Trait versus type

TL;DR

Trait Versus Type: Why Psychology Abandoned the Boxes

  • The type claim is testable: real types would produce two-peaked score distributions. They don't.
  • Most people score near the middle - exactly where type instruments draw their cut-line.
  • Retest instability follows mathematically: 39-76% of people get a different four-letter type within weeks.
  • Continuous measurement keeps the information the box throws away - including the middle, and the movement.

Two people sit one point apart on an introversion scale of a hundred. One scores 49, the other 51. A trait psychologist records them, accurately, as nearly identical. A type instrument sorts them into opposite species - one an I, one an E - and hands each a different career pamphlet.

That single manoeuvre - the cut through the middle of a distribution - is what academic psychology spent the late twentieth century testing, and rejecting. The story of how is one of the cleaner victories of evidence over intuition, and it explains more about the modern assessment industry than any single instrument review.

What is the difference between personality types and traits?

A type sorts people into categories - introvert or extravert, thinker or feeler - treating personality as a set of boxes you fall into. A trait measures position on a continuous scale, treating the same qualities as dimensions along which people vary by degree. The distinction is testable: if types were real, scores should pile up at the two ends with few people between. Every rigorous test has found the opposite - most people score near the middle, which is precisely where type instruments draw their cut-line.

The graph that ended the argument

The beauty of the type question is that it has a statistical signature. If humanity genuinely divides into introverts and extraverts, plotting everyone's scores should produce two humps - a crowd at each pole, a valley between. This is called bimodality, and finding it would vindicate the boxes.

It has been looked for, carefully, for decades. McCrae and Costa's 1989 study reinterpreted the MBTI through the lens of trait psychology and found its scales measure four relatively independent continuous dimensions - with, in their words, no support for truly dichotomous preferences. Bess and Harvey's 2002 analysis asked whether the bimodal patterns sometimes reported were fact or artefact, and answered: artefact of scoring. Stein and Swan's 2019 review in Social and Personality Psychology Compass gathered the record and concluded flatly that a taxonomic structure for the MBTI is not supported. Score distributions are centrally peaked - one hump, most people in the middle, thinning towards the poles. The shape of a crowd, not the shape of two camps.

Why is the MBTI not supported by psychologists?

Chiefly because its central claim - that people fall into discrete either/or types - fails statistically. McCrae and Costa showed in 1989 that MBTI scales measure four continuous dimensions with no support for genuine dichotomies; Bess and Harvey found the apparently two-peaked score distributions are a scoring artefact; and Stein and Swan's 2019 review concluded a taxonomic structure is not supported. Add retest instability - across studies, 39% to 76% of people get a different type within weeks - and the four-letter verdict rests on a cut through the middle of a bell curve.

The retest problem is not a separate flaw; it is the same flaw seen from the user's chair. If most people sit near the middle of each scale, then most people sit near the cut-line - and tiny, ordinary fluctuations in mood or context push them across it. The label flips not because the person transformed but because the box boundary runs through the densest part of the crowd. Pittenger reported about half of test-takers receiving a different type within five weeks; the range across studies runs 39% to 76%.

Meanwhile trait psychology consolidated. Goldberg's 1993 statement of the Big Five marked the field's consensus: five continuous dimensions, measured by degree, no boxes. Robert Hogan's summary of professional opinion on the MBTI was less diplomatic - most personality psychologists, he wrote, regard it as little more than an elaborate Chinese fortune cookie.

Why the boxes won anyway

Commercially, types never lost. The MBTI alone is taken by roughly two million people a year; the type format has since been cloned across the internet's most popular free tests. The reasons are practical: a label is memorable where a profile is homework; a tribe ("we INFJs") is shareable; and a binary is easy to train, easy to poster, easy to sell. Pittenger attributed the persistence to the beguiling, horoscope-like nature of the summaries and steady marketing. None of these virtues is measurement. All of them are product design - and the lesson for anyone building a serious instrument is that the product virtues and the measurement have to be earned separately.

What should I use instead of a personality type test?

Instruments that measure continuously and expect people to change. Academic psychology's consensus framework is the Big Five - five trait dimensions, measured by degree, with no type labels. For workplace use, the same logic favours tools that read position on a scale rather than assign a box, report tendencies rather than verdicts, and re-measure over time rather than filing one result - because the middle of the scale, where most people sit, is exactly what a binary label cannot represent.

What continuous measurement changes

This is not an academic preference; it changes what the instrument can see. The middle of a scale is not a failure to have a personality - for working preferences it is often the most useful reading, the person who moves between poles as the situation asks. A box format is structurally blind to its own most populated region.

Sariio MAPS is built on the trait side of this fifty-year argument, applied to work: 60 preference pairs, each read on a continuous scale, none collapsed into a category. Position matters - 49 and 51 read as neighbours, as they should - and so does movement, because the ten-minute survey is retaken at least twice a year and the delta between readings is treated as signal, not noise. There are no four-letter labels anywhere in the suite, by architecture rather than restraint: the AI works with positions and changes, and it has no box to put anyone in. The type era's instruments are reviewed individually elsewhere - Myers-Briggs, DISC, Insights Discovery - and the deeper history of what Jung actually claimed is in what did Carl Jung actually say about types.

See your own positions, not a box - the individual survey is free.

Sources