Department of Medical Statistics
University Medical Center Göttingen
Asymptotic Permutation Tests in General Factorial Designs
In general factorial designs where no homoscedasticity or a particular error distribution is assumed, the well-known Wald-type statistic (WTS) is a simple asymptotically valid procedure. However, it is well known that it suffers from a poor finite sample approximation since the convergence to its χ2 - limit distribution is quite slow. This gets even worse with an increasing number of factor levels. In this talk we discuss a modified permutation approach to improve the small sample behavior of the WTS maintaining its applicability to general settings as crossed or hierarchically nested designs. In particular, it is shown that this approach approximates the null distribution of the WTS not only under the null but also under the alternative yielding an asymptotically valid permutation test which is even finitely exact under exchangeability. Finally, its small sample behavior is compared with competing procedures in an extensive simulation study. A real data example illustrates the application of the proposed methods.
Sponsored by the Department of Mathematical Sciences
Host: Robert Serfling
Refreshments will be served in Room FO 2.610F 30 minutes before the talk begins.