My problems with the MBTI are similar to my problems with socionics:
1. Theories that have no supporting empirical evidence. First, it is assumed that because Jung said the auxiliary function is different "in every respect" from the primary function, that the auxiliary function has to be extraverted. If that's the case, why didn't Jung explicitly state that it's extraverted? Since what he said is ambiguous, the best way of verifying what he said would be to develop a test that measures the functions, and then determine if there is a trend as to which function is auxiliary. Otherwise, you would have to observe a large number of people, and infer that they exhibit this characteristic from their behaviour. Since they don't tell us how they came to their conclusions, we don't know if what they're saying is empirically true.
The second problem is that it's assumed that, for example, INTP = . Why couldn't INTP = ? It's explained that Katharine C. Briggs, who developed this theory, looked at the outer lives of her "meditative friends", and they all showed extraverted characteristics in their outer life. It says that she further verified that that an introvert with a dominant perceiving process has an auxiliary judging process, and vice versa. Finally, she says that she found that the auxiliary function gave way to the characteristic judging/perceiving behaviour What isn't stated is how she discovered much of this. Looking at the lives of your friends is hardly an accurate way of supporting a theory. Once again, it doesn't seem that any empirical testing was done to verify this. And, in fact, www.socionics.com seems to promote the opposite view. Surely, the people at that site must have some reason to believe what they believe. Without evidence of empirical verification, how can we say who is right?
2. There is no explanation of how a type uses all eight preferences. I don't believe that everyone doesn't use four functions at all. For example, if someone had no access to Fi, it seems to me that they wouldn't have any internal values. The only people who are like this are probably sociopaths or someone with brain damage. I think, in this sense, socionics is superior to MBTI theory. At least socionics can account for how all eight functions are used.
3. The last problem doesn't have to do with the system itself, but how people apply it. A lot of people tend to apply the rules in black and white terms. For example, some will tell you that since thinkers make more logical decisions than feelers, thinkers almost always have an easier time thinking logically than feelers. While there might be a correlation between these two factors, it shouldn't be treated as a rule. Consider this thread on an INFP forum: http://infp.globalchatter.com/messag...er=asc&start=0. While thinking types might tend to score higher, there were still people who scored 9 or 10 out of 10 on the test, and quite a few people did well; the trends don't apply to everyone.
People also have a tendency to think that the MBTI can explain any relevant behaviour. Sure, in general, thinkers might have an easier time with logic, but wouldn't someone's IQ be a better explanation? I don't believe that a feeler with an IQ of 150 would get a low score on that test. Therefore, IQ is probably a better explanation of someone's logical abilities, and while there is a very slight correlation between IQ and thinking, I've seen statistics that show that there are many feelers with high IQs.