A simple place to start is in PROC GLM or PROC GLIMMIX, depending on what assumptions you want to make about the error distribution of the Y variable, something you haven't mentioned. What is your Y variable? What distribution are the errors?
Anyway, if the variable Y has normally distributed errors, then use PROC GLM like this:
proc glm data=have;
class categoryvariablename1 categoryvariablename2
categoryvariablename3 binaryvariablename;
model y=continuousvariable categoryvariablename1
categoryvariablename2 categoryvariablename3 binaryvariablename/solution;
lsmeans categoryvariablename1 categoryvariablename2
categoryvariablename3 binaryvariablename;
run;
--
Paige Miller