of incomplete covariates, proportion of missingness, and covariates showing HTE. Convergence rates, bias, mean squared error, coverage probability, Type I error rate, and statistical power were evaluated to compare methods. Results indicated that including incomplete covariates rarely improved estimation or power and often introduced more problems than benefits. When only fully observed covariates were included, all four methods consistently converged and produced negligible bias. However, once incomplete covariates — particularly those involved in HTEs — were added, MI CE showed higher rates of nonconvergence, and LD, FIML, and IPW frequently yielded biased ATE estimates. Across all methods, coverage probabilities and Type I error rates were more likely to deviate from nominal levels in smaller samples, in scenarios with multiple incomplete covariates, or when missingness was substantial.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.