A Fortran Program for Hypothesis Testing in the General Linear Model

IN recent years, researchers in education and psychology have become aware of the fact that many statistical analyses (e.g. fixed model analysis of variance and covariance, unequal cell frequency experimental designs) can be more easily approached in terms of a full rank regression model than the traditional analysis of variance model. Cohen (1968) described tHe methods for expressing nominal scale (e.g. group membership) data as independent variables as well as procedures for handling interactions, curvilinearity, and missing data for either correlated or uncorrelated independent variables under the multiple regression format. The program to be described enables a researcher to test a wide class of statistical hypotheses in terms of the full rank regression model. Given the familiar linear model