TITLE: Principles for Covariate Adjustment in Analyzing Randomized Clinical Trials
INSTRUCTOR: Shao Jun, University of Wisconsin – Madison and Ting Ye, University of Washington


In randomized clinical trials, adjustments for baseline covariates at both design and analysis stages are highly encouraged by regulatory agencies. To gain credibility and efficiency, a recent trend in clinical trials is to apply covariate-adaptive randomization at the design stage and to use a modelassisted approach in analysis that further adjusts for covariate not used in design and produces asymptotically valid inference even when the model is incorrectly specified. In this tutorial we present concepts and methodologies that are crucial for model-assisted inference based on clinical data. In particular, we introduce and elaborate three principles: (1) guaranteed efficiency gain principle: a model-assisted method should often gain but never hurt efficiency compared with a benchmark method not utilizing covariates. (2) validity and universality principle: a valid procedure should be universally applicable to all commonly used randomization schemes, simple or covariate adaptive. (3) robust standard error principle: variance estimation should be heteroscedasticity robust. To fulfill these principles, we recommend a working model that includes all covariates utilized in randomization and all treatment-by-covariate interaction terms. Our conclusions are based on a general asymptotic theory that does not assume a correct model, is valid for most commonly used covariate-adaptive randomization schemes such as permuted block randomization and minimization, and reveals distinct results between cases of two-arm and multiple-arm trials. Numerical examples with data close to real trials are presented and discussed for illustration.

Instructors’ Biography:


1987: Ph.D. (Statistics), University of Wisconsin, Madison, Wisconsin, U.S.A.

1982: B.S. (Mathematics), East China Normal University, Shanghai, China


Aug. 1996{present: Professor, University of Wisconsin

Jan. 1994{Aug. 1996: Associate Professor, University of Wisconsin

May 1991{Dec. 1993: Associate Professor, University of Ottawa

Sep. 1989{May 1991: Assistant Professor, University of Ottawa

Aug. 1987{Mar. 1989: Visiting Assistant Professor, Purdue University


Aug. 1997|July 2004: Associate Chair of Department of Statistics

July 2005|July 2009: Chair of Department of Statistics


Fellow: American Statistical Association

Fellow: Institute of Mathematical Statistics

Board of Directors: International Chinese Statistical Association (2000-2002)

President: International Chinese Statistical Association (2007)

Member: Management Board of Statistics and Its Interface (June 2007{present)


Editor: Statistical Theory and Related Fields (Jan. 2017{present)

Editor: Journal of Nonparametric Statistics (Jan. 2016{Dec. 2018)

Coeditor: Journal of System Science and Complexity (June 2014{Sept. 2019)

Associate Editor: Statistica Sinica (June 1991{Dec. 2008)

Associate Editor: Journal of American Statistical Association

(March 1993{Dec. 1996; Jan. 1999{Dec. 2005)

Coeditor: Sankhya (Jan. 2002{Dec. 2007)

Coeditor: Jounral of Multivariate Analysis (Jan. 2002{Dec. 2005)


Doctoral Thesis: Forty one Ph.D. students graduated between 1996 and 2020

Four in progress


206 published papers in refereed journals.

7 published textbooks and research monographs.

Ting Ye is an Assistant Professor in the Department of Biostatistics at the University of Washington. Her research centers around addressing modern complications in randomized clinical trials and hidden biases in causal inference. In randomized clinical trials, she has developed pragmatic and robust statistical methods for delayed treatment effect in cancer immunotherapy trials, survival-time-dependent missing covariates, monotone order constraints in stratified phase II cancer trials, and covariate adjustment under covariate-adaptive randomization. In causal inference, she has expertise in instrumental variables, difference-in-differences, sensitivity analysis, and data integration.

This entry was posted in . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *