Corrections for FW error in regression


Fortran must die
There are various approaches for dealing with family wise error in regression. Does anyone know how these are implemented in SAS (I know how they are done in ANOVA but not regression in SAS)?


Not a robit
You are just asking about post hoc pairwise comparisons, correct?

You can just look up on the SAS source documentation how the unique corrections are calculated, and I think they are based on the same calculation regardless of the procedures, since most are just based on calculations on the alpha level or applied to the p-values. Am I missing what you are asking?


Fortran must die
I am actually asking about the code that generates the corrections (such as bonferoni or schafe) for regression :p I had always heard, though never seen it, that you should adjust the p values for regression when you arive at a final model by testing many other models first (which I always do). I am not talking here about the pairwise comparisons used in post hoc tests.