Chaos and Order

💡 왼쪽 원문을 읽으면서 오른쪽에 따라 써보세요. Tab 키로 힌트를 받을 수 있습니다.

Introduction — When Both Sentences Are True

People do not change. The longer someone has lived, the more readily they nod at that sentence. People can change. Every shelf of self-help stands on that one. The common treatment is to pick one of the two. This post does not pick, because the two sentences measure different things and are therefore true at the same time, and that tension is itself the most important fact to know about personality.

A great deal of actual life rides on this question. Whether you fold an attempt with the sentence I am just this kind of person or try once more rides on it; whether you pull a relationship along another few years on the expectation that the other person will change, or wind it up here, rides on it. The cynicism that nothing changes makes every effort meaningless, and the optimism that anything can change manufactures self-blame at every failure. Both are expensive wrong answers, so holding the resolution of the picture the actual data draws matters practically.

This is the post in the series I wrote most carefully. Personality research is a large and old field, and exaggeration here injures people more easily than on any other subject. So each section records the kind of evidence and its limits alongside the finding.

1. The Evidence on the Not-Changing Side — Rank Order Holds

The reference point when personality psychology talks about stability is the 2000 meta-analysis in which Roberts and DelVecchio synthesised 152 longitudinal studies and 3,217 test-retest correlations. What it measured is rank-order stability: the question of whether the person who is more extroverted than me within my peer group is still more extroverted than me ten years later.

The result was clear. The rank-order stability correlation rose from 0.31 in childhood to 0.54 in the college years and 0.64 around age thirty, peaking near 0.74 between the fifties and the seventies. The older you get, the more firmly the relative position of personality sets. The common sense that people do not change fits these figures well. The probability that a very introverted person is the centre of the party five years later is, honestly stated, low.

Think of a class reunion and the texture of these numbers comes through. Friends met after ten years are startlingly the same. The one who talked a lot still talks a lot; the one who planned still plans. That sameness is what rank-order stability is talking about. Only, a correlation of 0.74 is not 1.0, and that has to be read with the same weight. It means rank is considerably maintained, not that everybody is in place, and in fact there are one or two people at that reunion who are quite different from before. And the figure starting at 0.31 in childhood means that the younger you are, the more rank itself moves. Stability is not a constant you are born with but a variable that hardens with age.

The evidence class of this section is high. It is a large meta-analysis, and it broadly agrees with later research. Considerable stability of personality is a result repeatedly confirmed in this field.

2. The Evidence on the Changing Side — The Average Moves

But rank-order stability is not the same statement as no change. A class can keep the same height order for ten years and still grow as a class. In personality, exactly that happens.

In the 2006 meta-analysis in which Roberts, Walton and Viechtbauer synthesised 92 samples, people on average became more conscientious, more emotionally stable and more socially confident across their twenties to their forties. This is the so-called maturity principle. Passing through the responsibilities of work and relationships, the whole group moves in the same direction. That is, personality is not something fixed but something that moves slowly, in a predictable direction. Only, that movement runs on the scale of decades, and is gradual enough to be hard for an individual to feel.

That the direction is broadly a good one is worth writing down separately. Contrary to the common belief that people get duller and worse with age, the average trajectory the data draws runs toward becoming a more settled and more trustworthy person. For someone who does not much like their twenty-something self, this fact is a fairly substantial consolation: even if we do nothing, time is broadly on our side. The question the rest of this post takes up is whether the speed of that time can be brought forward at all.

That is the picture in the natural state. Rank holds, and the average moves slowly. The remaining question is this one. Can it be changed without waiting, deliberately, because you want it?

3. Does It Change Because You Want It To — Studies of Volitional Change

What took this question on head first is Hudson and Fraley's 2015 experiment. In two experiments spanning 16 weeks, when participants who had set a goal of changing a particular personality trait were made to draw up concrete implementation plans, self-reported personality scores and related behaviours actually moved in the intended direction.

There is a more important follow-up. The 2019 study from Hudson's team proposed weekly behavioural challenges to participants over 15 weeks. If you want to become more extroverted, a challenge like greeting an acquaintance first this week. The core of the result is this. Merely accepting the challenges did not change personality, and only the amount of challenges actually completed predicted change. The wish to change is a starting point; change was a function of executed behaviour.

This result contradicts one current of self-help discourse head on: the narrative centred on declaration and resolve, that the moment of deciding to change is what matters. In the data, resolve is not worthless, but by itself it predicted nothing, and all the predictive power sat in execution. Not the practice of wanting what you want more intensely, but actually doing this week what a person who wants it would do. If you want change, the question should not be how badly you want it but what this week's challenge is.

There is a verification at larger scale too. In the 2021 study in which Stieger's team tested a three-month smartphone-app-based intervention on 1,523 people, participants changed significantly in the intended direction (a self-report effect size of around 0.5), and the notable point is that the people close to them, that is, observers, also recognised the change. The change was maintained until three months after the intervention ended. This is a partial answer to the objection that it is an illusion of self-report alone.

The evidence class of this section is middling. The designs are experimental, but this is a line of research about a decade old, a substantial part of it has come from a small number of research teams, and measurement is centred on self-report. Reading it as promising but still young evidence is accurate.

4. Therapy as a Route — Where the Largest Change Is Recorded

The strongest record of volitional change is in the psychotherapy literature. In the 2017 review in which Roberts and colleagues synthesised 207 studies that included personality measurement before and after clinical intervention, personality traits changed by an average effect size of 0.37 across interventions averaging about 24 weeks, and what moved most was emotional stability. The change was maintained at follow-up as well.

An honest caveat is needed. Many of these studies were not designed for personality change but measured personality alongside treatment for anxiety or depression, so it is hard to separate cleanly the scores improving as symptoms lift from temperament itself changing. Even so, a size of 0.37 in 24 weeks is startlingly fast compared with the decades-long natural change of section 1. Let me put one practical line here as well. If attempts to change your mood or your behaviour fail repeatedly and it is intruding on daily life, getting help from a professional is usually a faster road than changing methods on your own.

5. Synthesis — Personality Is a Slowly Changing Bundle of Habits

Lay the evidence on top of itself and this picture comes out. Personality is not a fixed essence. Nor is it clothing you change according to how you make up your mind. The metaphor that fits current research best is a slowly changing bundle of habits. It is a set of defaults for interpreting situations and reacting to them; being defaults they do not change easily, and being only defaults they can change.

The shape of the practice follows from the metaphor as well, because the habit principles of the previous post apply as they are.

Pick a behaviour, not a trait. Not becoming a more extroverted person, but asking one question first in a meeting each day.
Hang it on a situation. In the form of an implementation intention with a when and a where attached. What predicted change in the Hudson experiments was execution in this form.
Shrink it to a completable size. The lesson of the 2019 study was that completed challenges, not accepted challenges, are what get counted.
Measure by the quarter. The time unit of personality is the season, not the week. Compare self-report a few months later and, where possible, the observation of someone close to you.

The environment counts as a lever too. As seen in the habits installment, behaviour is bound to context, and if personality is a bundle of habits then personality too is bound to the contexts that call those habits up. It is why the same person becomes a slightly different person in front of old friends and at a new job. Old contexts keep calling old defaults, so making one new environment, one new gathering, one new role during a period when you are attempting change is not a detour but the direct route.

Expectations come out of the metaphor as well. The change the studies report is an effect size of 0.3 to 0.5, a period of a few months, and a direction of one or two traits the person chose. Not the breadth in which an introverted person becomes a natural on stage, but the breadth in which the days you speak to someone first at a gathering noticeably increase. Stalls and regressions mixed in along the way are part of the normal process too. People change. Only not at the speed and the breadth people expect.

What Has Not Been Verified

Independent replication of the volitional change research — the Hudson line of studies is well designed but sits substantially within the same research current, and carries the limit of depending on self-report. The observer reports in the Stieger study make up for this in part, but this field as a whole is still young.
The interpretation of the therapy research — separating symptom relief from change in temperament is an unresolved point of contention. Reading the figure of 0.37 straight off as the speed at which personality changes is too much.
Permanence — follow-up periods are mostly a matter of months. Whether deliberately made change is maintained years later has not been confirmed.
Whether it works for anyone at all — participants in intervention studies are people who volunteered because they wanted to change. Evidence about how to change someone with no wish to change is not in this post, and probably is not anywhere.
The common belief that personality changes completely every five years — it runs head on against the stability data above. A narrative of wholesale replacement is merely easier to sell than data on partial movement.

Closing — From the Language of Essence to the Language of Defaults

The sentence I am just this kind of person sounds like a diagnosis but is actually a prediction, and as predictions go it is fairly accurate. Rank really is well maintained. But there is something that sentence misses: that what the word just points at is not an essence but a long-repeated default, and that defaults can be rewritten one at a time, in the form of behaviour, over time measured in seasons.

The same resolution helps with other people too. The sentence that person will never change and the sentence I will change them both run against the data. An adult's personality does not move under a few months of nagging, and at the same time, where the person themselves wants to change and changes their behaviour, measurable change over a few months is on record. There is no evidence about how to change someone else, and being the environment beside a person who is trying to change is about all we can do. This too is one of the honest answers this field gives.

So if there is something you want to change, aim not at personality but at the next behaviour. Personality follows after those behaviours have piled up, at the speed the statistics allow.

Self-development fundamentals series — Previous: The Power of Small Habits · Next: Metacognition — The Gap Between Feeling You Know and Knowing

References

All the materials below were accessed on 2026-08-16.