MBTI: A lukewarm analysis


MBTI, or the Myers-Briggs Type Indicator, is probably the most popular personality test. It contains four axes: Introverted/Extraverted, iNtuitive/Sensing, Thinking/Feeling, Judging/Perceiving. If you take the test, you may be assigned one of 16 personality types, for instance I would be INTJ.

The MBTI is regarded as pseudoscience, perpetuated by the popular consciousness and HR departments rather than academic research. One time I asked a personality psychologist and she said it was just so far off from reality that nobody even bothered talking about it. Psychologists prefer to talk about another personality model, called the Five Factor Model, also known as The Big Five. This has five axes, labeled Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism (OCEAN).

I’ve often remarked that although the Five Factor Model is supposedly more scientific, it’s clearly a lot less compelling. And isn’t that something? I couldn’t honestly say that I find astrology compelling, or ear candles compelling, but the MBTI, now that’s some yummy pseudoscience. I have some remarks on what makes MBTI a pseudoscience, what makes it compelling, and what its problems are.


MBTI as pseudoscience

Caveats: I do not have a psychology background, and due to constraints on my time I do as little research as possible. This is a casual chat, and I invite readers to make corrections and fill gaps.

You can design a questionnaire with any set of questions, and calculate a score from them. You don’t need any scientific training to do it. Just call it the Hermione/McGonagall score and brand it as a “What Harry Potter Character are You?” quiz. As ridiculous as it sounds, that quiz measures something “real”, even if only how you respond to that quiz.

The question is, how do you figure out what the “best” thing to measure is, and how do you pick the “best” questions to measure it? These are difficult questions, with no good answers. But at least we can throw a bunch of math at it and pretend we solved the problem.

The Five Factor Model is based on factor analysis. My understanding is that psychologists comes up with hundreds(?) of questions relating to personality–basically everything they can think of. After giving this questionnaire to a bunch of people, they use a statistical software package to assign each respondent a mere five numbers, and these numbers are optimized to predict how each respondent answered the hundreds of questions. Afterwards, they whittle down the list of questions to get a shorter questionnaire that suffices to measure the five numbers.

Nothing about this methodology suggests that the Five Factor Model is measuring any fundamental truths about the human brain or whatever. But it at least guards against the personal biases of the researchers. Different researchers can try it, and they’ll find more or less the same thing. (It seems to me that bias may enter through the design of the initial large questionnaire, but eh??)

In contrast, the MBTI is based on a fiat assertion about how personality works. Well sure, Myers and Briggs and Jung made some observations and applied their personal insights before making fiat assertions. But there aren’t really any guard-rails against the personal biases of its creators. MBTI is measuring something real to be sure, but there isn’t a good answer to “Why measure this thing and not some other thing?” Why MBTI and not Harry Potter characters or enneagrams?

What I have come to accept, is that the MBTI doesn’t need to be scientific to be useful on some level. It doesn’t need to be designed in a way that’s robust to the personal biases of its designers. Observations and personal insights followed by fiat assertion are “good enough”. Honestly I think HR departments could squeeze value out of Harry Potter quizzes too, if they set their minds to it.

I also have to say, some of the stuff that comes out of academic psychology is hardly any better. Consider Moral Foundations Theory, which describes people’s moral concerns in terms of five foundations: care, fairness, loyalty, authority, and purity. Many years ago, this was popularized in psychologist Jonathan Haidt’s book The Righteous Mind. But what is it based on? As far as I can tell, it comes from personal observations followed by fiat assertions, no better than MBTI.

MBTI as compelling

For all the science of the Five Factor Model, and pseudoscience of MBTI, oddly enough they converge. It turns out that Sensing is correlated with Openness, Thinking with Agreeableness, Perceiving with Conscientiousness, and Extraversion with Extraversion.

But MBTI still feels far more compelling, and I think it’s all in the names of the axes. Where MBTI has flattering names for both ends of each axis, The Big Five only name one end of each axis, and either use unflattering names (neuroticism) or names that are unflattering to the other side (agreeableness, conscientiousness, openness).

Perhaps this is because the MBTI is ideologically committed to the idea that personalities are value-neutral. You’re Valid no matter where you land. Psychologists, on the other hand, have no such commitment, and will happily test the hypothesis that some personalities lead to worse outcomes.

Although, I don’t think the unflattering names are based on any sort of empirical data. The names are totally made up. Factor analysis will give you a set of axes, but doesn’t tell you what to call them. You have to look at what questions were associated with each axis, and make names up. And psychologists are not very good at coming up with names. When I blogged about the Big Five back in 2012, I dug up a table listing various names that psychologists had used for the five factors. Apparently OCEAN is just the tip of the godawful names iceberg.

Funnily enough, last I checked Wikipedia, some genius decided to give each of the spectra more flattering descriptors for both sides. I’m not sure if that comes from a researcher or a Wikipedia editor, but bless them.

  1. Openness to experience (inventive/curious vs. consistent/cautious)
  2. Conscientiousness (efficient/organized vs. easy-going/careless)
  3. Extraversion (outgoing/energetic vs. solitary/reserved)
  4. Agreeableness (friendly/compassionate vs. challenging/detached)
  5. Neuroticism (sensitive/nervous vs. secure/confident)

What would be fun, is coming up with axes names that make both ends sound unflattering. Leave your ideas in the comments.

Deeper problems with the MBTI

On the internet, the MBTI is mostly meme fodder, and is harmless as meme fodder goes. But that’s not the only context where it is used. The thing to understand, is that the MBTI is a commercial product sold by the Myers-Briggs Company. You can become a certified MBTI practitioner by paying $2500 (in my area) for a four-day class. And it costs even more to administer tests, produce reports, and take refresher courses.

I mean, presumably that mostly comes out of the budgets of HR departments, so I guess it’s up to you how upset you feel about money moving from the coffers of one corporation to another. Personally I don’t care that much, I’m just submitting this for the reader’s consideration.

A long time ago I knew an MBTI practitioner, and she answered a few of my questions about it. One of my concerns was that MBTI would be used to select job candidates or the like. Apparently practitioners are taught that this is unethical, and good for them. It does leave me wondering what decisions, if any, are made on the basis of MBTI, and whether those decisions could possibly be ethical. It seems like the MBTI is caught between unethical and useless.

Another concern was that MBTI is treated as a typology, when it’s obviously not. Imagine saying that there are two height-types: above average and below average. Well, I guess that’s technically correct, but a misleading way to think about it, when the mode of the distribution is pretty close to the boundary. The 16 MBTI types are like that, but more so. What I’ve heard is that MBTI practitioners are aware of the issue, and say in their defense that it’s just a common misconception. But, this is clearly a “common misconception” that they are deeply invested in promoting themselves. The Myers-Briggs website says:

The 16 personality types of the Myers-Briggs Type Indicator® instrument are listed here as they are often shown in what is called a “type table.”

And you can find self-appointed defenders saying stuff like

Psychological Type is dynamic – each of the 16 Types are psychological systems of energy

Yeah, so that’s anti-convincing.

Comments

  1. says

    A couple additional remarks that I forgot to add:
    1. The design of the five factor model also implies that it has predictive power. That is, your answers to a short questionnaire predict your answers to the longer one.
    2. While the MBTI is not to be used for candidate selection, other measures like the five factor model definitely get used that way, which certainly raises some ethical concerns.

  2. Mara Jade says

    Thanks for the post? Made me curious about the Big 5. It makes me wonder why HR deptarments use MBTI. It sounds useless to them. I took it in a community college setting, but it was just something for students to ponder over. It sounds more useful to the individual taking it than it does to HR.

  3. Mara Jade says

    Whoops. That question mark shouldn’t be there. Editing problem. I sincerely meant, “Thanks for the post.” 🙂

  4. says

    @Mara Jade,
    I have no idea of the ratio of departments use MBTI vs other personality types. FWIW my current employer gave job candidates a personality test that I believe was based on the Big Five. I guess they can get away with that because the Big Five is ~scientific~. I think I’d rather see a company apply pseudoscience in a cautious and harmless way than see them apply a scientific measure in a way that has much greater potential to cause harm, and probably isn’t endorsed by the original researchers anyway.

  5. says

    I vaguely remember going to a workshop for student leaders in undergrad that was based around taking a personality test (can’t remember if it was MBTI or a different one), then talking about how different types can affect different workstyles, and what recommended workstyles might be for our newly-identified subtype, and then also information on how to best work with, encourage, and give feedback to others using what we know about their types (while we were all students, it looked like some of the materials were likely also used for managers).

    The theory from a managerial perspective was that knowing about people subtypes would help you understand how to set them up with a workstyle that might be more suited for them, or how to give feedback. So like, for someone who had a more logical type, you might give feedback “let’s not do x because it’s less efficient”; to a more emotional type you might give feedback like “let’s not do x because it might hurt people’s feelings”. Or you might have more or less frequent check-ins, or provide more or less detailed project instructions based on type. They gave us a whole printed packet about specific suggestions for how to work with people of each type, but unfortunately I didn’t keep it.

    I do remember that in the workshop they said that you aren’t supposed to use types to decide who you do/don’t hire or who you give what kinds or amounts of work….but based on the way they emphasized it, it definitely made it sound like the kind of disclaimer they have to give because so many people do exactly the opposite.

  6. Loren Petrich says

    I find the Big Five MUCH more intuitive than the MBTI. In fact, one has to ask if the MBTI is much more than (say) astrology, something passed down the generations without much assessment of it.

    There are a LOT more papers written about the Big Five than about the MBTI, and the Big Five seems to have real-world validity. For instance, Conscientiousness can be divisions into two subtraits, Industriousness and Orderliness, or else six, Competence, Order, Dutifulness, Achievement Striving, Self-Discipline, and Deliberation (I’ve yet to find a consensus on such trait divisions). It was thought to be a moral trait before a psychological one. C’ness is positively correlated with academic and career success, and that is because a big part of c’ness is diligence, and that’s what’s necessary to succeed.

  7. Loren Petrich says

    The Big Five has been extended to nonhuman species. I’d have to hunt down the link, but it starts with describing how a certain Suzie is grouchy and otherwise unpleasant, and stating “Suzie is a bear”.

    Outside our species, conscientiousness has only been found in chimpanzees. If one wants to make a termite fishing rod, one has to concentrate on that task for as long as it is necessary for making that rod.

    Agreeableness has been found in other mammalian species, like dogs. That’s a part of being social, I’m sure.

    Analogs of Extroversion (Boldness) and Neuroticism (Sensitivity to Threat) have been found in most of the species tested, including in guppies and octopuses.

    I’ve seen a theory of supertraits, along with associated neurotransmitters:
    * Plasticity – Dopamine – Extroversion, Openness?
    * Stability – Serotonin – Conscientiousness (motivational stability), Agreeableness (social stability), negative Neuroticism (emotional stability)

    Dopamine, serotonin, and similar neurotransmitters are found all over Bilateria, and that may explain why analogs of extroversion and neuroticism are similarly distributed.

  8. Loren Petrich says

    A common form of factor analysis is principal components analysis (PCA). It roughly consists of fitting the data points to a multidimensional ellipsoid and then finding the longest axes of that ellipsoid. One can then find out where the points are in this ellipsoid, so one can interpret its axes.

  9. Loren Petrich says

    An interesting application of the Big Five was by psychologists Rubenzer and Faschingbauer. They evaluated most US Presidents with it, using some 160 experts’ rather subjective comparisons of the presidents with what is typical of the general population. Despite this subjectivity, some results stand out.

    Presidents are usually high in conscientiousness, at least in diligence. That is what one expects of career successes. Exceptions are rare, like Ronald Reagan. He likes to get up at the crack of noon, and he liked his briefing papers short.

    Presidents are usually low in agreeableness. This makes it easy for them to select who they want to try to please.

    They were very variable in the other three traits. Thomas Jefferson was high in openness and Harry Truman low in it, for instance.

  10. Loren Petrich says

    I’ve found a blog post on a paper that correlates Big Five factors with words in speeches. Applied to Donald Trump and Hillary Clinton, Trump comes out low in conscientiousness and Hillary Clinton high in it.

    This high c’ness makes Hillary Clinton like many presidents, and it seems to me that some other notable ones share that feature. Nancy Pelosi has a legendary work ethic, Mitch McConnell calls his memoirs “The Long Game” after his strategizing, and Alexandria Ocasio-Cortez has shown a lot of diligence over her life — and she considers it a very positive quality in herself and others.

    Trump shows low c’ness in other ways. He likes his briefings short, and he spends much of his day in “Executive Time”, watching Fox&Friends and tweeting. Which is more evidence of what an anomalous president he has been.

  11. says

    @Loren Petrich,
    MBTI is too much to swallow, but you believe in the application of the five factor model to nonhuman mammals animals and to politicians? That sounds like extremely questionable research, with fatal flaws in their premises.

    Factor analysis and PCA are distinct analyses, although at the level of depth in this discussion they are the same.

  12. Sue says

    At my place of employment, personality type quizzes are given by HR. (It’s not Myer Briggs, though.) Then managers are encouraged to use the information to best interact with their employees. For example, if you have a personality type that responds well to inquiries about family, remember to ask about that with those “types.” For other types, give them space. These are small examples, but it seems like HR is *attempting* in a clumsy way to address the fact that people are different and managers should deal with that rather than forcing everyone to work the same way.

  13. Jazzlet says

    It’a been a while since I’ve done anything like this, but my problem with the ones I have been subject to that I recall is the insistence on an either/or answer to the questions, when for most of them I could give a 60/40% or 20/80% or 50/50%, but not pick one of the two options. It’s obviously for a far more limited purpose, but I have recurrent depression and regularly answer a suite of questions designed to give a snap-shot of my state of mind and nearly all of those give a sliding scale. I’m pretty sure that the results do reflect my state of mind fairly well and therefore help my therapists judge how I am doing rather better than any system that insists on binary answers.

  14. says

    @Jazzlet #13,
    I know that people who answer questionnaires don’t like to hear this, but the fact that a question doesn’t make sense or doesn’t provide enough possible answers does not necessarily make it a worse question. You can imagine that when scientists design that large questionnaire, they might put in some questions that make less sense than others, and they might determine, empirically, that those are the questions that give the strongest measure on the Big Five. Of course, if the survey isn’t designed with any sort of statistical analysis, then intuitively I would think that questions that make more sense would be better.

  15. Loren Petrich says

    https://gosling.psy.utexas.edu/wp-content/uploads/2014/09/CDPS99Xspeciesreview.pdf — that work on extending the Big Five model to other species. It required some interpretation of personality features that may seem rather farfetched. It started with “In a recent article in the Los Angeles Times, Robert Fagen, a professor of biometry, described Susie as irascible, irritable, grumpy, and manipulative. This is hardly newsworthy, except that Susie is a bear.”
    The interpretations:
    Chimpanzee:
    Neuroticism: Emotional Stability, Audiovisual Reactivity, Excitability-Agitation
    Agreeableness: Agreeableness, Aggression-Affinity
    Extraversion: Surgency, Affect-Extraversion, Social Play
    Openness: Openness
    Conscientiousness: Dependability, Task Behavior
    Two additional ones: Dominance, Activity

    Cat:
    Neuroticism: Emotional Reactivity
    Agreeableness: Affection
    Extraversion: Energy
    Openness/Conscientiousness: Competence

    Octopus:
    Neuroticism: Reactivity
    Extraversion: Bold vs. Avoiding

  16. Loren Petrich says

    A big problem with the MBTI is its binary divisions. There is zero evidence that human personality works that way. Whenever one projects personality-test answers onto some set of factors, one finds a continuous range of factor values. I once took a MBTI test that was honestly-designed enough to return these factor values. It reported that I was strongly introverted and that I’m almost halfway between Judging and Perceiving.

    Every Big Five test that I’ve ever taken has been careful to return such continuous values.

    https://arstechnica.com/science/2018/09/study-people-tend-to-cluster-into-four-distinct-personality-types/

    Notes that continuity, but discusses evidence of spots in Big-Five parameter space with extra concentration:
    * Average: Neu+ Ext+ Opn- Agr+ Cnc+
    * Self-Centered: Neu0- Ext+ Opn- Agr- Cnc-
    * Reserved: Neu- Ext0 Opn- Agr0+ Cnc0+
    * Role Model: Neu- Ext+ Opn0+ Agr+ Cnc+
    I myself am Neu- Ext- Opn+ Agr0 Cnc+ — I fit none of these clusters

  17. Jazzlet says

    Siggy @14
    It’s not that the questions don’t make sense that I mind so much as when I am asked to pick one of two answers as true, but I think both are true.

  18. anothersara says

    I love the analogy with Harry Potter character quizzes. I agree. I think personality quizzes/tests/etc sometimes can offer useful insights – and I think a well-made Harry Potter character quiz could work as well as any other well-made personality quiz/test/etc.

    By the way, I don’t recall hearing that you got hired, congratulations!

  19. anat says

    I recall a Harry Potter – themed MBTI quiz. Problem is, fans can’t agree which character matches which ‘personality type’.

    I have a colleague who is very much into personality typing – more into enneagrams than MBTI (in fact, working towards a certification to teach about enneagrams, no idea who certifies such things). She was able to describe to me the 4 binaries of MBTI in a way that made my scores at that time make some kind of sense to me, but then the description of my alleged ‘type’ was so not me. Whatever.

    Actually my husband and I found our enneagram-types to be more insightful – we learned what the worst problems in our respective approaches to life were, and some ways of mitigating them.

  20. Loren Petrich says

  21. Dunc says

    It makes me wonder why HR deptarments use MBTI. It sounds useless to them.

    Well, considering that HR departments are mostly useless anyway, I don’t think it matters. It gives them the illusion that they’re doing something…

    The theory from a managerial perspective was that knowing about people subtypes would help you understand how to set them up with a workstyle that might be more suited for them, or how to give feedback.

    Or you could try to actually get to know your employees… Sorry, crazy talk, I know. We must have a standardised test that allows us to allocate people into categories which we can then treat according to some rules-based system.

    Sorry, I appear to be suffering from a little excess cynicism today.

  22. says

    @Loren Petrich #15,
    I appreciate the citations, and at a glance it looks like legitimate research. But I maintain skepticism that this is a fruitful direction for research.

    @jazzlet #17,
    When I talk about questions that don’t make sense, I’m also thinking about questions where the response options are too limited.

    @Sara #18,
    Yep, I have been hired. Not sure I meant to let that slip…

    @anat #19
    Although I said MBTI feels compelling, I definitely can’t say the same of Enneagrams. The type descriptions felt like throwing a bunch of stuff at a wall. And that’s before even looking at the numerology. But if you found it insightful…

  23. Sue says

    @Dunc I totally agree. I think they want to sort people into types in order to “know them” without really trying. (How to succeed at business…)

  24. Korey says

    My attempt at unflattering labels:
    O: Adrenaline junkie vs stick in the mud
    C: Fusspot vs careless
    E: Obnoxious vs wallflower
    A: Brown noser vs hardhead
    N: Neurotic vs boring

Leave a Reply

Your email address will not be published. Required fields are marked *