ABSTRACT
The methodology for framing and testing theoretical predictions-subsuming tasks often identified as data analysis or hypothesis testing-has been a topic of intense concern throughout the 20th century. Following earlier insights (e.g., Student, 1907), Fisher (1935) systematized procedures that came to be known as experimental design and the analysis of variance. At the same time, parallel developments-designed to answer questions in genetics and biological inheritance-were being made in correlational methods and regression analysis. Since the 1970s, the underlying similarity of these approaches has been amply noted. Indeed, both the analysis of variance and regression analysis use variants of the general linear model, a very flexible representation of data that underlies many of the most commonly used statistical tests employed in psychology and the social sciences. The ways in which t tests, linear correlations, the analysis of variance, and multiple regression analysis can be formulated through the general linear model are well known, due to expositions by Cohen (1968; Cohen & Cohen, 1983) and others (e.g., Darlington, 1968).
