Review by Choice Review
The computer revolution has really hit statistics, not only in the obvious way of allowing more data to be processed, but in the sense of allowing entirely new processing methods. Traditional statistics allows one to answer one of a standard set of questions by computing some numbers from data, and then looking up the significance of these numbers in a precomputed table. Occasionally there is no standard statistical test available, or data clearly violate the assumptions used in constructing that test. Randomized techniques allow computer calculations of the equivalent of a statistical table for data by assuming that an appropriate null hypothesis holds, and then comparing actual data with simulated data. A tour of some of the usual domains of statistics shows how randomized tests can be used on each area, including comparisons, analysis of variance, and regression; and biological areas for which standard statistical methods are lacking, including distance matrices, distribution of organisms, existence of competition, and periodicity in time series. Each chapter includes Fortran programs. Although a first course in statistics is assumed, the author's clear writing style makes this book accessible to advanced undergraduates.-P. Cull, Oregon State University
Copyright American Library Association, used with permission.
Review by Choice Review
