Age Differences in Personality Traits From 10 to 65: Big Five Domains and Facets in a Large Cross-Sectional Sample

Abstract

Hypotheses about mean-level age differences in the Big Five personality domains, as well as 10 more specific facet traits within those domains, were tested in a very large cross-sectional sample (N = 1,267,218) of children, adolescents, and adults (ages 10-65) assessed over the World Wide Web. The results supported several conclusions. First, late childhood and adolescence were key periods. Across these years, age trends for some traits (a) were especially pronounced, (b) were in a direction different from the corresponding adult trends, or (c) first indicated the presence of gender differences. Second, there were some negative trends in psychosocial maturity from late childhood into adolescence, whereas adult trends were overwhelmingly in the direction of greater maturity and adjustment. Third, the related but distinguishable facet traits within each broad Big Five domain often showed distinct age trends, highlighting the importance of facet-level research for understanding life span age differences in personality.

Full text

Age Differences in Personality Traits From 10 to 65:
Big Five Domains and Facets in a Large Cross-Sectional Sample
Christopher J. Soto
Colby College
Oliver P. John
University of California, Berkeley
Samuel D. Gosling
University of Texas at Austin
Jeff Potter
Cambridge, Massachusetts
Hypotheses about mean-level age differences in the Big Five personality domains, as well as 10 more
specific facet traits within those domains, were tested in a very large cross-sectional sample (N⫽
1,267,218) of children, adolescents, and adults (ages 10 – 65) assessed over the World Wide Web. The
results supported several conclusions. First, late childhood and adolescence were key periods. Across
these years, age trends for some traits (a) were especially pronounced, (b) were in a direction different
from the corresponding adult trends, or (c) first indicated the presence of gender differences. Second,
there were some negative trends in psychosocial maturity from late childhood into adolescence, whereas
adult trends were overwhelmingly in the direction of greater maturity and adjustment. Third, the related
but distinguishable facet traits within each broad Big Five domain often showed distinct age trends,
highlighting the importance of facet-level research for understanding life span age differences in
personality.
Keywords: Big Five, five-factor model, facet traits, age differences, age trends
How does the personality of a typical middle-aged adult differ
from that of a typical young adult? From that of a typical adoles-
cent? Such questions concern mean-level age differences in per-
sonality traits, which occur when the average standing on a trait
differs as a function of age. Much recent research has examined
age differences in the Big Five trait domains: Extraversion, Agree-
ableness, Conscientiousness, Neuroticism, and Openness to Expe-
rience. The available data indicate that, from emerging adulthood
through middle age, Conscientiousness and Agreeableness show
positive age trends, Neuroticism shows a negative trend, and
Extraversion and Openness to Experience show flat trends (Alle-
mand, Zimprich, & Hendricks, 2008; Denissen, Geenen, van
Aken, Gosling, & Potter, 2008; Donnellan & Lucas, 2008; McCrae
et al., 1999, 2000; Roberts, Walton, & Viechtbauer, 2006; Srivas-
tava, John, Gosling, & Potter, 2003; Terracciano, McCrae, Brant,
& Costa, 2005).
These studies provide an informative sketch of mean-level age
differences in personality traits, but much further work is needed
to complete the picture. The present research was therefore con-
ducted to address two remaining issues, using data from a very
large cross-sectional sample. First, how do personality traits differ
by age across childhood and adolescence, and how do these
differences fit with adult trends—are trends at younger and older
ages similar, or do some traits show quite different trends in
childhood and adolescence versus adulthood? Second, do the Big
Five domains themselves capture all of the important information
about age differences in personality traits, or do the more specific
facet traits within these broad domains sometimes show substan-
tially different trends?
Note that these questions about mean-level age differences are
distinct from questions about rank-order changes. Rank-order
changes occur when the relative positioning of individuals on a
trait changes over time—for example, if the most conscientious
individuals in a group of young adults were no longer the most
conscientious individuals when the group reached middle age. The
present research focuses exclusively on mean-level age differ-
ences, rather than rank-order changes, consistent with its cross-
sectional design.
Age Differences in Personality Across Late
Childhood and Adolescence
Late childhood and adolescence are periods of rapid biological,
social, and psychological change. For example, the biological
changes that define puberty—accelerated growth, changes in body
shape, and the development of secondary sex characteristics—
typically begin around age 11 for girls and age 13 for boys
(Marshall & Tanner, 1986). Socially, there are normative changes
in youths’ relationships with adults and peers (Buhrmester, 1996;
This article was published Online First December 20, 2010.
Christopher J. Soto, Department of Psychology, Colby College; Oliver
P. John, Department of Psychology and Institute of Personality and Social
Research, University of California, Berkeley; Samuel D. Gosling, Depart-
ment of Psychology, University of Texas at Austin; Jeff Potter, Cambridge,
Massachusetts.
This research was supported by a faculty research grant from Colby
College awarded to Christopher J. Soto. We thank Christiana Lumbert and
Adam K. Thompson for their helpful comments on drafts of this article.
Correspondence concerning this article should be addressed to Christopher
J. Soto, Department of Psychology, Colby College, 5550 Mayflower Hill,
Waterville, ME 04901. E-mail: cjsoto@colby.edu
Journal of Personality and Social Psychology © 2010 American Psychological Association
2011, Vol. 100, No. 2, 330–348 0022-3514/10/$12.00 DOI: 10.1037/a0021717
330

Hunter & Youniss, 1982; Rice & Mulkeen, 1995) and in their
attitudes toward social values and norms (Colby, Kohlberg, Gibbs,
& Lieberman, 1983; Eisenberg & Morris, 2004). Psychologically,
youths work to establish coherent identities (Erikson, 1968), and
they develop more complex, abstract, and better differentiated
self-concepts (Byrne & Shavelson, 1996; Donahue, 1994; Harter,
1999, 2006; Harter & Monsour, 1992; Marsh, 1989; Marsh &
Ayotte, 2003; Montemayor & Eisen, 1977; Soto, John, Gosling, &
Potter, 2008).
Many of these changes have implications for personality traits,
and previous research has shown that youths can provide reliable
and valid Big Five self-reports (e.g., De Fruyt, Mervielde, Hoek-
stra, & Rolland, 2000; Measelle, John, Ablow, Cowan, & Cowan,
2005; Soto et al., 2008). Few studies, however, have examined age
differences in the Big Five across childhood and adolescence
(Caspi, Roberts, & Shiner, 2005; Klimstra, Hale, Raaijmakers,
Branje, & Meeus, 2009), and fewer still have examined how
differences across these years fit with adult trends. Moreover,
findings from the available studies often conflict with one another
(Klimstra et al., 2009). For example, most studies have found
positive age trends for Openness to Experience in adolescence
(Allik, Laidra, Realo, & Pullmann, 2004; Branje, van Lieshout, &
Gerris, 2007; McCrae et al., 2002; Pullmann, Raudsepp, & Allik,
2006), but some have found negative trends (De Fruyt et al., 2006;
Lamb, Chuang, Wessels, Broberg, & Hwang, 2002). Findings for
the other four domains have been even less consistent.
It is unclear whether these inconsistencies reflect differences in
the populations sampled, the particular age groups selected, the
personality measures used, or a combination of these factors
(Klimstra et al., 2009). One intriguing possibility is that some traits
might show curvilinear age trends, with opposite directions in
different parts of childhood and adolescence, and therefore that the
trends observed in a particular study will depend on the age groups
selected. For example, two previous studies that found negative
patterns for Openness to Experience examined age trends from
childhood into adolescence (De Fruyt et al., 2006; Lamb et al.,
2002), whereas four studies that found positive patterns for this
domain examined age trends from early adolescence into emerging
adulthood (Allik et al., 2004; Branje et al., 2007; McCrae et al.,
2002; Pullmann et al., 2006). One interpretation of these mixed
results is that they suggest a curvilinear pattern for Openness to
Experience, with a negative age trend from childhood to early
adolescence, then a positive trend into adulthood.
Unfortunately, the available data for the other four domains do
not converge on any particular age trends— either linear or curvi-
linear. For each of these domains, there has been an almost equal
split between studies finding significant positive trends, studies
finding significant negative trends, and studies finding null results
(e.g., for Agreeableness, compare Branje et al., 2007; McCrae et
al., 2002; and Pullman et al., 2006). Moreover, this between-study
variability does not appear to relate with the particular age ranges
examined; unlike the findings for Openness to Experience, neither
studies centered around early adolescence nor those centered
around late adolescence have consistently found trends in the same
direction. The overall pattern of results may therefore indicate (a)
a lack of substantial age differences on these domains across late
childhood and adolescence, (b) curvilinear age trends that previous
studies have not been well suited to capture, given that most have
included only a few age groups, or (c) methodological differences
across studies, such as differences in sampling and measurement
procedures.
Thus, one major goal of the present research was to test hypoth-
eses about age differences in Big Five domains and facets across
late childhood and adolescence and about how these fit with adult
trends. Importantly, we used a sample that allowed us to precisely
estimate age-specific means, year by year, and thereby identify
periods within childhood and adolescence when age trends are
especially pronounced, or even have opposite directions at differ-
ent ages. Moreover, we used a personality measure, the Big Five
Inventory (BFI; John, Donahue, & Kentle, 1991; John, Naumann,
& Soto, 2008; Soto & John, 2009a), that has been shown to
converge with a variety of other Big Five measures (Gosling,
Rentfrow, & Swann, 2003; John et al., 2008; John & Srivastava,
1999; Soto & John, 2009a) and to elicit reliable and valid Big Five
self-reports from children and adolescents. Specifically, a previous
study examined the measurement properties of the BFI for partic-
ipants at each year of age from 10 to 20 (Soto et al., 2008). Overall,
the internal consistency of the domain scales, differentiation
among the scales, and clarity of the BFI’s factor structure were
somewhat better at older ages. However, after controlling for
individual differences in response style, at each age the domain
scales’ alpha reliabilities averaged at least .71, and the BFI’s
intended five-factor structure was clearly recognizable: Congru-
ence coefficients with the age 20 structure were at least .93 for
each domain at each age, exceeding the standard of .90 that
indicates factor replication (Barrett, 1986; Mulaik, 1972). Taken
together, these findings show that respondents as young as age 10
can provide meaningful information about their own personalities
using the BFI.
Age Differences in Personality at the Level of
Big Five Facets
Each broad Big Five domain subsumes several more specific
traits, often referred to as “facets” (e.g., Costa & McCrae, 1992,
1995; John et al., 2008). For example, the Conscientiousness
domain encompasses the more specific traits of orderliness and
self-discipline, among others (Roberts, Chernyshenko, Stark, &
Goldberg, 2005). Such same-domain facet traits correlate with
each other across individuals: An orderly person is also likely to be
self-disciplined. However, these correlations are not perfect or
nearly so, with values typically averaging about .40 (Costa &
McCrae, 1992). Indeed, previous research has shown that the
distinctions between same-domain facets are important, because
each facet captures unique personality information (Costa &
McCrae, 1995; McCrae & Costa, 1992), and this unique informa-
tion predicts a variety of important behaviors and life outcomes—
beyond the level of prediction afforded by the five broad domains
themselves. For example, individual Big Five facets relate
uniquely with academic achievement (O’Connor & Paunonen,
2007), alcohol consumption and abuse (Ruiz, Pincus, & Dickin-
son, 2003), delinquent behavior (Heaven, 1996), eyewitness accu-
racy and suggestibility (Liebman et al., 2002), life satisfaction
(Herringer, 1998), personality disorders (Samuel & Widiger,
2008), and many other behaviors (Ashton, Jackson, Paunonen,
Helmes, & Rothstein, 1995; Paunonen & Ashton, 2001).
331
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

Because more specific facet traits can be distinguished within
each Big Five domain, and because these facets capture unique
information about behavior, it is important to ask whether the
facets within each domain show similar or different age trends. If
same-domain facets show identical or nearly identical trends, then
the five broad domains themselves would be sufficient to capture
all of the important information about age differences in person-
ality traits. If, however, different facets show quite different trends,
then facet-level research is needed to achieve a full understanding
of life span age differences in personality.
Considerable recent attention has been devoted to age differences
in personality traits at the level of the broad Big Five domains, but few
studies have examined differences at the facet level. The available
evidence suggests that, within at least some domains, different facets
show different trends. For example, a recent meta-analysis (Roberts et
al., 2006) distinguished between two facets of Extraversion: social
dominance (defined as assertiveness and self-confidence) and social
vitality (defined as gregariousness, positive affect, and energy level).
The results indicated important differences between these two facets:
Mean levels of social dominance increased from the college years
through early adulthood, whereas levels of social vitality were flat.
Similarly, a recent study examined age trends for the 30 facets
assessed by the NEO Personality Inventory–Revised (NEO PI-R;
Costa & McCrae, 1992) in a sample of mostly middle-aged and older
adults (Terracciano et al., 2005). Within most Big Five domains,
different facets showed different age trends. For example, within the
Agreeableness domain, Altruism showed a positive age trend,
whereas Modesty did not.
To our knowledge, no previous study has precisely tracked age
trends for Big Five facets from childhood through middle age.
Therefore, a second central aim of the present research was to test
hypotheses about age differences in Big Five facets across these
years. We tested these hypotheses using the BFI facet scales (Soto
& John, 2009a), which assess two facet traits within each broad
Big Five domain: Assertiveness and Activity within the Extraver-
sion domain, Altruism and Compliance within the Agreeableness
domain, Order and Self-Discipline within the Conscientiousness
domain, Anxiety and Depression within the Neuroticism domain,
and Openness to Aesthetics and Openness to Ideas within the
Openness to Experience domain. Table 1 presents example items
for the 10 BFI facet scales.
Although a consensus has not yet emerged about the ideal way
to define the facets of each Big Five domain, there is considerable
overlap among previously proposed facet-level structures (John et
al., 2008), and the BFI facet scales assess traits common to several
of these (Soto & John, 2009a). Specifically, each BFI facet cor-
responds with a conceptually similar construct from at least three
of the five following models: (a) the facets assessed by the NEO
PI-R (Costa & McCrae, 1992), (b) facets identified in analyses of
English and German trait adjectives (Saucier & Ostendorf, 1999),
(c) circumplex regions defined in analyses of English trait adjec-
tives (Hofstee, de Raad, & Goldberg, 1992), (d) aspects of the Big
Five identified in analyses of existing questionnaire scales
(DeYoung, Quilty, & Peterson, 2007), and (e) facets identified in
the item pool of the California Psychological Inventory (Soto &
John, 2009b). For example, the BFI Anxiety facet corresponds
with the Anxiety, Emotionality, IV-/IV-, Withdrawal, and Anxiety
facets of these five models, respectively.
Testing Hypotheses About Age Differences in Big Five
Domains and Facets From Late Childhood Through
Middle Age
The present research tested several specific hypotheses about
age differences in Big Five domains and facets across late child-
hood (approximately ages 10 –12), adolescence (ages 13–17),
emerging adulthood (ages 18 –25), early adulthood (ages 26 –35),
early middle age (ages 36 –50), and late middle age (ages 51– 65).
These hypotheses are summarized in Table 2.
Conscientiousness and Agreeableness
During childhood and adolescence, one important developmen-
tal task is for youths to become more autonomous in their social
Table 1
The BFI Facet Scales: Names and Example Items
BFI facet scale Example items
Extraversion
Assertiveness 1. Has an assertive personality. 2. Is sometimes shy, inhibited. (R)
Activity 3. Is full of energy. 4. Generates a lot of enthusiasm.
Agreeableness
Altruism 1. Is helpful and unselfish with others. 2. Is considerate and kind to
almost everyone.
Compliance 3. Has a forgiving nature. 4. Starts quarrels with others. (R)
Conscientiousness
Order 1. Tends to be disorganized. (R) 2. Can be somewhat careless. (R)
Self-Discipline 3. Perseveres until the task is finished. 4. Is easily distracted. (R)
Neuroticism
Anxiety 1. Worries a lot. 2. Remains calm in tense situations. (R)
Depression 3. Is depressed, blue. 4. Can be moody.
Openness to Experience
Openness to Aesthetics 1. Values artistic, aesthetic experiences. 2. Has few artistic interests. (R)
Openness to Ideas 3. Likes to reflect, play with ideas. 4. Is curious about many things.
Note. Reverse-keyed items are denoted by (R). The common stem for all BFI items is “I see myself as someone
who . . . ” BFI ⫽Big Five Inventory.
332 SOTO, JOHN, GOSLING, AND POTTER

values and behavior (Colby et al., 1983; Eisenberg & Morris,
2004; Larson, Richards, Moneta, Holmbeck, & Duckett, 1996;
Steinberg & Silverberg, 1986). Specifically, young children gen-
erally accept the values and norms conveyed to them by adult
authority figures and try to act accordingly (Colby et al., 1983;
Eisenberg & Morris, 2004). In contrast, adolescents seek greater
autonomy from authority figures, both by spending a greater
amount of time away from adult supervision (Patterson &
Stouthamer-Loeber, 1984) and by more frequently questioning and
resisting values, rules, and norms that they perceive as imposed on
them by adults. These changes create greater opportunities for
risky and rebellious behaviors, ranging from arguments with par-
ents and other authority figures (C. Jackson, 2002; Smetana, 1995,
2000) to unprotected sex and criminal activity (Centers for Disease
Control and Prevention, 2008; Federal Bureau of Investigation,
2007; Loeber & Farrington, 2000; Steinberg, 1986).
From late adolescence into adulthood, youths increasingly de-
velop and internalize abstract moral and social principles that
promote prosocial and responsible behaviors (Eisenberg, Carlo,
Murphy, & van Court, 1995; Eisenberg & Morris, 2004). They
also continue to develop self-regulatory skills that can help them
avoid risky behaviors in the interest of long-term goals (Dem-
etriou, 2000; Gestsdottir & Lerner, 2008; Murphy, Eisenberg,
Fabes, Shepard, & Guthrie, 1999). In terms of the Big Five, we
hypothesized that these changes would lead to curvilinear patterns
for Conscientiousness and Agreeableness across childhood and
adolescence. Specifically, we expected that mean levels of Con-
scientiousness, Agreeableness, and their more specific facets
would show negative age trends from late childhood into adoles-
cence, then positive trends from late adolescence into emerging
adulthood.
During early adulthood and middle age, many people work at
two important life tasks: achieving in a job or career and forming
satisfying and supportive close relationships (Erikson, 1968;
Hogan & Roberts, 2004). Because Conscientiousness and Agree-
ableness facilitate the successful pursuit of these goals (Hogan &
Roberts, 2004; Ozer & Benet-Martı´nez, 2006; Roberts & Wood,
2006), mean levels of overall Conscientiousness and Agreeable-
ness should show positive age trends from emerging adulthood
through middle age. At the facet level, because prosocial tenden-
cies, cooperation with others, and reliability are generally more
important for achieving these life goals than is preference for
organization and structure, we hypothesized that these trends
would be more pronounced for the Altruism and Compliance
facets of Agreeableness, and for the Self-Discipline facet of Con-
scientiousness, than for the Order facet of Conscientiousness (cf.
J. J. Jackson et al., 2009).
Neuroticism
We expected that overall Neuroticism, and its more specific
facets of Anxiety and Depression, would show different age trends
in childhood and adolescence versus adulthood, as well as different
trends for males versus females. Previous research has demon-
strated that, by early adulthood, levels of Neuroticism tend to be
higher among women than among men (e.g., Donnellan & Lucas,
2008), but less is known about the emergence of this gender
difference. During adolescence, girls are more likely than boys to
face a variety of important social and psychological difficulties,
including awareness of negative gender expectations and stereo-
types (Hill & Lynch, 1983; Wichstrøm, 1999), body image con-
cerns (Stice & Bearman, 2001; Stice, Hayward, Cameron, Killen,
& Taylor, 2000), and negative self-perceptions (Cole, Martin,
Peeke, Seroczynski, & Fier, 1999; Dweck, 1986, 1989). We there-
fore hypothesized that (a) gender differences in Neuroticism, Anx-
iety, and Depression would be first present in adolescence, such
that girls would show higher mean levels than would boys, and (b)
these gender differences would be primarily due to positive age
trends for Neuroticism, Anxiety, and Depression among girls,
rather than to negative trends among boys.
Over the course of adulthood, most people shift toward greater
use of emotion regulation strategies that effectively reduce nega-
tive affect (Helson & Soto, 2005; John & Gross, 2004; Labouvie-
Vief, Diehl, Jain, & Zhang, 2007). Moreover, most adults establish
close relationships that become more satisfying and supportive
over time (Anderson, Russell, & Schumm, 1983; Carstensen,
Table 2
Summary of Hypotheses
Big Five domain or facet
Expected age trend
Childhood and adolescence Adulthood
Conscientiousness Negative, then positive Positive
Self-Discipline Negative, then positive Positive
Order Negative, then positive Flat
Agreeableness Negative, then positive Positive
Altruism Negative, then positive Positive
Compliance Negative, then positive Positive
Neuroticism Positive for females, flat for males Negative
Anxiety Positive for females, flat for males Negative
Depression Positive for females, flat for males Negative
Extraversion Flat Flat
Activity Negative Flat
Assertiveness Flat Positive
Openness to Experience Positive Flat
Openness to Ideas Positive Flat
Openness to Aesthetics Flat Flat
333
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

1992; Gorchoff, John, & Helson, 2008; Kapinus & Johnson, 2003;
Lang & Carstensen, 2002; Rollins & Cannon, 1974) and poten-
tially buffer them from life stressors. Therefore, we expected that
levels of Neuroticism, Anxiety, and Depression would show neg-
ative age trends across early adulthood and middle age.
Extraversion
We expected that different facets of Extraversion would show
different age trends. Both psychologists’ models of child temper-
ament (e.g., Buss & Plomin, 1984; Rothbart, Ahadi, & Evans,
2000; Thomas & Chess, 1977) and nonpsychologists’ descriptions
of children’s personalities (Eaton, 1994; John, Caspi, Robins,
Moffitt, & Stouthamer-Loeber, 1994) often include the trait of
activity level as a central feature. In contrast, models of adult
personality usually relegate this trait to secondary status, as a facet
of Extraversion (e.g., Costa & McCrae, 1992; Saucier & Osten-
dorf, 1999). One plausible explanation for this difference is that
children typically show much higher levels of activity than do
adults. We therefore hypothesized that the BFI Activity facet of
Extraversion, which assesses energy and enthusiasm, would show
a pronounced negative age trend from late childhood into adoles-
cence, whereas the Assertiveness facet, which assesses social
dominance, talkativeness, and expressiveness, would not.
During early adulthood and middle age, many people strive to
increase their social status, both inside and outside of their work
role (Helson, Soto, & Cate, 2006; Kuhlen, 1968), and most people
achieve their highest level of status during middle age (Cameron,
1970; Helson & Soto, 2005; Lachman, Lewkowicz, Marcus, &
Peng, 1994; Roberts, 1997). We hypothesized that these normative
changes in status would be accompanied by a positive age trend for
Assertiveness from emerging adulthood into middle age (cf. Rob-
erts et al., 2006). We did not expect the Activity facet of Extra-
version to show a similarly positive trend.
Openness to Experience
We hypothesized that overall Openness to Experience, and
particularly the facet of Openness to Ideas, would show positive
age trends from adolescence into emerging adulthood, for two
reasons. The first concerned normative changes in youths’ cogni-
tive capacities. Compared with children, adolescents are better
able to think in terms of the abstract and hypothetical (Flavell,
Miller, & Miller, 1993; Inhelder & Piaget, 1958), and these ca-
pacities open up new possibilities for creativity and exploration.
The second reason was the transition to college. More than half of
all recent high school graduates attend college (U.S. Census Bu-
reau, 2006), and the college years typically expose students to a
variety of new ideas and people, and also provide greater freedom
(compared with earlier schooling) for students to pursue their
particular areas of intellectual interest.
Across early adulthood and middle age, the notion that people
become “set in their ways” seems to suggest that mean levels of
overall Openness to Experience, and particularly Openness to
Ideas, should show negative age trends. However, this may be a
caricature of adulthood, as many adults continue to experience
important transitions in work (e.g., job change), relationships (e.g.,
parenthood, divorce, remarriage), and education (e.g., college or
graduate school) across these years (U.S. Census Burearu, 2001,
2006), suggesting ongoing openness to new experiences (cf. Rob-
erts et al., 2006).
The Present Research Design
In the present research, we examined age differences in Big Five
domains and facets from late childhood through middle age, using
data from a very large cross-sectional sample (N⫽1,267,218)
assessed over the World Wide Web. Compared with previous
studies of Big Five age trends, the present design had four dis-
tinctive features. First, most previous studies have used samples
either of children and adolescents or of adults. In contrast, the
present sample included participants ranging in age from 10 to 65
years old. This allowed us to examine how age differences across
childhood and adolescence fit with adult trends, and specifically to
test whether some personality traits show quite different age trends
in childhood and adolescence versus adulthood.
Second, most previous studies have each estimated trait mean
levels for only a few age groups, or estimated relatively simple age
trends (e.g., linear, quadratic). In contrast, the present sample,
which included at least 945 participants at each individual year of
age within its range, allowed for much more fine-grained analyses
than have been previously possible. Specifically, we computed
trait mean levels year by year from ages 10 to 65 and used these
age-specific means to test hypotheses that some traits show com-
plex age trends that have different slopes, or even opposite direc-
tions, in different periods of life.
Third, only a small handful of previous studies have examined
age trends at the level of Big Five facets. Two of these studies have
each considered facets for only one of the Big Five domains:
Extraversion (Roberts et al., 2006) or Conscientiousness (Jackson
et al., 2009). The remaining studies have all used the same instru-
ment, the NEO PI-R (Costa & McCrae, 1992), highlighting a need
for research that tests the generalizability of their findings to other
measures. Toward this end, we used the BFI (John et al., 1991,
2008; Soto & John, 2009a) in the present research to measure 10
facet traits (see Table 1). This allowed us to test whether all— or
almost all—meaningful information about age differences in per-
sonality traits can be captured by the Big Five domains themselves
or whether a complete understanding of such differences requires
consideration of facet-level traits.
Finally, whereas most cross-sectional studies collect data over
the course of only a few weeks or months, the present data were
collected over a period of 7 years. This longer period of data
collection produced variation in participant year of birth beyond
that shared with age at time of participation. For example, the
sample included individuals who all participated at the age of 30
but were born in different years: 1973, 1974, 1975, 1976, 1977,
1978, and 1979. This variation in birth years, in turn, enhanced the
generalizability of our findings and allowed us to test whether
earlier born versus later born participants’ personalities differed
systematically from each other. We consider the possibility of such
birth-cohort effects, as well as other alternative explanations for
our findings, in the Discussion section.
Method
Participants. Participants were 1,267,218 residents of
English-speaking countries who volunteered to provide personality
334 SOTO, JOHN, GOSLING, AND POTTER

and demographic information over the World Wide Web as part of
the Gosling-Potter Internet Personality Project (see Srivastava et
al., 2003). Participants ranged in age from 10 to 65 years old (M⫽
25.7 years old, SD ⫽10.9 years), and 65% were female. Impor-
tantly, the sample included at least 945 participants at each year of
age, including at least 422 participants of each gender.
1
The sample was diverse in terms of ethnicity and nationality.
Regarding ethnicity, 70% of participants were White/Caucasian,
8% were Asian/Asian American, 6% were Black/African Ameri-
can, 6% were Hispanic/Latino, 1% were American Indian/Native
American, 5% reported their ethnicity as “other,” and 4% did not
report their ethnicity. Regarding nationality, 72% were residents of
the United States, 6% were residents of Canada, 6% were residents
of the United Kingdom, 1% were residents of Ireland, 3% were
residents of Australia, 1% were residents of New Zealand, and
11% did not report their nationality.
The sample was also diverse in terms of socioeconomic status.
When asked to describe their social class, 12% of participants
described themselves as working class, 10% as lower-middle class,
20% as middle class, 8% as upper-middle class, 1% as upper class,
35% as not financially independent (presumably, almost all of
these were students), and 14% did not describe their social class.
Regarding highest level of education completed, 20% had not (or
not yet) graduated high school, 43% were high school graduates
(including those who were currently attending college), 15% were
college graduates, 10% had completed graduate or professional
school, and 12% did not report their level of education.
Procedure. The data were collected using a noncommercial,
advertisement-free website, outofservice.com, that offers its visi-
tors free feedback on several surveys and personality measures.
Potential participants could reach this site in a number of ways,
including search engines, links from other websites, and informal
channels such as e-mail, online discussion forums, and word-of-
mouth. All participants anonymously completed an English-
language, Web-based version of the BFI (John et al., 1991, 2008;
Soto & John, 2009a). After submitting their responses, participants
received automatically generated, generally worded feedback
about their standing on each of the Big Five domains as well as
background information about the Big Five and suggestions for
ways to learn more about personality theory and research.
Relations to two previous studies. Age trends for the Big
Five domains in adulthood (ages 21 to 60) were examined in an
earlier study (Srivastava et al., 2003), using a sample of 132,515
adults assessed via outofservice.com between December 1998 and
August 2000. The sample analyzed here was assessed between
March 2003 and April 2009. Thus, the two samples do not overlap,
and our domain-level analyses from ages 21 to 60 should be
regarded as an independent replication of this earlier study. Sriv-
astava et al. (2003) did not examine age differences (a) in child-
hood and adolescence or (b) at the level of Big Five facets, and so
our use of outofservice.com data to address these issues is entirely
novel.
Another study (Soto et al., 2008) examined the measurement
properties of the BFI domain scales from late childhood into
emerging adulthood (ages 10 to 20), using a sample of 230,047
youths assessed via outofservice.com between December 1998 and
May 2004. Thus, a small portion of the present sample (N⫽
74,234, or 6%) overlaps with that analyzed by Soto et al. (2008),
and our analyses of the BFI domain scales’ measurement proper-
ties should be considered an extension of this previous study.
However, Soto et al. (2008) did not examine the measurement
properties of the BFI facet scales, nor, most importantly, did they
examine mean-level age trends for any of the BFI domain or facet
scales.
Measurement.
The Big Five Inventory (BFI). The BFI (John et al., 1991,
2008; Soto & John, 2009a) is designed to efficiently measure the
core aspects of each Big Five domain. Its 44 items are short,
easy-to-understand phrases that participants rate on a 5-point
agreement scale ranging from 1 (strongly disagree)to5(strongly
agree). These items can be used to score five domain scales, as
well as 10 facet scales (see Table 1) that were recently developed
to assess more specific personality traits within each broad domain
(Soto & John, 2009a). These traits were selected to overlap with
the facet constructs assessed by the widely used NEO PI-R (Costa
& McCrae, 1992) and are each common to several other Big Five
facet models (e.g., DeYoung et al., 2007; Hofstee et al., 1992;
Saucier & Ostendorf, 1999; Soto & John, 2009b).
The BFI was well suited to the present research for several
reasons. First, the BFI domain and facet scales have previously
demonstrated strong internal consistency, retest reliability, conver-
gence with longer Big Five measures, and self-peer agreement
(e.g., Benet-Martı´nez & John, 1998; DeYoung, 2006; John et al.,
2008; John & Srivastava, 1999; Rammstedt & John, 2007; Soto &
John, 2009a). Second, the BFI is easy to understand; its fifth-grade
reading level (Benet-Martı´nez & John, 1998) makes it accessible
to younger respondents (Soto et al., 2008) and to other individuals
with relatively little formal education. Fourth, the individual BFI
items assess common behaviors, thoughts, and feelings that should
be relevant to respondents of diverse ages and backgrounds. Fi-
nally, the BFI can be completed in less than 15 min, a clear
advantage for a study to which each participant was expected to
devote only a limited amount of time.
Controlling for individual differences in acquiescent response
style. Acquiescent response style is the tendency to consistently
agree (yea-saying) or consistently disagree (nay-saying) with test
items, regardless of their content. Uncontrolled individual differ-
ences in acquiescence can pose a serious threat to validity, espe-
cially for scales with an imbalance of true- and false-keyed items.
In particular, such differences bias scale means and interscale
correlations (McCrae, Herbst, & Costa, 2001; Soto et al., 2008).
They also distort factor structures, sometimes even resulting in the
emergence of an artifactual acquiescence factor (Soto et al., 2008;
Ten Berge, 1999).
The BFI domain and facet scales are not fully balanced, and
preliminary analyses indicated that acquiescence varied somewhat
by age in the present sample. Specifically, we indexed acquies-
cence as the within-person mean response to a set of 16 pairs of
BFI items with opposite implications for personality (see Soto &
John, 2009a; Soto et al., 2008). The average score on this index
showed a positive age trend across late childhood and adolescence,
then a slightly negative trend across adulthood. Moreover, indi-
1
The age range of 10 – 65 was selected on the basis of preliminary
analyses indicating that reduced age-specific sample sizes outside this
range resulted in less precise estimated means and therefore less systematic
age trends.
335
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

vidual differences in acquiescence were much more pronounced in
the youngest age groups, with almost twice as much variance in
acquiescence at age 10 as at ages 20 and older (cf. Soto et al.,
2008). Individual differences in acquiescence were therefore con-
trolled, through within-person centering around the acquiescence
index, prior to all analyses presented here.
2
Checking overall data quality. Previous research indicates
that psychological studies conducted over the Internet yield valid
data that replicate findings obtained using more traditional assess-
ment methods (Buchanan & Smith, 1999; Chuah, Drasgow, &
Roberts, 2006; Gosling, Vazire, Srivastava, & John, 2004;
McGraw, Tew, & Williams, 2000; Robins, Trzesniewski, Tracy,
Gosling, & Potter, 2002; Skitka & Sargis, 2006; Srivastava et al.,
2003). Nevertheless, two sets of preliminary analyses were con-
ducted to check the overall quality of the present data, and spe-
cifically to make sure that the sample did not include large pro-
portions of participants who responded randomly or who blatantly
self-enhanced.
First, the alpha reliabilities of the BFI domain and facet scales
were examined. If the present sample included a high proportion of
random responders, then these reliability coefficients would be
substantially lower than in other samples. This was not the case. In
the overall sample, the alpha reliabilities of the five domain scales
were .87 for Extraversion, .81 for Agreeableness, .85 for Consci-
entiousness, .84 for Neuroticism, and .78 for Openness (M⫽
0.83), values typical for these scales (John et al., 2008; John &
Srivastava, 1999). The alpha reliabilities of the 10 shorter facet
scales averaged .67, ranged from .82 for Assertiveness to .51 for
Depression, and were very similar to those found in two previous
samples (Soto & John, 2009a).
Second, discriminant correlations among the domain scales
were examined. If the present sample included a high proportion of
self-enhancers, then the magnitudes of these correlations would be
larger than usual, because self-enhancers would describe them-
selves as highly extraverted, agreeable, conscientious, emotionally
stable (i.e., not neurotic), and open to experience. This also was not
the case. In the overall sample, the average magnitude of the
interscale correlations was .19, which is very similar to the values
obtained in previous studies (Gosling et al., 2003; John et al.,
2008; John & Srivastava, 1999).
Examining the appropriateness of cross-age comparisons.
The strength of the conclusions that can be drawn from cross-age
comparisons of BFI scale scores depends on the extent to which
the BFI functions similarly at different ages. A previous study
found that the internal consistency of the BFI domain scales, and
differentiation among them, were better at older ages across late
childhood and adolescence but that the overall five-dimensional
structure of the BFI could be recovered as early as age 10 (Soto et
al., 2008). As a further check in the present sample, (a) alpha
reliabilities and interscale correlations were examined and (b)
principal-components analyses were conducted in each of 20 age
groups (every individual year of age from 10 to 20, then every 5
years from 21 to 65).
Consistent with the findings of Soto et al. (2008), the alpha
reliabilities of the BFI domain and facet scales were somewhat
higher at older ages. Specifically, the mean reliability of the
domain scales showed a positive age trend from .75 at age 10 to
.83 at age 18, and was either .83 or .84 in each of the older groups.
Similarly, the mean reliability of the shorter facet scales showed a
positive trend from .54 at age 10 to .67 at age 18, and was between
.67 and .69 in each older group. Correlations among the domain
scales also differed somewhat by age. Specifically, the mean
magnitude of the interscale correlations showed a negative age
trend from .22 at age 10 to .16 at age 18, then a positive trend to
.26 at ages 61– 65.
Despite these age differences in reliability and interscale corre-
lations, the BFI’s intended five-dimensional structure was clearly
recovered in principal-components analyses conducted in each age
group. In each group, a scree test indicated five meaningful di-
mensions, and five components were therefore extracted and
varimax-rotated. Congruence coefficients were then computed to
compare each component in each age group with the correspond-
ing component in the age-21-to-25 group; this was selected as the
reference group because the BFI’s intended five-dimensional
structure emerged very clearly in it, as in previous studies of
emerging-adult samples (e.g., Benet-Martı´nez & John, 1998).
These congruence coefficients averaged .99, and all 100 were at
least .96, easily exceeding the standard of .90 that indicates factor
replication (Barrett, 1986; Mulaik, 1972). Taken together, these
results indicate that the overall structure of the BFI was very
consistent across late childhood, adolescence, and adulthood, al-
though there was greater measurement error at younger ages.
Presentation of results. Preliminary analyses indicated that
(a) almost all of the domain and facet traits assessed by the BFI
showed substantial mean-level age differences, (b) some traits
showed age trends that differed considerably by gender, and (c) the
shapes of some trends were curvilinear and quite complex. These
trends could not be closely fit by regression models, even models
that included higher order polynomial age terms (quadratic, cubic,
quartic, quintic, etc.) and corresponding Age ⫻Gender interaction
terms. These preliminary analyses also indicated that the size of
the present sample rendered conventional tests of statistical sig-
nificance meaningless; for example, in the full sample, a correla-
tion of .003 between two variables would be statistically signifi-
cant at the ␣⫽.001 level, and a correlation of .02 would be
significant at the ␣⫽1⫻10
⫺100
level.
We therefore present our results with a focus on the patterns and
sizes of age and gender differences. In describing these patterns,
we refer to “age trends” and “age differences”—rather than devel-
opmental “changes,” “increases,” or “decreases”— due to the
present research’s cross-sectional design. For the same reason,
note that we use prepositional phrases such as “across middle age,”
“by late adolescence,” and “from adolescence into adulthood” to
refer to cross-sectional age ranges, rather than periods of devel-
opment over time.
2
There was strong overall agreement between age trends for the BFI
domain and facet scales in the raw and centered data. For each of the 30
combinations of gender with a domain or facet scale, we computed a
correlation comparing the set of 56 age-specific means from the raw data
with the set of 56 age-specific means from the centered data. These 30
correlations averaged .99, and all were at least .93. Differences between the
raw and centered trends were attributable to age differences in acquies-
cence inflating mean scores on scales with a majority of true-keyed items
(and suppressing mean scores on scales with a majority of false-keyed
items) during late adolescence and emerging adulthood, relative to older
and younger age groups.
336 SOTO, JOHN, GOSLING, AND POTTER

We use the T-score metric to index effect sizes; Tscores are
standard scores with a mean of 50 and standard deviation of 10. In
terms of Cohen’s (1988) now conventional guidelines for inter-
preting effect sizes, a difference of 2 T-score points represents a
small effect, a difference of 5 points represents a medium effect,
and a difference of 8 points represents a large effect. To control for
age and gender effects when converting to Tscores, we computed
the overall mean of each BFI domain and facet scale by first
computing its mean in each of the 112 age- and gender-specific
samples (56 years of age ⫻2 genders) and then averaging these
112 group means. Similarly, we computed the overall standard
deviation of each scale as the square root of the pooled within-
group variance term from a two-way (56 years of age ⫻2 genders)
analysis of variance—an estimate that controls for between-group
variance. This process produced a T-score distribution uninflu-
enced by the age and gender differences that are the subject of the
present research. In the present sample, all pairwise age differ-
ences of at least 2 T-score points were statistically significant at the
␣⫽.00001 level.
Results
Mean scores on the BFI domain and facet scales, by age and
gender, are shown in Figures 1–5. We calculated these group
means separately for each individual combination of age and
gender (age-10 males, age-10 females, age-11 males, age-11 fe-
males, etc.); no smoothing functions have been applied. In each
figure, single lines show the means for males, and double lines
show the means for females. In Figures 1B, 2B, 3B, 4B, and 5B,
black lines show the means for one facet of a domain, and gray
lines show the means for the other facet.
Conscientiousness. Mean levels of Order, Self-Discipline,
and overall Conscientiousness are shown in Figure 1. As hypoth-
esized, overall Conscientiousness showed very different age trends
in late childhood and adolescence versus adulthood. Specifically,
as shown in Figure 1A, Conscientiousness showed a negative age
trend from late childhood into adolescence; the total difference
was approximately 3 T-score points, or one third of a standard
deviation unit. Conscientiousness then showed a pronounced pos-
itive trend from adolescence through emerging adulthood, with a
total difference of approximately 7 T-score points. This trend was
even more pronounced for females than for males, such that by
emerging adulthood females were slightly more conscientious, on
average, than were males (by approximately 2 T-score points).
At the facet level, Figure 1B illustrates that Self-Discipline and
Order showed age trends across these years similar to overall
Conscientiousness, with negative trends from late childhood into
adolescence (with total differences of 2 or 3 T-score points), then
pronounced positive trends through emerging adulthood (with total
differences of 5 or 6 T-score points). Gender differences in these
two facets, however, related differently with age. Like overall
Figure 1. Means for overall Conscientiousness (A) and its facets (B), by age and gender. Single lines show the
means for males, and double lines show the means for females. In Panel B, black lines show the means for
Self-Discipline, and or gray lines show the means for Order.
337
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

Conscientiousness, a small gender difference in Self-Discipline
was first present in emerging adulthood. In contrast, females were
more orderly than males, on average, at each age from 10 to 65.
Overall Conscientiousness showed a further, although less pro-
nounced, positive trend across early adulthood and middle age,
with age differences of approximately 5 more T-score points for
both males and females. At the facet level, as hypothesized, these
differences were substantial for Self-Discipline (approximately 6
T-score points), but only trivial for Order (approximately 1 T-score
point). Altogether, the total age differences from adolescence
through middle age were approximately 11 T-score points for
overall Conscientiousness and Self-Discipline. These are very
large cross-sectional age effects that equal the difference between
scores at the 50th and 86th percentiles of a normal distribution.
The total age difference for Order was approximately 7 T-score
points, a substantial effect that equals the difference between
scores at the 50th and 76th percentiles of a normal distribution.
Agreeableness. Mean levels of Altruism, Compliance, and
overall Agreeableness are shown in Figure 2. As hypothesized,
overall Agreeableness showed age trends similar to Conscientious-
ness, although the trends for Agreeableness were somewhat less
pronounced. As shown in Figure 2A, Agreeableness showed a
negative trend from late childhood into adolescence (with a total
difference of approximately 2 T-score points), a positive trend
from adolescence into emerging adulthood (with a total difference
of approximately 3 T-score points), and a further positive trend
across early adulthood and middle age (with a difference of ap-
proximately 3 more T-score points). The total difference of 6
T-score points from adolescence through middle age represents a
substantial age effect, equal to the difference between scores at the
50th and 73rd percentiles of a normal distribution.
At the facet level, Figure 2B illustrates that Altruism and Com-
pliance showed age trends similar to overall Agreeableness, with
only one qualification: Compliance did not show a negative trend
from late childhood into adolescence for males. Instead, males
showed low levels of Compliance even at age 10, a finding
unlikely to surprise elementary-school teachers. Regarding gender
differences more generally, at each age females were somewhat
more agreeable, altruistic, and compliant, on average, than were
males (by approximately 2 T-score points).
Neuroticism. Mean levels of Anxiety, Depression, and over-
all Neuroticism are shown in Figure 3. There were only trivial
gender differences in these traits at age 10 (less than 1 T-score
point), but, as hypothesized, there were very different age trends
for males versus females across late childhood, adolescence, and
adulthood. For females, Anxiety and overall Neuroticism showed
positive trends into adolescence (with total differences of approx-
imately 3 T-score points), flat trends through emerging adulthood,
and then negative trends across early adulthood and middle age
(with total differences of approximately 5 T-score points). In
Figure 2. Means for overall Agreeableness (A) and its facets (B), by age and gender. Single lines show the
means for males, and double lines show the means for females. In Panel B, black lines show the means for
Altruism, and gray lines show the means for Compliance.
338 SOTO, JOHN, GOSLING, AND POTTER

contrast, males’ mean levels of Anxiety and overall Neuroticism
showed slightly negative trends from late childhood through mid-
dle age (with total differences of approximately 2 T-score points).
Substantial gender differences in both of these traits (of approxi-
mately 5 T-score points) were present by midadolescence, such
that females were more prone to anxiety and other negative emo-
tions, on average, than were males. The magnitudes of these
gender differences diminished across early adulthood and middle
age, although a small difference (of approximately 2 T-score
points) was present even at the end of middle age.
Age trends for Depression— by which we mean general suscep-
tibility to sad affect, rather than any clinical disorder— differed
from those for Anxiety and overall Neuroticism in several ways.
First, and unexpectedly, mean levels of Depression showed two
distinct peaks for females— one in adolescence and one in early
adulthood; Depression then showed a negative trend (with a total
difference of approximately 5 T-score points) across middle age.
Second, for males, levels of Depression showed a positive trend
(with a total difference of approximately 3 T-score points) from
late childhood into early adulthood, then a slightly negative trend
(with a total difference of approximately 2 T-score points) across
middle age. Third, although a substantial gender difference in
mean levels of Depression (of approximately 4 T-score points) was
present by midadolescence, the size of this difference diminished
with age, and men and women reported equal levels of sad affect
by late middle age.
Extraversion. Mean levels of Assertiveness, Activity, and
overall Extraversion are shown in Figure 4. Figure 4B illustrates
that, as hypothesized, Activity and Assertiveness showed different
age trends in late childhood, adolescence, and emerging adulthood.
Activity showed a marked negative trend from late childhood
through adolescence, as well as a less pronounced negative trend
across emerging adulthood. The total age difference across these
years was approximately 7 T-score points, a large effect that equals
the difference between scores at the 50th and 24th percentiles of a
normal distribution.
In contrast, Assertiveness and overall Extraversion showed only
modestly negative trends from late childhood into adolescence,
then flat trends through emerging adulthood. Unlike Activity, the
negative trends for Assertiveness and overall Extraversion were
somewhat more pronounced for males (with total differences of 4
or 5 T-score points) than for females (with total differences of 2 or
3T-score points), such that by midadolescence females were more
talkative, expressive, and generally extraverted than were males
(by approximately 2 T-score points). Contrary to expectations,
Assertiveness did not show a positive age trend across early
adulthood. Instead, Assertiveness, Activity, and overall Extraver-
sion showed flat trends from early adulthood through middle age.
Openness to Experience. Mean levels of Openness to Aes-
thetics, Openness to Ideas, and overall Openness to Experience are
shown in Figure 5. We had expected that overall Openness, and
particularly Openness to Ideas, would show positive age trends
Figure 3. Means for overall Neuroticism (A) and its facets (B), by age and gender. Single lines show the means
for males, and double lines show the means for females. In Panel B, black lines show the means for Anxiety,
and gray lines show the means for Depression.
339
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

from adolescence into emerging adulthood, then flat trends
through late middle age. As Figure 5 illustrates, this hypothesis
was only partially supported. Openness to Ideas and overall Open-
ness showed negative trends from late childhood into early ado-
lescence (with total differences of approximately 2 T-score points),
and further negative trends across adolescence for females (with
differences of approximately 2 more T-score points). These traits
then showed positive trends across emerging adulthood for both
genders (with total differences of 2 or 3 T-score points), and
further slightly positive trends through middle age (with differ-
ences of approximately 2 more T-score points). Small gender
differences were present by emerging adulthood, such that males
were somewhat more open, on average, than were females (by 2 or
3T-score points).
Mean levels of Openness to Aesthetics showed a rather different
pattern. For males, Openness to Aesthetics showed a positive age
trend from late childhood through emerging adulthood (with a total
difference of approximately 3 T-score points), then a flat trend
across early adulthood and middle age. For females, Openness to
Aesthetics showed a flat trend from late childhood through early
adulthood, then a slightly positive trend across middle age (with a
total difference of approximately 3 T-score points). At each age,
females were more open to aesthetics, on average, than were
males. This difference was only trivial (approximately 1 T-score
point) in early adulthood and early middle age, but was more
substantial (3 or 4 T-score points) in late childhood, adolescence,
and late middle age.
Discussion
The present results support several conclusions about age dif-
ferences in personality traits from late childhood through middle
age, in terms of the Big Five domains and their more specific
facets. They also highlight some key issues in need of further
investigation.
Age differences in personality traits across late childhood
and adolescence. How do the mean levels of Big Five domains
and facets differ across childhood and adolescence? In the present
sample, Agreeableness, Conscientiousness, and their more specific
facets showed curvilinear, nonmonotonic age trends, with negative
trends from late childhood into early adolescence, then pronounced
positive trends into emerging adulthood. Extraversion showed a
negative trend from late childhood into adolescence; facet-level
analyses indicated that this was mainly due to age differences in
Activity, rather than Assertiveness.
Age trends for Neuroticism and Openness to Experience dif-
fered both by gender and by facet. Within the Neuroticism domain,
Anxiety and Depression showed marked positive trends from late
childhood into adolescence among females; Depression— but not
Anxiety—then showed a negative trend into the college years. In
contrast, among males, Anxiety showed a negative trend across
late childhood and adolescence, whereas Depression showed a flat
trend. Among both males and females, Openness to Ideas showed
a negative trend into adolescence, then a positive trend across the
college years. In contrast, Openness to Aesthetics showed a pos-
Figure 4. Means for overall Extraversion (A) and its facets (B), by age and gender. Single lines show the means
for males, and double lines show the means for females. In Panel B, black lines show the means for
Assertiveness, and gray lines show the means for Activity.
340 SOTO, JOHN, GOSLING, AND POTTER

itive trend from late childhood into emerging adulthood among
males only.
Taken together, these results indicate that late childhood and
adolescence are key periods for understanding life span age dif-
ferences in personality traits, although we acknowledge that the
overall uncertainty of the present findings is greater at younger
ages, due to greater measurement error. Some traits showed sub-
stantial age differences only across late childhood and adoles-
cence, some showed especially pronounced trends, and some even
showed curvilinear trends with opposite directions at different
ages. Moreover, several of these curvilinear patterns—including
those for Agreeableness, Conscientiousness, and Openness to
Ideas—illustrate that life span age differences in personality traits
do not simply represent a monotonic trend toward greater and
greater psychosocial adjustment. Instead, they indicate that the
biosocial changes and challenges of early adolescence (“storm and
stress;” Hall, 1904) are often accompanied by negative personality
trends.
The curvilinear patterns found in the present research may help
explain some of the seemingly inconsistent findings from previous
studies that have examined age trends for the Big Five in adoles-
cence. For example, some studies have found negative trends for
Openness to Experience in early adolescence (e.g., De Fruyt et al.,
2006), whereas others have found positive trends from adolescence
into emerging adulthood (e.g., McCrae et al., 2002). Both sets of
results are consistent with the present finding that Openness shows
a curvilinear pattern from late childhood through emerging adult-
hood, with a negative trend at younger ages and a positive trend at
older ages.
Such curvilinear trends may also help explain why a recent
meta-analysis suggested that mean-level age trends for the Big
Five domains may be less pronounced across adolescence (defined
as ages 10 –18) than across early adulthood (defined as ages
22–30; Roberts et al., 2006). For example, averaging negative age
trends in early adolescence with positive trends in late adolescence
would lead to the conclusion that there were only modest age
differences overall. Thus, the present results highlight the special
importance of research that can capture curvilinear patterns by
precisely estimating trait mean levels year by year across late
childhood and adolescence.
A final point regarding the importance of late childhood and
adolescence is that mean-level gender differences in several traits
were first present across these years. The social and biological
changes surrounding puberty are often experienced in quite differ-
ent ways by boys versus girls (e.g., Cole et al., 1999; Dweck, 1986,
1989; Hill & Lynch, 1983; Stice & Bearman, 2001; Stice et al.,
2000; Wichstrøm, 1999), and these differences appear to have
gender-specific implications for adolescents’ personalities. Most
strikingly, we found pronounced positive age trends for both the
Anxiety and Depression facets of Neuroticism among adolescent
Figure 5. Means for overall Openness (A) and its facets (B), by age and gender. Single lines show the means
for males, and double lines show the means for females. In Panel B, black lines show the means for Openness
to Ideas, and gray lines show the means for Openness to Aesthetics.
341
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

girls, but not among boys. These patterns parallel epidemiological
findings that gender differences in rates of clinical depression
(Angold & Worthman, 1993; Hankin et al., 1998) and anxiety
(Hale, Raaijmakers, Muris, van Hoof, & Meeus, 2008) are present
by adolescence. The convergence of these findings highlights
adolescence as a key period for understanding—and potentially
reducing— gender differences in anxious and sad affect.
Age differences in personality traits across adulthood.
How do the mean levels of Big Five domains and facets differ
across adulthood? In the present research, the most pronounced
age trends were positive trends for Agreeableness and the Self-
Discipline facet— but not the Order facet— of Conscientiousness.
Specifically, our results indicate that a typical 65-year-old is more
self-disciplined than approximately 85% of early adolescents, and
more agreeable than approximately 75% of early adolescents. Both
of these findings agree well with previous cross-sectional (Don-
nellan & Lucas, 2008; Jackson et al., 2009; McCrae et al., 1999,
2000; Srivastava et al., 2003) and longitudinal (Roberts et al.,
2006; Terracciano et al., 2005) studies. They should also be
welcome news to young people (and their parents), especially
those who question whether they will have the psychosocial ma-
turity needed to successfully negotiate important life tasks, such as
pursuing a career, establishing close and supportive relationships,
or contributing to the welfare of the next generation (e.g., Erikson,
1968; Hogan & Roberts, 2004).
Anxiety, Depression, and overall Neuroticism showed negative
age trends across early adulthood and middle age. These were
more pronounced among women than men, such that by late
middle age, mean-level gender differences were either small (for
Anxiety and overall Neuroticism) or nonexistent (for Depression).
Previous research has consistently found negative trends for Neu-
roticism across adulthood (McCrae et al., 1999, 2000; Roberts et
al., 2006; Srivastava et al., 2003; Terracciano et al., 2005). More-
over, the pattern of smaller gender differences in Neuroticism,
Anxiety, and Depression at older ages converges with epidemio-
logical findings that women show higher rates of clinical depres-
sion than men in early adulthood, but not in middle age (e.g.,
Bebbington et al., 1998; Bland, Newman, & Orn, 1988). The
present results thus highlight a need for additional research exam-
ining gender differences in Neuroticism and its facets across
adulthood.
Extraversion and Openness to Experience showed the smallest
age differences across adulthood. Previous results regarding these
domains have been mixed, with some studies finding lower levels
at older ages (e.g., McCrae et al., 1999, 2000), but others finding
flat age trends (e.g., Roberts et al., 2006; Srivastava et al., 2003).
Taken together, the available evidence now indicates that there are
likely no more than modest age differences in overall Extraversion
and Openness across early adulthood and middle age. In our view,
however, the possibility of positive trends for the Assertiveness
facet of Extraversion warrants further investigation (cf. Helson &
Soto, 2005; Roberts et al., 2006; Terracciano et al., 2005).
Age differences and levels of analysis: Domains versus fac-
ets. Do the related but distinguishable facet traits within each
broad Big Five domain show similar or different age trends? The
present results indicate that, within most domains, different facets
show different trends. For some domains, these differences were
relatively subtle, but for others—particularly Extraversion, Con-
scientiousness, and Neuroticism—they were quite pronounced.
This conclusion is further supported by the few previous studies
that have examined facet-level age trends within at least some
of the Big Five domains (Jackson et al., 2009; Roberts et al., 2006;
Terracciano et al., 2005).
Importantly, this growing body of findings indicates that con-
ceptualizing traits at the level of Big Five facets is necessary for a
full understanding of life span age differences in personality;
research at the domain level can provide a rough sketch of these
differences, but not a complete picture. Looking ahead, future
research at the facet level will benefit from continued efforts to
identify the most important traits within each broad Big Five
domain (e.g., DeYoung et al., 2007; John et al., 2008), in the same
way that consensus around the broad domains themselves has
facilitated progress toward tracing the general contours of age
differences in personality.
Considering alternative explanations: Cohort, self-selection,
and social desirability. In the present research, we examined
data from a large and diverse cross-sectional sample. Compared
with previous studies, this design had the important advantage of
allowing us to estimate mean levels of Big Five domains and facets
year by year from late childhood through middle age. However, it
also had some limitations.
First, the cross-sectional nature of the design raises the possi-
bility that some of the observed age trends might reflect birth-
cohort effects—the effects of older and younger participants being
born during earlier and later years, respectively—rather than, or in
addition to, aging effects (Schaie, 1977). For example, if individ-
uals born in the 1940s were more strongly socialized, in childhood
and adolescence, to be altruistic and polite than were individuals
born in the 1970s, this could produce positive cross-sectional age
differences in Agreeableness across adulthood— even if individu-
als did not typically become more agreeable as they aged.
An unusual feature of the present research, compared with
previous cross-sectional studies, is that the data were collected
over a 7-year period. Therefore, each specific age group in the
sample (e.g., age 30) included members of several different birth
cohorts (e.g., individuals born in 1973, 1974, 1975, 1976, 1977,
1978, and 1979). Including a range of birth years in each age group
provides a degree of generalizability across cohorts. Moreover,
this variation in birth years, within the age groups, allowed us to
directly test for some cohort differences between earlier- and later
born participants. Specifically, we conducted a series of 15 anal-
yses of variance; each analysis included year of participation, age,
and gender as between-subjects factors, and one of the 15 BFI
domain or facet scales as the dependent variable.
If there were pronounced differences between earlier- and later-
born participants, then (a) the estimated marginal means by year of
participation (controlling for age and gender) should differ sub-
stantially from each other and (b) the estimated marginal means by
age and gender (controlling for year of participation) should differ
substantially from the observed means shown in Figures 1–5.
Neither of these expectations was met. Regarding the estimated
marginal means by year of participation, all 105 of these (7 years
of participation ⫻15 traits) were between 49.40 and 51.27—very
close to their expected value of 50.00. Regarding the estimated
marginal means by age and gender, these were virtually identical
to those shown in Figures 1–5. Specifically, each of the 1,680
estimated means (56 ages ⫻2 genders ⫻15 traits) was within 0.24
T-score points of its corresponding observed mean.
342 SOTO, JOHN, GOSLING, AND POTTER

These results argue against the influence of pronounced cohort
effects, a conclusion further supported by two other aspects of the
present findings. First, the observed adult age trends—particularly
the positive trends for Conscientiousness and Agreeableness, and
the negative trend for Neuroticism—agree well with the available
longitudinal data (cf. Roberts et al., 2006), which are not suscep-
tible to cohort effects. Second, several personality traits—such as
Agreeableness, Conscientiousness, Depression, Activity, and
Openness to Ideas—showed age trends from late childhood into
emerging adulthood that were pronounced and curvilinear. To
explain these trends, cohort effects would need to be similarly
curvilinear and even more pronounced (given that each age-
specific mean included data from members of several different
birth cohorts). This seems quite unlikely. Even so, further re-
search— especially studies that repeatedly assess members of mul-
tiple birth cohorts—is needed to fully separate cohort and aging
effects on personality traits.
A second limitation is that our participants were volunteers,
raising the possibility of differential self-selection effects—
individuals at different ages volunteering to participate for differ-
ent reasons associated with their personality traits. For example,
one possible explanation for the slightly negative age trend for
Openness to Experience from late childhood into adolescence, and
for the slightly positive trend across middle age, is that children
and middle-aged adults who choose to complete questionnaires in
exchange for personality feedback may be especially concerned
with understanding themselves, an aspect of Openness (Beitel &
Cecero, 2003; Costa & McCrae, 1992). In contrast, actively seek-
ing out personality feedback may be a more common behavior
among adolescents and emerging adults (Erikson, 1968). There-
fore, participants at the tails of the present age distribution may
have been somewhat self-selected for Openness to Experience,
resulting in elevated mean levels of Openness at these ages.
We tested for possible self-selection effects within the present
sample by examining the variability of Big Five domains and
facets in different age groups. If our youngest and oldest partici-
pants were indeed self-selected on the basis of their personality
traits, then the observed variability of scores on those traits would
be largest in adolescence and emerging adulthood (where age-
specific samples would be most representative) and smallest in late
childhood and late middle age (where age-specific samples would
be most self-selected).
To identify any such trends, we computed the standard deviation
of each domain and facet scale in (a) late childhood (ages 10 –12),
(b) adolescence and emerging adulthood (ages 13–25), and (c) late
middle age (ages 51– 65), controlling for gender differences. Then,
for each scale, we computed two ratios to compare its standard
deviation (a) in late childhood versus during adolescence and
emerging adulthood and (b) in late middle age versus in adoles-
cence and emerging adulthood. Whereas ratios substantially
smaller than 1.00 would indicate much less variability in the
youngest or oldest age groups—and thus the possibility of differ-
ential self-selection—the obtained ratios were consistently close to
1.00. Specifically, the 15 late-childhood ratios averaged 0.99, and
only two were less than 0.90 —those for Assertiveness (0.89) and
overall Extraversion (0.88). The 15 late-middle-age ratios aver-
aged 1.01, and all were greater than 0.90.
In summary, the variability of each domain and facet scale was
quite consistent across age groups, and the scales that showed the
largest age differences in variability—Assertiveness and overall
Extraversion— did not also show the largest age differences in
mean levels. These results suggest that the present findings are not
likely due to differential self-selection effects, but replication using
different methods of sample recruitment remains important.
A final limitation is that our data are personality self-reports.
Although previous research has demonstrated substantial self-peer
agreement for the BFI domain and facet scales (e.g., Benet-
Martı´nez & John, 1998; DeYoung, 2006; John et al., 2008; John &
Srivastava, 1999; Rammstedt & John, 2007; Soto & John, 2009a),
agreement between self-reports, peer reports, and observer reports
of behavior is imperfect (e.g., Gosling, John, Craik, & Robins,
1998). A particular concern here is that the youngest participants
in the present sample may be inaccurate or unreliable judges of
their own personalities. For example, negative age trends for
self-reported Agreeableness and Conscientiousness from late
childhood into adolescence might reflect unrealistically positive
self-views in childhood compared with more accurate views in
adolescence (cf. Robins et al., 2002).
One way to test this possibility is to examine the individual BFI
items. Because these items differ in their social desirability, we can
compare age differences in the average responses to more and less
desirable items. Specifically, if age differences in personality self-
reports across late childhood and adolescence are mainly due to
children’s inflated self-views, then average responses to the most
socially desirable items should be much higher at younger ages,
responses to the least desirable items should be much lower at
younger ages, and responses to more evaluatively neutral items
should show relatively small age differences.
For each of the 44 BFI items (and separately for males and
females), we therefore computed an age-difference index equal to
the average response in late childhood (ages 10 –12) minus the
average response in early adolescence (ages 13–15); positive val-
ues on this index indicate a higher average response in childhood
than in adolescence, and negative values indicate the opposite. The
social desirability of each item was indexed as its mean standard-
ized rating, by 15 judges, on a scale ranging from 1 (extremely
undesirable)to9(extremely desirable)(␣⫽.96; see Robins,
Tracy, Trzesniewski, Potter, & Gosling, 2001). We then correlated
the mean social desirability ratings with the age-difference index.
We did this separately (a) for males and for females and (b) across
the 26 BFI items keyed in a socially desirable direction (i.e., the
true-keyed Extraversion, Agreeableness, Conscientiousness, and
Openness items, and false-keyed Neuroticism items) and across
the 18 items keyed in a socially undesirable direction, resulting in
a total of four correlations.
If responses to the most and least socially desirable items
showed especially large age differences, then all of these correla-
tions should be positive, and substantial in size. This, however,
was not the case. Instead, only one of the four correlations was
positive (for females, across the socially undesirable items), and
this correlation was small and did not approach statistical signif-
icance (r⫽.18, p⫽.47). These results indicate that the age
differences across late childhood and adolescence found here were
not mainly due to children’s inflated self-views. However, we
readily acknowledge the importance of replicating these findings
using nonself-report data. Although replication using multiple data
sources is always an important issue, it is particularly important
when considering personality self-reports from children and ado-
343
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

lescents, which are less reliable than adults’ reports (Soto et al.,
2008). Thus, a key goal for future research is to examine age
differences in personality traits across childhood and adolescence
using informant reports and observational data (see also Branje et
al., 2007).
From description to explanation. As research findings con-
tinue to show mean-level age differences in personality traits
across the life span, they also raise questions about the causes of
these differences. For example, why might middle-aged adults
generally be more agreeable, self-disciplined, and emotionally
stable than young adults?
We caution that the present results should not be taken as
evidence for particular causes of age differences in personality
traits. For example, any pattern of mean-level differences could be
attributable to either social or biological factors (McCrae et al.,
2000). With this caveat in mind, however, there are several ways
in which the present findings can inform future research designed
to directly investigate the causes of age differences in personality
traits. First, the finding that Altruism, Compliance, and Self-
Discipline all showed similar age trends (see Figures 1B and 2B)
suggests that these three traits might be influenced by similar
social and biological factors. For example, these traits’ positive
trends from adolescence through middle age might all result from
investment in a common set of social roles—such as student,
worker, and parent—that frequently call for prosocial behavior and
impulse regulation (Helson, Kwan, John, & Jones, 2002; Hogan &
Roberts, 2004; Roberts & Wood, 2006), or from the ongoing
development of neurobiological systems that help regulate such
behavior (DeYoung, Peterson, & Higgins, 2002).
Conversely, the findings that Order and Self-Discipline (see
Figure 1B), Anxiety and Depression (see Figure 3B), Assertive-
ness and Activity (see Figure 4B), and Openness to Aesthetics and
to Ideas (see Figure 5B) showed quite different age trends indicate
that different facet traits within the same broad Big Five domain
are sometimes influenced by different mechanisms. For example,
investment in student, worker, and parent roles may typically lead
to greater persistence and self-control, but not to greater orderli-
ness.
Finally, the shapes of the age trends for some traits suggest key
periods in the life span. For example, the gender-specific age
trends for Depression (see Figure 3B) suggest that positive trends
in sad affect might be caused by at least two distinct sets of
mechanisms: one that typically affects adolescent girls, but not
boys, and one that affects both men and women in emerging and
early adulthood. One possible explanation for this pattern is that
high levels of sad affect among adolescent girls might be due to
negative gender expectations and stereotypes that are especially
salient across these years (Hill & Lynch, 1983; Wichstrøm, 1999),
whereas high levels among early-adult men and women might
reflect common physical and emotional strains linked with the
transition to parenthood (Walker, 1977).
The most direct approach for testing these hypotheses, and
others like them, is through longitudinal studies in which person-
ality traits—and causal factors that might influence them—are
assessed repeatedly over time. Such studies could not only esti-
mate mean-level age trends but also distinguish between individ-
uals who show especially positive or negative changes on a par-
ticular trait. They could therefore identify social and biological
factors that predict individual differences in subsequent personal-
ity change—a necessary condition for inferring causation. For
example, a previous longitudinal study of women’s adult devel-
opment examined the effects of work investment on personality
change across middle age (Helson & Soto, 2005). It identified
women who were more and less invested in their careers during
their early 50s and found that the women who were most invested,
at this age, subsequently showed especially steep declines in
assertiveness during the years surrounding retirement, as work
pressures eased. This pattern suggests that individual differences in
work investment led to differences in the experience of retirement,
and ultimately to differences in personality change. Such studies,
however, remain rare, and new research could contribute im-
mensely to our understanding of the processes underlying age
differences in personality traits.
References
Allemand, M., Zimprich, D., & Hendriks, A. A. J. (2008). Age differences
in five personality domains across the life span. Developmental Psychol-
ogy, 44, 758 –770. doi:10.1037/0012-1649.44.3.758
Allik, J., Laidra, K., Realo, A., & Pullman, H. (2004). Personality devel-
opment from 12 to 18 years of age: Changes in mean levels and structure
of traits. European Journal of Personality, 18, 445– 462. doi:10.1002/
per.524
Anderson, S. A., Russell, C. S., & Schumm, W. R. (1983). Perceived
marital quality and family life-cycle categories: A further analysis.
Journal of Marriage & the Family, 45, 127–139. doi:10.2307/351301
Angold, A., & Worthman, C. W. (1993). Puberty onset of gender differ-
ences in rates of depression: A developmental, epidemiologic and neu-
roendocrine perspective. Journal of Affective Disorders, 29, 145–158.
doi:10.1016/0165-0327(93)90029-J
Ashton, M. C., Jackson, D. N., Paunonen, S. V., Helmes, E., & Rothstein,
M. G. (1995). The criterion validity of broad factor scales versus specific
facet scales. Journal of Research in Personality, 29, 432– 442. doi:
10.1006/jrpe.1995.1025
Barrett, P. (1986). Factor comparison: An examination of three methods.
Personality and Individual Differences, 7, 327–340. doi:10.1016/0191-
8869(86)90008-5
Bebbington, P. E., Dunn, G., Jenkins, R., Lewis, G., Brugha, T., Farrell,
M., & Meltzer, H. (1998). The influence of age and sex on the preva-
lence of depression conditions: Report from the National Survey of
Psychiatric Morbidity. Psychological Medicine, 28, 9 –19. doi:10.1017/
S0033291797006077
Beitel, M., & Cecero, J. J. (2003). Predicting psychological mindedness
from personality style and attachment security. Journal of Clinical
Psychology, 59, 163–172. doi:10.1002/jclp.10125
Benet-Martı´nez, V., & John, O. P. (1998). Los Cinco Grandes across
cultures and ethnic groups: Multitrait-multimethod analyses of the Big
Five in Spanish and English. Journal of Personality and Social Psychol-
ogy, 75, 729 –750. doi:10.1037/0022-3514.75.3.729
Bland, R. C., Newman, S. C., & Orn, H. (1988). Period prevalence of
psychiatric disorders in Edmonton. Acta Psychiatrica Scandinavica, 77,
33– 42. doi:10.1111/j.1600-0447.1988.tb08545.x
Branje, S. J. T., van Lieshout, C. F. M., & Gerris, J. R. M. (2007). Big Five
personality development in adolescence and adulthood. European Jour-
nal of Personality, 21, 45– 62. doi:10.1002/per.596
Buchanan, T., & Smith, J. L. (1999). Using the Internet for psychological
research: Personality testing on the World Wide Web. British Journal of
Psychology, 90, 125–144. doi:10.1348/000712699161189
Buhrmester, D. (1996). Need fulfillment, interpersonal competence, and
the developmental contexts of early adolescent friendship. In W. M.
Bukowski, A. F. Newcomb, & W. W. Hartup (Eds.), The company they
keep: Friendship in childhood and adolescence (pp. 158 –185). New
York, NY: Cambridge University Press.
344 SOTO, JOHN, GOSLING, AND POTTER

Buss, A. H., & Plomin, R. (1984). Temperament: Early developing per-
sonality traits. Hillsdale, NJ: Erlbaum.
Byrne, B. M., & Shavelson, R. J. (1996). On the structure of social
self-concept for pre-, early, and late adolescents: A test of the Shavelson,
Hubner, and Stanton (1976) model. Journal of Personality and Social
Psychology, 70, 599 – 613. doi:10.1037/0022-3514.70.3.599
Cameron, P. (1970). The generation gap: Which generation is believed
powerful vs. generational members’ self-appraisal of power? Develop-
mental Psychology, 3, 403– 404. doi:10.1037/h0030050
Carstensen, L. L. (1992). Social and emotional patterns in adulthood:
Support for socioemotional selectivity theory. Psychology and Aging, 7,
331–338. doi:10.1037/0882-7974.7.3.331
Caspi, A., Roberts, B. W., & Shiner, R. L. (2005). Personality develop-
ment: Stability and change. Annual Review of Psychology, 56, 453– 484.
doi:10.1146/annurev.psych.55.090902.141913
Centers for Disease Control and Prevention. (2008). Youth risk behavior
surveillance, United States, 2007. Washington, DC: U.S. Department of
Health and Human Services.
Chuah, S. C., Drasgow, F., & Roberts, B. W. (2006). Personality assess-
ment: Does the medium matter? No. Journal of Research in Personality,
40, 359 –376. doi:10.1016/j.jrp.2005.01.006
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
Colby, A., Kohlberg, L., Gibbs, J., & Lieberman, M. (1983). A longitudinal
study of moral judgment. Monographs of the Society for Research in
Child Development, 48, 1–124. doi:10.2307/1165935
Cole, D. A., Martin, J. M., Peeke, L. A., Seroczynski, A. D., & Fier, J.
(1999). Children’s over- and underestimation of academic competence:
A longitudinal study of gender differences, depression, and anxiety.
Child Development, 70, 459 – 473. doi:10.1111/1467-8624.00033
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality
Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI)
professional manual. Odessa, FL: Psychological Assessment Resources.
Costa, P. T., Jr., & McCrae, R. R. (1995). Domains and facets: Hierarchical
personality assessment using the Revised NEO Personality Inventory.
Journal of Personality Assessment, 64, 21–50. doi:10.1207/
s15327752jpa6401_2
De Fruyt, F., Bartels, M., van Leeuwen, K. G., De Clercq, B., Decuyper,
M., & Mervielde, I. (2006). Five types of personality continuity in
childhood and adolescence. Journal of Personality and Social Psychol-
ogy, 91, 538 –552. doi:10.1037/0022-3514.91.3.538
De Fruyt, F., Mervielde, I., Hoekstra, H. A., & Rolland, J. (2000). Assess-
ing adolescents’ personality with the NEO PI-R. Assessment, 7, 329 –
345. doi:10.1177/107319110000700403
Demetriou, A. (2000). Organization and development of self-
understanding and self-regulation: Toward a general theory. In M.
Boekaerts, P. R. Pintrich, & M. Zeidner (Eds.), Handbook of self-
regulation (pp. 209 –251). San Diego, CA: Academic Press. doi:
10.1016/B978-012109890-2/50036-6
Denissen, J. J. A., Geenen, R., van Aken, M. A. G., Gosling, S. D., &
Potter, J. (2008). Development and validation of a Dutch translation of
the Big Five Inventory (BFI). Journal of Personality Assessment, 90,
152–157. doi:10.1080/00223890701845229
DeYoung, C. G. (2006). Higher-order factors of the Big Five in a multi-
informant sample. Journal of Personality and Social Psychology, 91,
1138 –1151. doi:10.1037/0022-3514.91.6.1138
DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2002). Higher-order
factors of the Big Five predict conformity: Are there neuroses of health?
Personality and Individual Differences, 33, 533–552. doi:10.1016/
S0191-8869(01)00171-4
DeYoung, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets
and domains: 10 aspects of the Big Five. Journal of Personality and
Social Psychology, 93, 880 – 896. doi:10.1037/0022-3514.93.5.880
Donahue, E. M. (1994). Do children use the Big Five, too? Content and
structural form in personality description. Journal of Personality, 62,
45– 66. doi:10.1111/j.1467-6494.1994.tb00794.x
Donnellan, M. B., & Lucas, R. E. (2008). Age differences in the Big Five
across the life span: Evidence from two national samples. Psychology
and Aging, 23, 558 –566. doi:10.1037/a0012897
Dweck, C. S. (1986). Motivational processes affecting learning. American
Psychologist, 41, 1040 –1048. doi:10.1037/0003-066X.41.10.1040
Dweck, C. S. (1989). Motivation. In A. Lesgold & R. Glaser (Eds.),
Foundations for a psychology of education (pp. 87–136). Hillsdale, NJ:
Erlbaum.
Eaton, W. O. (1994). Temperament, development, and the five-factor
model: Lessons from activity level. In C. F. Halverson Jr. & G. A.
Kohnstamm (Eds.), The developing structure of temperament and per-
sonality from infancy to adulthood (pp. 173–187). Hillsdale, NJ: Erl-
baum.
Eisenberg, N., Carlo, G., Murphy, B., & van Court, P. (1995). Prosocial
development in late adolescence: A longitudinal study. Child Develop-
ment, 66, 1179 –1197. doi:10.2307/1131806
Eisenberg, N., & Morris, A. S. (2004). Moral cognitions and prosocial
responding in adolescence. In R. M. Lerner & L. Steinberg (Eds.),
Handbook of adolescent psychology (2nd ed., pp. 155–188). Hoboken,
NJ: Wiley.
Erikson, E. H. (1968). Identity: Youth and crisis. New York, NY: Norton.
Federal Bureau of Investigation. (2007). Uniform crime report for the
United States. Washington, DC: U.S. Government Printing Office.
Flavell, J. H., Miller, P. H., & Miller, S. A. (1993). Cognitive development
(3rd ed.). Englewood Cliffs, NJ: Prentice Hall.
Gestsdottir, S., & Lerner, R. M. (2008). Positive development in adoles-
cence: The development and role of intentional self-regulation. Human
Development, 51, 202–224. doi:10.1159/000135757
Gorchoff, S. M., John, O. P., & Helson, R. (2008). Contextualizing change
in marital satisfaction during middle age: An 18-year longitudinal study.
Psychological Science, 19, 1194 –1200. doi:10.1111/j.1467-
9280.2008.02222.x
Gosling, S. D., John, O. P., Craik, K. H., & Robins, R. W. (1998). Do
people know how they behave? Self-reported act frequencies compared
with on-line codings by observers. Journal of Personality and Social
Psychology, 74, 1337–1349. doi:10.1037/0022-3514.74.5.1337
Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief
measure of the Big-Five personality domains. Journal of Research in
Personality, 37, 504 –528. doi:10.1016/S0092-6566(03)00046-1
Gosling, S. D., Vazire, S., Srivastava, S., & John, O. P. (2004). Should we
trust Web-based studies? A comparative analysis of six preconceptions
about Internet questionnaires. American Psychologist, 59, 93–104. doi:
10.1037/0003-066X.59.2.93
Hale, W. W., III, Raaijmakers, Q., Muris, P., van Hoof, A., & Meeus, W.
(2008). Developmental trajectories of adolescent anxiety disorder symp-
toms: A 5-year prospective community study. Journal of the American
Academy of Child & Adolescent Psychiatry, 47, 556 –564. doi:10.1097/
CHI.0b013e3181676583
Hall, G. S. (1904). Adolescence. New York, NY: Appleton.
Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., &
Angell, K. E. (1998). Development of depression from preadolescence
to young adulthood: Emerging gender differences in a 10-year longitu-
dinal study. Journal of Abnormal Psychology, 107, 128 –140. doi:
10.1037/0021-843X.107.1.128
Harter, S. (1999). The construction of the self: A developmental perspec-
tive. New York, NY: Guilford Press.
Harter, S. (2006). The self. In W. Damon & R. M. Lerner (Series Eds.) &
N. Eisenberg (Vol. Ed.), Handbook of child psychology: Vol. 3. Social,
emotional, and personality development (6th ed., pp. 505–570). Hobo-
ken, NJ: Wiley.
Harter, S., & Monsour, A. (1992). Developmental analysis of conflict
345
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

caused by opposing attributes in the adolescent self-portrait. Develop-
mental Psychology, 28, 251–260. doi:10.1037/0012-1649.28.2.251
Heaven, P. C. L. (1996). Personality and self-reported delinquency: Anal-
ysis of the “Big Five” personality dimensions. Personality and Individ-
ual Differences, 20, 47–54. doi:10.1016/0191-8869(95)00136-T
Helson, R., Kwan, V. S. Y., John, O. P., & Jones, C. (2002). The growing
evidence for personality change in adulthood: Findings from research
with personality inventories. Journal of Research in Personality, 36,
287–306. doi:10.1016/S0092-6566(02)00010-7
Helson, R., & Soto, C. J. (2005). Up and down in middle age: Monotonic
and nonmonotonic changes in roles, status, and personality. Journal of
Personality and Social Psychology, 89, 194 –204. doi:10.1037/0022-
3514.89.2.194
Helson, R., Soto, C. J., & Cate, R. A. (2006). From young adulthood
through the middle ages. In D. K. Mroczek & T. D. Little (Eds.),
Handbook of personality development (pp. 337–352). Mahwah, NJ:
Erlbaum.
Herringer, L. G. (1998). Facets of extraversion related to life satisfaction.
Personality and Individual Differences, 24, 731–733. doi:10.1016/
S0191-8869(97)00194-3
Hill, J., & Lynch, M. (1983). The intensification of gender-related role
expectations during early adolescence. In J. Brooks-Gunn & A. Petersen
(Eds.), Girls at puberty (pp. 201–228). New York, NY: Plenum Press.
Hofstee, W. K. B., de Raad, B., & Goldberg, L. R. (1992). Integration of
the Big Five and circumplex approaches to trait structure. Journal of
Personality and Social Psychology, 63, 146 –163. doi:10.1037/0022-
3514.63.1.146
Hogan, R., & Roberts, B. W. (2004). A socioanalytic model of maturity.
Journal of Career Assessment, 12, 207–217. doi:10.1177/
1069072703255882
Hunter, F. T., & Youniss, J. (1982). Changes in functions of three relations
during adolescence. Developmental Psychology, 18, 806 – 811. doi:
10.1037/0012-1649.18.6.806
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from
childhood to adolescence. New York, NY: Basic Books. doi:10.1037/
10034-000
Jackson, C. (2002). Perceived legitimacy of parental authority and tobacco
and alcohol use during early adolescence. Journal of Adolescent Health,
31, 425– 432. doi:10.1016/S1054-139X(02)00398-1
Jackson, J. J., Bogg, T., Walton, K. E., Wood, D., Harms, P. D., Lodi-
Smith, J., . . . Roberts, B. W. (2009). Not all conscientiousness scales
change alike: A multimethod, multisample study of age differences in
the facets of conscientiousness. Journal of Personality and Social Psy-
chology, 96, 446 – 459. doi:10.1037/a0014156
John, O. P., Caspi, A., Robins, R. W., Moffitt, T. E., & Stouthamer-Loeber,
M. (1994). The “Little Five”: Exploring the nomological network of the
five-factor model of personality in adolescent boys. Child Development,
65, 160 –178. doi:10.2307/1131373
John, O. P., Donahue, E. M., & Kentle, R. L. (1991). The Big Five
Inventory—Versions 4a and 54. Berkeley: University of California,
Berkeley, Institute of Personality and Social Research.
John, O. P., & Gross, J. J. (2004). Healthy and unhealthy emotion regu-
lation: Personality processes, individual differences, and life span de-
velopment. Journal of Personality, 72, 1301–1334. doi:10.1111/j.1467-
6494.2004.00298.x
John, O. P., Naumann., L. P., & Soto, C. J. (2008). Paradigm shift to the
integrative Big Five trait taxonomy: History, measurement, and concep-
tual issues. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.),
Handbook of personality: Theory and research (3rd ed., pp. 114 –158).
New York, NY: Guilford Press.
John, O. P., & Srivastava, S. (1999). The Big-Five trait taxonomy: History,
measurement, and theoretical perspectives. In L. A. Pervin & O. P. John
(Eds.), Handbook of personality: Theory and research (2nd ed., pp.
102–139). New York, NY: Guilford Press.
Kapinus, C. A., & Johnson, M. P. (2003). The utility of family life cycle
as a theoretical and empirical tool: Commitment and family life-cycle
stage. Journal of Family Issues, 24, 155–184. doi:10.1177/
0192513X02250135
Klimstra, T. A., Hale, W. W., III, Raaijmakers, Q. A. W., Branje, S. J. T.,
& Meeus, W. H. J. (2009). Maturation of personality in adolescence.
Journal of Personality and Social Psychology, 96, 898 –912. doi:
10.1037/a0014746
Kuhlen, R. G. (1968). Developmental changes in motivation during the
adult years. In B. L. Neugarten (Ed.), Middle age and aging (pp.
115–136). Chicago, IL: University of Chicago Press.
Labouvie-Vief, G., Diehl, M., Jain, E., & Zhang, F. (2007). Six-year
change in affect optimization and affect complexity across the adult life
span: A further examination. Psychology and Aging, 22, 738 –751.
doi:10.1037/0882-7974.22.4.738
Lachman, M. E., Lewkowicz, C., Marcus, A., & Peng, Y. (1994). Images
of midlife development among young, middle-aged, and older adults.
Journal of Adult Development, 1, 201–211. doi:10.1007/BF02277581
Lamb, M. E., Chuang, S. S., Wessels, H., Broberg, A. G., & Hwang, C. P.
(2002). Emergence and construct validation of the Big Five factors in
early childhood: A longitudinal analysis of their ontogeny in Sweden.
Child Development, 73, 1517–1524. doi:10.1111/1467-8624.00487
Lang, F. R., & Carstensen, L. L. (2002). Time counts: Future time per-
spective, goals, and social relationships. Psychology and Aging, 17,
125–139. doi:10.1037/0882-7974.17.1.125
Larson, R. W., Richards, M. H., Moneta, G., Holmbeck, G., & Duckett, E.
(1996). Changes in adolescents’ daily interactions with their families
from ages 10 to 18: Disengagement and transformation. Developmental
Psychology, 32, 744 –754. doi:10.1037/0012-1649.32.4.744
Liebman, J. I., McKinley-Pace, M. J., Leonard, A. M., Sheesley, L. A.,
Gallant, C. L., Renkey, M. E., & Lehman, E. B. (2002). Cognitive and
psychosocial correlates of adults’ eyewitness accuracy and suggestibil-
ity. Personality and Individual Differences, 33, 49 – 66. doi:10.1016/
S0191-8869(01)00135-0
Loeber, R., & Farrington, D. P. (2000). Young children who commit crime:
Epidemiology, developmental origins, risk factors, early interventions,
and policy implications. Development and Psychopathology, 12, 737–
762. doi:10.1017/S0954579400004107
Marsh, H. W. (1989). Age and sex effects in multiple dimensions of
self-concept: Preadolescence to early adulthood. Journal of Educational
Psychology, 81, 417– 430. doi:10.1037/0022-0663.81.3.417
Marsh, H. W., & Ayotte, V. (2003). Do multiple dimensions of self-
concept become more differentiated with age? The differential distinc-
tiveness hypothesis. Journal of Educational Psychology, 95, 687–706.
doi:10.1037/0022-0663.95.4.687
Marshall, W. A., & Tanner, J. M. (1986). Puberty. In F. Falkner & J. M.
Tanner (Eds.), Human growth (Vol.11, pp. 171–209). New York, NY:
Plenum Press.
McCrae, R. R., & Costa, P. T. (1992). Discriminant validity of the NEO-
PIR facet scales. Educational and Psychological Measurement, 52,
229 –237. doi:10.1177/001316449205200128
McCrae, R. R., Costa, P. T., de Lima, M. P., Simoes, A., Ostendorf, F., &
Angleitner, A. (1999). Age differences in personality across the adult life
span: Parallels in five cultures. Developmental Psychology, 35, 466 –
477. doi:10.1037/0012-1649.35.2.466
McCrae, R. R., Costa, P. T., Ostendorf, F., Angleitner, A., Hrebickova, M.,
& Avia, M. D. (2000). Nature over nurture: Temperament, personality,
and life span development. Journal of Personality and Social Psychol-
ogy, 78, 173–186. doi:10.1037/0022-3514.78.1.173
McCrae, R. R., Costa, P. T., Jr., Terracciano, A., Parker, W. D., Mills, C. J.,
De Fruyt, F., & Mervielde, I. (2002). Personality trait development from
age 12 to age 18: Longitudinal, cross-sectional and cross-cultural anal-
yses. Journal of Personality and Social Psychology, 83, 1456 –1468.
doi:10.1037/0022-3514.83.6.1456
346 SOTO, JOHN, GOSLING, AND POTTER

McCrae, R. R., Herbst, J. H., & Costa, P. T., Jr. (2001). Effects of
acquiescence on personality factors structures. In R. Riemann, F. M.
Spinath, & F. Ostendorf (Eds.), Personality and temperament: Genetics,
evolution, and structure (pp. 217–231). Berlin, Germany: Pabst Science
Publishers.
McGraw, K. O., Tew, M. D., & Williams, J. E. (2000). The integrity of
Web-delivered experiments: Can you trust the data? Psychological Sci-
ence, 11, 502–506. doi:10.1111/1467-9280.00296
Measelle, J. R., John, O. P., Ablow, J. C., Cowan, P. A., & Cowan, C. P.
(2005). Can children provide coherent, stable, and valid self-reports on
the Big Five dimensions? A longitudinal study from ages 5 to 7. Journal
of Personality and Social Psychology, 89, 90 –106. doi:10.1037/0022-
3514.89.1.90
Montemayor, R., & Eisen, M. (1977). The development of self-conceptions
from childhood to adolescence. Developmental Psychology, 13, 314 –
319. doi:10.1037/0012-1649.13.4.314
Mulaik, S. A. (1972). The foundations of factor analysis. New York, NY:
McGraw-Hill.
Murphy, B. C., Eisenberg, N., Fabes, R. A., Shepard, S., & Guthrie, I. K.
(1999). Consistency and change in children’s emotionality and regula-
tion: A longitudinal study. Merrill-Palmer Quarterly, 45, 413– 444.
O’Connor, M. C., & Paunonen, S. V. (2007). Big Five personality predic-
tors of post-secondary academic performance. Personality and Individ-
ual Differences, 43, 971–990. doi:10.1016/j.paid.2007.03.017
Ozer, D. J., & Benet-Martı´nez, V. (2006). Personality and the prediction of
consequential outcomes. Annual Review of Psychology, 57, 401– 421.
doi:10.1146/annurev.psych.57.102904.190127
Patterson, G. R., & Stouthamer-Loeber, M. (1984). The correlation of
family management practices and delinquency. Child Development, 55,
1299 –1307.
Paunonen, S. V., & Ashton, M. C. (2001). Big Five factors and facets and
the prediction of behavior. Journal of Personality and Social Psychol-
ogy, 81, 524 –539. doi:10.1037/0022-3514.81.3.524
Pullmann, H., Raudsepp, L., & Allik, J. (2006). Stability and change in
adolescents’ personality: A longitudinal study. European Journal of
Personality, 20, 447– 459. doi:10.1002/per.611
Rammstedt, B., & John, O. P. (2007). Measuring personality in one minute
or less: A 10-item short version of the Big Five Inventory in English and
German. Journal of Research in Personality, 41, 203–212. doi:10.1016/
j.jrp.2006.02.001
Rice, K. G., & Mulkeen, P. (1995). Relationships with parents and peers:
A longitudinal study of adolescent intimacy. Journal of Adolescent
Research, 10, 338 –357. doi:10.1177/0743554895103003
Roberts, B. W. (1997). Plaster or plasticity: Are work experiences associ-
ated with personality change in women? Journal of Personality, 65,
205–232. doi:10.1111/j.1467-6494.1997.tb00953.x
Roberts, B. W., Chernyshenko, O. S., Stark, S., & Goldberg, L. R. (2005).
The structure of Conscientiousness: An empirical investigation based on
seven major personality questionnaires. Personnel Psychology, 58, 103–
139. doi:10.1111/j.1744-6570.2005.00301.x
Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of
mean-level change in personality traits across the life course: A meta-
analysis of longitudinal studies. Psychological Bulletin, 132, 1–25. doi:
10.1037/0033-2909.132.1.1
Roberts, B. W., & Wood, D. (2006). Personality development in the
context of the neo-socioanalytic model of personality. In D. K. Mroczek
& T. D. Little (Eds.), Handbook of personality development (pp. 11–39).
Mahwah, NJ: Erlbaum.
Robins, R. W., Tracy, J. L., Trzesniewski, K., Potter, J., & Gosling, S. D.
(2001). Personality correlates of self-esteem. Journal of Research in
Personality, 35, 463– 482. doi:10.1006/jrpe.2001.2324
Robins, R. W., Trzesniewski, K. H., Tracy, J. L., Gosling, S. D., & Potter,
J. (2002). Global self-esteem across the life span. Psychology and Aging,
17, 423– 434. doi:10.1037/0882-7974.17.3.423
Rollins, B. C., & Cannon, K. L. (1974). Marital satisfaction over the famil
life cycle: A reevaluation. Journal of Marriage and the Family, 31,
271–282. doi:10.2307/351153
Rothbart, M. K., Ahadi, S. A., & Evans, D. E. (2000). Temperament and
personality: Origins and outcomes. Journal of Personality and Social
Psychology, 78, 122–135. doi:10.1037/0022-3514.78.1.122
Ruiz, M. A., Pincus, A. L., & Dickinson, K. A. (2003). NEO PI-R
predictors of alcohol use and alcohol-related problems. Journal of Per-
sonality Assessment, 81, 226 –236. doi:10.1207/S15327752JPA8103_05
Samuel, D. B., & Widiger, T. A. (2008). A meta-analytic review of the
relationships between the five-factor model and DSM-IV-TR personality
disorders: A facet level analysis. Clinical Psychology Review, 28, 1326 –
1342. doi:10.1016/j.cpr.2008.07.002
Saucier, G., & Ostendorf, F. (1999). Hierarchical subcomponents of the
Big Five personality factors: A cross-language replication. Journal of
Personality and Social Psychology, 76, 613– 627. doi:10.1037/0022-
3514.76.4.613
Schaie, K. W. (1977). Quasi-experimental research designs in the psychol-
ogy of aging. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the
psychology of aging (pp. 39 –58). New York, NY: Van Nostrand Rein-
hold.
Skitka, L. J., & Sargis, E. G. (2006). The Internet as psychological
laboratory. Annual Review of Psychology, 57, 529 –555. doi:10.1146/
annurev.psych.57.102904.190048
Smetana, J. G. (1995). Parenting styles and conceptions of parental au-
thority during adolescence. Child Development, 66, 299 –316. doi:
10.2307/1131579
Smetana, J. G. (2000). Middle-class African American adolescents’ and
parents’ conceptions of parental authority and parenting practices: A
longitudinal investigation. Child Development, 71, 1672–1686. doi:
10.1111/1467-8624.00257
Soto, C. J., & John, O. P. (2009a). Ten facet scales for the Big Five
Inventory: Convergence with NEO PI-R facets, self-peer agreement, and
discriminant validity. Journal of Research in Personality, 43, 84 –90.
doi:10.1016/j.jrp.2008.10.002
Soto, C. J., & John, O. P. (2009b). Using the California Psychological
Inventory to assess the Big Five personality domains: A hierarchical
approach. Journal of Research in Personality, 43, 25–38. doi:10.1016/
j.jrp.2008.10.005
Soto, C. J., John, O. P., Gosling, S. D., & Potter, J. (2008). The develop-
mental psychometrics of Big Five self-reports: Acquiescence, factor
structure, coherence, and differentiation from ages 10 to 20. Journal of
Personality and Social Psychology, 94, 718 –737. doi:10.1037/0022-
3514.94.4.718
Srivastava, S., John, O. P., Gosling, S. D., & Potter, J. (2003). Develop-
ment of personality in early and middle age: Set like plaster or persistent
change? Journal of Personality and Social Psychology, 84, 1041–1053.
doi:10.1037/0022-3514.84.5.1041
Steinberg, L. (1986). Latchkey children and susceptibility to peer pressure:
An ecological analysis. Developmental Psychology, 22, 433– 439. doi:
10.1037/0012-1649.22.4.433
Steinberg, L., & Silverberg, S. B. (1986). The vicissitudes of autonomy in
early adolescence. Child Development, 57, 841– 851. doi:10.2307/
1130361
Stice, E., & Bearman, S. K. (2001). Body-image and eating disturbances
prospectively predict increases in depressive symptoms in adolescent
girls: A growth curve analysis. Developmental Psychology, 37, 597–
607. doi:10.1037/0012-1649.37.5.597
Stice, E., Hayward, C., Cameron, R. P., Killen, J. D., & Taylor, C. B.
(2000). Body-image and eating disturbances predict onset of depression
among female adolescents: A longitudinal study. Journal of Abnormal
Psychology, 109, 438 – 444. doi:10.1037/0021-843X.109.3.438
Ten Berge, J. M. F. (1999). A legitimate case of component analysis of
ipsative measures, and partialling the mean as an alternative to ipsati-
347
AGE DIFFERENCES IN BIG FIVE DOMAINS AND FACETS

zation. Multivariate Behavioral Research, 34, 89 –102. doi:10.1207/
s15327906mbr3401_4
Terracciano, A., McCrae, R. R., Brant, L. J., & Costa, P. T., Jr. (2005).
Hierarchical linear modeling analyses of the NEO-PI-R scale in the
Baltimore longitudinal study of aging. Psychology and Aging, 20, 493–
506. doi:10.1037/0882-7974.20.3.493
Thomas, A., & Chess, S. (1977). Temperament and development. New
York, NY: Brunner/Mazel.
U.S. Census Bureau. (2001). Number, timing, and duration of marriages
and divorces: 2001. Washington, DC: U.S. Department of Commerce.
U.S. Census Bureau. (2006). School enrollment in the United States: 2006.
Washington, DC: U.S. Department of Commerce.
Walker, C. (1977). Some variations in marital satisfaction. In R. Chester &
J. Peel (Eds.), Equalities and inequalities in family life (pp. 127–139).
London, England: Academic Press.
Wichstrøm, L. (1999). The emergence of gender difference in depressed
mood during adolescence: The role of intensified gender socialization.
Developmental Psychology, 35, 232–245. doi:10.1037/0012-
1649.35.1.232
Received October 2, 2009
Revision received July 7, 2010
Accepted August 17, 2010 䡲
348 SOTO, JOHN, GOSLING, AND POTTER