HEXACO vs Big Five: what the sixth factor adds

Two candidates for a job that controls a departmental budget fill in the same five-factor questionnaire, and the results come back almost indistinguishable. Both moderately extraverted, both above average on conscientiousness, both calm, both easy to like.

A year later a difference shows up that was nowhere on paper. James hands the unspent money back to the budget even though he could have quietly cycled it through his own department. Daniel takes a private commission from a supplier he recommended himself and mentions it to nobody. The questionnaire missed that difference. Not because it was a weak questionnaire, but because it never measured the trait in the first place.

That missing piece is what the argument between the two most widely used scientific models of personality is really about. One describes five factors, the other six. The sixth is the reason a comparison is worth the trouble.

The same method, a different count

Both models rest on one idea and one piece of statistics. The idea is the lexical hypothesis: anything about people that matters enough for others to notice will sooner or later get a word of its own. So if you want to know how people differ, the dictionary is a reasonable place to start.

What follows is tedious, which is exactly why it is trustworthy. Every adjective in the language that describes a person gets pulled out, hundreds of people rate those words on themselves or on people they know, and factor analysis works out which words travel together. Nobody decides in advance how many clusters should come out. As many come out as the data contain.

Run on English word lists, the procedure kept producing five factors (Goldberg, 1990; Costa and McCrae, 1992), and those five became the standard you now know by the acronym OCEAN. What each of the five scales covers is laid out in the guide to the Big Five personality model.

Around the turn of the millennium the Canadian psychologists Michael Ashton and Kibeom Lee did something that sounds unremarkable: they ran the same procedure on other languages. Korean, Italian, Hungarian, Polish, Dutch, French, German. Again and again the data gave them six clusters instead of five.

Here is the most surprising part of the story. The sixth factor did not appear because Koreans or Hungarians are a different kind of person from Americans. It appeared because every adjective list is only a sample of a language, and the English lists the five-factor model grew out of happened to be thin on words from one particular family. Back then the model was not measuring the world. It was measuring the vocabulary it had been handed.

The sixth factor: honesty-humility

That family of words looks like this: sincere, honest, fair, modest, unassuming at one end, and sly, calculating, greedy, self-important at the other. In the six-factor model it turned into a scale of its own, honesty-humility, marked with the letter H.

It covers four things: whether someone plays straight even when a lie would go unnoticed, how much it bothers them to gain at another person's expense, how strongly money and luxury pull at them, and how far they see themselves as someone who is simply owed more than the rest.

The name of the scale invites a misreading. Humility here has nothing to do with shyness or with talking your own abilities down. Someone high on H can be a confident specialist who knows exactly how good they are and just does not turn that into a claim on special treatment. The reverse holds too: low H often arrives wrapped in considerable charm, because anyone who wants to win people over has to be pleasant first. In the data, dishonesty rarely sits next to rudeness.

The five-factor model does not lose this content completely. It dissolves it. Part of the fairness and the reluctance to exploit others lands under agreeableness, part of the rule-following under conscientiousness, and the rest scatters far enough that it can no longer be read off as a single number. That is why Daniel from the opening can come out of a five-factor profile looking perfectly ordinary. He is organised, agreeable, calm. The scale he would differ on is not there.

Low H is not a label for a fraudster. It describes a tendency, not a deed, and in ordinary life it shows up in far more mundane ways: relationships get read as an exchange, status and visible rewards carry more weight, and a rule looks like an obstacle you can walk around when it pays and nobody is watching. The scale gets a closer look in the guide to the H factor.

Emotionality is not neuroticism

The second difference is subtler, and it catches out even people who know both models. The letter E in the six-factor model does not mean what N means in the five-factor one, even though both scales revolve around emotion.

Neuroticism is a proneness to negative emotion in general: anxiety, sadness, irritability, anger, swings of mood. HEXACO untied that bundle. Emotionality holds fear, worry, the need to lean on people who are close, and sensitivity to their distress. Anger is not part of it.

It also contains something you will not find inside neuroticism: attachment to other people. High emotionality describes someone who is afraid for the people they care about and needs someone to lean on. That is not a synonym for emotional instability, even though the result often gets read that way.

Take two colleagues. Emma starts dreading a presentation two days out, does not sleep the night before, and needs to hear that she will be fine before she walks into the room. Rachel hosts the same event without blinking, but when a silly objection comes up in the discussion she cuts the questioner down so hard that nobody comes to his defence. A five-factor questionnaire returns high neuroticism for both. On six scales, Emma is high on emotionality, while Rachel is low on agreeableness.

Try to picture someone you have described as oversensitive. Was it a person who startles easily and takes things to heart, or the one who blows up the moment something goes against them? On five scales the two end up in the same place. On six they end up nowhere near each other.

Where the anger went

The answer to that question is also the third difference. In HEXACO anger moved into agreeableness, or more precisely into its lower end.

Agreeableness here means forbearance: the capacity to forgive, to let something go, to accept a compromise, to not hit back. Score low on this scale and you hold a grudge longer, criticise hard, concede reluctantly and lose your temper fast. Where the five-factor model records a short fuse as emotional instability, the six-factor one reads it as an inability to put up with other people.

The consequence is practical. Two people with equally high neuroticism can be completely different colleagues. One withdraws from a conflict and then replays it at three in the morning; the other starts it and cannot work out the next day why everyone is looking at him strangely. Six scales pull those two apart. Five leave them together.

A map between the two models

The table shows where each five-factor dimension moves to in the six-factor model. These are not equals signs, more a rough pointer to the nearest counterpart.

The table makes it clear that the difference between five and six is not just one extra number. Three factors stay essentially as they were, one is added and two get rearranged. Take both tests and compare the results, and openness, conscientiousness and extraversion should come out very similar, while the numbers for emotion and agreeableness may well diverge. That is not measurement error, only a different set of borders between the scales.

Worth knowing too: neither model stops at its six or five numbers. Every factor splits further into narrower traits, and only down there does it become clear why two people with the same score live differently. Two equally high conscientiousness scores can rest on tidiness and precision in one case, on hard work and persistence in the other. Decide anything on the broad factors alone and you are reading headlines instead of the article.

What the sixth factor adds

The usefulness of a new factor comes down to one question: does it tell us anything the five scales fail to catch? For honesty-humility the answer is yes.

The work of Ashton and Lee and the research that followed link low H with a readiness to bend a rule when it pays, with a willingness to manipulate people, and with behaviour at work that harms the employer, from petty misuse of resources to serious fraud. The scale also overlaps substantially with the traits known as the dark triad, most of all with Machiavellianism.

Keep it grounded, though. These are statistical patterns across thousands of people, not a prediction about any one person; every comparison of this kind is approximate. A personality scale describes a tendency, not a deed, and differences inside groups tend to be larger than differences between them. H is no basis for deciding whether to hire a candidate or for confirming a suspicion about a colleague. Its usefulness lies elsewhere: it helps explain why working with someone who seems pleasant and reliable still leaves you with an odd feeling after every meeting.

When to reach for which

The Big Five makes sense as a general map. Decades have piled up incomparably more studies, norms and derived instruments around it, so if you want a result you can hold up against the literature or against another questionnaire, five factors are the safe choice. It also fits when your main interest is how you handle stress, because neuroticism answers that question more directly: it keeps anxiety and anger on one scale. Our Big Five test works with fifty items from the public IPIP item pool.

The six-factor model earns its place the moment trust is at stake. Typically:

• Roles that decide about money, access or authority without oversight.
• When rules keep getting bent in a team and yet nobody looks like the culprit.
• Self-knowledge, at the point where people around you say you act harder than you admit to yourself.

The last reason is the most common one, and it is a simple one: you walked away from a five-factor result feeling that something essential was missing and could not put a name to it. The sixth scale is often the missing word.

Our HEXACO test has sixty questions and takes about eight minutes. Alongside the six scores it also shows an approximate comparison with a five-factor profile, so you can see from the result where the two models meet and where they part ways. What the individual scales mean in practice is summed up in the guide to what HEXACO is.

Neither model wins

Five factors or six is a legitimate scientific debate, not a clash of brands. The five-factor camp argues from parsimony and from the mountain of data gathered over decades. The six-factor camp argues from replication across languages and from the fact that the sixth scale explains behaviour five-factor profiles walk past. Both sides have data, both publish, and neither claims the other is measuring nonsense.

The practical gap is also smaller than the academic argument suggests. Both models work with scales rather than types, so they put you in no box and no number in them is better than another. Both rest on self-report, which means they measure how you see yourself more than how you behave, and both can be skewed when someone fills in a questionnaire for other people's eyes. And both stop being useful the moment a description is turned into a verdict.

Try a small test with no questionnaire. Write down three people who let you down in the past year, and one sentence each about what they actually did. "He never finished what he promised" points to conscientiousness. "She exploded over nothing" points to agreeableness. "He knew it would hurt me and did it anyway because it paid off" fits neither. That sentence is what the sixth scale describes, and you will not find it in a five-factor profile.