We discuss the use of selected modern graphical methods for enhancing the use of residual and added variable plots to assess the adequacy of a linear regression model. Emphasis is placed on rotation in up to four dimensions. It is shown that three- and four-dimensional residual plots can be an effective supplement to standard two-dimensional diagnostic plotting. Situations in which the standard plot of residuals versus fitted values can be expected to diagnose model deficiency are studied as well.
Bibliographical noteFunding Information:
This work is supportedi n part by National ScienceF oundation Grants.
- Added variable plots
- dynamic graphics
- linear models