I think most people, when requesting a citation in this context, would also be fine with a description of an experiment (preferably a repeatable one) and a record of the empirical result of that experiment. That's the "base case" you're talking about in the otherwise infinite regression of citations.

But this offered nothing other than some slight preference with a compelling argument that the differences in spending are more obvious. Yet, even in stating this, the conceit is held that "higher mpg is always better."

So, my citations desire would best be settled by any sort of empirical study showing that they actually affected buyers in the desired way.