Leverage Plots for General Linear Hypotheses

Abstract

Abstract Leverage plots are a generalization of partial-regression leverage plots, extending the idea to apply to general linear hypothesis tests. Leverage plots can show the point-by-point composition of the sum of squares for a hypothesis test. They are valuable in revealing the degree of fit, the parameter estimates, the residuals, a measure of the variance of the fit, influential points, nonfitting points, nonlinearities, and even collinearity.