I don't have a dog in this fight, but it seems to me that this would be extraordinarily simple to test (at least statistically): take the performance from any number of games (same players), and see how many runs are produced when the order is changed. Wouldn't that provide a reasonable test? Of course, you could argue that each AB is situational, so that changing the order of 'what happens' isn't valid. But if you argue that, you are really assuming your conclusion: the argument (you would be saying) is untestable. I'm sure you've gone over this before--sorry, I haven't been following it-- but which of those is your position? (1) That the test (rearranging results) do indeed show a significant difference in batting order? or (2) that it is finally untestable?