Partly. The four scales behind the Myers-Briggs approach do measure real differences between people, and they line up with four of the Big Five traits. What research does not support is the step that turns those scales into sixteen separate types. Most people score somewhere in the middle, not in one of two boxes.
So the honest answer depends on which claim you mean. The scales pass; the types do not.
Which MBTI Claims Hold Up
Most articles on this question give one verdict for the whole thing. It is clearer to split MBTI into its separate claims and look at each one.
| Claim | What the evidence says | Verdict |
|---|---|---|
| The four scales measure stable differences between people | They track extraversion, openness, agreeableness and conscientiousness from the Big Five | Supported |
| People fall into two clear groups on each scale | Scores bunch in the middle, with no natural gap between the two letters | Not supported |
| There are sixteen distinct kinds of people | Taxometric studies found continuous dimensions, not categories | Not supported |
| Your type is fixed and comes out the same each time | People near the middle of a scale often get a different letter on retest | Weak |
| Type predicts how good you will be at a job | Not shown well enough for hiring or selection | Do not use it this way |
The Scales Measure Real Traits
The strongest point in MBTI’s favour comes from its critics. McCrae and Costa (1989) gave both the MBTI and a Big Five questionnaire to 468 adults, along with ratings from peers. The four MBTI indices did measure aspects of four of the five major personality dimensions. Extraversion lines up with the Big Five trait of the same name, sensing and intuition with openness, thinking and feeling with agreeableness, and judging and perceiving with conscientiousness. The Big Five trait it misses is emotional stability.
So if your result says you lean introverted, that score usually reflects something true about you. The problem starts when the score becomes a label.
Why The Sixteen Types Do Not Hold Up
The same 1989 study found no support for the idea that MBTI measures truly opposite preferences or qualitatively distinct types. It looked instead like four fairly independent dimensions.
Two other lines of work point the same way:
- No gap in the middle. If types were real, scores should form two humps with few people in between. Earlier reports of that pattern turned out to be an artifact. Bess and Harvey (2002) rescored data from about 12,000 people with a better method and found the scores were heaviest in the centre, not split into two groups.
- No hidden categories. Arnau and colleagues (2003) used taxometric analysis, a set of statistical methods built to tell categories apart from smooth scales, on three type questionnaires including the MBTI. None showed a true category underneath. The preferences behaved like continuous dimensions.
Pittenger’s 2005 review for consulting psychologists also urged caution about reading the results as fixed types.
What Our Own Test Shows About Type Labels
To show why a cut-off loses information, we worked through the scoring of our 16 personality types test. Each of its four scales has 12 statements answered on a five-point scale, which gives 49 possible scores per scale. A score of 50% or more gives one letter, and anything below gives the other.
We wrote a short script that applies that scoring to the extreme cases. These figures describe what the scoring allows, not what our visitors scored.
| Comparison | Gap on each scale | Gap in answer changes, all four scales |
|---|---|---|
| Two people with the same type, as far apart as possible | 50 points | 96 one-step changes |
| Two people with opposite types on every letter, as close as possible | About 2 points | 4 one-step changes |
In other words, two people labelled with the same four letters can answer very differently, while someone labelled INFP and someone labelled ESTJ can be four small answer changes apart. The letters hide that. Your four percentage scores do not, which is why our test shows them alongside the type.
So Is MBTI Pseudoscience?
Calling it pseudoscience goes too far for the scales, and not far enough for some of the ways it is sold. The questionnaire measures real traits, if with less precision than a dedicated trait test. The theory of sixteen fixed types is the part that lacks evidence, and claims about matching types to careers or partners go well beyond the research.
How To Use A Type Result Sensibly
- Read your scores, not just your letters. A score of 52% is a mild lean, not an identity. Our guide to what the MBTI letters mean shows how to read each level.
- Treat close scores as both. If a scale sits within about five points of the middle, read both type descriptions.
- Use it for reflection, not decisions. A type is a prompt for thinking about how you work, not evidence for hiring, ruling out a career or judging a relationship.
- Cross-check with a trait test. The Big Five personality test uses the same kind of scales without forcing a cut-off.
- Expect some change on retest. Our post on getting a different type on a retest shows how often a borderline letter flips.
For a wider look at how far any online test can be trusted, see our guide on how accurate online personality tests are.
Frequently Asked Questions
Is MBTI accurate?
The four scales give a reasonable rough picture of where you lean. The four-letter type is less dependable, especially for anyone whose scores sit near the middle, because a small change in answers can switch a letter.
Why do psychologists criticise MBTI?
Mainly because it sorts people into opposite categories when the evidence shows smooth scales, and because it leaves out emotional stability, one of the best-studied personality traits. Researchers generally prefer trait measures such as the Big Five.
Is MBTI the same as the Big Five?
No, but they overlap. Four MBTI scales correspond closely to four Big Five traits. The Big Five keeps everything as a score, adds emotional stability and does not create types.
Should employers use MBTI to hire people?
No. The evidence does not support using type to select candidates. It is better suited to team discussions where people choose to share their own results.