AN INVESTIGATION OF THE ACCURACY OF THE PEARSON SELECTION FORMULAS

Abstract
The Pearson formulas for correcting correlation coefficients for restriction of range are based on crucial assumptions of linearity of regression and homoscedasticity of the error distributions. Some small previous studies have suggested that these formulas are reasonably accurate providing extreme selection is not involved. These studies tend to suggest that the formulas typically provide undercorrections in most instances. The present study involved two vary large data sets and attempted to verify the accuracy or inaccuracy of these formulas and the assumptions on which they are based for both moderate and extreme selection.

This publication has 0 references indexed in Scilit: