It seems to me that the LSMEANS statement in GLM would be better suited to giving you the values you are looking for. In order to do this, you would need to go back to the original Race/Sex/Education variables and put them on the CLASS statement (e.g. do not create dummy variables). Once you have done that, you should be able to easily combine the LsMeans results using the example linked below as a guide.
30698 - How do I combine the LSMEANS and differences in LSMEANS from Proc GLM using Proc MIANALYZE? (sas.com)
... View more
