Abstract
Analysis of variance, simple correlation, and multiple regression, though ordinarily construed as distinctly different statistical strategies, are demonstrated to be mathematically equivalent, insofar as each method arrives at the same variance accounted for. Conceptually, the three methods are only different procedures to divide the same total variation among a set of observations. As a practical matter, the methods differ in their suitability to analyze data formatted in a particular manner. Underlying these approaches and many of the statistics used in the behavioral sciences is the general linear model, which means that data are organized in a linear fashion. A linear rule is the most parsimonious reduction of complex data but at the same time tends to ignore information not predicted by the rule.
Get full access to this article
View all access options for this article.