The Science Behind MBTI: What Holds Up and What Doesn't

The Myers-Briggs Type Indicator is easy to like. You answer a set of questions, you get four letters, and the description that comes with them often reads as if someone had been taking notes on you. Researchers have also been arguing about it for decades. Both can be true at once. The more useful question isn't whether MBTI is "real" but what the four letters measure, how steady they are, and what they're good for. What follows is the evidence on each side, with links to every source so you can check it yourself.

Where the four letters came from

The ideas start with Carl Jung, whose book Psychological Types appeared in 1921, the earliest entry on the MBTI publisher's own timeline, and was translated into English in 1923. According to StatPearls, Jung's "typology of introversion and extraversion, along with his theory of psychological functions, laid the groundwork for modern personality typologies, including the Myers-Briggs Type Indicator."

According to the Myers & Briggs Foundation, Katharine Cook Briggs had already begun developing her own personality system by the time Jung's book appeared in English, and her daughter, Isabel Briggs Myers, spent decades turning Jung's theory into a questionnaire. Myers, a member of Swarthmore College's class of 1919, had also published a book called Murder Yet to Come in 1930. Neither woman had a PhD or worked in academia, which the Foundation states plainly in its response to critics. The Myers-Briggs Company dates the development of the indicator to 1943, during World War II. The psychologist David Pittenger, in a 1993 critique, wrote that Myers began it in the early 1940s "as a test to be used for personnel selection." He also writes that the fourth pair, the Judging and Perceiving preferences, "were invented by Briggs and Myers."

Its road to the mainstream wasn't smooth. Pittenger's account notes that the Educational Testing Service began distributing the test for research in 1957 and, after an unfavorable internal review, ended the relationship. In 1975, he writes, Consulting Psychologists Press acquired the right to sell it. The Myers-Briggs Company, formerly CPP Inc., lists 1975 as the year it became the exclusive publisher, when the assessment "becomes widely available to the psychological community for the first time." One version, Form M, is a 93-item questionnaire, according to the publisher's product page.

What the four pairs are supposed to describe

The Myers & Briggs Foundation describes four pairs of opposite preferences:

• Extraversion (E) or Introversion (I): opposite ways to direct and receive energy. Do you prefer to focus on the outer world or on your own inner world?
• Sensing (S) or Intuition (N): opposite ways to take in information. Do you prefer to focus on the facts or the big picture?
• Thinking (T) or Feeling (F): opposite ways to decide and come to conclusions. Do you prefer an objective approach or an empathetic one?
• Judging (J) or Perceiving (P): opposite ways to approach the outside world. Do you prefer to seek closure or to stay open to new information?

One letter from each pair gives sixteen combinations. The Foundation is careful about what a result means: the assessment "sorts for preferences; it does not measure traits, ability, skill, or character," and there are "no better or worse types."

A few things follow from the definitions themselves. Extraversion and Introversion, in MBTI's sense, are about where your attention and energy go, which is a different question from how shy or confident you are (the piece on introverts and extroverts goes into that difference). Thinking and Feeling are about what you weigh when you decide, not whether you have emotions. And Judging has nothing to do with being judgmental; it's a preference for getting things settled.

The case against: what researchers found

The letters often change on a retest

If a type is a fixed part of who you are, you should get the same one when you take the test again a few weeks later. Often you don't. In 1991 a National Research Council committee, on a project supported by the Army Research Institute, reviewed the evidence on the MBTI. It found the underlying scores reasonably consistent: internal reliability figures for them typically ranged from .75 to .85, which the committee called acceptable. But it added that those figures "do not address the key theoretical issue of the MBTI, which is stability of type." Across 11 samples reported in the test's own manual, between 24 and 61 percent of people kept the same four-letter type when retested, with a median of 40 percent. In a 1983 study by McCarley and Carskadon that the committee cited, only 47 percent kept their type over five weeks. Its conclusion: changes in type "of these magnitudes suggest caution in classifying people in these ways and then making decisions that would influence their careers or personal lives."

Most people land near the middle

A type system assumes people come in kinds. If Extraversion and Introversion were two separate groups, scores on that scale should bunch up at the two ends with a dip in between. Pittenger's 1993 critique found "no evidence of bimodal distributions for the MBTI. Instead, most people score between the two extremes." A 2002 analysis by Tammy Bess and Robert Harvey of about 12,000 people in leadership development programs found that earlier reports of two-humped score distributions were an artifact of a default setting in the scoring software. With a better setting, the distributions "became strongly center-weighted," a result the authors described as not supportive of the type-based view.

That helps explain the retest problem. When most people sit near the dividing line, a couple of answers going the other way can carry someone across it, and two people with nearly identical scores can walk away with different letters. Pittenger made the same point: because the test uses a cutoff, "it is possible for people with relatively similar scores to [be] labeled with much different personalities." Robert McCrae and Paul Costa reached a similar conclusion in a 1989 study of 468 adults: they found "no support for the view that the MBTI measures truly dichotomous preferences or qualitatively distinct types." Instead, they wrote, the instrument "measures four relatively independent dimensions."

The theory, and the career claims

A 2019 paper by Randy Stein and Alexander Swan examined the theory behind the types, the idea that each person has a "true type" across four dichotomies, and concluded that it "lacks agreement with known facts and data, lacks testability, and possesses internal contradictions." They note that "despite its immense popularity and impressive longevity," the MBTI "has existed in a parallel universe to social and personality psychology." The 1991 committee, which observed that the MBTI was becoming increasingly popular as a tool for vocational guidance, was direct about that use: "At this time, there is not sufficient, well-designed research to justify the use of the MBTI in career counseling programs."

The case for: what holds up

None of that means the MBTI measures nothing. The scores themselves hold together well. Besides the committee's figures, a 2002 meta-analysis by Robert and Mary Margaret Capraro, which pooled reliability results across studies, found that the MBTI and its scales "yielded scores with strong internal consistency and test-retest reliability estimates, although variation was observed." A 2025 review by Erford and colleagues of 193 studies that used Form M between 1999 and 2024 found internal consistency of 0.845 to 0.921, and evidence of agreement with similar measures that was "robust across six personality instruments." The same review found that "structural validity and test-retest studies were absent" from the 25 years of research it covered, so it had no retest results for Form M to pool.

The Myers & Briggs Foundation answers the critics directly. In its view the dispute comes down to two different ways of measuring personality: trait models such as the Big Five report how much of a trait you have, while type theory sorts you into a preference category and "does not measure how much of that preference a person possesses." That difference, the Foundation says, is "at the heart of the matter." It also cites the publisher's 2018 manual for the newer Global Step I and Step II versions, which reports that in "a global sample of 1,721 adults who took the MBTI Step I twice between 6 and 15 weeks," "the test-retest reliability coefficients were 0.81 to 0.86 across all four scales." Those coefficients describe how closely the scores line up from one sitting to the next. They aren't the same measure as the share of people who keep all four letters, which is the figure the 1991 committee focused on.

Read side by side, the two sets of findings mostly fit together. The scores on each scale are fairly consistent. What wobbles is the letter you get when a score near the middle is cut in two. That's a problem with the sorting more than with the scales.

The four scales also line up with the model academic researchers use most. Even the Foundation says that the Five Factor Model, better known as the Big Five, "is the most used personality model and assessment in academia," while the MBTI is the most used outside it. The Big Five, described in StatPearls, a medical reference indexed in PubMed, as "one of the most influential and empirically supported trait theories," sums up personality in five broad traits: Openness to experience, Conscientiousness, Extraversion, Agreeableness and Neuroticism. McCrae and Costa found that the four MBTI scales "did measure aspects of four of the five major dimensions of normal personality." A 2003 study of 900 people by Furnham and colleagues replicated the pattern: Extraversion-Introversion went with Extraversion, Sensing-Intuition with Openness, Thinking-Feeling with Agreeableness, and Judging-Perceiving with Conscientiousness. The fifth trait, Neuroticism, the tendency toward emotional instability, is the one left out of those pairings.

Not every scale is equally strong, though. Correlations reported in the test's own manual, as summarized by the 1991 committee, pointed to good construct validity for the Extraversion-Introversion scale and "relatively weak validity" for the Sensing-Intuition and Thinking-Feeling scales, which showed only moderate correlations with other instruments built to measure similar things. The committee added that those findings came mostly from high school and college students.

Why your description can feel so accurate

So why does a type description so often feel like it was written about you? Part of the answer is above: the scales track real tendencies, and the description is built from your own answers, so it hands back what you told it in friendlier words. Part of it is the descriptions themselves. Pittenger argued that they "are generally flattering and sufficiently vague so that most people will accept the statements as true of themselves," the pattern psychologists call the Barnum effect. There's more on how that works in why personality tests feel so accurate.

Using a type without getting boxed in

Even the people who stand behind the MBTI treat a result as a starting point. The Foundation's ethical guidelines say that type preferences "are tendencies, inclinations, and preferred ways of doing things and being in the world," that "they are not absolutes," and that "results are never to be used to limit anyone." They also state that the assessment "is not designed to be used for hiring purposes." And the official process doesn't end with the printout: the Foundation says professional ethical guidelines require an interpretive feedback session in which you settle on a "best-fit type," defined as "the four-letter type that you think best describes you after learning about type and reviewing your MBTI results." In the Foundation's words, "Ultimately, you are the only one who can decide which of the 16 types you prefer."

Personality also shifts over a lifetime. A 2006 meta-analysis of 92 samples by Roberts, Walton and Viechtbauer found that people tend to increase in social dominance (a facet of extraversion), conscientiousness and emotional stability, especially between the ages of 20 and 40. The 1991 committee drew a similar lesson for questionnaires like this one: they are "likely, at best, to capture a person's current state."

A few ways to use a type that fit the evidence:

• Check how close each pair was. A narrow lead says much less than a wide one, and it's the easiest kind to tip the other way next time.
• Read both sides of a close pair. If you sit near the middle, parts of both descriptions will probably fit.
• Use it as vocabulary, not a verdict. The letters are handy for talking about differences, like one person wanting a plan settled while another wants to keep options open. They make a poor reason to skip things that don't come naturally.
• Keep it out of decisions about other people. Hiring someone, or sizing up a partner, by four letters asks more of the test than the evidence supports.

How the quiz here handles it

The personality type quiz on this site uses the same four pairs. It isn't the official Myers-Briggs Type Indicator, and it hasn't been tested the way research questionnaires are. Because of the findings above, it shows how your answers split on every pair and flags a close split instead of hiding it behind a letter. The quiz page explains the exact rule, and how we build quizzes covers the method.

Correction, October 8, 2026. An earlier version of this article attributed a five-week retest result to "a 1991 study." The five-week figure cited now comes from a 1983 study by McCarley and Carskadon, as reported in the 1991 National Research Council review. The earlier version also named only two of the four pairs as matching Big Five traits; research has matched all four. An opening anecdote that could not be verified has been removed.