Michael Eriksson
A Swede in Germany
Home » Humans | About me Impressum Contact Sitemap

Musings on the Myers-Briggs Type Indicator


This article deals with a number of topics relating to MBTIw, including distribution, relative value of various types, and problems with the tests and the classification.

The article began with the section on frequencies, and has since expanded as my knowledge grew. If and when I revisit the subject again a further growth and improvement is likely, including refinements and corrections of opinions (considering how complicated the field is below the immediate surface, I fully assume that my opinions will change, at least in detail).

A side-effect of this growth is that the contents are a bit haphazard at the moment.


Making some language fixes in 2023, many years after the last prior changes, I find that I have forgotten virtually everything that I once knew—to the point that I could not now explain the reason for many of the below claims. (The more the shame, as some of them must be read with an implied disclaimer of “holds if the assumptions of MBTI and the accepted wisdom within the MBTI crowd are true” and look almost like astrology without that disclaimer and the contextual knowledge that I once had: Capricorns naturally tend to X, Y, and Z, while Pisces [etc.].)

Any non-trivial revision would require a considerable refreshing of knowledge, and, contrary to my original optimism, is unlikely to happen. If in doubt, the personality frameworks that I have encountered so far have been lacking in stringency and objectivity. (Note e.g. some discussion of problems with MBTI below.) They might well bring value, but not at the level of e.g. an IQ test, and a further investment of my time does not seem worthwhile.

(In an interesting complication, I suspect that the idea of multidimensional frameworks is problematic and that it would be better to have more focused approaches, each investigating just one specific aspect of personality.)

Relative frequencies

On Myers-Briggs_Type_Indicatorw I found the following listing of the frequencies of the various four letter combinations (see that link for meaning and background information; all numerical values in percent):


I noted with some minor surprise that my own category (INTJ) was not the smallest (ENTJ and INFJ being even smaller), and with major surprise that “the I’s have it”. That introverts (the I’s) are in a majority, contrary to most other claims, is likely rooted in the technical meaning of “introvert” in this specific context which deviates from the “everyday use”; however, I cannot rule out a sample skew or other problem.


This section was written when I had no deeper understanding of the individual profiles and I relied on a test. After reading up (which is reflected in other sections of this page), I have some doubts as to whether I am a J—the INT part, however, is unambiguous.

Intrigued with this, I used an Awk snippet to calculate some accumulated proportions (listed below). A “.” implies any character; the number is the sum of the corresponding actual values; numbers in parenthesis are the result of multiplying the individual probabilities of letter combinations; numbers in brackets are the theoretical value assuming that all letters have equal probabilities and that the dichotomies are entirely uncorrelated:

INT.5.4( 5.5)[12.5]
IN.J3.6( 7.4)[12.5]
.NTJ3.9( 5.8)[12.5]
INTJ2.1( 3.0)[ 6.25]

(With reservations for arithmetic and transcription errors.)

Notable observations include that introverts and extraverts are equally common (again, note the difference in definition compared to everyday use), and that only the S/N division shows a considerable deviation from 50–50. What really fascinates me is that, contrary to my expectations, the individual dichotomies seem to be comparatively independent of each other: Of the investigated minority of all combinations, only the .N.J combination (and, naturally, the other combinations that include .N.J) shows a major deviation from the multiplied probabilities. (Unfortunately, I do not have the math tools needed for a proper regression analysis at the moment, but such an analysis would be likely to show numerous weak dependencies. A complication could be that the I/E division affects the meaning of the P/J division, cf. the Wikipedia link.)

This exercise was in many ways an eye-opener to me, because I had a strong pre-conceived opinion that there would be very clear patterns, e.g. that I...’s would be disproportionally likely to be I.T.’s, or that .NTJ’s almost always would be INTJ’s.

An interesting observation is that the .NT. combinations amount to only 10.4 %—to contrast with 25 % in a perfectly equal and random distribution (i.e. each letter of a two-letter choice has a 50 % probability of matching, and each two-letter choice is independent of the others). This is critical, because it can be argued that .NT. is an objectively superior combination, at least in a post-neolithic context: N is better than S, because the former allows a greater and freer thought, while the latter is too restricted to the immediately observable and tangible; T is better than F, because it represent rational thought with a higher degree of objectivity and less disturbance through emotional noise. (A similar objective classification does not appear possible for the I/E and J/P dichotomies.) In effect, the best modern humans (with regard to one particular aspect; other aspects, e.g. IQ, exist) remain at 1 in 10, when they should statistically be 1 in 4—and would be 9 in 10, were humans more highly evolved. Add to this the Tall Dancer syndrome, and quite a lot of the problems in the modern world can be explained in one stroke.


Again, it is important to keep the technical meanings of intro-/extravert in mind: Using the more everyday meanings, a strong case can be made that introverts are superior.

Hiring and promoting

MBTI is a commonly used criterion when considering hiring and promoting decisions. Below some approximate guide-lines are given. Beware, however, that it can be argued that MBTI is too primitive an instrument to be reliable for these purposes; that the test results are not always clear-cut and stable over time; and that they indicate preference, not actual ability. Someone who is an INT. by preference, but has an IQ of 80, is unlikely to make a good decision maker; while an ESF. with an IQ of 150 could well be an excellent choice.

For qualified positions with all other factors reasonably equal:

It is only rarely correct to choose an ..F., instead try to go with a ..T.: Rational thinking is of paramount importance; and a focus on emotions is a liability in most qualified positions, while sometimes an adequate substitute in other positions (say, in primary or secondary education).


It is easy to go down the road of “..F.’s are better at dealing with feelings; ergo, they are better at dealing with humans; ergo, they make good managers.” or similar.

This argument, however, is naive on several counts, including that a focus on emotions does not automatically give someone social skills (but, on the contrary, ..T.’s are often better at them); that a match in type is often more valuable than social skills, and in e.g. an engineering or software development context (with INT.’s) an ..F. has a built-in handicap; and that, as argued elsewhere, selecting managers based on social skills can be a dire long-term mistake—if more social skills are needed, it is better to hire a “communicator” to assist the manager.

The dichotomy .N../.S.. is less clear-cut, but the more qualified the position, the more important it is to chose an .N.. Also bear in mind that by hiring too many .S..’s in lower positions, the pool of good candidates for later promotions can be diminished. A particular danger is that the stereotypical .ST. (beware of individual variations) is too rule-oriented and inflexible; and thus, while valuable as follower, can be very problematic as a leader: Rules are ranked above reason, position in a hierarchy has greater weight than actual competence, opinions once established are immutable, etc., which can be extremely frustrating for those more intelligent or with an .NT. mentality—and prevents new ideas, improvements in procedure, and similar, from coming to fruit. Similarly, they are often hell-bent on being considered right on every issue, regardless of counter-arguments, which brings similar problems.


A corollary of these two rules of thumb: Decision makers should be .NT.

The two remaining dichotomies will often have no objectively superior choice, but the choice must take the specifics of the position (and potentially later held positions) into account. In particular, a position requiring thought and decision making is usually best left to an I... (in particular, an INT.); a position that requires more “doing” is often better given to an E...: Go to an INT. to find out what to do; then relay his recommendation to an ENT. to actually get it done.

A particular danger here is that E...’s are chosen, with some justification, for positions in middle or project management; and that promotions to upper management, where I...’s are preferable, is done from among the E...’s in middle and project management.


Beware that the E/I and P/J dichotomies can vary considerably in their effect, based on what the other letters are—and that the choice must, therefore, also take these into consideration. In particular, I do not feel confident to make blanket recommendations regarding them at the time of writing.

A problem, as discussed elsewhere, is that ESF.’s and close-by personalities are often over-promoted by corporations, because they tend to be better at networking, making superficial impressions, and similar (and, obviously, Tall Dancer). This is highly unfortunate, because they tend to be poor decision makers.

Problems with questions in tests

Deficient questions are common in various tests (not restricted to MBTI). Consider a typical example for the I/E dichotomy: Do you find it easy to make friends?

This question has at least the following issues:

  1. MBTI focuses on preference of ability, not level of ability, which implies that the answer will only be an indirect indicator.

  2. Most tend to over-estimate their own abilities—and the more incompetent and unintelligent to a higher degree. Correspondingly, the answers to this question will likely have systematic errors, in that there are too many positive answers and that there is a negative causal component between the answer (not necessarily ability) and competence/intelligence.

  3. It overlooks that the level of ability can depend heavily on non-inherent factors, say amount of socializing in the past, age, amount of relevant literature studied, etc. Few INT.’s, e.g., are likely to go through the effort of learning how to gain friends—but those who do can easily reach the ability of the average E...

Improvements that can be made include speaking of preference (“Do you like to make new friends?”) and focusing on the natural ability (e.g. by explicitly saying “natural(ly)” or referring to behaviour/preference/ability at a younger age).

Other common complications include ambiguous questions; questions that can often go either way depending on a momentary perception, current mood, mood at the time of the event, ...; and questions where most will misjudge their own strength or weakness entirely.

Notably, one the best question there is for judging character has always been absent: When reading a question, do you often feel an urge to point out problems with or otherwise analyze the question, it self, rather than just answering it?

Problems with MBTI

MBTI is a much too restrictive and complicated tool; in particular as it does not factor in e.g. intelligence and psychiatric disorders. Consider e.g. that an unemotional thinker with lower intelligence might be a psychopath and have a mentality that is good for intrigue, fraud, and similar; while one with high intelligence will typically have very highly developed moral standards. (Note that these are not orthogonal criteria that can be left out to leave an over-simplification, but that actually severely influence how an MBTI personality functions.)

The types are not all of equal value

Usually, those involved with MBTI go to great lengths to stress that all types are of equal value: They are different, but not worse or better than each other. This, however, is likely driven more by political correctness, prejudice, or a wish for success, than actual insight:

It is true that situations can always be found where an apparently inferior type comes out ahead; it might further be true that different types have been absolutely more successful at various times than today. However, it is also true that certain categories, e.g. the INT.’s, have a better chance at reaching the highest levels of self-actualization pyramids and hierarchies (e.g. Maslow’sw and Dabrowski’sw), making scientific breakthroughs, making great contributions to art, becoming good decision makers, ...; while others, e.g. ESFJ’s, tend to remain intellectual dwarfs and often are more trouble than use.

For another take on this, consider the Keirsey Temperament Sorterw, and its (approximate) mapping to MBTI categories: A “teacher” or a “provider”, e.g., fill valuable roles; however, these roles are secondary. In contrast, roles like “mastermind” or “inventor” are much more primary and harder to replace. If we remove the power-cord from a computer, it will become near useless—nevertheless, the CPU is a far more important component.

A good analogy could be to differ between three types of physicians who respectively treat symptoms, cure diseases, and invent new cures; or between philanthropists who respectively fish for a hungry man, teach him to fish for himself, and invent a better way to fish: The second is far more valuable than the first, and the third far more valuable than the second. (Assuming that the respective success-rates are not too far apart.)

For a third, consider that S and F is something that almost all children default too, and which they grow out of to varying degrees as they become more intelligent, gain more perspective, etc.: A sufficiently small child is driven entirely by what is immediately present and what their own immediate wishes are. (Note that it does not make sense to speak of .SF. in this case, because the personality types as a whole are far more complex than just an addition of the individual characteristics. However, if I were to speculate on a complete default type for a very young, but mobile, child, ESFP would be my guess.)

A few caveats:

The above is an “all other factors equal” discussion: A personality type that is better in principle cannot overcome, say, a thirty point difference in IQ.

Intelligence and giftedness has a non-trivial correlation with MBTI type, which can make .NT.’s appear in a more favorable light, due to their higher average intelligence, than can be warranted from the personality type alone. (Then again, it is possible that this correlation exist because the more intelligent someone is, the more likely he is be rational, think in the long term/big picture, etc.)

The Tall Dancer problem can, conversely, give a faulty overestimation of the value of e.g. ESFJ’s—or give small minority groups unnecessary problems with regard to their personal development, career, restriction on contributions due to being misunderstood, ...


How do I know that I am not, myself, misjudging others in a tall-dancery manner? Perfect certainty is not possible; however, I note that I am aware of the risk, have thought this article through in the light of that risk, and that I have rational reasons to see a difference in value.

One related area where the above could be vulnerable to attack, however, is priorities: If someone feels that a kindergarten teacher is more valuable than a theoretical physicist, e.g, then we will simply have to agree to disagree on very basic values. (Some related opinions are discussed in an article on humans vs. animals).