Problem Note 7028: Incorrect results from IPF estimation if response variable has more than
256 levels
If you use the IPF estimation method (by specifying the ML=IPF option in
the MODEL statement) and if you have a response variable with more than
256 levels, then the results, including the Likelihood Ratio test,
parameter estimates (if requested by the ML=IPF(PARM) option), and
predicted values, may be incorrect. Improper ordering of the
populations in the "Response Profiles" or "Maximum Likelihood Predicted
Values for Frequencies" tables is evidence of the problem.
A Technical Support hot fix for Release 8.2 TSLEVEL TS2M0 for this
issue is available at:
http://www.sas.com/techsup/download/hotfix/82_sbcs_prod_list.html#007028
Operating System and Release Information
| SAS System | SAS/STAT | Microsoft Windows 2000 Server | 8 TS M0 | 9 TS M0 |
| Microsoft Windows 95/98 | 8 TS M0 | |
| Microsoft Windows 2000 Datacenter Server | 8 TS M0 | 9 TS M0 |
| Microsoft Windows 2000 Professional | 8 TS M0 | 9 TS M0 |
| Microsoft Windows 2000 Advanced Server | 8 TS M0 | 9 TS M0 |
| ABI+ for Intel Architecture | 8 TS M0 | |
| OpenVMS VAX | 8 TS M0 | |
| OS/2 | 8 TS M0 | |
| OpenVMS Alpha | 8 TS M0 | 9 TS M0 |
| Tru64 UNIX | 8 TS M0 | 9 TS M0 |
| Microsoft Windows NT Workstation | 8 TS M0 | 9 TS M0 |
*
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: | Problem Note |
| Priority: | alert |
| Topic: | Analytics ==> Regression
SAS Reference ==> Procedures ==> CATMOD
Analytics ==> Categorical Data Analysis
Analytics ==> Longitudinal Analysis
Analytics ==> Multivariate Analysis
Analytics ==> Analysis of Variance
|
| Date Modified: | 2002-07-23 09:42:06 |
| Date Created: | 2002-03-01 10:04:31 |
