@gollum,
I think some judges have done this, but not in any kind of systematized way. It is more in the nature of a judge saying "I've presided over (x number) of jury trials where I disagreed with the jury's verdict." Judges, however, are certainly not required to report how many times they would have ruled differently from juries. The most definitive indication of disagreement, as
sturgis points out, is when the judge overrules the jury by means of a
judgment N.O.V., but I doubt if there are any statistics on those. If you're looking for a study where an analysis is made between how the jury ruled and how the judge would have ruled, I don't think it exists.