The Case for Non-Zero Restrictions in Statistical Analysis

Abstract

The focus of this paper will be on the utility of making a non-zero restriction. Why the zero restriction occurs so frequently will be questioned and hopefully researchers and statisticians will see how the zero restriction limits the conclusions of the research. The argument will be made for making non-zero restrictions, resulting of course from "non-zero" research hypotheses The argument will be made for each of these statistical procedures: two group t test, Pearson correlation, single population mean, one-way analysis of variance, and interaction.