I hesitate to even post in this thread because it's going to perpetuate it, but I suppose it's already resurrected so why the hell not.
--
We use type test statistics and correlation with other measures (IQ, % population, etc.). We also state how the type tests are terrible a measure of actual type.
We then claim the former to be meaningful anyway, disregarding the latter assertion when it suits our purposes to do so.
We're fantastic. Go us!
If (if!) we believe in JCF extensions to MBTI, and if
(if!) we actually give a fuck about this question, the actual answer lies in facing this problem and reconciling it.
Off the top of my head, here: It's a matter of conditional probability. We have an answer to the question--"What's the correlation between tested type and IQ?" We don't have an answer to the question--"Given that one tests as a certain type, what's the probability that he
is that type?" Depending on
that probability, virtually every conclusion we make regarding IQ, population, etc. based on type statistics could be pretty meaningless, regardless of what the correlation table between type and IQ tells us.
For the record, the
g factor has some merit.