multiple comparison issue in linear regression


New Member
I am puzzled by the following question:

If you have a 3-level categorical predictor in a linear regression model, you will get 2 estimated coefficients for: say, level 2 vs level 1 and level 3 vs level 1, and you will have a p-value for each coefficient. It seems to me that this is multiple comparison just like doing multiple pairwise comparisons after ANOVA. But people usually just report the p-values associated with the coefficients without correction for multiple comparison. Is this right? If not, how should we adjust for multiple comparison in this setting? Any input will be greatly appreciated!