MBTI was the personality test of the office Christmas party for forty years. It is now the personality test of the corporate onboarding portal, the dating profile, and the LinkedIn bio. There is a whole industry of certified administrators trained to interpret your four-letter type, and they bill by the hour.
The test was written by a woman named Katharine Cook Briggs and her daughter Isabel Briggs Myers. Neither was a trained psychologist. They built the instrument in the 1940s on a reading of Jung that Jung himself never validated. It has been the subject of serious peer-reviewed critique since at least 1993, and the critique is consistent: the test has poor test-retest reliability, the four-letter type system collapses bimodal distributions that are actually continuous, and the predictive validity for any outcome anyone cares about is roughly the same as picking randomly.
If you are reading this and your immediate reaction is 'but my type really fits me', I would gently suggest that the test is well-tuned to give people the satisfying sense that the type fits. That is a feature of the test as a marketing product. It is not a feature of the test as an instrument of measurement.
What replaced it, scientifically
The serious personality science of the last forty years converged on a small number of dimensions that survive factor-analytic scrutiny across cultures, languages, and time. The Big Five is the most widely known: Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism. It is much better than MBTI on all the boring measurement-quality grounds.
The Big Five has a younger and more interesting cousin called HEXACO, introduced by Kibeom Lee and Michael Ashton. It adds a sixth dimension that the Big Five missed: Honesty-Humility. This dimension predicts a long list of things that other tests cannot, including white-collar crime, integrity violations under pressure, and behaviour during morally ambiguous bargaining games.
If you ever sat in a hiring interview thinking 'this candidate looks great on paper but something about them feels off' and you could not articulate what, the something is generally a low score on the H factor that HEXACO measures and that the Big Five does not.
Why did the science take so long to get here
Two reasons. The first is that personality research used to be done largely on white American undergraduates, who turn out to have a peculiar variance structure that washes out the H factor. The second is that the older instruments had no good way to detect respondents who were just answering in the socially desirable direction; the H factor effectively measures the trait that produces socially-desirable answering, which means once it exists, the instrument is self-cleaning.
Lee and Ashton ran the analyses on data sets from a dozen languages and cultures and the H factor survived. That is when it became a real thing rather than a methodological artefact.
Pittenger's 1993 takedown, in three lines
MBTI test-retest reliability is so low that approximately half of people who take the test twice get a different type. The four-letter type system imposes categories on continuously-distributed traits. The predictive validity for job performance, marital satisfaction, leadership ability, and educational attainment is at chance.; Paraphrased from Pittenger, 'Cautionary Comments Regarding the Myers-Briggs Type Indicator' (1993)
This is not a fringe view. Pittenger's paper is the standard cite. It has been the standard cite for thirty-two years. The reason MBTI persists is that it tells a satisfying story, and the satisfying story is a commercially viable product, and the commercial viability has nothing to do with the science.
What to use instead, in descending order of seriousness
For a clinical-grade assessment, look at the Hogan Assessment system. It is the test that serious organisational psychologists use and it costs serious money to administer.
For a peer-reviewed but accessible self-assessment, take the HEXACO-PI-R online. It is the test maintained by the model's authors themselves. It takes about twelve minutes. The report you get is six numbers on six scales. It will not tell you a satisfying story about your archetype; it will tell you where you actually score.
I run a platform that wraps the HEXACO test in a friendlier consumer surface and adds the kind of interpretation report a layperson can read. The science underneath is unchanged. The wrapper is for the people who want to use this material without spending six weeks learning the literature first.
The harder question
It is fair to ask why any of this matters. If MBTI works as a conversation starter at a corporate retreat, why drag it through a peer-reviewed beating?
Because it is being used to make hiring decisions. It is being used to coach people in directions that have no empirical basis. It is being used to type teenagers in school career counselling sessions. The accumulated cost of an entire industry running on a 1940s parlour game written by two women without psychology degrees is non-trivial, even if no single use of the test does much damage.
There is a better instrument. It has been there for the last quarter-century. The only thing keeping the old one alive is inertia and a very effective marketing operation. The science part of the situation is settled.