• Print |
• E-mail  |
• Bookmark

If the same variable list is used on both the MODEL and RANDOM
statements, then PROC MIXED will produce incorrect results for the
parameter estimates of any covariates that are listed on the MODEL
statement after the variable list.  For example, if the MODEL and RANDOM
statements are as follows:

   model y=x1-x5 z1 z2 z3 / s;
   random int x1-x5 / subject=a;

then the parameter estimates for Z1, Z2, and Z3 will be incorrect.

Additionally, the variance component for the intercept term on the
RANDOM statement will be incorrect.

To circumvent this problem, list the variables individually on the
RANDOM statement, or download the available hotfix corresponding to
the SAS Release you are running.

A Technical Support hot fix for Release 8.2 (TS2M0) for this
issue is available at:

http://www.sas.com/techsup/download/hotfix/82_sbcs_prod_list.html#011541

A Technical Support hot fix for SAS 9.1 (TS1M0) for this
issue is available at:

http://www.sas.com/techsup/download/hotfix/b9_sbcs_prod_list.html#011541

Operating System and Release Information

A fix for Release 8.2 (TS2M0) for this issue is available at:

A fix for SAS 9.1 (9.1 TS1M0) for this issue is available at: