GLM Contrast/Estimate Adjust alpha

[Kastchei]

Hello!

I was wondering if there is anyway to apply multiple comparison adjustments to the F and t test done by the CONTRAST and ESTIMATE statements in PROC GLM.  Options like bonferroni, tukey, scheffe, dunnett are available in the MEANS and LSMEANS statements, but not with the CONTRAST and ESTIMATE statements.

Thanks,

Michael

[SteveDenham · Accepted solution]

I would strongly recommend that you shift over to PROC GLIMMIX or PROC MIXED.  The multiple comparison adjustments are available with the LSMESTIMATE statement.

Steve Denham

[AnalytX]

Hello!

One other option would be to use PROC MULTTEST which addresses the multiple testing problem: SAS/STAT(R) 9.3 User's Guide

Best,

[SteveDenham]

As a follow-up, the MULTTEST approach would be an extension of Example 61.5 Inputting Raw p-Values.  You will need to output all of the contrast or estimate values to a dataset (ODS OUTPUT is your friend), and rename the probt variable to raw_p.  Specific multiple comparison methods, such as Dunnett's, Tukey's and Scheffe's methods are not available in MULTTEST, and only Scheffe, Sidak, and Dunnett (using the simulate option, and correct specification of the LSMESTIMATE statement) are available in GLIMMIX through the LSMESTIMATE statement.  Tukey's adjustment isn't available in either procedure.

Steve Denham

[Kastchei]

Thanks, gents for the very quick replies.  This is exactly what I need.

Michael Cooney