23 Graphical methods for linear models

Publisher Summary Graphical displays are indispensable tools for data analysis in general and for linear models in particular. Graphical displays and numerical analyses enhance each other. They are complementary and one should not be substituted for the other. A data analysis is rarely thorough and complete without graphical displays. This chapter presents several graphical displays that guide data analysts while going through the iterative and dynamic regression process and also presents several graphical displays under two headings: (a) graphs before fitting a model to the data and (b) graphs after fitting a model to the data. Several examples illustrating the purpose and usefulness of each graphical display are provided in the chapter. The emphasis is on the way to read and interpret graphs. These graphical displays can serve one or more of the following purposes : uncovers errors in the data, recognizes patterns in the data, discover new phenomena, explores relationships among variables, assesses the adequacy of the model fitted to the data, suggests remedial actions, conveys the properties of a fitted regression model, and enhances numerical analyses in general.