Book Review: Non-parametric analysis of longitudinal data in factorial experiments
Abstract
Statistical Methods in Medical Research 2005; 14: 603-604 Book reviews Brunner E, Domhof S, Langer F 2001: Non-parametric analysis of longitudinal data in factorial experiments. New York: John Wiley. 288 pp.US $115.00 (HB). ISBN: 0 471 44166 X. This book's main theme is the nonparametricanalysis of designed experiments with replicated measurements on the same subject. It is a nice andreadable book. The authors try to make the results of contemporary research accessible to scientistsfrom various fields. The methods presented have wide applicability in different experimental disci-plines, and these methods are well illustrated by real data examples.Not surprisingly, nonparametric foundations of the presented methods are based on several notionsclosely related to ranks, yielding desirable insensitivity to model assumptions. The most innovativepart comes from the fact that the approach is used not only for testing, but also for the estimationof nicely interpretable quantities. To this end, the `relative treatment effect' (RTE) plays a central role.The RTE based comparisons are not too complicated computationally; they are also rather generaland unit-free. One can plot and compare them in a meaningful way. The book offers a whole strategyfor their practical usage, which is not dissimilar to the strategy of using classical linear models