Psychology · 6 min read
What Longitudinal Research Shows About Personality Change in Adulthood
What repeated measurements establish about adult trait stability, average change, individual variation, and unresolved causes.

What Longitudinal Research Shows About Personality Change in Adulthood
Personality traits are meant to describe relatively enduring patterns, but enduring does not mean immovable. Research that repeatedly measures the same people shows two results that can sound contradictory: adults often retain their position relative to their peers, and their average trait levels can still change over time. A person may remain more conscientious than most people in the same cohort while becoming more conscientious than they were ten years earlier.
That distinction corrects two common overstatements. Adult personality is neither fixed at a particular birthday nor freely remade by any experience. The evidence supports patterned average change, substantial stability, and meaningful differences between individuals. It is much less decisive about why a given person changes.
Stability and change are different measurements
Personality researchers usually organize broad tendencies with the Big Five: extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience. Each label covers narrower facets. Extraversion, for example, can separate social vitality from social dominance; two people with the same broad score may differ in sociability and assertiveness.
Several measurements answer different questions. Rank-order stability asks whether people preserve their relative ordering within a group. Mean-level change asks whether a group's average rises or falls. Individual change asks whether particular people follow or depart from the average trajectory. Structural continuity asks whether the traits retain the same pattern of relationships. A finding about one kind of continuity cannot automatically answer the others.
This is why the phrase “personality changes” needs a denominator and a timescale. A small average shift across thousands of people may coexist with large changes in some participants and little movement in others. Repeating a questionnaire also introduces measurement error, mood effects, and changes in how people interpret questions. Strong longitudinal designs use multiple observations and statistical models that separate persistent change from temporary fluctuation, but no model removes every ambiguity.
What coordinated longitudinal samples find
In a 2006 Psychological Bulletin meta-analysis, Brent Roberts, Kate Walton, and Wolfgang Viechtbauer combined 92 longitudinal samples spanning ages 10 to 101. They found statistically detectable mean-level change in all major trait domains during some part of the life course. Social dominance, conscientiousness, and emotional stability tended to increase most in young adulthood. Social vitality changed less and tended to decline later; openness rose earlier and declined in old age in the available samples. The results challenged the claim that mean trait levels stop moving after age 30.
A later coordinated analysis led by Eileen Graham used the same statistical approach across 16 longitudinal samples, mostly of middle-aged and older adults. Coordinated analysis matters because a single pooled dataset can conceal differences in measures, sampling, and follow-up. The 16 studies did not yield one universal curve. On average, neuroticism declined through much of adulthood before increasing in older age, while conscientiousness and agreeableness showed modest gains and then later declines. Yet the size and even direction of slopes varied among samples.
That heterogeneity is evidence, not merely noise. Most long-running cohorts were assembled in particular countries, years, and social conditions. They differ in education, health, survival, and attrition. People who remain in a study into advanced age are not a random cross-section of everyone originally recruited. An average late-life decline can reflect aging processes, illness, selective dropout, cohort history, or a combination.
Rank-order findings add the other half of the picture. Roberts and Wendy DelVecchio's 2000 meta-analysis found that relative consistency generally rises from childhood into adulthood and reaches high levels in middle age. High rank-order stability means that knowing an adult's trait score is informative about their later position compared with peers. It does not mean that the person's absolute score stays unchanged.
Average trajectories do not describe each person
Group averages are poor biographies. In repeated-measure studies, individuals vary around the average slope: some become more conscientious, some less, and some show no reliable movement. Researchers test whether this spread exceeds what imperfect questionnaires alone would produce. Growth models across several waves generally find genuine variation in both direction and rate.
Life events are associated with some of that variation, but the associations are not simple instructions for changing a trait. Work investment, partnership transitions, unemployment, illness, and caregiving have each been linked to later personality measures in some cohorts. The “social investment” account proposes that commitment to adult roles rewards greater reliability, emotional regulation, and cooperation. It fits several average patterns, particularly in young adulthood.
Selection complicates the account. More conscientious people may enter and persist in structured work; emotionally stable people may form durable relationships. If a later trait score differs, the role may have influenced the person, the person's earlier traits may have influenced the role, or both may be responding to another condition. Studies that measure traits before and after a transition improve the chronology but still cannot randomly assign marriages, layoffs, or parenthood.
Intervention research tests a narrower causal question. A 2017 meta-analysis by Roberts and colleagues examined 207 intervention studies and reported trait changes over an average of about 24 weeks, especially in emotional stability and extraversion. Most interventions were clinical rather than programs designed specifically to alter personality. Participants seeking treatment, the absence of untreated comparisons in some studies, and overlap between symptom measures and trait measures limit what can be generalized. The review indicates that measured traits can move during intervention; it does not show that any short exercise can redesign personality on demand.
Why causes remain difficult to identify
Biology and environment are not competing complete explanations. Twin and family studies find genetic contributions to both stability and change, while environmental experiences contribute too. “Heritable” describes variation in a population under particular conditions; it does not assign a fixed percentage of one person's trait to genes. Genetic influence is compatible with developmental change because gene expression, environments, and the choices people make are interdependent across time.
Measurement remains a second constraint. The Big Five provide a widely replicated descriptive framework, but inventories rely heavily on self-report. Observer ratings can confirm some change and also disagree with self-reports. The meaning of an item such as “I make plans and follow through” may shift after retirement or illness even if the underlying disposition does not change in the same way. Different instruments also divide broad traits into facets differently.
Cultural coverage is improving but remains uneven. Much of the longest longitudinal evidence comes from North America, Europe, and Australia. Cross-cultural studies find some recurring age patterns, yet institutions shape the timing and meaning of work, family, and retirement. A trajectory found in one cohort should not be presented as a biological timetable for all adults.
The boundary of the evidence
Longitudinal research establishes that adult personality combines continuity with change. Relative position becomes fairly stable for many adults, while average levels and individual scores continue to move. The most consistent mean-level shifts occur in young adulthood, and later-life trajectories are more mixed. Individuals regularly depart from those averages.
What remains contested is the causal share assigned to maturation, chosen roles, involuntary events, health, culture, and measurement. Existing studies can identify chronology and association more securely than they can isolate a mechanism. The defensible conclusion is therefore bounded: adulthood does not close the possibility of personality change, but population trajectories do not predict how much a particular person will change or provide a recipe for producing it.
Sources
4 sources, every one with a link you can follow.
- Roberts, Walton, and Viechtbauer, “Patterns of Mean-Level Change in Personality Traits Across the Life Course,” Psychological Bulletin 132 (2006), doi:10.1037/0033-2909.132.1.1.
- Roberts and DelVecchio, “The Rank-Order Consistency of Personality Traits from Childhood to Old Age,” Psychological Bulletin 126 (2000), doi:10.1037/0033-2909.126.1.3.
- Graham et al., “Trajectories of Big Five Personality Traits,” European Journal of Personality 34 (2020), doi:10.1002/per.2259.
- Roberts et al., “A Systematic Review of Personality Trait Change Through Intervention,” Psychological Bulletin 143 (2017), doi:10.1037/bul0000088.
