People make a big deal over how humans are too complicated to ever be described by mathematics. (I don’t agree.  Topic for another time.) So it’s astonishing that so many people give credence to the Myers-Briggs Type Indicator test.


The MBTI classifies people into sixteen categories, based on their proclivity to:

  1. Extraversion
  2. Facts vs. Ideas
  3. Heart vs. Head
  4. Like things settled vs. Like things open-ended

Each dimension varies on a sliding scale, so one’s score on a 100-question test is like [0,1] × [0,1] × [0,1] × [0,1] (actually it’s sparser in practice and could be several dimensions smaller, due to overlap).

MBTI < [0,1]^4 ... how bad is it?

Now that’s a shocker!  I know that we’re supposed to stick to simple models to describe things, but four parameters is the same as a bloody electron!  How are these four static numbers supposed to capture such a large chunk of human complexity?

Measures the Wrong Stuff, and Measures Unreliably

The model discards many characteristics that could be relevant—nice, creative, clever, honest, hard-working, religious, favorite sandwich—that’s a hack list but the point is, I don’t think they’re measuring the principal components of personality.

Not only that — but the test isn’t even sound.  One, MBTI asks people to describe themselves, with no external check.  Two, the best validation of it is a 90% consistency only if testing the same people soon after.  Let’s turn that around and say that in 10% of cases the test fails a basic scoff test.  Three, Wikipedia claims (with citation) that merely 36% of adults remain the same type after more than nine months.

I’m being harsh, but there really are some people who take the MBTI too far — using it to plan corporate teams, screening job candidates because they “don’t fit” according to a far-out extrapolation of the theory.

I remember in my first sales job the trainer used a stripped-down version of MBTI (4 types) to help us characterize customers quickly.  It was fine for that task—but shouldn’t be used for serious stuff like career counseling.  I’m thinking of books like Do What You Are — as if who you are doesn’t change over time, or wouldn’t be influenced by the job market!  Jeesh.


So where does the MBTI come from, anyway?  The first three parameters come from schlockmeister Jung, the same one who gave us “cultural memory” and the book Synchronicity.  Well my trust in those is blown.  [Persi Diaconis, the statistician’s statistician (and magician), refuted the improbable nature of the coincidences that comprise Synchronicity.] 

Read the Wikipedia article for more of this history.  But essentially the practical purpose that brought MBTI typology into common use was its application in the US during WW II, to match women to jobs while the men were at war.  It wasn’t intended to find them a fulfilling career for life — merely to point people who had worked exclusively at home toward a factory or office that they were more likely to enjoy.

Now that is an appropriate sized job for the MBTI.  It shouldn’t be used for important stuff like screening employees, managing them, choosing a career, or making definitive judgments about anyone.  The typology is too small, people change, the test is unreliable, and these four factors probably aren’t the principal components.

10 notes

  1. isomorphismes posted this