Mean-level vs rank-order change: two ways personality changes

What do mean-level change and rank-order change mean?

Say conscientiousness is measured in the same group of people at 20 and again at 40. Two different questions can be asked of that data, with two different answers.

The first question is about the group: did the average go up, down, or stay put? That is mean-level change, sometimes called normative change. The second is about individuals: did the person who scored highest at 20 still score near the highest at 40? That is rank-order change, usually reported as a stability correlation, called rank-order consistency in the research literature (Roberts & DelVecchio, 2000).

The two are independent. A trait can shift on average while everyone keeps roughly the same place in line, or the average can barely move while individuals swap positions. Reading a change study well means asking which of the two questions it is answering.

How stable is a person's rank order as they age?

A review of 152 longitudinal studies, covering 3,217 test-retest correlations, gives the fact sheet most researchers cite. Rank-order consistency rose steadily with age: about .31 in childhood, .54 during the college years, .64 around age 30, then a plateau near .74 between ages 50 and 70, with the gap between testing held at roughly 6.7 years (Roberts & DelVecchio, 2000).

Two things follow. Consistency keeps rising well past 30, so the idea that personality locks into place by then is not what the correlations show. And even at its highest, .74 is not 1.0, so the order is not fixed. Some people still rise or fall relative to their peers, and that room is where individual change lives.

Which traits show mean-level change, and when does it happen?

A meta-analysis of 92 longitudinal samples found average levels of social dominance, a facet of extraversion, along with conscientiousness and emotional stability, rising across adulthood, fastest between 20 and 40. Agreeableness moved mainly later, chiefly in old age, and social vitality and openness rose in adolescence before declining in old age (Roberts, Walton & Viechtbauer, 2006).

A much larger, cross-sectional study of 132,515 adults aged 21 to 60 found conscientiousness and agreeableness rising across the entire span, with neuroticism declining with age among women but not men, evidence against the old 'set like plaster by 30' idea (Srivastava et al., 2003).

Both studies are about the average. A mean-level rise in conscientiousness means more people ended up somewhat more conscientious than less, not that every single person moved by the same amount.

Do major life events change the average, the order, or both?

A panel study of 14,718 German adults modelled both kinds of change at once across adulthood. Rank-order stability rose into midlife then declined for emotional stability, extraversion, openness and agreeableness, an inverted U, while conscientiousness kept rising across the range studied. Personality also predicted major life events and shifted afterward in response to them, with effects specific to the event rather than a general 'good things help, bad things hurt' pattern (Specht et al., 2011).

That single study illustrates the two-question idea well: one dataset produced a mean-level answer and a rank-order answer, and neither stood in for the other.

How can everyone change and the order stay the same?

Picture a group of friends taking the same conscientiousness scale at 20 and again at 30. Suppose every one of them becomes somewhat more organised and reliable over that decade, the pattern the mean-level research describes (Roberts, Walton & Viechtbauer, 2006). The friend who was most conscientious at 20 gains a bit, and so does the friend who was least conscientious. If the gains are roughly even, the leader at 20 still leads at 30, even though the group's average has moved up noticeably.

That is how both findings are true at once: the average rises, a fact about the group, while the order among these friends barely changes, a fact about their relative standing. Stability would fall only if the gains were uneven enough to reshuffle who was ahead, and in practice they are, which is why real stability correlations sit well under 1.0 (Roberts & DelVecchio, 2000).

A rise in the group average tells you almost nothing about whether your own position, relative to people you know or a wider sample, has moved. A result reporting only one of them answers only one of the two questions.

How does Wellington measure this?

Wellington is a science-based personality assessment from Therabot Labs LLC. It measures personality as continuous traits, built on the Big Five and HEXACO models, and writes the results back as a personal report rather than a type.

The free Snapshot asks 60 questions across 6 dimensions, Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness to Experience and Honesty-Humility, and can be retaken from your dashboard at any time. Every trait is reported as a percentile against the reference sample. The three bands (Quiet below the 30th percentile, Balanced to the 70th, High above) and the Portrait names exist to make the result easier to talk about; the percentile is the measurement, and it is always shown alongside. Norms are provisional while the norming sample is completed.

Comparing two sittings shows both kinds of change at once. A percentile is a position against the reference sample, so a percentile that barely moves is the personal version of the rank-order question, and a percentile that moves says your position has shifted. What a percentile cannot show on its own is mean-level change, since a whole sample drifting the same way leaves your position where it was. The Wellington Membership, which reaches 90 traits and covers all 255 with retests in year two, is built around that comparison, keeping each sitting on file so a later one can be read against it. Wellington is a wellness tool for reflection, not a medical or psychological diagnosis, and it does not replace care from a qualified professional. It is not designed for hiring, and you can export or delete your data at any time.

Questions people ask

Is rank-order change the same thing as mean-level change?

No. Mean-level change is the group average moving over time. Rank-order change is whether individuals keep their position relative to each other. An average can rise sharply while the order barely shifts, because most people move in the same direction by a similar amount (Roberts, Walton & Viechtbauer, 2006; Roberts & DelVecchio, 2000).

At what age is personality most stable in rank order?

Stability climbs through early adulthood and levels off later: about .31 in childhood, .54 in the college years, .64 around age 30, and a plateau near .74 between ages 50 and 70 (Roberts & DelVecchio, 2000). It levels off rather than reaching a ceiling, since .74 still leaves room for movement.

Does rank-order stability mean personality cannot change?

No. A correlation of .74, the highest typically reported, still means people change position relative to others; it is well short of a perfect correlation, and describes how much reshuffling happens on average, not whether any one person's standing is fixed (Roberts & DelVecchio, 2000).

Which traits change most on average through adulthood?

Conscientiousness and emotional stability show clear average rises, especially between 20 and 40, alongside social dominance, a facet of extraversion (Roberts, Walton & Viechtbauer, 2006). Conscientiousness and agreeableness also keep rising across a wide age range in a large sample (Srivastava et al., 2003), while agreeableness in the meta-analysis moved mainly later in life.

Can a single life event explain a change in my personality score?

Rarely on its own. A panel study of 14,718 German adults found personality predicted major life events and shifted after them, but effects were specific to each event rather than a general rule that good or bad events change traits the same way (Specht et al., 2011). Treat one event as a plausible contributor, not a full explanation.