do the results of your methods vary more than the respective uncertainties of each method? I do not know what kind of instruments you were using and to which detail you worked into the quantification algorythms, but usually this is very tricky.
these methods also can have very different penetration depths. You should check that also...
The first obvious answer that comes to my mind is the different penetration depth of those techniques. XPS is surface-sensitive, so the signal that you get comes basically from the uppermost 1-2 nm of the sample. The other techniques penetrate deeper into the volume of the sample, depending on the detailed conditions of your experiment. The question that you should consider is if it is conceivable that your sample might have different composition in depth.