The Mysterious Dark Arts of the MBTI

Recently I read a response by Rich Thompson, Director of Research at CPP to Adam Grant’s paper Say Goodbye to MBTI, the Fad That Won’t Die. In defence of his conviction, Rich doubled-down on the issue that thousands of peer-review journals and case studies have documented the MBTI “instrument’s” validity and reliability. Also, almost flabbergasted, he finds himself scratching his head when confronted by Adam Grant’s criticism of the MBTI?

This feels very familiar.  In all my dealings with either officials at CPP, the exclusive publisher of the Myers-Briggs instrument, or their “licensed” cadres across the world I find this reaction typical. I too have done the assessment (my frontal cortex ended up as an I, my brain stem an S, my corpus collosum a T and my thalamus a J) and even got certified on its products but somehow, I suspect, I’ve missed out on the inner-circle handshake that gains me access to their item-level repository of raw data for further research?  I just cannot for the life of me figure out what the gate-keepers like Rich et. al. are feverishly protecting. Any legitimate attempts at adding further value to their in-house MBTI research conclusions are simply stymied. Considering Rich’s assertion that “CPP, Inc., freely makes its supporting data publicly available”, I reached out to him to provide me with the raw, item-level, data set that they used to validate the MBTI in order to independently do some research of my own. I even offered to share the results with him and his team. His response was to point me to a website where I could make the request for $250.

So, here’s why I feel “academically justified” to point out the following:

  1. One Psychologist doth not a Model make

The theories of Carl Jung that were used by mother and daughter combination (Katharine and Isabel) to develop the MBTI are just that – theories. It is not a comprehensive model of personality.

  1. Counting is NOT measuring

The MBTI is not a measure. It is a collection of four apple-pear type arbitrarily joined dimensions namely Extraversion / Introversion (E/I), Sensing / Intuition (S/N), Thinking / Feeling (T/F), Judging / Perceiving (J/P). Much like gluing a ruler end-to-end to a thermometer (Figure 1) and measuring either height or temperature – depending on which one pitches up during the process (Table 1).

Here each item has binary choices – Yes or No (0 or 1) – with the underlying theory being that you are either Extraverted or Introverted. This means that in the event of you selecting “gives you more energy” and “a good mixer”, etc., by default, you are NOT Introverted and vice-versa. And this, despite the fact that you may at times find it a bit “draining “or possibly experience the odd time where you feel “quiet and reserved”!  As a consequence, in the MBTI’s altered universe, this enables you to not only establish that you’re around 1.76 meters tall but simultaneously determine that your internal temperature is approximately 32.1 degrees C!! Even more disconcerting, with this configuration, is the fact that you can end up being neither Extraverted nor Introverted midway up this peculiar scale (Figure 3).

So, why is this the case with the MBTI. It’s really all about the rules we lay down for our measure. How we employ these rules determine how acceptable our measures are. For example, when we measure the length of an object, there is an expectation that the ruler and the object to be measured need to be aligned so that they lay parallel to one another. We know that there has to be a starting point and units to count along the measure has to be calibrated. In addition, our rules need to make allowance for conditions of coincidence and attained interpolation. And most important, we can only measure ONE thing at a time. Without stoically maintaining these rules the results of our measurement will be disorderly and unreliable.

With the MBTI, this fundamental prerequisite of counting less of and more of the underlying construct (like the common ruler we reference above) breaks down before we even start constructing the measure. Its lack of linearity – having two uniquely different constructs combined as one measure – means that halfway up the measure we magically stop measuring one thing and switch to another! Like measuring your height and around about your hips switching to measuring your body temperature?

The E/I example below (Figure 3) highlights just why this unfortunately cannot work – no matter how hard we try to re-frame it. Cast your mind back to those “rules of measurement” and consider the following:

  1. Do we have a single construct? To which we have to answer an emphatic, NO.
  2. Is there a single, calibrated measure? To which we can again say an emphatic, NO.
  3. Does the measure have a starting and end point with equal measuring units along the scale? To which we can only answer an emphatic, NO.
  4. Can we be guaranteed that the measure will produce reliable and consistent results. To which we once again have to answer an emphatic, NO.

So, no matter how hard we try to duck-and-dive the issue, all measurement at its most fundamental level has to be singular and uni-dimensional. We live this on a daily basis. For example, despite a solid cube having a number of attributes namely, height, length, width, volume, density, etc. when we analyse it, we do so one attribute at a time. When we measure its length, we only measure length and no other attribute, etc. The MBTI fails this fundamental test. It rams together pairs of opposite dimensions and subsequently concocts from it four alien, single, bi-polar (Figure 1) “measures”. Essentially, four sets of two “things” arbitrarily concatenated and mysteriously called either “The Favourite World” Preference or some equally peculiar thing!!!?

To further nail this coffin shut, while counting is the first step of fundamental measurement necessary and sufficient for developing a measure the fact remains that simply tallying raw scores and using them as an indicator of the strongest endorsement for a particular item is presumptive. Presumptive in the sense that counting raw item rating scores alone are unable to factor in the inherent lack of linearity both within and across the various items measuring a specific construct.  This means that even if the MBTI’s dimensions and items were constructed properly (in a technically correct way), it would still lack the prerequisite linearity and conjoint additivity expectation of measurement!

  1. Where does that leave the MBTI?

Even at its most fundamental expectation of measurement, namely counting, the MBTI is deficient. We simply cannot use it to measure anything because it is firstly, not uni-dimensional and secondly, not reliable. Not because we hate Katharine and Isabel, but because they unwittingly used a flawed theory. Good psychological models like the Big Five-based assessments and the Occupational Personality Questionnaire of SHL can generally be revisited using proper measurement development tools like Rasch analysis because they are fundamentally sound from a construct perspective. They have the fundamentals of proper psychological measures in place and are not constrained by a single psychologist’s perspective on personality.

Also, instruments like the MBTI have had unchecked proliferation and PR over decades without serious internal scrutiny – much like the counsel of the proverbial tyrant’s inner circle. The resulting tragedy here is that there are now just too many collectives and individuals that are academically and economically vested in maintaining the MBTI in its current state.

I do believe that with a small measure of honesty it should be simple to revisit and re-frame the MBTI but…, what does one do when you have millions of sold assessments, thousands of associated products, and hundreds of endorsements by career psychologists? In the words of Adam Grant, a “wildly popular personality test, which is taken by more than 2.5 million people a year, and used by 89 of the Fortune 100 companies”.

Therein lies the rub” …

So finally, despite the hegemony of the MBTI over the last three decades or so let’s, in the light of the aforementioned, guard against blindly holding on to pet theories by keeping in mind a very infamous quote: “By means of shrewd lies, unremittingly repeated, it is possible to make people believe that heaven is hell – and hell heaven. The greater the lie, the more readily it will be believed.”

[1] The theoretical Big 5 are as follows: Surgency or Extraversion (Extraversion as defined in the TPQ), Agreeableness (Sociability as defined in the TPQ), Neuroticism (Constancy as defined in the TPQ), Conscientiousness and, Intellect, Imagination, or Openness (Originality as defined in the TPQ) –

For more information or assistance, please contact Gary Townsend at Skillworx Africa (Pty) Ltd, mobile 0824915392,  Or feel free to browse our website on