I am somewhat ignorant of how/if voting data on SO can be mined, so I was wondering if it's possible to run a search of voting patterns to see if people tend to up-vote less when the author is in their approximate rep bracket (either sub-consciously, or on purpose).
If you ask "why would anyone do something ilke that" - I actually have a plausible theory.
Joel at least once mentioned, when promoting SO, that the SO rep these days is sometimes used as a sort of score when job hunting (I know I'd at least pay attention to that datum on a resume when I interview people, although it won't be even remotely most important).
So people may treat their rep standing (if not exact value) as something of a scarce resource and - possibly sub-consciously - try to deny that resource to closest competition (as in, you don't have to be faster than a bear, just faster than the fellow next to you :)
Jon Skeets of the world are way out of reach anyway so up-voting them does no harm as far as one's own relative rep standing, whereas up-voting much lower rep users is safe enough since they will be unlikely to catch up to you. But up-voting someone with score similar to you may feel like helping the competition.
I don't know if it's possible to mine for that, but I have observed a couple of times when people "accepted" obviously not-the-best answer from either very low or very high rep user yet left alone a much better one from similar reps (that's what started me wondering).
EDIT: Just to be clear, in response to Brad's note, I don't expect (or want :) to see the voting data - I assume that very few people (may be only 2) have access to it.