11

Personality Development in
Adolescence and Young Adulthood
Theo A. Klimstra, Jeroen Borghuis
and Wiebke Bleidorn

The phrase ‘personality development’ would importance of studying how multiple charac-
have looked like a ‘contradictio in terminis’ teristics develop together and how their con-
to many researchers until approximately 30 figuration within an individual may change.
years ago, as personality was widely believed Therefore, in this chapter, we will refer to
to be a stable construct. In contrast, it is now changes in both single personality traits and
widely acknowledged that there is much in personality profiles.
change in personality throughout the lifes- A further consideration in studying per-
pan, with the most pronounced changes sonality is that there are different kinds of
occurring during adolescence and young traits representing different levels and dif-
adulthood. This chapter focuses on these key ferent functions in a broader personality
periods of personality development, roughly system. Within such a system, dispositions
spanning ages 12 to 40 years. encompassing a wide range of behaviors and
Personality is a broad concept and was cognitions, such as the Big Five personal-
originally defined as ‘the dynamic organiza- ity traits, are usually considered the core of
tion within the individual of those psycho- personality (e.g., Asendorpf and van Aken,
physical systems that determine his unique 2003; McAdams and Olson, 2010; McCrae
adjustments to his environment’ (Allport, and Costa, 2008). However, personality also
1937, p. 48). This definition emphasized that encompasses more process-like constructs
personality is not static and thus may change, like attitudes, goals, motives, and self-related
and therefore already pointed to the impor- constructs (e.g., narrative identity, self-
tance of studying personality development. In esteem). In addition, Big Five trait domains
addition, the definition refers to personality are defined by more specific traits referred
as a configuration of attributes within a par- to as facets. For example, the trait domain
ticular individual, and therefore points to the neuroticism can be divided into facets of
182 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

anxiety, angry hostility, depression, self- gross motor skills, such as running. Thus, there
consciousness, impulsiveness, and vulner- may be heterotypic continuity, in the sense that
ability (McCrae et al., 2005). In the present the same underlying asset (i.e., gross motor
chapter, the main focus will be on core char- skills) may manifest itself differently.
acteristics, typically summarized in the Big This heterotypic continuity is also an issue
Five personality trait domains, but we will in the study of personality traits from child-
briefly touch upon facet traits. hood through adolescence into adulthood, in
Our chapter is divided into four sections. the sense that stable individual differences
In the first section, we will discuss develop- are typically examined along temperamental
mental trends in personality traits by refer- dimensions in childhood and along personal-
ring to findings regarding structural stability, ity traits in adulthood. This raises the question
rank-order stability, mean-level change, and when one should stop studying temperament
individual-level change. The second section and start studying personality traits, and thus
is focused on the correlates of personal- whether studying personality traits in adoles-
ity development. We will distinguish broad cence is appropriate. Based on previous work
mechanisms and narrow mechanisms in in which the overlap between temperament
this section. The third section summarizes and personality dimensions was examined
findings obtained with person-centered (e.g., Caspi and Shiner, 2006; De Pauw et al.,
approaches (i.e., approaches focused on pro- 2009), Soto and John (2014) developed an
file stability and typological approaches). integrative model called the Little Six. This
Finally, the fourth section summarizes the model adds an activity factor to the Big Five
states of research on adolescent and young to provide a more encompassing view on
adult development and provides potential important individual differences from child-
future directions for this field of research. hood to late adolescence, thereby breaking
the trend to only consider Big Five traits
when studying personality trait development.
Soto and John (2014) also show that the
DEVELOPMENTAL TRENDS IN meaning of traits may change with age. For
PERSONALITY TRAITS FROM example, they showed that conscientious-
ADOLESCENCE THROUGH YOUNG ness was increasingly defined by socially
ADULTHOOD responsible versus deceitful behavior. This
finding points out that it is crucial to for-
mally test for such shifts in meaning in the
Structural Stability
study of adolescent and young adult person-
A key consideration before studying develop- ality trait development. Such tests, referred
mental trends in any construct is whether the to as measurement invariance tests, are
construct itself is relevant and valid for indi- increasingly often conducted (e.g., Lucas
viduals of all ages included in your design. and Donnellan, 2011), but are still not com-
Much like it is useful to study crawling in mon practice. Given that such tests were
infants, but not in adolescents, psychological usually not conducted in studies that were
constructs may lose or gain relevance. summarized in influential meta-analyses
Constructs that lose direct relevance are defi- (e.g., Roberts et al., 2006), much of what
nitely not meaningless for predicting future we appear to ‘know’ about personality trait
behavior, as certain characteristics may just development might not be entirely accurate.
change in the way they manifest themselves. That is, what appear to be changes in lev-
To stick with the crawling example, individual els of an entire trait domain might actually
differences in this behavior may predict indi- be due to changes in the mean-level scores
vidual differences in other abilities related to of only some parts (i.e., items or facets) of
 Personality Development in Adolescence and Young Adulthood 183

that trait domain. The mean levels of other that trait stability increases from childhood
parts of that trait might remain the same or through middle adulthood has been referred
even decrease. Thus, we argue that testing to as the cumulative continuity principle of
for measurement invariance is crucial, espe- personality development (Roberts and
cially in research on personality trait change Mroczek, 2008).
focusing on developmental periods charac- However, the stability and change of per-
terized by rapid changes in cognitive capac- sonality traits varies across adolescence and
ity and social skills. Such tests should feature young adulthood. Existing meta-analyses
prominently, and not be buried away in one have aggregated rank-order stability find-
sentence in the method section with results ings across relatively broad age categories
being ‘available from the first author upon (e.g., ages 12 to 18), and therefore less is
request’ (e.g., Klimstra, Hale et al., 2012). known about differences in rank-order sta-
bility across narrower age categories. Two
studies that attempted to address this gap
found different results with respect to sta-
Rank-Order Stability
bility and change in the one-year rank-order
Rank-order stability refers to the mainte- stability of Big Five traits in adolescence
nance of individuals’ relative standing on a and young adulthood. One study found
trait dimension within a population over that rank-order stability increased through-
time. Rank-order stability is commonly out adolescence (Klimstra et al., 2009).
assessed by means of a test–retest correlation However, another study found that rank-
or a stability coefficient in a path or struc- order stability only increased from early
tural equation model. Previous research through middle adolescence, and remained
clearly shows that not all people change to stable in late adolescence and early adult-
the same extent and in the same direction. hood (Borghuis et al., 2017).
Therefore, rank-order stability of personality A handful of studies have also examined
trait dimensions is by no means perfect. rank-order stability in facets. A study using
Although rank-order stability tends to parent reports found that in early adolescence,
decrease as intervals between assessments rank-order stability was similar for facets and
increase, there is even substantial rank-order domains (De Fruyt et al., 2006). This result
stability over decades (Fraley and Roberts, was more or less confirmed in more recent
2005). This suggests that some portion of the work, which found high rank-order stability
between-person variance in personality traits of facets in mother-reports on children and
is truly stable. Indeed, research has shown adolescents, but did not compare these coef-
that some personality or temperament traits ficients to those for domains (de Haan et al.,
in childhood show small, positive correla- 2017). One study examining a sample that
tions with personality traits in adulthood was heterogeneous in terms of age, but also
(Asendorpf et al., 2008; Block, 1993; Kagan included young adults, found that rank-order
and Moss, 1962; Shiner et al., 2003). In addi- stability was only slightly lower for facets
tion, research on lifespan personality trait than for trait domains (Bleidorn et al., 2009).
development shows that 6-year rank-order A study specifically focused on young adult
stability of personality or temperament traits college students confirmed in two samples
is moderately high in preschool years that rank-order stability of facets is indeed as
(r ≈ .50), increases until middle adulthood high as rank-order stability of trait domains
(r ≈ .70), and then plateaus (Anusic and (Klimstra et al., 2016). In general, individual
Schimmack, 2016; Bazana and Stelmack, differences in personality facets seem to be
2004; Briley and Tucker-Drob, 2014; Roberts about as stable as individual differences in
and DelVecchio, 2000). The robust finding personality trait domains.
184 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

Mechanisms Explaining Change One mechanism that may drive increas-

in Rank-Order Stability ing personality stability with age is increas-
ing commitment to, and thus maintenance
Relatively little is known about the exact of, an identity (Roberts and Caspi, 2003;
mechanisms that drive age-graded increases Roberts et al., 2008). Identity maintenance
in the rank-order stability of personality has been argued to stimulate personality sta-
traits. Nevertheless, meta-analytic research bility because an identity provides clear refer-
has provided important hints about the nature ence points for making life decisions (Caspi
of these mechanisms. First, Briley and et al., 2005), facilitates ‘strategic’ informa-
Tucker-Drob (2014) conducted a meta- tion selection and processing, resulting in the
analysis of genetically informed longitudinal maintenance of consistent self-perceptions
(mostly twin) studies to illuminate the envi- (Roberts and Caspi, 2003), and results in
ronmental and genetic contributions to per- consistent role demands and social expecta-
sonality continuity across the lifespan. They tions (Roberts et al., 2008). This reasoning
found that both genetic and environmental is consistent with identity theory, as defined
influences on personality increase in stability by Erikson (1950, 1968) and Marcia (1966).
with age. However, because the relative However, despite a growing number of studies
influence of genetic effects on personality examining linkages between personality traits
traits has been found to decrease with age, or types on the one hand and identity forma-
the environmental contributions to stability tion processes on the other hand (e.g., Hatano
were found to increase with age. Therefore, et al., 2017; Luyckx et al., 2014), the role of
increasing rank-order stability of personality identity formation in predicting the stability
with age is largely accounted for by the of personality still awaits empirical scrutiny.
increasingly stable environmental effects,
rather than by the increasingly stable genetic
effects (see also Bleidorn et al., 2014).
Mean-Level Change
Second, a meta-analysis by Anusic and
Schimmack (2016) focused on the extent Mean-level change refers to the change in the
to which personality traits, across the life­ average trait levels of a population over time.
span, are influenced by stable effects (i.e., Mean-level change is conceptually independ-
all influences on personality traits that never ent from rank-order change because rank
change) and by changing effects (i.e., all orders can be perfectly maintained in groups
influences on personality traits that change that change on average and rank orders might
over time). Examples of the former are per- be completely reordered in groups that do not
manent internal biological influences and change in their average trait level (Roberts
environmental experiences with permanent et al., 2006).
effects; examples of the latter are temporary Previous research has provided a clear pic-
effects of genes and social role transitions. ture of how mean levels of broad personality
Importantly, they found that the increase in traits change. A meta-analysis found that, in
the rank-order stability of personality traits young adulthood, individuals show signifi-
was attributable to an increase in the rela- cant mean-level increases in the Big Five traits
tive influence of stable factors on personal- agreeableness, emotional stability, and consci-
ity, rather than to an increase in the stability entiousness (Roberts et al., 2006). These mean-
of change factors. Taken together, these two level increases have been referred to as the
meta-analyses suggest that the increasing maturity principle of personality development
stability of personality traits is driven by (Roberts and Mroczek, 2008). That is because
increasingly stable environmental influences being agreeable, conscientious, and emotion-
on personality traits. ally stable corresponds closely to definitions of
 Personality Development in Adolescence and Young Adulthood 185

maturity that emphasize functioning in society studies (Jackson et al., 2009; McCrae et al.,
and social relationships, such as being liked, 2004; Soto et al., 2011). This research pro-
respected, and admired (Roberts and Mroczek, vided more evidence for diverging mean-
2008; Roberts et al., 2008). level trends for facets belonging to the same
In contrast with the period of young adult- domain. The most consistent patterns were
hood, there is convincing evidence that, in found for extraversion and conscientiousness.
adolescence, the mean levels of most Big Five Specifically, across different extraversion
trait domains tend to first decrease and then facets, the mean-level change was more pro-
increase (i.e., U-shaped change). Specifically, nounced for the facets of excitement seeking
both a recent meta-analysis (Denissen et al., and positive emotions than for facets like activ-
2013) and a large-scale cross-sectional study ity and assertiveness. For conscientiousness,
(Soto et al., 2011) found evidence for tempo- mean levels of dutifulness and self-­discipline
rary mean-level decreases in conscientious- facets increased more than mean levels of
ness, openness, extraversion, and emotional orderliness-related facets. Longitudinal stud-
stability (among girls) in early adolescence, ies distinguishing facets in early adult samples
whereas they found evidence for mean-level (Klimstra et al., 2016; Soto and John, 2012)
increases in conscientiousness, emotional yielded relatively similar findings. Notably,
stability, and openness in late adolescence Klimstra et al. (2016) included college sam-
and early adulthood. In addition, though ples from the US and Belgium and found
contrary to Denissen et al. (2013), Soto et al. subtle differences between those samples. In
(2011) also found evidence for U-shaped particular, students in the US changed toward
change in agreeableness. a more laid-back and intellectually curious
These mean-level trends at the trait-domain profile whereas Belgian students changed
level do not necessarily directly translate into toward a more anxious profile. These find-
findings at the facet level. That is, De Fruyt ings may well be attributable to contextual
et al. (2006) showed that in early adolescence differences in how competitive selection pro-
mean-level changes in facets belonging to the cedures are and how much value a particular
same domain (e.g., the self-confidence and college degree has for finding a job. Overall,
anxiety facets of the neuroticism domain) did the pattern of findings across studies suggests
not always correspond. For example, mean that considering facets in addition to broad
levels of self-confidence did not change sig- trait domains may provide a few nuances that
nificantly whereas anxiety decreased. De would be overlooked if only the domain level
Haan et al. (2017) showed that differences in of personality were to be considered.
mean-level changes between facets belong-
ing to the same trait domain are typically
not that large, but do seem to become larger Individual-Level Change
as individuals grow older (i.e., enter middle
adolescence). A large-scale cross-sectional Perhaps even more important is to note that the
study (Soto et al., 2011) had similar find- aforementioned research on personality trait
ings but, within the neuroticism domain, they change has been concerned more with describ-
did show that mean levels of anxiety in late ing mean-level trends of personality traits than
adolescent females remained relatively sta- with accounting for individual differences in
ble whereas their mean levels of depression development. This holds especially for the
dropped. Thus, in adolescence there is only period of adolescence. Mean-level change
some evidence for different mean-level trends coefficients summarize the average develop-
for facets belonging to the same domain. ment in a population and can therefore not be
Several studies on young adults exam- used to make inferences about development at
ined mean-level age trends in cross-sectional the individual level. Considering the individual
186 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

Figure 11.1 Graphical representation of the magnitude of individual differences in boys’ person-
ality trait change in conscientiousness; the regression curves represent development of individu-
als across age. Regression curves (N = 500) were drawn from a simulated multivariate normal
distribution based on the parameter estimates of a latent growth curve model (N = 2,230).
Source: adapted from Borghuis et al. (2017)

level is important, as adolescents and young (cf. Klimstra et al., 2013a). We will first dis-
adults differ substantially with respect to their cuss and define these broad mechanisms, and
personality trajectories. As Figure 11.1 (adapted then move to the more specific mechanisms.
from Borghuis et al., 2017) illustrates with
respect to boys’ conscientiousness, some ado-
lescents increase, some decrease, and others Broad Mechanisms
remain stable in their personality trait level.
Attempts at identifying broad predictors of
This suggests that mean-level trajectories do
personality change in adolescence and young
not always provide accurate descriptions for
adulthood have been guided by several theo-
individuals’ personality change. A deeper
ries and principles (for reviews, see Bleidorn,
understanding of personality development
2015; Specht et al., 2014). Two of these are
requires moving beyond stability and change at
particularly prominent, and these are five-
the population level in order to understand and
factor theory (FFT, McCrae and Costa, 2008)
account for individual variation in developmen-
and the social investment principle (Roberts
tal trajectories (Asendorpf, 1992; Lönnqvist
et al., 2005). FFT posits that personality traits
et al., 2008; Roberts and Mroczek, 2008).
are mainly influenced by evolved, genetically
controlled biological processes, and that life
experiences play only a negligible role (unless
CORRELATES OF PERSONALITY TRAIT these alter gene expression). The Social
DEVELOPMENT Investment Principle holds that social role
transitions (e.g., marriage, becoming a parent,
In examining predictors and correlates of getting a ‘real’ job) do play a significant role
individual differences in change, both broad in personality maturation, especially if one
and narrow mechanisms can be distinguished psychologically commits to a new social role.
 Personality Development in Adolescence and Young Adulthood 187

In order to examine whether personality matter (Turkheimer and Waldron, 2000;

trait development is driven by broad mech- Vinkhuyzen et al., 2012). Despite this,
anisms driving change in multiple traits or Briley and Tucker-Drob (2014) have drawn
narrow mechanisms driving change in spe- the encouraging conclusion that a substan-
cific traits, several studies have addressed tial amount of environmental variance is not
correlated change among personality traits attributable to random measurement error
(for a review, see Allemand and Martin, and is stable across time, and should there-
2016). Research on this phenomenon cov- fore, in principle, be measurable.
ering adolescence and young adulthood Somewhat consistent with the proposi-
(Klimstra et al., 2013a) found that there tions of FFT, several cross-cultural studies
is indeed evidence for correlated change have found relatively similar mean-level
among specific pairs of Big Five traits. personality changes across different cul-
Specific pairs of traits that have been tures with different societal structures,
hypothesized to be affected by the same neu- different political systems, and different his-
robiological system (i.e., the serotonergic torical backgrounds (Bleidorn et al., 2013;
or dopaminergic system) or same develop- McCrae and Terracciano, 2005). However,
mental principle (i.e., the social investment Bleidorn et al. (2013) also found substantial
principle) produced the highest correlated cultural differences in the timing and mag-
change estimates. Especially, changes in nitude of mean-level age trends in the Big
extraversion and openness, which are the Five. Providing some evidence for the rel-
two traits that have been associated with evance of environmental influences, these
the dopaminergic system (DeYoung et al., cultural differences in personality change
2002), were strongly correlated with each were related to cultural differences in the
other. Thus, there was some evidence for timing of social role transitions (e.g., gradu-
broad mechanisms affecting personality ation from school, marriage, parenthood).
trait development. However, this approach Thus, there appears to be evidence for both
may contribute to uncovering whether or genetic effects and environmental effects
not broad mechanisms are involved and on age-related differences in personality
may also provide hints regarding what trait development (for a review, also see
these mechanisms are, but it does not actu- Bleidorn, 2015).
ally test mechanisms triggering personality Although FFT and the social investment
trait development. principle have also been used to explain
One method that explicitly tests concrete mean-level changes in adolescent personal-
broad mechanisms, especially those related ity traits, little is known about the mecha-
to FFT, is longitudinal behavioral genetic nisms that drive these changes. Soto and
research. Such studies can help to quantify Tackett (2015) speculated that the temporal
the proportion of individual differences dips in personality maturity (as defined by
that are due to genetic and environmental conscientiousness, agreeableness, and emo-
influences. Reviews suggest that the rela- tional stability) during early adolescence are
tive influence of genetic influences on per- related to the rapid and significant biologi-
sonality traits (i.e., heritability) decreases cal, psychological, and social changes that
from childhood to old adulthood. By impli- characterize this period. Other possible rea-
cation, this suggests that environmental sons for the U-shaped mean-level changes
influences on individual differences in per- are that adolescents temporarily adhere to
sonality traits increase with age (Bleidorn less mature peer norms (Moffitt, 1993) and
et al., 2014; Briley and Tucker-Drob, 2014). that they experience initial difficulties in
Yet, it remains a mostly unanswered ques- adjusting to increasingly mature expecta-
tion which genes and environmental factors tions (Denissen et al., 2013).
188 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

Narrow Mechanisms: Correlates regarding directionality of effects between

Within the Broader Personality identity processes and personality traits.
Trait System That is, infant and child behavior is thought
to be largely driven by basic tendencies (i.e.,
Instead of broad mechanisms that are hypoth- personality traits). In late childhood, motiva-
esized to affect multiple traits, narrow mech- tions and goals are added to the equation, and
anisms are supposed to affect single traits. behavior becomes more goal-directed. From
Moreover, narrower mechanisms do not refer late adolescence onwards, individuals start
to broad biological or sociological processes, behaving in a way that is in line with the way
but rather to specific variables such as rela- they perceive themselves. Thus, their (narra-
tionship satisfaction, self-esteem, or anxiety tive) identity starts to play an important role
symptoms. When examining narrow mecha- in their behavior. At first, these self-views are
nisms of personality trait change, one direc- largely built upon the way individuals typi-
tion is to consider other variables within the cally behaved up to that point, making traits
broader personality system. Several theories as well as goals and motivations better pre-
emphasize the importance of this. For exam- dictors of identity at that point. Once identi-
ple, the social investment principle, which is ties become solidified, they also start to affect
often interpreted as predicting that it is ‘just’ traits. Typically, this should happen from
transitions into social roles of adult life that young adulthood onwards. Hence, McAdams
trigger personality trait change, emphasizes would predict that personality traits are good
the importance of committing to these social predictors of identity processes throughout
roles. The commitment processes referred to development, whereas identity processes
for the social investment principle are identi- only become better predictors of personality
cal to the commitment processes that feature traits in young adulthood.
in Erikson’s (1950, 1968) developmental Longitudinal research on the linkages of
theory (Lodi-Smith and Roberts, 2007). identity formation processes with personal-
Hence, such processes feature in models of ity traits is rather scarce. Overall, these stud-
identity formation (e.g., Crocetti et al., 2008; ies suggest that personality traits are often
Luyckx et al., 2008). These models also better predictors of identity processes than
emphasize the importance of exploration the other way around (e.g., Luyckx et al.,
processes. The social investment principle 2014). In adolescence, identity processes
would predict that the aforementioned iden- rarely predict changes in personality traits
tity processes, but especially commitment, (Luyckx et al., 2014), but this is more com-
would predict changes in personality traits. mon in young adults (Luyckx et al., 2006),
FFT (McCrae and Costa, 2008) and the confirming McAdams’ theorizing. However,
core versus surface trait model (Asendorpf recent research on Japanese early and mid-
and van Aken, 2003) emphasize a distinction dle adolescents questions this conclusion
between basic tendencies (i.e., Big Five per- by demonstrating bidirectional identity–­
sonality traits) and more specific ‘character- personality trait linkages from early adoles-
istic’ adaptations such as identity formation cence onwards (Hatano et al., 2017). This
processes. This model would predict that Big latter study also considered personality facets
Five traits and identity processes are associ- underlying broader trait domains and showed
ated, and that the effects of Big Five traits on that facets belonging to the same trait domain
identity processes are larger than those in the were often differentially associated with
inverse direction. identity processes.
McAdams’ multilayered model (e.g., Other studies provided more specificity on
McAdams and Olson, 2010) presents a sort of the identity side by considering life-domain-
compromise between the previous standpoint specific identity processes. These studies also
 Personality Development in Adolescence and Young Adulthood 189

found few effects of these processes on per- research can also provide additional insight
sonality traits, but the few effects of identity in the personality trait–self-esteem link.
on personality that were found tended to be in There are few longitudinal studies exam-
line with the social investment principle. That ining linkages between personality traits and
is, educational and relational commitment pre- self-esteem. A study on adolescents showed
dicted relative increases in emotional stability that neuroticism mattered in how much not
and conscientiousness, respectively (Klimstra, meeting expectancies regarding peer accept-
Luyckx et al., 2012; Kimstra et al., 2013b). ance affected self-esteem development
Overall, these studies show that linkages (Poorthuis et al., 2014). That is, adolescents
between identity processes and personality who were neurotic and saw their expectan-
traits are complex. There appears to be some cies regarding peer acceptance not being
evidence for each of the theoretical perspec- met experienced a decrease in self-esteem.
tives on these linkages, but none of the evi- Furthermore, the importance of adolescent
dence is overwhelming. Yet, the development personality development was underscored, as
of identity processes seems to be associated a more favorable personality profile in ado-
with the development of adolescent and lescence was predictive of higher self-esteem
young adult personality traits. later in adulthood (Blatný et al., 2015).
Another variable in the self-system that Research on young adults (Klimstra et al.,
has been associated with personality trait 2016; Lönnqvist et al., 2009; Wagner et al.,
development is self-esteem. Self-esteem 2013) provides some evidence for high lev-
reflects an overall evaluation of one’s worth els of favorable personality traits (especially
and positive versus negative feelings about emotional stability) predicting high levels of,
the self (e.g., Orth et al., 2008). Large-scale or even increases in, self-esteem. However,
cross-sectional studies research in young another study (Erol and Orth, 2011) found
adults (e.g., Erdle et al., 2009; Robins et al., that Big Five traits were related to self-
2001b, 2001c) and adolescents (e.g., Mlacic esteem concurrently but did not predict
et al., 2007; Vaszonyi et al., 2015) generally change in self-esteem. Thus, the predictive
found strong negative associations of self- role of personality traits for the develop-
esteem with extraversion and conscientious- ment of self-esteem is still somewhat unclear.
ness, and strong negative associations with Furthermore, a recent study showed that self-
neuroticism. In addition to self-esteem levels, esteem can also affect the development of
self-esteem stability is also associated with personality traits (Klimstra et al., 2016), as
Big Five personality trait domains in young high self-esteem predicted decreases in facets
adulthood. Especially high emotional sta- of neuroticism in two independent samples.
bility, agreeableness, and conscientiousness These studies show that the development
are associated with high self-esteem stabil- of self-related processes and core personal-
ity (Zeigler-Hill et al., 2015). In fact, young ity trait domains are closely intertwined.
adults who had a high level of self-esteem However, the development of personality
and who were also stable had the most opti- domains is obviously also linked to the devel-
mal personality profile, characterized by opment of variables that are less directly
high emotional stability, agreeableness, and related to the self.
conscientiousness. Individuals with stable
low self-esteem had low levels of openness.
This suggests that considering aspects of Narrow Mechanisms: Correlates
self-esteem beyond the level (e.g., also con- Outside of the Personality System
sidering stability of self-esteem) can produce
further insights into its relation with person- Adolescent and young adult personality
ality trait domains. In addition, longitudinal development affects, and is affected by,
190 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

changes in a wide range of other phenomena. examining peer support in addition to paren-
Especially, interpersonal relationships, psy- tal support suggested that conscientiousness
chopathology symptoms, and academic was the only trait domain that prospectively
experiences have received the attention of predicted support from both the father and
various researchers. the mother (Asendorpf and van Aken, 2003).
Studies that consider interpersonal rela- However, conscientiousness did not predict
tionships in adolescence examined asso- peer support (but see Jensen-Campbell and
ciations with parenting. In a study covering Malcolm, 2007). For that, especially, extra-
childhood and adolescence, parenting had version and agreeableness seemed impor-
few effects on the child’s personality (van tant trait domains (Asendorpf and van Aken,
den Akker et al., 2014). The child’s per- 2003; Jensen-Campbell et al., 2002).
sonality did have more of an effect on par- Toward late adolescence and into young
enting, with agreeableness and openness adulthood, romantic relationships become
eliciting positive parenting, whereas extra- more important and more common.
version seemed a mixed blessing. That is, Personality traits are known to predict rela-
extraversion triggered warm parenting, but tionship development (e.g., Ahmetoglu et al.,
also more overreactivity from parents, pos- 2010; Lehnart and Neyer, 2006), including
sibly because extraverted children are charm- who is more likely to get into a relationship
ing yet expressive. Their expressiveness may in the first place (e.g., Neyer and Lehnart,
cause more conflicts with parents. Most con- 2007). That is, more emotionally stable, extra-
vincing was the evidence for parallel devel- verted, and conscientiousness young adults
opment between parenting and personality, seemed more likely to end up in a relation-
with increases in agreeableness, conscien- ship. Once in a relationship, emotional sta-
tiousness, and openness running parallel bility, agreeableness, and conscientiousness
with increased positive parenting. Thus, it are associated with better relationship qual-
is somewhat unclear what the directionality ity. Romantic relationships – especially the
between parenting and personality is, but it is more serious and enduring ones – also have
clear that the two are intertwined. an effect on personality development (e.g.,
The quality of adolescent relationships Lehnart et al., 2010; Neyer and Asendorpf,
with their family members has been some- 2001; Robins et al., 2002). That is, entering a
what more extensively studied in association relationship has been most clearly associated
with personality. An extensive study exam- with increases in emotional stability and con-
ining entire families across time showed scientiousness. Thus, personality traits affect
that, especially, agreeableness was of major relationships, but relationship experiences
importance in this regard. That is, more also have a considerable effect on personality
agreeable individuals perceived more sup- trait development (Mund and Neyer, 2014).
port from others, and others also perceived Another potential major influence on per-
more support from them (Branje et al., sonality development is experiences related
2004). However, the development of other to psychopathology. The linkages between
Big Five personality traits was also some- personality traits and psychopathology are
what associated with changes in support. A complex and can take many shapes (Durbin
study that focused on the relative contribu- and Hicks, 2014), but there certainly are
tions of adolescents’ and parents’ personali- strong associations. Research on adoles-
ties to the quality of their mutual relationship cents has suggested that psychopathology
showed that the older the adolescents got, symptoms have an effect on, and are affected
the more the quality of the relationship was by, personality trait development (Klimstra
determined by their personality instead of et al., 2010a). However, these associations
the parents’ (Denissen et al., 2009). A study may be due to psychopathology symptoms
 Personality Development in Adolescence and Young Adulthood 191

being manifestations of very high, or very emotional stability, agreeableness, and con-
low, scores on particular personality traits scientiousness, whereas social adjustment
(De Bolle et al., 2012). In this regard, espe- predicted increases in extraversion and con-
cially, emotional stability and extraversion scientiousness. Thus, attitudes and affective
are consistently associated with internal- experience toward education may trigger per-
izing symptoms (e.g., depressive and anxi- sonality maturation.
ety symptoms), whereas emotional stability Actual academic performance (e.g., Grade
and agreeableness are the most consistently Point Average; GPA) has rarely been consid-
associated with adolescent externalizing ered as a predictor of personality development.
symptoms (e.g., aggression, delinquency, We are aware of only one study examining
substance abuse). In young adulthood, the such an effect. This study showed that GPA
pattern of associations is pretty much the had no effect on the differential stability of
same (Mezquita et al., 2015). However, it personality traits in adolescence (Pullmann
should be noted that these linkages may be et al., 2006). More research is needed to con-
slightly different depending on the particu- firm whether or not academic performance
lar kind of type of psychopathology that is can affect personality trait development.
examined within the broader internalizing
and externalizing domains (Klimstra et al.,
2014). In sum, it remains a question whether
personality and psychopathology should be TRUE PERSONALITY DEVELOPMENT:
seen as separate constructs, but the develop- TAKING A PERSON-CENTERED
ment of the two is clearly interconnected. APPROACH
Academic experiences are also strongly
associated with personality trait development. The aforementioned research on personality
A study by Bleidorn (2012) suggested that development focused on trait domains, or on
personality trait development is triggered by the facets underlying these domains.
nearing the transition from high school to col- However, personality refers to a dynamic
lege, probably mostly due to students prepar- organization of traits within an individual (cf.
ing for their high school exams. That is, mean Allport, 1937). The study of personality
levels of conscientiousness clearly increased development should therefore also focus on
in the final year of high school, whereas no changes in the configuration of traits, also
such changes were found in the year before referred to as profiles. To examine develop-
that. Moreover, these increases in conscien- ment at the profile-level, one can examine
tiousness were associated with increases in profile stability or examine change from a
academic achievement behavior. Somewhat typological perspective on personality. In
related to this are findings suggesting that both approaches, the configuration of varia-
education commitment predicts increases in bles within an individual is the study object,
emotional stability (Klimstra, Luyckx et al., which is why these are referred to as person-
2012). However, it should be noted that there centered approaches.
were many more effects of personality traits
on changes in educational commitment.
Finally, adjustment to one’s educational insti- Profile Stability
tution also seems to matter for personality
maturation, as a recent study (Klimstra et al., Profile stability provides information on the
2016) found that academic and social adjust- stability of a constellation of traits for every
ment to college predicted relative increases in single person in a research sample. To assess
several personality trait domains and facets. profile stability, one computes the within-
Academic adjustment predicted increases in individual consistency of the mean scores on
192 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

personality traits. It is possible, for example, from each other, it turns out that it is not the
that a person is more conscientious than agree- stability of the distinctive qualities of one’s
able, more agreeable than open to experience, profile that drives the association with adjust-
more open to experience than emotionally ment, but mainly the similarity of one’s pro-
stable, and more emotionally stable than extra- file to the norm (i.e., average) in a sample
verted. The extent to which this pattern of trait (Klimstra et al., 2010c). Thus, individuals
scores of this person remains the same across with a stable profile are not necessarily the
time indicates that person’s profile stability better-adjusted ones. Instead, the individuals
(e.g., Furr, 2008). Thus, profile stability is who are more similar to the average person in
about the extent to which, for example, the tidy, the sample appear to be better adjusted.
adventurous, and friendly person is still like Profile stability is one way to approach
that at a later point in time. personality development from a person-
The calculation of profile stability for centered perspective, but it has its limitations.
every person within a research sample is The main limitation is that it remains unclear
done by correlating the individual’s set of what exactly remains stable and what changes
personality trait scores at one time point in an individual’s profile (cf. Bleidorn et al.,
with that same person’s set of personality 2012). To obtain greater insight into what
trait scores at the next time point. Every indi- changes about personality profiles, and to
vidual gets a stability score. Therefore, this answer questions about whether the general
profile stability score has a sample mean. direction of development is favorable or not,
Using this approach, relatively high profile typological approaches are needed.
stability has been demonstrated in early ado-
lescents (De Fruyt et al., 2006). Using data
on multiple measurement occasions (i.e., Typological Approaches
five), and therefore multiple time lags (i.e.,
four), Klimstra et al. (2009) were able to The study of personality types can be more or
demonstrate that personality profile stabil- less traced back to ancient Greece and China,
ity increased substantially from early to late but the most relevant to the contemporary
adolescence. Other studies (Donnellan et al., view on personality types is inspired by the
2007; Roberts et al., 2001; Robins et al., work that Block and Block started in the
2001a) have shown that profile stability 1970s (Block, 1971; Block and Block, 1980).
remains high in young adulthood. Block (1971) initially identified separate
High profile stability has been linked to typologies for men and women using a
more desirable personality traits and well- Q-sorting procedure. These typologies were
being in several studies (e.g., Donnellan actually development trajectories, as data on
et al., 2007; Lönnqvist et al., 2008; Roberts multiple measurement occasions were used.
et al., 2001), and it should therefore be indic- From the early 1980s onwards, the focus
ative of positive adjustment. However, profile shifted to three replicable types that were
stability coefficients suffer from conceptual derived on one measurement occasion and
and statistical problems arising from the were distinguished on the temperamental
confounding effect of profile normativeness. traits of ego-control (comparable to impulse
That is, profile stability might not only arise control) and ego-resiliency (the ability to
from people’s tendency to retain their idi- adapt one’s level of ego-control to fit the
osyncratic, distinctive qualities over time, but environmental demands). Specifically, Block
could also arise from their tendency to con- and Block (1980) distinguished Resilients
sistently resemble the normative personality (high ego-resiliency, flexible levels of ego-
profile (e.g., Furr, 2008). When these two dif- control), Undercontrollers (low ego-resiliency,
ferent sources of consistency are disentangled low ego-control), and Overcontrollers
 Personality Development in Adolescence and Young Adulthood 193

(low ego-resiliency, high ego-control). These from Undercontroller to Resilient. This tran-
types have distinguishable Big Five profiles sition pattern and the transition patterns found
(e.g., Asendorpf and van Aken, 1999; Robins in other person-centered studies are thus in
et al., 1996). Resilients have high levels of all line with the maturity principle, which sug-
Big Five traits if emotional stability is con- gests that individuals, on average, become
sidered instead of its counterpart neuroticism. more agreeable, conscientious, and emotion-
Undercontrollers are mostly distinguishable ally stable (e.g., Klimstra et al., 2009; Roberts
by their particularly low levels of agreeable- et al., 2001).
ness and conscientiousness, coupled with Studying transitions between types is only
high extraversion. Overcontrollers do have one way to approach personality develop-
high levels of agreeableness and conscien- ment from a typological perspective. Another
tiousness, but are low on extraversion and way is to use clustering techniques that take
emotional stability. development into account when assigning
Several studies used these types to exam- individuals to a certain class. This technique
ine personality development in adolescence. was used in Block’s (1971) groundbreak-
Somewhat older studies (van Aken and Dubas, ing study on personality development. He
2004; Akse et al., 2007) found moderate sta- distinguished developmental trajectories
bility of type membership, with about 50% of of personality from junior high school into
adolescents remaining in the same personal- adulthood. Five male and six female tra-
ity type across time. This is partly due to the jectories were distinguished. These were
analysis technique (i.e., cluster analysis) used described along a large number of attributes
in these studies. Cluster analysis does not obtained by using a q-sort procedure, and a
account for measurement error or classifica- description of these types would therefore
tion inaccuracy. This classification inaccuracy be too much for the purpose of this chapter.
is an important issue, since personality types The key message obtained from the Block
should not be viewed as a fully categorical (1971) study is that individuals not only dif-
variable, such as sex. Instead, they are best fer remarkably in their personality in junior
viewed as ‘fuzzy’ categories with somewhat high school, but also in their developmental
unclear boundaries, or zones of classification trajectory thereafter. To give one example,
uncertainty (Asendorpf et al., 2001). That the belated adjusters Block (1971) identi-
is, some individuals fall almost perfectly in fied ended up being (almost) as adjusted as
between being, for example, a Resilient or an the ego resilients, even though they experi-
Overcontroller. With contemporary techniques enced much more turbulence in adolescence.
to create types, such as latent profile analy- This provides evidence for a developmental
sis, there are several options to account for phenomenon referred to as equifinality (i.e.,
this classification inaccuracy (e.g., Vermunt, ending up in the same way from a different
2010). Hence, studies using these techniques starting point; Cicchetti and Rogosch, 1996).
find much higher stability of personality type More recently, Morizot and Le Blanc (2005)
membership, with over 70% being classified used a similar technique (i.e., longitudinal
in the same type across time (Meeus et al., cluster analysis) to distinguish developmen-
2011). If adolescents do change from one type tal trajectories from middle adolescence into
to another, they tend to change from the Over- adulthood. A key finding of this study was
or Undercontroller types to the Resilient type. that large individual differences in extraver-
Specht et al. (2014) mainly found evidence sion may come into existence as individuals
for stability in personality type membership (i.e., men) enter young adulthood, providing
in young adulthood. Only one transition from evidence for multifinality (starting at approx-
one type to another occurred more often than imately the same point but ending up at dif-
would be expected by chance: the transition ferent points; Cicchetti and Rogosch, 1996).
194 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

This shows the potential of longitudinal clus- and that individuals grow toward a more
tering techniques to capture important devel- mature personality profile. Yet, there are
opmental phenomena. important individual differences in the amount
More recent studies use techniques like of change and stability of personality. We still
Latent Class Growth analysis and General know little about the conditions, correlates,
Mixture Modeling (e.g., Nagin, 2005) to and consequences of these individual differ-
provide a more advanced spin on Block’s ences. Below, we provide some suggestions
(1971) classification procedure. Using such regarding future directions that could help to
procedures, Resilients, Undercontrollers, fill this relative gap of knowledge.
and Overcontrollers have been replicated A first suggestion is to study personality
as developmental trajectories of personality development in an even more detailed man-
from childhood into adolescence (de Haan ner. We already reviewed some literature on
et al., 2013), from early to middle adolescence personality facets underlying broader trait
(Klimstra et al., 2010b), from middle to late domains, with, for example, depression and
adolescence (Luyckx et al., 2014), and from anxiety underlying the domain neuroticism.
early to late adolescence (Branje et al., 2010). Recently, evidence has been found for the
Across studies, these three types remained validity of traits that are even more specific
distinguishable from each other across time than facets. These traits are called nuances
but, especially in the study by de Haan et al. (Mõttus et al., 2017). Usually measured with
(2013), between-type differences on particular few or single items, nuances represent one of
traits became larger (for conscientiousness) the lowest levels in personality and are much
or smaller (especially for agreeableness). closer to actual behavior than trait domains
Overall, these studies clearly show the poten- are. Examples of such nuances are liking
tial for using techniques to classify individu- parties with many people, making a detailed
als based on developmental trends in multiple plan before going on holiday, or liking loud
traits simultaneously. Such procedures are music. The idea that personality traits are
more in line with the way personality was hierarchically ordered, and that traits further
originally defined than studies distinguishing down than the facet level in the trait hierarchy
different developmental trajectories on single are also important, is not new (e.g., Eysenck,
personality traits (e.g., Durbin et al., 2016; 1990). Yet, research on personality devel-
Johnson et al., 2007). However, the latter stud- opment tends to overlook the nuance level.
ies are useful for visualizing and quantifying It could be that changes at the trait-domain
hetereogeneity in development. level can be traced back to changes (much)
lower in the trait hierarchy. Future research
considering changes in all levels of the trait
hierarchy simultaneously may eventually
SUMMARY AND FUTURE DIRECTIONS teach us much about how personality trait
development actually comes about.
In this chapter, we reviewed research on per- A second suggestion concerns implement-
sonality development in adolescence and ing measurements at various timescales.
young adulthood. A large number of cross- There is growing consideration of the state
sectional and longitudinal studies have pro- level in contemporary models of person-
vided convincing evidence that personality ality (Fleeson and Jayawickreme, 2015).
traits undergo important changes during the Such personality states refer to patterns of
period of adolescence and early adulthood. thinking, feeling, and behaving at a par-
We know now that rank-order stability ticular moment. Therefore, these states can
increases as individuals grow older, that mean be measured several times a week or even
levels of personality traits typically increase, several times a day. Due to the widespread
 Personality Development in Adolescence and Young Adulthood 195

availability of smartphones, researchers can part of the population (e.g., only including
now ask participants to report either actively the higher end of the distribution on con-
(by prompting individuals to report on their scientiousness as a result of focusing on the
behavior, feelings, or thoughts) or passively higher educated), or using a measure that
(by recording their behavior) on their person- operationalizes Big Five trait domains in a
ality states (Wilson et al., 2017). Considering slightly different manner when compared to
traits and states in one longitudinal design the measures that were used in other studies
will eventually allow us to trace back the ori- that did find Resilients, Undercontrollers,
gins of personality development to changing and Overcontrollers. One way to proceed
patterns of daily activity. would be to use latent profile analysis and
A third, broader, point is that much of the not get too hung up on replicating the exact
research that claims to examine personality three types (and only these types) all the
development is actually examining person- time. Latent profile analysis comes with an
ality trait development. Studying traits in additional advantage, as the results of these
isolation can be useful, but to gain a better analyses no longer have to be used in a fully
understanding of true personality develop- categorical manner. Individuals get relative
ment, the configuration of traits within par- class assignment, which means that they can,
ticular individuals and changes herein need for example, be classified as having a 40%
to be studied over time. Studies examining chance of being Resilient and a 60% chance
changes in personality trait profiles across of being an Overcontroller. These chances
time are by far outnumbered by studies exam- can be taken into account in follow-up analy-
ining change in a trait-by-trait fashion (for ses (e.g., Vermunt, 2010), thereby account-
exceptions, see for example, Meeus et al., ing for the fact that types may be fuzzy
2011 and Specht et al., 2014). Moreover, instead of truly categorical (Asendorpf et al.,
research examining longitudinal associations 2001). Moreover, this approach potentially
of changes in trait profiles with changes in addresses the issue of the limited predictive
other variables is scarce. Thus, little is known power that is inherent to fully categorical
about the antecedents, correlates, and effects approaches when compared to continuous
of changes in personality types. traits (cf. Asendorpf and Denissen, 2006).
Among the reasons why personality types Specifically, a recent study using Latent
are underexamined are concerns regard- Profile Analysis found clear evidence for an
ing their replicability and their predictive increase in explained variance in outcome
power (Costa et al., 2002). Regarding rep- variables if relative classification (i.e., clas-
licability, it should be noted that the litera- sification probabilities) instead of absolute
ture suggests that there are usually at least classification (i.e., using a classic categori-
three types that appear (i.e., Resilients, cal distinction) was used (Hadiwijaya et al.,
Undercontrollers, and Overcontrollers) but 2015). Thus, new data-analytical advance-
that sometimes other types are found too (cf., ments have opened the door for the possible
Chapman and Goldberg, 2011). Note that resurrection of psychological types, such as
this is partly inherent to the techniques that personality types.
are used nowadays to identify types. Such Fourth, much of the work that has been
techniques (e.g., Latent Profile Analysis) are done concerning personality development
exploratory in nature and (perhaps therefore) has relied on quantitative data. To attain a bet-
sensitive to the peculiarities of each sample. ter understanding on the actual processes and
Thus, different types can be found for sev- experiences that go along with personality
eral reasons, including method effects (e.g., development, it would be useful to comple-
using self-reports instead of peer-reports, or ment this quantitative data more often with
the other way around), examining a particular qualitative data. So far, the narrative identity
196 THE SAGE HANDBOOK OF PERSONALITY AND INDIVIDUAL DIFFERENCES

research tradition has done the best job in Anusic, I., & Schimmack, U. (2016). Stability
supplementing quantitative data on person- and change of personality traits, self-esteem,
ality trait development with qualitative data and well-being: Introducing the meta-
(for an illustration, see Pals, 2006). However, analytic stability and change model of retest
there appears to be no published research yet correlations. Journal of Personality and Social
Psychology, 110, 766–781.
on adolescents or young adults that includes
Asendorpf, J. B. (1992). Beyond stability: Pre-
multiple measurement occasions of narrative dicting inter-individual differences in intra-
accounts and personality traits. Such research individual change. European Journal of
likely provides important insights into how Personality, 6, 103–117.
personality development comes about and Asendorpf, J. B., Borkenau, P., Ostendorf, F., &
should therefore be prioritized. van Aken, M. A. G. (2001). Carving person-
We only provided four possible future ality description at its joints: Confirmation of
directions for the study of adolescent and three replicable personality prototypes for
young adult personality development, but both children and adults. European Journal
there obviously are many more possible direc- of Personality, 15, 169–198.
tions. Overall, this chapter shows that our Asendorpf, J. B., & Denissen, J. J. A. (2006).
Predictive validity of personality types versus
knowledge of personality development has
personality dimensions from early childhood
rapidly expanded over the past 15 to 20 years. to adulthood: Implications for the distinction
We know much more about the direction of between core and surface traits. Merrill-
personality development and individual dif- Palmer Quarterly, 52, 486–513.
ferences around the general patterns, while Asendorpf, J. B., Denissen, J. J. A., & van Aken,
our insight into the possible mechanisms, M. A. G. (2008). Inhibited and aggressive
predictors, and consequences of personality preschool children at 23 years of age: Person-
development is also growing. Together with ality and social transitions into adulthood.
the many available possible future directions, Developmental Psychology, 44, 997–1011.
this suggests that research on adolescent and Asendorpf, J. B., & van Aken, M. A. G. (1999).
young adult personality development is thriv- Resilient, overcontrolled and undercontrolled
personality prototypes in childhood: Replica-
ing and will continue to thrive.
bility, predictive power, and the trait-type
issue. Journal of Personality and Social Psy­
chology, 77, 815–832.
Asendorpf, J. B., & van Aken, M. A. G. (2003).
REFERENCES Personality–relationship transaction in adoles-
cence: Core versus surface personality charac-
Ahmetoglu, G., Swami, V., & Chamorro- teristics. Journal of Personality, 71, 629–666.
Premuzic, T. (2010). The relationship between Bazana, P. G., & Stelmack, R. M. (2004). Stabil-
dimensions of love, personality, and relation- ity of personality across the life span: A
ship length. Archives of Sexual Behavior, 39, meta-analysis. In R. M. Stelmack (Ed.), On
1181–1190. the psychobiology of personality (pp. 113–
Akse, J., Hale, W. W., III, Engels, R. C. M. E., 144). New York, NY: Elsevier.
Raaijmakers, Q. A. W., & Meeus, W. H. J. Blatný, M., Millová, K., Jelínek, M., & Osecká,
(2007). Stability and change in personality T. (2015). Personality predictors of succesful
type membership and anxiety in adolescence. development: Toddler temperament and
Journal of Adolescence, 30, 813–834. adolescent personality traits predict well-
Allemand, M., & Martin, M. (2016). On corre- being and career stability in middle adult-
lated change in personality. European Psy­ hood. PLoS ONE, 10, e0126032.
chologist, 21, 237–253. Bleidorn, W. (2012). Hitting the road to adult-
Allport, G. W. (1937). Personality: A psycho­ hood: Short-term personality development
logical interpretation. New York, NY: Holt, during a major life transition. Personality and
Rinehart, & Winston. Social Psychology Bulletin, 38, 1594–1608.