Westerns are semi-quantitative and I agree that they are MUCH more sensitive. IHC can be pseudo-semi-quantitative however you must be careful as often times people over-develop their slides and are no longer in a linear range of enzymatic activity and subtle differences between staining intensities may be lost.
Also westerns applications call for processing proteins very differently than for IHC so perhaps there are subtle differences in the antigen binding sites between the two processes that give you the discrepant results. Are you using the same primary?
If you truly want to find out if you have a difference you may want to go to a truly quantitative method such as ELISA etc.[/color]