How Many Traits Make Up Your Personality?
- Markus Over
- May 7
- 12 min read
Short of time? Read the key takeaways.
🧠 Some personality models are empirically derived. The Big Five personality model analyzes personality in terms of:
Openness (to Experience),
Conscientiousness,
Extraversion,
Agreeableness, and
Neuroticism (also known as ‘emotional instability’)
This model emerged from the lexical hypothesis, which claimed if a personality difference matters, languages will eventually invent words for it. Factor analysis of thousands of personality adjectives across many datasets, tended towards five stable groupings, which resulted in the Big Five personality model.
📐 The HEXACO personality model adds a sixth dimension: Honesty-Humility. This trait captures sincerity, fairness, and and a low appetite for exploiting others. This content is already somewhat present in Big Five's Agreeableness and Neuroticism dimensions.
📉 Factor analysis reveals a natural break at five. In the data from our study, we found a fairly striking drop in the amount of variance explained, when moving from the 5th factor to the 6th. The 5th factor still explained over 5% of additional variance, whereas the 6th explained only about half as much. And after that, the 6th, 7th, and 8th factors were all relatively similar to each other in usefulness, suggesting five is a natural stopping point.
🎯 HEXACO's predictive edge is narrow, not general. Both models showed similar predictive accuracy across 66 life outcomes. HEXACO tended to outperform Big Five on outcomes where Honesty-Humility is directly relevant, such as cheating, arrests, or substance addiction, but more outcomes were predicted better by the Big Five than by HEXACO.
🗺️ Five dimensions may be an optimal personality model. Neither the Big Five nor HEXACO model captures the fullness of human complexity, but our study suggests that five factors strike the best balance: informative enough to be useful, compact enough to remain practical for describing and predicting behavior.
There's a debate that has raged in academic journals and among personality researchers about the nature of humans: how many dimensions does it take to best represent a person's personality? Or, put another way: how many attributes would you have to score someone on to have a good sense of what they are like?
In this article (which continues below), we dive into the topic, but if you prefer, you can instead watch our video on the subject (or you can go even deeper with our full research report):
Imagine the following scenario: one day, you're introduced to two new coworkers, let's call them Sarah and Dave. Sarah walks into the office with enthusiasm and energy. Within five minutes, she knows everyone's name and is volunteering to present the team update. Dave, on the other hand, hangs back, listens closely, takes some time to think, and later gets back to you with a well-structured text message.
Just based on these brief interactions, you can probably make some reasonable predictions about how Sarah and Dave may react to certain situations - perhaps that Sarah may be energized by a busy open office, while Dave will do better with deep tasks that he can work on independently. And you would have a decent chance, albeit no guarantees, of being right.
This is the core promise of personality psychology: that people differ in stable, measurable ways, and that those differences are informative. Many of us form such impressions quickly and intuitively, but psychologists have long tried to systematize this intuitive concept. If we can describe a person using a handful of traits, we get a compressed representation of who they are. Imperfect, but useful for understanding them and anticipating how they might react to stressful situations, how they relate to others, and what they'd enjoy.
But what is that "handful of traits"? And how many of them are there?
One of the most influential answers to this question is the Big Five model, which says there are five broad dimensions of personality:
Openness (to Experience),
Conscientiousness,
Extraversion,
Agreeableness, and
Neuroticism (also known as ‘emotional instability’).
A rival model, HEXACO, looks very similar at first glance, but argues that there should be a sixth dimension, in addition to these five: Honesty-Humility.
We recently ran a study to investigate these questions. In this article, we'll explain how these models ended up with their dimensions, how to evaluate which one is more "correct" (and what that would even mean), and dig into the surprisingly clear results that our data suggest.
The Origins of Big Five
Before we get into the debate between five and six dimensions, it's worth taking a closer look at where the Big Five came from in the first place. As we'll see, the five dimensions this model relies on were not an arbitrary design decision, but empirically derived.
The starting point was the lexical hypothesis, which is the claim that if a personality difference matters, languages will eventually invent words for it. If making distinctions about people, such as being "reliable," "rude," "bold," or "petty," is useful, we'll end up encoding them in vocabulary. In that sense, dictionaries become large datasets of trait concepts.
But how do you get from thousands of adjectives to an informative personality model? Researchers approached this by asking lots of people to rate themselves (or others) in terms of which adjectives applied to them, and then used a statistical method known as factor analysis, searching for the most informative "hidden factors".
For instance, the data may show that if someone is rated as "talkative," they're also more likely to be rated as "energetic" and "friendly," and less likely to be rated as "reserved." In other words, the adjectives naturally correlate and group together. Factor analysis tries to find the smallest set of underlying "dials" that explains as much of the correlation structure as possible.
What factor analysis produces, however, is not a set of concrete labels like "Extraversion" or "Agreeableness", but rather a mathematical structure. Specifically, it produces a set of factors along with numbers (called "factor loadings") that represent how strongly each adjective is associated with each factor. The factors themselves don't have names at this stage; researchers have to examine the structure (e.g., noticing that "talkative", "energetic", and "friendly" all load most on the same factor) and decide what each factor might represent. That's how we end up with labels like "Extraversion".
The basic process behind personality models like Big Five: Survey people on personality questions, run factor analysis, and then make sense of the results by finding meaningful labels for the "hidden factors" factor analysis identified.
However, this does not yet explain why we should end up with five factors in particular. While factor analysis helps find the n most informative factors, it does not tell you how many factors to choose, and it will happily tell you any number of most informative factors.
The reason researchers converged on five factors is that in study after study, a solution with about five big groupings kept showing up as particularly stable across many different datasets. These five factors captured a substantial amount of the variation in how people responded to personality questions, and this high degree of "compression" combined with the stability across many studies and datasets is what made the Big Five model as influential as it is today.
Enter HEXACO
The Big Five, as described above, came out of the idea that important personality differences leave footprints in language, but its origins are mostly in the English language. So a natural follow-up question is: what happens if you stop relying mostly on English?
Researchers eventually did exactly that. They repeated the same process - trait words, lots of ratings, factor analysis - across a wider range of languages. And what some researchers found is that a particular solution with six factors often fits the data at least as cleanly as a five-factor solution. Later research suggested that this six-factor solution may actually hold up similarly well in English datasets after all, indicating that prior research may have too quickly settled on a five-factor model.
In the early 2000s, Kibeom Lee and Michael C. Ashton turned these six factors into a full-fledged model: HEXACO. The five dimensions from Big Five are, for the most part, still present, with a new dimension emerging, which they labeled Honesty-Humility. High Honesty-Humility involves sincerity, fairness, and a low appetite for exploiting others; low Honesty-Humility involves manipulativeness, entitlement, arrogance, and a willingness to bend rules when it pays. Combined with the other factors, this gives us:
Honesty-Humility
Emotionality
eXtraversion
Agreeableness
Conscientiousness
Openness
At first glance, HEXACO can look like "Big Five plus one", and in some ways, it basically is. If you line the two up, you get a very familiar mapping: Openness, Conscientiousness, and Extraversion land in almost the same territory in both models. Agreeableness is also there in both, but it is one of the places where the borders don't quite match. And Big Five's Neuroticism maps only partially onto HEXACO's Emotionality.
A rough representation of the relations between Big Five and HEXACO traits, based on our understanding of the literature and incorporating feedback from an expert academic. While there are many parallels, some things get moved around, and Agreeableness in particular has a different flavor in the two models. Note that this is a simplification, and in reality, the relation between the models is even less clean, and the details even depend on which exact "version" of the two models one uses.
Seeing these models side by side, several questions arise. Does adding a sixth dimension to the model, and thereby reorganizing some of the surrounding trait space, actually give us a better high-level map of personality? And if so, is it sufficiently better to justify the added complexity of an extra factor? And finally, is there any "objective" way to answer these questions at all, or is the choice between the five-factor and six-factor models ultimately just a matter of preference?
What We Tested
To examine these questions, we set up a direct comparison between the two models. Note that we explained all this in greater detail in our full research report.
We built a questionnaire that allowed us to score our 343 study participants on both Big Five and HEXACO, while trying to make sure the comparison was as fair as possible. The Big Five side drew on prior studies we ran and used 102 survey questions, while for HEXACO, we constructed a comparable six-factor measure based on another 102 questions, some of which overlapped with the Big Five questions, yielding a total of 149 unique questions in our study. (The reason for the overlap is that, as shown in the chart above, there is significant overlap in what the Big Five and HEXACO measure).
The resulting questionnaire allowed us to score each participant on Big Five and HEXACO traits. We also asked all participants 66 questions about their lives ("life outcomes", such as how many deep friendships a person has, whether they're religious, and whether they get promoted at work), which were not used for personality tests themselves, but which we could then use as a measure of predictive performance of both models.
We then investigated three questions with our new data:
What's the general relation between Big Five and HEXACO traits?
How good are both models at "compressing" the data from our 149 personality questions?
How good are they at predicting the 66 life outcomes?
What We Found
So, what did we find? Let's go through the three questions one by one.
1) What's the relation between Big Five and HEXACO traits?
Given the many parallels between the factors of Big Five and HEXACO, it makes sense to first ask whether the similarly named factors actually measure similar things. For this, we compared the Big Five scores our study participants received to their HEXACO scores. Here's what we found:
Overall, the answer is: yes, to a large extent, they do. Openness, Conscientiousness, and Extraversion line up strongly across the two frameworks, and Big Five Neuroticism has a strong relation to HEXACO Emotionality.
But looking further, we see that Agreeableness seems to have been partly split up and redistributed. While Big Five Agreeableness still correlates most strongly with HEXACO Agreeableness, it also shows substantial overlap with HEXACO Emotionality and Honesty-Humility. This indicates that our sixth factor is not an entirely "new dimension" of personality, but at this level of granularity, we start splitting and "reshuffling" existing factors - a finding that confirms other research.
But that's just looking at the "shape" of the two models; it doesn't really tell us how useful they are. So let's see how well they performed when we put them to the test.
2) How good are both models at "compressing" the data from our 149 personality questions?
Our second question was how well different models "compress" the underlying personality data. Since our participants answered 149 personality-related questions, one way to evaluate a model is to ask how much of the variation in those responses can be captured by the given number of underlying factors.
For this, we didn't just compare the two models directly, but used factor analysis to check how much information could at most be captured by any number of factors, given our dataset. This would then give us some indication of the "true number of dimensions" in our data (or, at least, how much compression we can get using a different number of factors).
Interestingly, factor analysis gave us some evidence in favor of an answer to the dimensionality question. Namely, that there was a fairly striking drop from the 5th factor to the 6th. The 5th factor still explained over 5% of additional variance, whereas the 6th explained only about half as much. And after that, the 6th, 7th, and 8th factors were all relatively similar to each other in usefulness:
This suggests that the first five factors are in some way "different" from the ones that come after. It's not that a 6th factor is useless; it clearly adds something. But if you do decide to go beyond five, it becomes much harder to explain why you would stop at six in particular, rather than also include a seventh or eighth factor. By contrast, stopping at five seems to have a much clearer justification in our data. If you take the above chart at face value and use it as a basis to judge "how many dimensions" the human personality data from our study has, then 5 looks like a pretty compelling answer to get as close to the truth as a single number possibly could.
3) How good are the models at predicting the 66 life outcomes?
Of course, compressing personality survey responses is only one kind of test, and there's a sort of circularity to it: above, we just "predicted" the data that was used for the personality models. If that is the only thing personality data could do, then it would not be very useful. So, a more important question may be whether these models help predict things beyond the personality items themselves. So we also used Big Five and HEXACO scores to predict 66 life outcomes that were measured separately from the personality questions. Here's what we found:
Overall, both models had a modest amount of predictive power. Across the 66 outcomes, the average out-of-sample R score (a measure of predictive accuracy, ranging from 0 = no predictive power at all, to 1 = perfect predictions) was 0.191 for Big Five and 0.198 for HEXACO. So HEXACO did perform slightly better overall. But the improvement was very small, and in our paired analysis across outcomes, the difference in prediction accuracy between the Big Five and HEXACO was not statistically distinguishable from zero.
If we look at how well HEXACO predicts life outcomes compared to Big Five on a per-outcome basis, it becomes even more apparent that there doesn't seem to be a clear winner here:
Each of these bars represents one of the 66 life outcomes, and we ordered them by how much better HEXACO did compared to Big Five. And while the aggregated height of the orange bars (where HEXACO did better) exceeds that of the blue bars (where Big Five did better), we also see that the number of blue bars is slightly larger, meaning that more life outcomes, out of our 66, were predicted better by Big Five than by HEXACO.
That said, HEXACO did seem to help on certain kinds of outcomes in particular. E.g., it performed better on outcomes relating to having used physical force, having been arrested in the past 10 years, substance addiction, and having cheated on a partner - outcomes where it is plausible that Honesty-Humility captures something useful that would get lost or watered down in the Big Five traits.
Importantly, the above findings all hinge on what life outcomes are being measured and predicted. We did our best to come up with a varied set of life outcomes without biasing these towards any model (you can find the full list of life outcomes and how both models performed on them in the appendix of our extended research report). But clearly, other selections of life outcomes could lead to different results and conclusions, which is an important limitation of many such studies. For specific applications or uses of personality tests, one may want to focus more on some outcomes and ignore less relevant ones.
Taken together, our findings suggest that HEXACO does add some value, but not in a way that would make it a generally better personality model. It seems useful in some narrower contexts, especially where the Honesty-Humility dimension is particularly relevant. But as a general-purpose personality model, it did not produce a clearly detectable overall improvement over Big Five in our data in terms of predicting (at least our set of) life outcomes, despite being more complex.
So, how many dimensions does personality have?
On the one hand, HEXACO does seem to capture something more about people. Its sixth factor, Honesty-Humility, is not meaningless, and in some contexts it appears useful and predictive. On the other hand, our data do not suggest that moving from five factors to six yields a large or general improvement in how well we can describe or predict people.
What stood out to us most was not just that HEXACO failed to clearly outperform Big Five, but that five factors seemed to occupy a special role in the structure of the data. If personality were simply a smooth continuum of ever finer distinctions, then a sixth factor should be just slightly less useful than the fifth one. But that is not what we found. Instead, there seems to be a reasonably large gap between the first five factors and the ones that come after them.
That does not mean that personality is "really" five-dimensional in any final or metaphysical sense. Human beings are obviously more complicated than any five numbers could capture. As a general rule, the more numbers a model uses, the more accurate it can get on average. But if the goal is to build a compact, useful map of personality, then our results suggest that five factors may be a particularly good stopping point: complex enough to be informative, but still simple enough to remain practical.
If you'd like to dig into the details behind that conclusion, including the full methodology, additional analyses, and deeper insights into the data, we've written all of that up in the longer report as well.
Supplementary Materials
To download the anonymized data from our study, showing each of the outcome values for each participant and each of the sub-scale values for each participant (i.e., all 5 of their big 5 scorers, all 6 of the big six scores), click here.
To download a data dictionary containing the names of all variables in the file above and a plain English description of each one, click here.
