Ah. I wasn't aware you were looking for anecdotal evidence.
Thanks for offering, but actually I prefer statistical regressions. By the standards of anecdotal evidence, even astrology can be useful. I have seen New-Age-ish couples successfully use astrological compatibility charts to maintain their relationships. Nevertheless, we both agree astrology is garbage. Granted, I would expect that Myers Briggs work better, and I've said as much in my first post to the other thread. Unlike astrology, Myers-Briggs does pay attention to input relevant to your personality, which should at least to some
good. Still, anecdotal evidence alone is fairly weak.
My confidence in Myers-Briggs would be much improved if you could point me to studies where, say, the performance of high-school students in particular classes is correlated to their Myers-Briggs types. For example, if the test can demonstrate with statistical significance that INTPs are better math students than ESFJs, and that ESFJs are more likely to be class presidents, that would help.
Or better yet, make it a prediction, not a correlation. Divide the testers into two groups. Have Group 1 administer the tests to the students. Then give the results to Group 2, and let the psychologists in it predict
who the best three math students are and who the class president is. I know psychology isn't physics, and I know that their predictions can't be perfect. But if together the Myers-Briggs testers can split the difference between perfection and a coin toss, that would be clear and convincing evidence that the test works.