MBTI and the Big Five: Where They Agree and What the Letters Lose

The Myers-Briggs Type Indicator and the Big Five usually turn up in different rooms. Adrian Furnham of the Norwegian Business School, who has compared the two tests in several studies, sums up the split in a 2022 paper. Asked which test is "the most valid and validated," he writes, the answer would be the NEO-PI-R, a Big Five questionnaire. Asked which is "the most marketed to and used by practitioners (e.g. trainers)," the answer would be the MBTI.

Different rooms don't mean different subjects. When the same people take both tests, the MBTI's four preference pairs tend to line up with four of the five Big Five traits, some closely and some loosely, though one large recent study found the links far weaker than the older ones did. This article looks at how big those overlaps are, which parts of each trait they involve, what the Big Five measures that the MBTI leaves out, and what is lost when a score becomes a letter. The MBTI is a popular framework whose scientific standing is contested. Its history, its retest record and the arguments over its theory are in the science behind MBTI, so they aren't repeated here.

Scores against scores

Both tests produce numbers. The NEO-PI-R, as Furnham describes it, is a 240-item questionnaire that scores five broad domains (Neuroticism, Extraversion, Openness, Agreeableness and Conscientiousness) and six narrower facets within each. The MBTI yields a score on each of its four dimensions too, and in Furnham's words those scores "are usually discussed in terms of types based on cut-off scores." A 1997 conference paper by Randolph Arnau, Bruce Thompson and David Rosen describes how the cut works. Each MBTI question is a forced choice between the two sides of a pair, the answers for each side are counted, and the side with more of them becomes the preference.

The figures below are correlations between the MBTI's four scales and the Big Five traits. A correlation near zero means two scores have little to do with each other. The closer it gets to 1 or to negative 1, the more tightly the two move together, and a negative sign means one goes up as the other goes down.

The 1989 figures

Much of the later work goes back to a 1989 study by Robert McCrae and Paul Costa. They evaluated the MBTI against the five-factor model as measured by self-reports and peer ratings on the NEO Personality Inventory, with data from 267 men and 201 women aged 19 to 93. Their abstract gives the conclusions but not the coefficients. Arnau and his coauthors report the main ones. MBTI Extraversion correlated with NEO-PI Extraversion at .74 for men and .69 for women. For the rest they give only the men's figures, because the correlations for men and women were very similar:

• Sensing-Intuition with Openness, .72. Intuition went with higher Openness.
• Thinking-Feeling with Agreeableness, .44. Feeling went with higher Agreeableness.
• Judging-Perceiving with Conscientiousness, negative .49. Perceiving went with lower Conscientiousness.
• Judging-Perceiving with Openness, .30. Perceiving also went with somewhat higher Openness.

Two things stand out. The first is the spread. Extraversion and Openness track their MBTI counterparts far more tightly than Agreeableness and Conscientiousness track theirs, so two of the four MBTI scores overlap heavily with a single Big Five trait while the other two overlap more loosely and leave a good deal unexplained. The second is the peer ratings. According to the same summary, every one of these correlations except Judging-Perceiving with Openness was also statistically significant when the NEO-PI scores came from peer ratings, though the correlations were somewhat smaller. That matters, because two questionnaires a person fills out about themselves can agree partly because the same person is answering both. Agreement with how other people see that person is harder to explain away.

Which part of each trait

Each Big Five trait is broad, and its facets show which part of it an MBTI scale overlaps. According to Furnham's 2022 paper, his own 1996 study looked at the correlations between the MBTI scales and all 30 facets of the five-factor model, and the highest correlations fell here:

• Extraversion-Introversion. Gregariousness, Warmth and Positive Emotions, all facets of Extraversion.
• Sensing-Intuition. Ideas, Fantasy and Aesthetics, all facets of Openness.
• Thinking-Feeling. Tender-Mindedness, Trust and Altruism, all facets of Agreeableness.
• Judging-Perceiving. Order, Deliberation and Self-Discipline.

The paper doesn't name a domain for the Judging-Perceiving facets. The International Personality Item Pool's table of the 30 NEO-PI-R facets puts all three under Conscientiousness, and lists Feelings, Actions and Values as the other three Openness facets. Read that way, the Intuition side of the MBTI has most in common with the parts of Openness about ideas, imagination and art, and the Judging side with the orderly, careful parts of Conscientiousness.

A large sample that found much less

Several later samples showed the same four pairings. Furnham's 2022 paper cites Canadian university students tested by MacDonald and colleagues (1994), 435 Austrian adults tested by Renner and colleagues (2014) and his own earlier work, each pairing Extraversion with Extraversion-Introversion, Openness with Sensing-Intuition, Agreeableness with Thinking-Feeling and Conscientiousness with Judging-Perceiving. Not every sample was that tidy: the same paper describes a Polish study whose authors matched only two Big Five domains to MBTI preferences, and it sums up the earlier work as showing both "consistency and inconsistency."

His own 2022 study told a different story. It drew on 7,538 men and 1,949 women, all employed and many of them managers, tested by a British psychological consultancy at assessment centers. Most of the correlations were small. The two Extraversion scores correlated at .20, the highest of Extraversion-Introversion's links but still a weak one. The link between Sensing-Intuition and Openness that earlier studies found didn't appear. Conscientiousness correlated most with Thinking-Feeling rather than Judging-Perceiving, and Judging-Perceiving correlated most with Openness. The pairing that held up best was Agreeableness with Thinking-Feeling, which Furnham calls the one "most strongly and consistently associated with an MBTI trait." Of the 140 correlations between the MBTI scores and the Big Five domains and facets, only two were above .25 (Furnham, 2022).

The paper's abstract reports very little relationship between the two sets of scores "despite issues of shared method variance," the agreement two self-report questionnaires can show simply because the same people answer both. Furnham writes that many readers "may be surprised at the low correlations between the two tests," and reads them as "suggesting little empirical as well as conceptual overlap." He also points to a difference in theory. As he summarizes it, Jungians hold that extraverts draw energy from an external world of interaction and doing and introverts from an internal world of thought and reflection, while Eysenck's 1967 theory of cortical arousal holds that extraverts are under-stimulated and seek out arousal, and introverts are over-stimulated and tend to avoid it.

One large study doesn't cancel several smaller ones that agree with each other, and they don't cancel it either. The samples also differ: working adults at assessment centers in 2022, adults aged 19 to 93 with peer ratings in 1989. A fair reading is that the pairings usually appear, that their size varies a good deal from sample to sample, and that the 1989 figures are one estimate rather than a fixed property of the two tests.

What the Big Five adds

A scale for Neuroticism

The plainest gap is Neuroticism, which Benjamin Lahey defines in a 2009 review as "relatively stable tendencies to respond with negative emotions to threat, frustration, or loss." The MBTI has no scale for it, a point McCrae and Costa made among their criticisms of the test, as Furnham's paper recounts. The trait still leaves traces in MBTI scores. Furnham's 1996 study found Neuroticism correlated with both Extraversion-Introversion and Thinking-Feeling, and in his 2022 data Neuroticism's highest correlation was with Thinking-Feeling.

It's a large gap. In Lahey's review, published in American Psychologist, neuroticism is "a robust correlate and predictor of many different mental and physical disorders," and of how often people use mental and general health services. Those are findings about large groups, not a verdict on anyone. If worry, low mood or stress is getting in the way of everyday life, the right person to talk to is a doctor or a licensed mental health professional, not a personality quiz.

Furnham sees a clash of outlooks here as well. He notes that the MBTI and the people who use it carry a message of "gifts differing" and the "acceptability" of every profile, while the Big Five literature suggests that people with high Neuroticism scores are prone to problems like depression, anxiety and psychosomatic illness. A system in which no profile is worse than another has no obvious place for a trait with a risky end.

How far, not only which side

A Big Five score says how far toward one end of a trait someone sits, and the NEO-PI-R splits each trait into six facets. The MBTI's publisher has a finer-grained version of its own. In the Step II assessment, "each of the four preference pairs has five facet pairs that sit within that preference pair," 20 facets in all, according to a 2023 article on The Myers-Briggs Company's site. For Extraversion-Introversion they include Initiating or Receiving and Gregarious or Intimate, and someone who prefers Extraversion overall can come out "out-of-preference" on one of them. Even so, each facet is framed as a pair of opposites. The publisher's reports also give each preference a clarity category, from Slight to Very Clear, but an article on its site says that preference clarity "doesn't measure how much of a preference a person has." The site's article on close splits covers what clarity does describe.

More evidence on what the scores predict

The Big Five also has more evidence behind what its scores predict. Furnham contrasts "abundant evidence" of criterion validity for Big Five measures, the NEO-PI-R in particular, with "scant evidence for the MBTI." When Christopher Soto tried to repeat 78 published links between Big Five traits and life outcomes, in preregistered replications with a median sample of 1,504 people, 87 percent of the attempts were statistically significant in the expected direction, though the effects were typically 77 percent as strong as the originals. And in a 1997 paper, McCrae and Costa compared data from six translations of the Revised NEO Personality Inventory, given to German, Portuguese, Hebrew, Chinese, Korean and Japanese samples of 7,134 people in all, with the American factor structure, which came out closely reproduced when targeted rotations were used. The authors concluded that the data "strongly suggest that personality trait structure is universal."

Big Five items are free to use, too. The International Personality Item Pool holds more than 3,000 items and more than 250 scales built from them, including measures of the five major personality factors, and says its items and scales are in the public domain and can be used "without asking permission and without paying a fee."

What the type boundaries lose

Turning a score into a letter keeps the direction and drops the distance. Furnham calls the MBTI "notorious for its scoring procedure that eliminates/ignores data by converting an interval scale to a binary ordinal scale." The statistics of that step, called dichotomization, are set out in a 2002 paper in Psychological Methods by Robert MacCallum, Shaobo Zhang, Kristopher Preacher and Derek Rucker. The site's article on close splits looks at what a cut does to one person's result. The question here is what it does to the overlap between two tests, and to the two usual defenses of a cut.

Weaker links

For a normally distributed measure split at its mean, MacCallum and colleagues show that its correlation with another measure falls, in the population, to .798 of its original size, and the variance the two share falls to .637 of what it was. The further the cut sits from the mean, the bigger the loss. MBTI scores won't always meet those exact conditions, so the figures don't carry over directly, but the same arithmetic suggests that a letter shares less with the Big Five than the score it came from.

Less dependable, not more

One defense of a cut, which MacCallum and his coauthors say they often heard from colleagues, is that a raw score may be imprecise but can at least be trusted to say whether someone is high or low. They tested that argument under classical measurement theory and concluded that dichotomizing a continuous variable "does not refine the original measurement, but on the contrary, makes reliability substantially worse" (MacCallum and colleagues). The article on the science behind MBTI shows the same pattern in MBTI retest research, where scores hold up better than four-letter types.

Kinds of people, or degrees

The other defense is that the groups are real: people come in kinds, and the cut finds them. MacCallum and his coauthors reply that even then a split is the wrong tool, since it assumes there are exactly two groups and draws arbitrary classes. Whether real groups exist is a question for taxometric methods, which look for distinct classes in the data, and those have been run across much of personality and clinical research. A 2012 review by Haslam, Holland and Kuppens covered 311 findings from 177 articles, with a combined sample of more than half a million participants, and found that normal personality yielded "little persuasive evidence of taxa," the technical word for real categories. Most of the variables that interest psychiatrists and personality and clinical psychologists, the review concluded, are dimensional.

Reading four letters in Big Five terms

If you already know your type, the research above gives a rough translation. It describes tendencies across groups, and with correlations of these sizes plenty of individuals won't fit the pattern. The pairings and figures are the 1989 ones as reported by Arnau and colleagues, with the facets and the 2022 results from Furnham's paper.

• Extraversion or Introversion. The Extraversion side goes with higher Big Five Extraversion, especially Gregariousness, Warmth and Positive Emotions. The closest match of the four in the 1989 figures for men, by a hair, and .20 in the 2022 sample.
• Sensing or Intuition. Intuition goes with higher Openness, mainly Ideas, Fantasy and Aesthetics. Nearly as close as the first in 1989, and the link the 2022 sample didn't confirm.
• Thinking or Feeling. Feeling goes with higher Agreeableness, especially Tender-Mindedness, Trust and Altruism. A looser link in 1989, and the one that held up best in 2022.
• Judging or Perceiving. Judging goes with higher Conscientiousness, especially Order, Deliberation and Self-Discipline, and Perceiving with somewhat higher Openness. In 2022 this pair's strongest link was with Openness.
• Neuroticism. No MBTI scale covers it. It has turned up in correlations with Extraversion-Introversion and Thinking-Feeling, but four letters say little about it.

For the Big Five side of the picture, the International Personality Item Pool publishes free scales for the five factors. Whichever framework you use, a score that shows how far you lean says more than a letter that only shows which way.

Where the quiz here fits

The personality type quiz on this site scores the four Myers-Briggs preference pairs, so it sits on the MBTI side of this comparison. It isn't the official MBTI, it isn't a Big Five measure, and it has no Neuroticism scale. It does keep part of what a bare letter throws away. The result shows the count behind each pair, so a narrow lead looks narrow, and the quiz page explains when a close split is reported as balanced instead of being forced into a letter, with figures worked out from its own questions.