Transcription
A debate has been raging for ages among personality researchers about a fundamental question related to human nature. How many dimensions of personality are there? In other words, how many different personality traits do you need people to be scored on in order to have an accurate sense of what they're like? Do you need three numbers to represent a person's personality? Or could it be more like four, five, six, 10, or even 100?
This might even seem to you like the sort of question that has no answer, but surprisingly it does. And today we're going to explore what that answer is.
One of the early attempts to answer this dates all the way back to 1944 and it relates to what is today the most famous personality test, the Myers-Briggs Type Indicator, also known as MBTI. It scores people on four traits or spectrums. In particular, it evaluates where each person falls between being an introvert and an extrovert, where they fall between being a feeler and a thinker, between being intuitive and a sensor, and between being a judger and a perceiver.
Most MBTI-style tests then dichotomize the results. If you're above average on extroversion, they call you an extrovert. Otherwise, they call you an introvert. This dichotomization makes the tests far less accurate because on each trait, they put you in just one of two buckets. To give an example, most people fall near the middle on most of these traits. For instance, on sensing versus intuition, if you dichotomize the data, then you force lots of people who are right near the boundary to go on one side or the other, which makes it less accurate.
If we use the same underlying framework, but we don't dichotomize people and instead represent them as having a score on each of the four MBTI-style spectrums, we can see these suggest that there are four important dimensions of human personality. But is it really true that personality has four dimensions? And how would we actually know?
After the MBTI was released, many other theories of personality were put forward. Cattell developed a test based on 16 traits around 1950. Eysenck developed a test with just two traits around 1960.
If we fast forward a bit, a major new personality test contender came on the scene known as the Big Five. Whereas MBTI-style tests are based on a psychological theory about human nature, the Big Five started with empirical data. Researchers took all of the adjectives in the English language that could be used to describe a person, duplicated the list, and asked people, "Which of these adjectives apply to you?" They noticed the adjectives tended to group together. If someone said they were organized, they'd usually also say they were perfectionistic. If someone said they were social, they'd usually also say they were talkative, and so on.
These groupings were then made more rigorous using a method known as factor analysis. The idea is that you run an algorithm that tries to find hidden dimensions in your data. One of these dimensions might capture, for instance, how conscientious you are, which could include both your level of organization and perfectionism in one factor. When they ran these factor analysis algorithms, they found there to be five personality dimensions, hence the name, the Big Five.
Other researchers replicated the five dimensions on different populations using different personality questions, getting a similar result. There's some debate over whether you get these five dimensions in all cultures. In small-scale societies, for instance, some people argue you get fewer factors. But across most countries that were studied, these same five dimensions of personality were found.
So the Big Five says there are five primary dimensions of personality, whereas MBTI-style tests say there are four dimensions. How can we tell which is right?
We ran our own study aiming to answer this question. The idea was to see whether using five dimensions rather than four enabled more accurate predictions about people. In the study, we asked people 37 questions about their lives, such as how many close friends they have, how satisfied they are with their life, and whether they received a promotion at work. The goal was to predict the answers to these 37 questions. We also administered each person a Big Five test and an MBTI-style test based on the same underlying theory created by Briggs and Myers. We could then compare how accurately we could predict the answers to the 37 questions using each of the two personality tests.
The results were quite remarkable. The Big Five test had a prediction accuracy more than 50% higher on average than the MBTI-style test. But how do we know that this is because of having five factors rather than four? Well, we can tell because we can see what happens when we drop from the Big Five one of its factors in this case, neuroticism. Neuroticism is the tendency towards negative emotionality such as depression, anxiety, and moodiness. Even with that factor gone, the Big Five, or maybe we should call it the Big Four in this case, still made more accurate predictions than the MBTI-style test by about 20%. If we add that fifth factor back to the Big Five, it helps even more, boosting the accuracy by another 27%. In other words, the five dimensions of personality win out over the four. You get a lot more predictive accuracy using five rather than four factors.
But is five factors really enough to capture the major aspects of personality? In the 2000s, another contender came on the scene known as HEXACO. Its creators Lee and Ashton argued based on looking at international data sets that there was actually a sixth factor of personality. This sixth factor is known as H, which stands for Honesty-Humility. Low scorers in this dimension tend to be manipulative, arrogant, greedy, and lawbreaking. The HEXACO model also organizes the other dimensions of personality a bit differently than the Big Five. As we can see in the diagram here, Openness and Conscientiousness stay intact, but some other components get moved around.
We decided to put the HEXACO framework to the test by running another study. To do so, we developed a set of questions that reflected the scales used in HEXACO. We then ran factor analysis to confirm that this set of questions grouped the way that HEXACO said they should, and indeed they did. So, HEXACO passed the very first test. We then had people in the US take both this test and our 102-question Big Five test. That way, we could directly pit them against each other.
So, how well did each test fare? Well, there are two different ways of looking at this. First is to see how well each framework did at predicting the answers to the personality questions themselves. For example, knowing how conscientious someone is, you can predict how they likely answered questions related to being organized and perfectionistic. We can capture this idea in terms of the percent of variance explained by each personality factor. When we do this, we get the following results. What we can see is that across the full set of personality questions from both tests, each of the five factors from the Big Five captures quite a bit of variance in people's responses. The first factor captures the most. The second factor captures the second most, and so on. But all five of the first factors capture quite a lot of variance. But after the first five, something weird happens. We get a substantial drop-off. The sixth factor corresponding to H or Honesty-Humility from HEXACO captures a lot less variation in responses than each of the Big Five factors do. To make matters worse, the seventh factor is only a little bit worse than the sixth factor. So while it seems natural to stop after five, before the big drop-off, if you are to go further and keep six, it seems very hard to explain why you aren't also keeping seven and eight, and so on. This is not looking good for HEXACO so far.
But we still have another important way to look at this question. Ultimately, to be useful, personality tests shouldn't just model how you answer those exact questions in the personality test. They should be able to predict other things about your life. In this case, we collected 66 facts about people's lives. A big increase from our last study. And we attempted to use the five scores from the Big Five as well as the six scores from HEXACO to predict each of these 66 answers.
So, how well did each test do? Remember when we did a similar test comparing the MBTI-style test to the Big Five, we found that that extra fifth factor added a lot of predictive accuracy. So is that what we found in this case? Actually, no. The two tests did about the same at predictive accuracy. HEXACO was only a tiny bit more accurate. You might think that using six factors rather than five should always improve your prediction accuracy because you literally know more about a person. You have six numbers about them rather than just five. But in this case, that sixth number didn't do much for you. It led to a negligible increase in prediction accuracy.
But let's look deeper into this. What specific facts about a person were we better able to predict using HEXACO than using the Big Five? Since HEXACO adds the H factor, Honesty-Humility, we might expect that it did especially well at predicting factors related to morality. And indeed, this was the case. It had an advantage relative to the Big Five at predicting whether a person had used physical force against someone else, whether they'd been arrested in the last 10 years, and whether they cheated on a romantic partner. Interestingly, as shown in this diagram, there were actually more facts where the Big Five did a better job of predicting them than HEXACO did than there were the reverse. The Big Five did a better job of predicting who had donated to charity in the past year, who had been fired from a job, and who'd received an award.
We believe that our study on HEXACO shows that for people in the US at least, it adds little to no value above and beyond the Big Five unless your concern is specifically on topics related to honesty or humility, such as moral behavior. This doesn't rule out the possibility that HEXACO could be useful in other countries, but it does cast doubt on its universality.
So how many major dimensions of personality are there truly? Five appears to be the best answer as science understands it today in terms of getting the biggest bang for your buck. Once you know how open, conscientious, extroverted, agreeable, and neurotic a person is, you know the broad strokes of their personality.
But there's a very important caveat here. This does not mean that there are only five dimensions of personality that exist. What do I mean by that? Well, suppose, for instance, that you're particularly interested in risk-taking behavior. It's known that this is linked to the extraversion dimension of the Big Five. But if what you really care about is risk-taking, you can improve your accuracy at predicting it by adding another scale to measure risk-taking in particular. This will do a better job at predicting risk-taking than if you just use the broad Big Five dimensions.
You can actually take this logic even further. If your goal is simply to make the most accurate predictions possible, you don't care about compressing or representing that information, then you don't even need the Big Five. You can train a model to predict things using all the questions in a personality test. And in fact, it tends to be more accurate than using the Big Five. The Big Five does not capture everything there is to know about personality. There are aspects of personality that are missing from the Big Five. And you can subdivide each of the Big Five into narrower facets and then measure those to get more nuanced results. In fact, we have a project to try to find every facet of personality that's missing from the Big Five. So, I'll have more to say about that soon, hopefully.
What's so special then about the Big Five? It's that these five factors of personality capture a ton of information about people in just five numbers. Those five scores in our study, for instance, captured 40% of the variation in people's responses on individual questions. Those first five factors seem to be special based on how much predictive juice you get from so little squeeze. In a sense, there are an unlimited number of factors of human personality, but five provides a surprisingly accurate model that's hard to improve on.
If you want to know more about your personality, take our free Ultimate Personality Test. It will give you your Big Five results as well as your results from other frameworks. If you found this interesting, I'd really appreciate if you subscribe. We do regular deep dives into psychological topics and have lots of new research coming out.