For what it's worth, the bad stats method generated a much more readable table for those of us using screen readers. Info presentation may possibly be more important than correct use of statistical calculation methods, particularly if the difference is at most 0.15%. I'm just sayin'