Roger, at my suggestion, did something like plotted M6+ quakes >x thousand km away from each M6+ quake (I think we don't use aftershocks). There are some number of M6+ quakes every year not counting close aftershocks (100+?). So, there is some probability that a M6 will be followed by another M6+ within 24 hours. Probably relatively simple statistics for Roger to see if the actual occurrence is more clustered than expected by chance.
Chris