ABSTRACT
This chapter discusses: three key aspects of model checking - evaluation of ideal inference conditions, identification of regression outliers, and identification of influential cases; why these key aspects need to be addressed before interpreting results; advantages and disadvantages of three types of model residuals that can be used as diagnostic tools for model checking; potential difficulties associated with using hypothesis tests to evaluate the ideal inference conditions of normality and equality of variances; how graphical displays can informally evaluate ideal inference conditions; a nine-step approach for evaluating ideal inference conditions for a simple linear regression analysis with an illustrative example and SAS programs; measures of influence and a SAS program that enables identification of potentially influential cases.
