Is the Big Five scientific? What the research shows

When you take a personality test, you want to know whether the result means anything. For the Big Five, the answer is largely yes. It is the model personality researchers work with most, precisely because it keeps holding up.

But being scientific also means being honest about limits. This page covers what the research shows, what the model cannot do, what the criticism is and why models built on types do less well.

Why is the Big Five considered scientific?

The Big Five counts as scientific because the five traits were found rather than invented, and because they keep turning up again. Researchers collected the words people use to describe each other and had large groups rate themselves and others on them. Different statistical methods kept producing the same five broad groups, both in self-descriptions and in descriptions by others ^[1].

A second reason is that the structure carries over into other languages. When questionnaires were translated into languages from five different language families, including German, Hebrew, Chinese and Japanese, the five traits broadly reappeared ^[2].

That is the heart of good science: a finding that stands when others try again, with different people and different methods. What the five traits are is explained in what is the Big Five.

How reliable is a Big Five score?

A Big Five score is reliable when the questions for a trait measure the same thing and when a new measurement puts you in roughly the same place. With a good test and enough questions, that is the case. Still, every measurement contains some noise: your mood that day, how you read a question, a chance choice between "agree" and "strongly agree".

The number of questions matters a great deal. Research on very short questionnaires, with one or two questions per trait, shows that they considerably underestimate how personality relates to behaviour. Even a slightly longer list helps clearly ^[3]. That is why a short test gives a first picture and a long test a more precise one.

A second check is whether other people recognise you in your result. Research comparing self-ratings with ratings by family, friends and colleagues shows that others can judge your personality fairly well. The better they know you, the better they do, especially for traits that are hard to see from the outside, such as Emotional stability ^[4].

For your own result, this means reading a score close to 50 as "close to the middle", not as an exact position. On a new measurement you might land just on the other side. Further from the middle, the direction is much clearer.

How stable is personality?

In adults, personality is fairly stable but not fixed. A large meta-analysis of long-running studies found that people's positions relative to one another become steadily more stable through childhood and adolescence, then level off in young adulthood ^[5]. After about age 25, the researchers found little evidence that this stability keeps increasing ^[5].

At the same time, the average does shift. With the years, people tend to become somewhat more mature, alongside that growing stability. Emotional stability in particular rose steadily across the whole of life, more than earlier research had suggested ^[5]. How that unfolds is covered in does personality change with age.

Why do people differ in the first place? A meta-analysis of twin, family and adoption studies estimated that about 40 percent of the differences in personality between people are linked to genetic make-up, and about 60 percent to environment and experience ^[6]. That is a figure about differences across groups, not a percentage of you. And it does not mean a trait is set in stone.

What does the Big Five explain, and what not?

The Big Five reliably describes how people usually differ, and across groups of people those differences are linked to outcomes in life. Whether those links are real has been tested thoroughly. A large replication project retested 78 previously reported links between traits and life outcomes, with large samples. About 87 percent held up in the expected direction, although the links were on average somewhat weaker than in the original studies ^[7].

That also shows what the model does not do. The links are usually modest: a trait makes an outcome somewhat more likely across a group of people, not certain. For any one person, circumstances, skills, choices and chance play at least as large a role. So a Big Five profile does not predict whether you will succeed at work or be happy in a relationship.

The model also describes mainly how you are, not why. It says little about your values, your goals, your life story or what drives you. It is a solid basis for getting to know yourself, not a complete description of a person.

What are the criticisms of the Big Five?

The main criticisms concern four things: self-description, culture, the number of traits and how broad the traits are.

Self-description. A test measures how you see yourself. People may describe themselves more kindly or more harshly than others see them. That is why researchers sometimes add ratings by other people, which research shows are a valuable complement ^[4].

Culture. The model was developed mainly in Western, highly educated, wealthy countries. In a study of interviewer-led surveys among almost 95,000 people in 23 low- and middle-income countries, the usual questions often failed to measure the intended traits. Among people from the same countries who filled in a questionnaire online, they worked much better ^[8]. So a questionnaire does not automatically work in every setting or with every way of asking.

Five or six? Some researchers find six broad traits rather than five. The best-known alternative is the HEXACO model, which adds honesty and humility as a separate trait. According to the researchers behind it, that sixth trait explains several things the Big Five captures less well ^[9].

Breadth. Five traits are broad strokes. Two people with the same Extraversion score can differ in the narrower sides within it, the facets and aspects.

Why do type models do less well?

Type models sort people into boxes, while research shows that personality varies gradually. A review of 102 studies that used statistical methods to look for genuine personality types found that newer, stricter studies almost never found them. The conclusion: real personality types are extremely rare or do not exist ^[10].

That has consequences. A type model has to draw a line somewhere. Someone just below the line and someone just above it get different labels, even though they answered almost the same. And someone just above the line gets the same label as someone who scores extremely high. A scale keeps that information. More on this in Big Five vs MBTI.