Are there really four personality types?

Source: Photo of Heinz Teh / Shutterstock

We like to classify ourselves and others into personality types. Maybe you are a "Type A." Or perhaps the Meyers-Briggs-type indicator, which many versions have been published on Facebook, calls you a "Protector" (ISFJ). But what is the difference between these types? Is everyone clearly type A or type B, with no one in between? Or do people fall on a spectrum? If there is a spectrum, are there two groups with a few people or a group in the middle with a few people at the extremes? Or something else? An article published today in Nature Human Behavior finds "solid evidence for at least four distinct personality types". But before you call Carrie, Samantha, Miranda or Charlotte (or Vincent, Drama, Turtle or E), see what they really have found.

Decades of research have revealed at least five independent personality traits: openness to new ideas, professional conscience, extraversion, amenity and neurosis. You can be up or down on each one. This is what is called the five-factor model of personality. Meanwhile, research on personality types – clusters of traits – has been more controversial. Some studies find three types, called "resilient", "over-controlled" and "under-controlled", but others do not. The new study analyzed a much larger dataset than previous studies – 1.5 million personality test scores, compared with less than a thousand for most studies – to try to solve the problem.

The researchers first used a published dataset of 145,000 notes from a 300-question online personality test. (They do not say where the participants lived.) They gave everyone a score for each of the five personality traits. Then they applied a clustering algorithm, a software program that aimed to find clusters of nearby data points, how you could look at a map with a dot for each person, and draw circles around cities.

The method highlighted four groups, which the researchers have named: "average" (average across the five dimensions of personality), "role model" (low on neurosis, grand opening, professional awareness, extraversion and amenity) . weak on openness, professional conscientiousness and friendliness) and "reserved" (low on openness and neurosis).

They repeated the analysis with three other large datasets and found similar patterns. They also found correlations between types and demographics. The models are usually over 40 years old. It is unlikely that women over 60 are egocentric.

But how big are the four groups? What is the archetype of types? In the metaphor of the population, are there dense cities surrounded by desert or is it a homogeneous suburban land with several areas barely densified? "We can get a more nuanced view of the degree of clustering," the authors write, "by visually exploring densities in suitably defined 2D hyperplanes (Fig. 5)." Clusters exist in five dimensions, one for each trait, which is difficult to visualize, so they took a two-dimensional slice of this 5D space and have it presented in an additional document to the article. Image a slice of bread with raisins from a bread.

Source: Nature Human Behavior

Unfortunately, the figure does not give much insight. On the left you see the raw data. More vivid colors indicate a greater concentration of people. As expected, people are concentrated in the middle of the 2D plot (which is near the middle of the 5D space); more people are close to the average on each line than on the extremes. The three black circles are supposed to describe three of the four groups. In my eyes, they are invisible. A suburban land. The image on the right shows the difference between the found distribution and what is expected of a normal distribution. Here, one can count between three and fifteen clusters, and none of them is centered on the three types of labeled personality.

(Also note that the four groups are related to what one would expect from random data in which everyone is on a smooth bell curve.Imagine it in two dimensions, like a real bell, with a size indicating the number of people concentrated in the center, these groups are only bumps on the surface of the bell, so there is a very important type of personality: completely average.

Seeking intuition about the rudeness of personality types, I sent an email to the authors last week. Martin Gerlach, a physicist from Northwestern University, responded, "I think it's fair to say they're certainly not well separated." There is no visual way to represent the disorder, I may be asking him about density piles elsewhere. "I'm scared, I can not put a number on it," he said. "The main point of view of our study is that there is a solid signature on lumps." In suburban countries, these slightly denser areas are probably slightly more dense.

So, although people like to find models in the world, classify them carefully. That means do not assume that if someone matches a guy in some ways, he will do it in others. Let's say that someone seems weak in openness and awareness. You can assume that it is also weak with respect to the pleasantness, as these three traits are grouped into an "egocentric" type. But the likelihood that someone with low open-mindedness and dedication is also low in amenity seems hardly more likely than the likelihood that the person will be very satisfied. Your hypothesis could easily mislead you.

Rather than suggesting that we categorize people, what this study reminds us is that people are not always up to the bins we expect.