Take an MBTI test twice and there's a fair chance you get two different answers. In retest studies, between 39% and 76% of people came out as a different four-letter type when they took the test again within five weeks. That isn't because you're unusually unstable. It's what happens when a test squeezes someone who behaves differently at work, at home, and with friends into one label. Here is what's actually moving, what isn't, and how to tell the difference.
Retest studies are the honest measure. Reviews of them report that between 39% and 76% of people receive a different type when retested within five weeks 1. Each axis on its own is reasonably stable. The trouble is the label: a type is four cut-offs stacked together, so one axis near the middle is enough to change the whole four-letter code.
That's why "which letter changed" is more useful than "my type changed". It's almost always the axis where your score was closest to 50/50, and for most people it's the same axis every time.
Your average behavior moves with the situation. Your pattern of movement doesn't. Personality researchers call this an if-then signature: if it's a meeting, then decisive; if it's family dinner, then soft; if it's a new subject, then scattered 4. Two people with the same average can have opposite signatures, and the signature is the part that stays recognizable for years.
A single-type test throws the signature away. It averages your four selves into one label, and on a different day a slightly different average produces a slightly different label. That is the whole mechanism behind "my MBTI keeps changing".
The 4-scenario test asks the same axes in four areas of life, daily habits, social situations, work, and learning, and scores each area on its own. You get four scenario types, a composite core type built from all of them, the letters that flip between areas, and a consistency score that says how much your behavior depends on context.
A worked example: someone whose core type is INTJ often reads as INTP in daily life (more reading than acting), INFJ in relationships (they slow down and listen), ENTJ at work (deadlines force the decisions), and INTJ while learning. None of those is wrong, and none of them is "the real one". The core type is the center they orbit; the flips are the signature.
Your letters can, and often do, because they are cut-offs on continuous scores. Deep change in how you tend to behave happens slowly, over years. A flip between two tests a week apart is almost always the test, the situation you had in mind, or an axis where you sit near the middle.
Because a four-letter type is four cut-offs stacked together. If you are close to the middle on even one axis, small differences in mood, wording, or which part of your life you picture while answering are enough to change the code. Retest studies find that 39% to 76% of people get a different type within five weeks.
Usually the question is wrong rather than the results. If one result came from thinking about work and another from thinking about home, both are real for that setting. Look at which axis flipped and how close its score was to 50/50; that tells you whether you are looking at a preference or a coin toss.
Yes. INFJ and INFP differ on the J/P axis, and if your score there is near the middle you will get both. That is a fact about the scale, not a flaw in you. A scenario-based result usually shows the J and the P living in different areas of your life.
Scores drift a little over decades; people tend to become somewhat more organized and less reactive as they get older, but the movement is gradual. If your type changed since last month, age is not the reason.
For any one person, the axis where their score is closest to the middle. There is no universal answer; the useful move is to find your own borderline axis and stop treating that letter as fixed.