flavio,
Your math and your logic make perfect sense, except for one assumption you've made that I think can't be overlooked. You assume that someone voting "no" means they really want Davis as their first choice. On the surface, that seems like a reasonable assumption... but is it really?
I can't help but wonder: how many Californians voted "no" for the recall simply because they think recalling a govenor in the middle of the term just isn't the right thing to do? I could see myself voting in such a manner - disagreeing with the recall on ethical or other grounds, regardless of whether I liked Davis, hated him, or was indifferent.
Therefore, it is entirely possible that some voters did want Arnie as their first choice, but thought the recall was in general not the right thing to do; thus, they could have voted "no" and still voted "arnie" in case the recall did happen.
Since there doesn't seem to be any way to really know how many of the voters actually wanted Davis as their first choice, or how many just disagreed with the recall in general even if they would have rather had another govenor, I don't see how we can take your assumption as any objective indication of how many voters preferred Davis or Arnie.
I simply can't see any clear way to get that information out of the results, due to the way the voting ballot was set up. Also, I have to really wonder why Davis didn't want his name among the recall candidates (wasn't that his choice?). Surely someone informed him of the mechanics of such a dual vote, and explained that not having his name among the recall candidates put him at a disadvantages. OTOH, if it wasn't his choice, if he wasn't "allowed" to have his name on that list, then that was a pretty botched up election.