Previous Page  Next Page 
Fit Analyses

Summary of Fit for Generalized Linear Models

For generalized linear models, the Summary of Fit table, as illustrated by Figure 39.12, includes the following:

Mean of Response
is the sample mean, {\overline{y}}, of the response variable.

SCALE
is the constant scale parameter specified in the method dialog or a value of 1.0 for maximum-likelihood estimation for Poisson or binomial distributions.

SCALE (MLE)
is the maximum-likelihood estimate of the scale parameter for normal, gamma, and inverse Gaussian distributions.

SCALE (Deviance)
is the scale parameter estimated by the mean error deviance.

SCALE (Pearson)
is the scale parameter estimated by the mean Pearson \chi^2.

Deviance
is the error deviance.

Deviance/DF
is the mean error deviance, the error deviance divided by its associated degrees of freedom.

Pearson ChiSq
is the Pearson \chi^2 statistic.

Pearson ChiSq / DF
is the mean Pearson \chi^2, the Pearson \chi^2 divided by its associated degrees of freedom.

When the scale parameter is a constant specified in the method dialog, or when the response has a Poisson or binomial distribution, the table also contains the scaled deviance and the scaled Pearson \chi^2:

Scaled Dev
is the error deviance divided by the dispersion parameter.

Scaled ChiSq
is the Pearson \chi^2 divided by the dispersion parameter.

fit12.gif (10549 bytes)

Figure 39.12: Summary of Fit and Analysis of Deviance Tables for Generalized Linear Models

Previous Page  Next Page  Top of Page

Copyright © 2007 by SAS Institute Inc., Cary, NC, USA. All rights reserved.