Who created HEXACO and how the sixth factor was found

In 1884 the British polymath Francis Galton sat down with a dictionary and started counting words. He wanted to know one thing: how many expressions English has for describing human character. He arrived at roughly a thousand, and out of that tally came an idea that most serious personality tests still rest on today.

The idea is unassuming. Whatever people notice about each other often enough and long enough eventually gets a word. A quality nobody cares about stays unnamed. A dictionary, then, is not a random heap of vocabulary but a record of what has been worth talking about for centuries.

This is known as the lexical hypothesis. It led to the Big Five, and a few decades later to a model with six factors instead of five.

Eighteen thousand words, copied out by hand

The first serious attempt to turn Galton's hunch into data came in 1936. The American psychologists Gordon Allport and Henry Odbert went through the unabridged English dictionary of the day and wrote out every term that could be used to tell one person apart from another. They collected around eighteen thousand entries.

Try to picture that job without a computer. No database, no search function. Two men, four hundred thousand dictionary entries and index cards on which the words were sorted by hand. From those eighteen thousand, Allport and Odbert set aside roughly four and a half thousand that describe lasting traits of character, and shelved the rest as moods, judgments and passing states.

The pile then landed with the British psychologist Raymond Cattell. He merged the words into clusters of related meaning, ran factor analysis on them and in the 1940s pulled out sixteen personality factors. His number did not survive in later data, but his method did. Take the words people use to describe people, find out which ones travel together, and let the structure emerge from statistics rather than from theory.

The five that kept coming back

In 1961 Ernest Tupes and Raymond Christal went back to data gathered in the US Air Force and found five recurring dimensions in the ratings of pilots and cadets. Warren Norman repeated the result two years later on a different sample. It still took two more decades before Lewis Goldberg gave them the name Big Five in 1981, and before they became the standard that academic personality psychology leans on to this day.

The agreement looked like a closed case. Different researchers, different samples, different decades, and always five. Yet all those analyses shared one detail that went unremarked for a long time: they ran on English adjectives.

The sixth factor turned up in languages nobody had bothered with

In the 1990s the lexical method began to be applied to other languages. Not in hope of a discovery, more out of a duty to verify. If the model describes human character in general, the same five should fall out of Hungarian, Italian or Korean words too.

They did not. In Hungarian the structure behaved differently from what the researchers expected. In the Italian data a separate cluster appeared that fitted none of the five pigeonholes. A Korean study from the late 1990s, co-authored by a young psychologist named Kibeom Lee, came out much the same way. Then the same thing happened in Dutch, Polish, German and French.

The cluster held similar words every time. On one side sincerity, modesty, fairness and a lack of self-interest; on the other boastfulness, greed, pretence and a willingness to exploit others for personal gain. In the English analyses those terms dissolved into agreeableness and conscientiousness without forming a dimension of their own. As soon as the method moved elsewhere, they stood apart.

Part of the explanation goes back to Allport and Odbert. When they sorted their eighteen thousand entries, they set aside terms that judge a person rather than describe one. But a great many words about honesty carry judgment by their very nature. Calling someone calculating is not a neutral observation, and expressions of exactly that kind were dropping out before they ever reached an analysis. Other research traditions were looser in their selection, and that stretch of the vocabulary stayed in their data.

It is a strange turn of events. A model that now ranks among the most cited in personality psychology exists because the research stopped being done only in English. Had the lexical studies stayed inside one language, we would probably still be tripping over the sixth factor as an awkward residue in the numbers.

Two Canadians and six letters

Around 2000, two Canadian psychologists pulled the scattered findings together. Michael Ashton of Brock University in Ontario and Kibeom Lee, now at the University of Calgary, first showed that the sixth factor was not a fluke of one country, and then assembled a coherent model out of it. In 2004 the Journal of Personality and Social Psychology published their joint paper with further researchers from Europe and Korea, summarising lexical data from seven languages at once. A six-factor solution came out of them again and again.

The model took the name HEXACO from the initials of the six dimensions and from the Greek word for six.

This is the point to correct a widespread mistake. HEXACO is not the Big Five with one factor bolted on. Emotionality and agreeableness are rotated relative to the original five: the tendency towards anger and irritation, which sits under neuroticism in the Big Five, attaches in the six-factor arrangement to the opposite pole of agreeableness. So four letters overlap closely with the Big Five, two only roughly, and one is new. The piece on what HEXACO is takes that apart in detail.

What changed once integrity became measurable

Before the H factor, character integrity was an elusive quantity in research. Everybody could describe it, but there was no room for it among five factors, so it kept leaking out of the results. The sixth dimension gave it coordinates, and with them opened up a set of questions that had been hard to study.

Low H has repeatedly been linked with things that ordinary personality tests tended to miss:

• a willingness to cheat when detection seems impossible
• counterproductive behaviour at work, from petty theft to abuse of authority
• the dark triad, meaning Machiavellianism, narcissism and psychopathy, maps in full onto the opposite pole of this factor
• property and status matter more than the relationships spent to get them

For work psychology this was interesting news. Conscientiousness long counted as the best personality predictor of job performance, yet it mainly says whether someone finishes what they start, not how they behave when an opportunity presents itself. An employee high in conscientiousness and low in honesty tends to be meticulous about covering their tracks as well.

Take two buyers with identical scores on five factors. Both hard-working, both sociable, both on time. One receives an offer from a supplier that need never be mentioned anywhere, and turns it down before thinking it through to the end. The other weighs it for a while, looking for a way it could be defended. Five dimensions miss that difference because it does not fit inside them. The sixth aims straight at it.

Can you think of a colleague who ticked every box on paper and was still a problem? Usually it was not about pace or ability. What exactly the H factor measures, and why it stings to read your own score on it, is the subject of a separate article on the honesty-humility dimension.

Where the model stands today

Ashton and Lee also published their own questionnaire for the model, and over two decades it has been translated into dozens of languages. A steady stream of studies now appears around the six factors, the model shows up in personality psychology textbooks, and personnel research treats the H factor as a variable in its own right.

Not everyone considers the matter settled. The Dutch psychologist Boele De Raad and a group of co-authors compared lexical data from further languages in 2010 and concluded that across cultures it is three factors rather than five or six that replicate reliably. The argument over the right number of dimensions continues, and six is not the last word.

In public awareness HEXACO still sits in the shade. The Big Five has a head start of several decades, and the sixteen-type typology has an appeal an academic model struggles to match: it condenses into a four-letter badge for a résumé. Six numbers on a scale travel worse than one acronym. If you are curious what your own profile looks like across all six dimensions, you can work through our HEXACO test.

One more thing from this whole story is worth keeping. Nobody invented the sixth factor. It had been lying in Hungarian, Korean and Italian dictionaries the entire time, waiting for someone to look somewhere other than English. How many further dimensions of human character already have names in a language the research has not reached yet, nobody can say.