Article,
Thesis chapter 1: Introduction
Thesis Chapters, (July 2022)translation.
Abstract
Introduction to Thesis by Dietmar G. Schrausser, Permutation tests: Theoretical and practical functioning of permutation methods in the independent 2-sample problem.
To draw statistically reliable conclusions on parameters in one or more samples, one has been dependent on the logical concept of asymptotic approximation. However, the associated assumption about the distribution forms of characteristic values from samples of a population places high requirements on the data material: The sample data must (a) follow an at least symmetrical, if not normal, distribution, (b) must be homogeneous in their variances (c) be of sufficient size (d) and through a random process materialized. If these conditions are not met, probability statements about the occurrence of parameters or parameter differences are no longer logically defined. This fact is of crucial importance in psychological research, since one rarely deals with raw values that meet the requirements of the asymptotic approach. Ways out represent non-parametric procedures or correction algorithms, the application of which, however, puts the significance of the results into perspective. However, a way of statistical inference that has been neglected up to now is increasingly coming to the fore-the probability definition of location parameter differences via permutation of the raw data.
