← All questions

Who you are

a straight answer, then a way to find out

Why do I get a different MBTI result every time?

Because you are near the middle on at least one dimension and a four-letter system has no way of saying so. Scores pile up in the centre rather than splitting into two camps, which puts an enormous number of people within a few points of a cut-off — and a few points is exactly what a bad night, a work mood or three differently worded questions will move. The letter changed. You did not.

INFP in March. INFJ in June. ENFP on a site your friend sent you, taken on a train. The natural reading is that one of them is right and the others were taken badly, or worse, that you do not know yourself well enough to answer twenty questions about your own life consistently.

Neither is what happened. The instability is a property of the model, not of you, and once you can see the mechanism the whole thing stops being unsettling and becomes quite specific information about yourself.

What is usually going on

You are near the middle, and the model has to round. This is the whole answer and everything else is detail. Each of the four dimensions is scored as a continuous quantity and then cut at a line, and the letter you get is which side of that line you landed on. A 52 against 48 split produces a capital letter with exactly the same confidence as an 89 against 11 split. Two answers change on a Tuesday, the score moves four points, the letter flips — and because the letter is part of a four-letter name, the entire profile you get handed flips with it. You went from a detailed description of one kind of person to a detailed description of a different kind of person because of a handful of points on one dimension out of four.

Scores cluster in the middle, so this happens to a lot of people. If the types were genuine categories, the scores would be two-humped and almost nobody would sit near the cut. They are not. On every dimension the distribution is a single mound with the bulk of people somewhere in the centre, which means near-the-line is not an edge case; it is the ordinary case. Roughly speaking, the more average you are on a dimension, the less your letter means and the more it moves.

One specific dimension is doing the flipping, and it is worth knowing which. Line up your results and you will usually find three letters that never move and one that does. That unstable one is where you are near the middle, and it is a real description of you: being genuinely balanced between thinking and feeling, or between planning and improvising, is a way to be. It is just a way the model has no notation for, so it alternates between two wrong answers instead.

You answered as a different person each time. Most of these instruments do not tell you which context to answer for, so you choose without noticing — and the you at work and the you on a Sunday do not give the same answers about planning, socialising or how decisions get made. People who take a test at their desk and again on holiday frequently get different types, and both sittings were honest.

State moved the needle. A stressful fortnight, three bad nights of sleep, the week after a breakup — these shift self-report on questions about energy, sociability and structure by more than four points, easily. The instrument is measuring how you answered today, and today is a variable.

The items were not the same items. Every free version of this on the internet uses different questions to measure the same four dimensions, with different wording and different numbers of items. Someone far from the middle gets the same letter from all of them. Someone near the middle is genuinely at the mercy of which fifteen sentences a particular site chose.

And some of the movement was you being more honest. This one is uncomfortable and it is real. People answer the first sitting somewhat aspirationally — as the person they are working on being — and later sittings, especially after a hard stretch, somewhat more plainly. If your type changed in the direction of a less flattering profile, it may not be noise.

What to do about it

  • Look for the number, not the letter. If a test gives you four letters and no indication of how strong each preference was, it is withholding the only part that would let you interpret it. That is a reason to distrust the test rather than yourself.
  • Pick a context and say so. Answer for me outside work and keep it. A result that averages your office self and your Saturday self describes a person who does not exist in either place.
  • Treat the flipping dimension as a finding. Being mid-range on thinking and feeling is not a failure to have a personality; it means you weigh both and that neither of the two profiles will fit you well. That is more useful than a letter would have been.
  • Do not retake it after a bad week. You will get a valid measurement of a bad week.
  • If you want stability, use a continuous instrument. The Big Five reports five numbers and rounds nothing, which is why its scores hold up across months where a type label does not. What a percentile actually means covers what those numbers actually mean once you have them.
  • Expect a continuous instrument to move a little as well, and know how much. Trait scores drift for reasons that have nothing to do with typology, and why a personality test result keeps changing has the longitudinal figures for how much of that is ordinary.

What it depends on

How much your type moves is decided by one thing: how far from the middle you sit on each of the four dimensions. That is a measurable quantity and you can simply have it. Someone at the far end of a dimension has a letter that has been the same since school. Someone in the middle has a letter that is, in the most literal sense, being generated by rounding.

Our version of the test reports that position for each dimension alongside the four letters, for exactly this reason, and says where the four-letter model is weak rather than hiding it behind a label. If you want the longer version of why the categories behave like this, how accurate the MBTI actually is goes through the retest and distribution research, and the type pages show what a described type looks like when the dimensional scores are put back in. It is the same reason no two people with the same four letters are alike: the letters record which side of each line you landed on and say nothing at all about how far.

Which one are you? Find out in two minutes

The number underneath the letter is the thing that answers this, and most free tests do not show it. Ours reports a dimensional percentile for each of the four, so a preference that came out at 51 percent is visible as a preference at 51 percent rather than as a capital letter pretending to be a fact.

Want to talk it through?

Luna reads your results first, so the advice is about you, not about people in general. Five free messages a day.

✦ Talk to Luna

People also ask

Which of my two results is the real one?

Neither is more real, and that is not a dodge. The dimensional scores behind both sittings are nearly identical; the two labels are two different roundings of the same position. If you have to carry one, carry the one from the sitting where you were rested and answering about the same context.

Does your personality type change as you get older?

Traits do move, slowly and in a consistent direction — most people become somewhat more conscientious and emotionally stable across adulthood, and the change is measured in decades. A letter that flips in six weeks is not that. Confusing the two is how a rounding artefact gets read as personal growth.

Am I an ambivert?

Almost certainly, in the sense that most people are. The word names the middle of a distribution, and the middle of a distribution is where most of the area is. It is less a rare third category than the ordinary state that a two-box system has no room for.

Can two people with the same four letters be nothing alike?

Routinely, and this is the quiet cost of the label. Two people reported as the same type can sit at opposite ends of the same dimension in strength — one barely over each line, one far out on all four — and share very little in practice. The letters say which side; they do not say how far, and how far is most of the person.

Related questions

Written by MoodyWonder from the attachment, conflict, trust, occupational and sleep literature. Describes patterns; does not diagnose. If you are unsafe or everything has gone grey, talk to someone you trust or a clinician today.