Thursday, June 28, 2012

Personality tests: MBTI and FFM

The Myers-Briggs Type Indicator (MBTI) is a personality test with four different axes: Introverted/Extraverted, Intuitive/Sensing, Thinking/Feeling, Judging/Perceiving.  It was created by two non-psychologists, based on ideas published by Carl Jung in 1921. 

The Five Factor Model (FFM) is a personality test with five different axes: Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.  It is based on empirical research and factor analysis.  Basically, they asked people a bunch of personality questions, and used quantitative analysis to make the most predictive model using the smallest number of factors. 

Years ago, I used to be annoyed by MBTI, because it's so obviously unscientific.  I would give FFM as an example of a personality test that is actually science-based.  But these thoughts never made it to my blog, probably because they were half-baked.  More recently, I've reevaluated MBTI upwards, and FFM downwards.  I no longer think that the MBTI needs to be science-based to be useful, in the sense of resonating with us, and describing patterns that we deem important.  As for FFM, we'll get to that later.

Skeptoid summarized the criticisms of MBTI well, but I should summarize them separately here, along with my thoughts.

1. MBTI isn't psychology.  Psychologists don't take MBTI very seriously, and the scientific study of personalities is very far from MBTI.  Carl Jung was a psychologist but he belongs to the proto-scientific era of psychology (eg he believed in anecdotes over statistics).

2. MBTI does not describe underlying psychology.  Obviously, since MBTI isn't psychology.  But I wanted to include this point because people jump so quickly from "patterns that resonate with us" (ie "usefulness") to "the way stuff works".  There isn't any "extraversion" chemical which is more concentrated in extraverts' brains, but extraversion could still be an emergent pattern.  And if a pattern resonates with us, or seems important, that fact emerges from even more complicated stuff like society and culture.  It's likely that a useful personality model would bear no relation to the underlying psychology, and a personality model based on underlying psychology would not be very useful.

Note that FFM, though it comes from psychological research, is not based on underlying psychology either.  It's a phenomenological model.

3. MBTI is not a typology.  Though people's results are often summarized by listing their "types" (eg INTJ, ENFP), each of the four axes is just a normal distribution centered on the dividing line.  That means most people sit on the dividing line or close to it.  Calling it a typology suggests that people would cluster into the 16 different types, with a lower density of people on the dividing lines.

Note that MBTI practitioners mostly agree that MBTI is not a typology, even though that is what the T stands for.  I'm not sure about FFM, but as far as I know it does not purport to be a typology.  So the fact that MBTI isn't a typology isn't so much a criticism as a comment.

4. MBTI is not self-consistent.  Meaning, people who retake the test tend to get different results.  Or so Skeptoid claims.  I generally trust Skeptoid, but to properly support this claim, I need a broad literature survey, which I'm unwilling to do.

5. MBTI may be influenced by the Forer Effect.  The Forer Effect is when people accept vague positive descriptions of themselves as highly accurate, even though they could describe nearly anyone.  I think the names of the four axes are neutral, but the 16 "types" often come with additional flattering descriptions.  If the Forer Effect influences MBTI, people may accept results as highly accurate even if the descriptions were all randomized.

Earlier, I described "usefulness" as a subjective quality.  If a personality test describes patterns that resonate with us, then it is useful.  The Forer Effect is a way for a personality test to appear useful, but not actually be any more useful than a test that spits out random flattery (eg astrology).  However, it is not clear if the Forer effect is at work here.  I think that would be interesting to research.


Okay, so here's what annoys me about FFM.  FFM is predictive, meaning it makes predictions that could be true or false.  So it's not really affected by the Forer Effect.  We could call the five axes in FFM 1, 2, 3, 4, and 5, and it would still succeed just as well in its purpose in making predictions.  However, psychologists decided to tack on additional descriptions of the five axes, calling them Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.  Of those five words, three are normatively positive (Openness, Conscientiousness, and Agreeableness), one is normatively negative (Neuroticism), and one is neutral (Extraversion).

Couldn't psychologists have come up with less normative descriptions?  It could be the case that Agreeableness really is correlated with characteristics that nearly everyone deems positive, but surely this is something to demonstrate with research, not an assumption to sneak into the axis name.  Scientists have been known to do things like describe butterflies with same-sex behavior as immoral, and I think this is similar.

I am not really sure where the names came from.  But I spotted this history in a 1990 paper:

(click for a bigger picture)

Psychologists are laughably inept at coming up with non-normative names.  "Likeability",  "Culture",  "Positive emotionality", "Psychoticism"?  Srsly.


drransom said...

I've never believed the Forer Effect has a major effect on the MBTI. It's possible, but the ex ante plausibility seems very low. Hand me the ESfJ and INTP descriptions and I have no doubt which one I'd pick.

I'm also not terribly convinced that being unscientific is a killer here. I see the MTBI as more of a humanities thing--a way of looking at yourself and other people that provides some insight.

kyrilluk said...

The MTBI is as usefull as the Astrological pages in the Cosmopolitan...

Lars said...

The Forer effect is part of EVERYTHING that gives positive or very general descriptions, so its also part of the MBTI. The important question is: how big is this part?

Of course you can distinguish ESFJ and INTP. But if you present someone who doesn't know about the MBTI and scores INTP, the rEsults of INTP, INTJ, ENTP and ISTP, i doubt hi woul easily choose the §right" profile.

drransom said...

Lars, I agree that people might have difficulty distinguishing between the "correct" profile and profiles that are one step away. (The extent might depend on which set of descriptions was being used!) But the fact that the claim needs to be restricted in this way is indicative of the point that the MBTI categories are not meaninglessly general.

Compare, say, the completely worthless personality system of astrological profiles.

Anonymous said...

The MBTI actually has some predictive validity in that people often have the type of jobs and education that the theory predicts.

And the Big Five isn't all that predictive. It's popularity seems to be more out of a lack of theories within personality research than anything else. Some facets of Big Five are more predictive though.

miller said...

If you showed me astrological profiles for Scorpio and Taurus, there's absolutely no doubt as to which one I'd pick. But I still think astrology is scorpion frass.