Penguinville

hayekd:

science:

Unfortunately, I feel the need to issue an all-out bullshit alert. The popular Myers-Briggs personality test is bullshit. The Big Five, however, are not, but we’ll get to that.

You’d think that a psychological test that is popular everywhere except in psychology would raise some eyebrows, but no matter. Normally rational, scientifically-minded people seem to fall as hard for this as everyone else. The Myers-Briggs Type Indicator (MBTI) is deeply flawed in all the important ways.

MBTI divides human personality into four binary dimensions, giving a total of sixteen personality types. These dimensions weren’t chosen based on evidence, they were pure speculation. Carl Jung, who’s got one thing going for him (he was, at least, a psychologist) started speculating, and then two women with no background in psychology, Katharine Cook Briggs and her daughter Isabel Briggs Myers, extrapolated from that speculation, adding their own unfounded speculation on top. Given this highly speculative, unempirical methodology, there is no reason to suspect the result to be accurate. Nevertheless, it could be that they just happened to find some distinctions that actually mattered based on pure luck. Unfortunately for MBTI, this is the opposite of what studies have found.

MBTI isn’t very reliable: people who take the test repeatedly with some time in between often report different results, despite not actually undergoing personality changes. The reason is understandable when you consider this: in the MBTI, every dimension is binary, not continuous and normally distributed. If this were true, we would expect to find a bimodal statistical distribution (two peaks, little overlap between the two clusters). But when we look at actual empirical data, we tend to find a normal distribution. MBTI chops the bell curve in two right down the middle, so people who are very close can be assigned opposite categories. Since this is true of all four categories, two people who have almost identical personalities, but happen to fall on either side of the big dividing line, can actually get opposite scores.

Nor do the four dimensions stand up to factor analysis, suggesting they are not four independent and exhaustive dimensions of personality. And to top it off, even though the MBTI is widely used (or believed to be applicable) in career planning, studies don’t find that scores predict careers. There aren’t more “Caregivers” in caregiving vocations than there are in the general population. Myers-Briggs doesn’t accurately predict behavior.

At this point, we might be ready to give up on personality tests. But there is hope. If we discard our biases and instead look at the data to see what we might find, a different picture emerges. The introversion-extraversion axis, found in MBTI and many other personality tests, stands up to scrutiny. The others, not so much. Instead, we find five big axes: openness to experience, conscientiousness, agreeableness, extraversion and neuroticism. There’s some debate about what to call some of these, as they don’t map very neatly to individual words.

This scheme, the Big Five personality traits, is derived from data, and not surprisingly, it generally shows good predictive power. For example, a meta-analysis found that “each [personality] disorder [in the DSM] displays a five-factor model profile that is meaningful and predictable given its unique diagnostic criteria.” And unlike MBTI, not every personality type is positively loaded: people who score high on neuroticism and low on agreeableness are more likely to be diagnosed with personality disorders. The Big Five generally predict behavior in the areas you’d expect personality to be a factor.

My personal (speculative!) theory why tests like Myers-Briggs capture the popular imagination in ways that the Big Five do not is that the Big Five don’t come with a story. Indeed, that’s the major criticism of the Big Five: there is no theory. The Five are derived from empirical data, and so don’t come with a built-in explanation. Why do these personality dimensions exist, and why do they influence behavior the way they do? This is still an open question for research. Myers-Briggs, on the other hand, starts with the story. The explanation precedes the discovery: data are shoehorned in to match the theory. Humans are attracted to stories, rather than to impersonal trends derived from large datasets.

In addition, I suspect MBTI abuses the Forer effect. It’s a well-known, documented fact that when people hear descriptions of their personality that are supposedly tailored to them specifically, but actually vague and fit most people, they tend to rate the descriptions as highly accurate. This is especially likely if the description is positive. Seriously, read some of the official descriptions of the 16 types and tell me you don’t agree with more than one of them.

No wonder I’ve gotten a different result every time I’ve taken the Myers Briggs.