An E-mail from Rob Low
A brief comment on Ferguson's formula. It is obviously open to
fairly silly misuse. For example, you quote a negative correlation between Mobius Test
scores and Mega scores: it would be easy to combine test scores on the two and get a
ridiculously inflated result simply by failing to take into account the fact that the mean
of the samples was up in the region of IQ 150, and that the computation had to be done
taking this into account.
That said, it shouldn't be difficult to give an explicit means of
combining all the acceptable tests--with only three tests, there are only four
combinations to worry about, and the data must exist for doing something moderately
sensible regarding the correlations.