Usage Note 22910: Some possible reasons for differing results from PROC GLM and PROC REG
In general, results from PROC GLM should be identical to those from PROC REG. Below are the most common reasons why differences occur:
- Failing to create dummy variables in PROC REG to represent CLASS variables specified in PROC GLM.
- Incorrect coding of dummy variables. See "Parameterization of GLM Models: Main Effects" in the Details section of the GLM documentation.
- Not specifying the same model in the two procedures. Note that you must create new variables in your data set for use in PROC REG to represent interaction or nested effects such as A*B or B(A) specified in PROC GLM. See "Parameterization of GLM Models" in the Details section of the GLM documentation.
- Rarely, results may differ due to differences in the check for singularity between the two procedures.
Operating System and Release Information
*
For software releases that are not yet generally available, the Fixed
Release is the software release in which the problem is planned to be
fixed.
| Type: | Usage Note |
| Priority: | low |
| Topic: | SAS Reference ==> Procedures ==> REG Analytics ==> Regression SAS Reference ==> Procedures ==> GLM
|
| Date Modified: | 2019-07-18 14:53:37 |
| Date Created: | 2002-12-16 10:56:40 |