Precision Loss for Repeated Measurement Designs due to Subject Dropout: A Design Selection Guidance

by Shi Zhao.

Abstract: Crossover designs are examined under the situation that subjects may drop out before completing the treatment sequences.A design class, called Uniformly Balanced Repeated Measurement Designs Balanced for Loss are introduced. We calculate the precision loss for uniformly balanced crossover designs under different model assumptions, i.e., fixed subject effects, random subject effects and autoregressive variance structure of random errors. They are considered to serve as design selection guidance for experimenters.

Key Words: Repeated measurement design, Crossover study, Latin square, Subject dropout, Precision loss

Author:
Shi Zhao, shi.zhao@xerox.com

Editor: Weiming Ke, Weiming.Ke@sdstate.edu

NOTE: This article was revised on September 24, 2010.

READING THE ARTICLE: You can read the article in portable document (.pdf) format (207478 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 1852 times since SEPTEMBER 24, 2010.


Return to the InterStat Home Page.