Calling it a paradox makes it sound as though something difficult to understand is happening. In most cases it is simply necessary to make sure that you are comparing 'like with like', that is, comparing things that are of the same class and that are actually comparable.

Of course t=it is still valuable to be reminded that averages can make it seem that things are of the same class when they are in fact not. So it gets one lonely upvote from me.