Name: Voss

Firstname: Dan

Title: Professor, Statistic Program Director

Institution: Wright State University

Street: 3640 Colonel Glenn Highway

City: Dayton

Zip-Code: 45435-0001

Country: USA

Phone: 937-775-2958

Fax: 937-775-2081

Email: dan.voss@wright.edu

Authors: Weizhen Wang and Dan Voss

Title: Control of error rates in adaptive analysis of saturated factorial designs

Abstract: Many methods have been proposed for the analysis of saturated factorial designs. These include some nonadaptive methods which provide strong control of error rates for testing and confidence interval estimation for individual and simultaneous inference, but similar results for adaptive methods are desired. In a landmark paper, Russell Lenth (1989) proposed a method that is appealing and popular because it utilizes an adaptive estimate of dispersion which should be more robust to the presense of a few large effects. It remains a challenging problem to investigate the strong control of error rates for such adaptive methods. Modest progress has been made in this regard. Wang and Voss (2001) established control of error rates for a rather restrictive class of adaptive methods. This result and subsequent extensions will be presented.

References: Lenth, R. V. (1989). Quick and easy analysis of unreplicated factorials. Technometrics, 31, 469-473.

Wang, W. and Voss, D. T. (2001). Control of error rates in adaptive anqalysis of orthogonal saturated designs. Annals of Statistics, 29, 1058-1065.