What about cases where there was a circuit split and the Supreme Court took a case from a different circuit? Those should count. Say there is a split with all the even circuits on one side, and all the odd circuits on the other. If the Supremes take a case from the Second Circuit, and reverse that Court, that should count in the plus column for the all the odd numbered circuits

Great point - I think you're right. That analysis would be more difficult to perform (which isn't to say that it wouldn't be worthwhile) just because the data would be harder to collect. Imagining the research involved, to see where Circuits aligned and departed, and if they broke into two or three or four camps on an issue, and deciding which Circuits "won" if the Court adopted rationales from one or more of them in whole or in part, is making my head hurt a little just thinking about the scope of the problem. Still, you're right to note that mere arithmetic analysis of review and reversal stats is an imperfect tool.