## Residuals-by-Predicted Plot

SAS/INSIGHT software provides many diagnostic tools
to help you decide if your regression model fits well.
These tools are based on the
*residuals* from the fitted model.
The residual for the *i*th observation
is the observed value minus the predicted value:

The plot of the residuals versus the predicted values is
a classical diagnostic tool used in regression analysis.
The plot is useful for discovering poorly specified models
or heterogeneity of variance (Myers 1986, pp. 138 -139).
The plot of **R_GPA** versus **P_GPA** in
Figure 14.5 indicates no such problems.
The observations are randomly scattered above and below
the zero line, and no observations appear to be outliers.

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.