Stability and Change of Personality Traits, Self …...personality traits is attributable to stable influences (83%). Changing factors had a greater influence on Changing factors had

Stability and Change of Personality Traits, Self-Esteem, and Well-Being:Introducing the Meta-Analytic Stability and Change Model of

Retest Correlations

Ivana AnusicMichigan State University

Ulrich SchimmackUniversity of Toronto Mississauga

The stability of individual differences is a fundamental issue in personality psychology. Althoughaccumulating evidence suggests that many psychological attributes are both stable and change over time,existing research rarely takes advantage of theoretical models that capture both stability and change. Inthis article, we present the Meta-Analytic Stability and Change model (MASC), a novel meta-analyticmodel for synthesizing data from longitudinal studies. MASC is based on trait–state models that canseparate influences of stable and changing factors from unreliable variance (Kenny & Zautra, 1995). Weused MASC to evaluate the extent to which personality traits, life satisfaction, affect, and self-esteem areinfluenced by these different factors. The results showed that the majority of reliable variance inpersonality traits is attributable to stable influences (83%). Changing factors had a greater influence onreliable variance in life satisfaction, self-esteem, and affect than in personality (42%–56% vs. 17%). Inaddition, changing influences on well-being were more stable than changing influences on personalitytraits, suggesting that different changing factors contribute to personality and well-being. Measures ofaffect were less reliable than measures of the other 3 constructs, reflecting influences of transient factors,such as mood on affective judgments. After accounting for differences in reliability, stability of affect didnot differ from other well-being variables. Consistent with previous research, we found that stability ofindividual differences increases with age.

Keywords: meta-analysis, stability, personality, life satisfaction, self-esteem

Supplemental materials: http://dx.doi.org/10.1037/pspp0000066.supp

The stability and change of personality and well-being has beena controversial issue in personality psychology, just like thenature–nurture debate has been a source of heated debate. Thenature–nurture debate has quieted down because quantitative be-havioral genetics studies provide evidence that most personalitycharacteristics are influenced by genetic factors and environmentalfactors. Moreover, behavioral genetics models have been used toquantify the contribution of genes and environment to variouspersonality characteristics, and the question is no longer whethergenes or environment matter but how much genes and environ-ment matter.

We propose that controversies about stability and change ofpersonality would equally benefit from quantitative models ofstability and change. Ample evidence shows that personality char-

acteristics are neither fixed nor rapidly changing from moment tomoment (Caspi & Roberts, 2001; Conley, 1984; Ferguson, 2010;Roberts & DelVecchio, 2000; Terracciano, McCrae, & Costa,2010; Trzesniewski, Donnellan, & Robins, 2003). Thus, personal-ity traits like the Big Five and personality characteristics likewell-being are stable, and they change over time (Roberts, Wood,& Caspi, 2008; Specht et al., 2014). To move research on stabilityand change forward, it is necessary to quantify the degree ofstability and change.

This focus on quantifying the extent to which personality char-acteristics change or remain stable over time is consistent withrecent calls in psychology to move from testing of the null hy-pothesis to parameter estimation (Cumming, 2014). The null hy-pothesis is particularly uninformative for questions about variancecomponents. The probability that personality has zero stability ornever changes is very small, and empirical data can never provethat it is zero. Thus, it is more productive to quantify the amountof stability and change in personality traits and other individualdifferences rather than simply test whether change happens.

Stability and change are most commonly quantified by test–retest correlations. Interpretation of retest correlations is madedifficult by two methodological problems. First, retest correlationsare attenuated by random measurement error. Thus, observedretest correlations underestimate stability and overestimate change.One solution to this problem is to use internal consistency as anestimate of reliability. However, internal consistency can be in-

This article was published Online First November 30, 2015.Ivana Anusic, Department of Psychology, Michigan State University;

Ulrich Schimmack, Department of Psychology, University of TorontoMississauga.

This research was supported by a Master’s and Doctoral SSHRC schol-arships awarded to Ivana Anusic and a SSHRC standard research grantawarded to Ulrich Schimmack.

Correspondence concerning this article should be addressed to IvanaAnusic, Department of Psychology, Michigan State University, 316 Phys-ics Road, East Lansing, MI 48824. E-mail: [email protected]

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Journal of Personality and Social Psychology © 2015 American Psychological Association2016, Vol. 110, No. 5, 766–781 0022-3514/16/$12.00 http://dx.doi.org/10.1037/pspp0000066

766

flated by systematic measurement error and attenuated by itemheterogeneity. Furthermore, it is not possible to calculate internalconsistency for single-item measures. A more direct way to esti-mate random measurement error is to use retest correlations overshorter time periods. However, retest correlations can overestimaterandom measurement error if the construct actually changes duringthe retest interval (e.g., well-being may actually change in a1-month period).

The second obstacle is that the amount of change that can beobserved depends on the time interval between repeated measure-ments. Change is unlikely to occur during a short time interval ofa few weeks, but the likelihood of personality change increaseswith time (e.g., an extravert is likely to be an extravert a monthlater, but may become an introvert in 20 years). As a result, asimple retest correlation does not quantify stability or changebecause it is dependent on the time interval between retests. Thus,a single retest correlation provides insufficient information aboutthe amount of stability or change of a personality characteristic.

To estimate stability and change, it is necessary to create amodel that takes into account both stability and change, fit ob-served retest correlations to the model, and interpret the modelparameters. In a seminal yet often neglected article, Conley (1984)made a first attempt to quantify stability and change of intelligencetests, personality traits, and self-evaluations (self-esteem, life sat-isfaction). He conducted a meta-analysis of retest correlations andplotted retest correlations as a function of the retest interval. Thisresulted in a nonlinear decay function where retest correlationsbecame smaller as retest intervals lengthened. He then fittedHeise’s (1969) autoregressive state model to the retest correlationsand found that intelligence was more stable than personality traits(extraversion and neuroticism), which were more stable than self-evaluations (self-esteem and life satisfaction). The rate of changeof variance at the between-person level for intelligence was lessthan 2% per year. The rate of change for extraversion and neurot-icism was 4% per year, and the rate of change for self-evaluationswas 12%. As change accumulates, these differences lead to muchlarger differences over longer time periods. For example, only18% of the between-person variance in intelligence changes overa 10-year period. For extraversion and neuroticism, 33% of thevariance changes over a 10-year interval, and 71% of the variancein self-evaluations changes over a 10-year period.

Conley’s (1984) findings provided the first empirical evidencethat some personality characteristics are more stable than others.The results also provided the first evidence that personality traitsand well-being are not merely fluctuating around a fixed set point,but can change for extended periods of time. Unfortunately, thisimportant finding was neglected in personality theories of well-being that exaggerated the importance of stable dispositions andignored the importance of environmental factors (see Lucas, 2007,for a review).

Although Conley made an important first step toward a quanti-tative model of stability and change, his model had some limita-tions. The most serious limitation was the choice of a simpleautoregressive function to model change. This model assumes thatall factors that contribute to variation in personality are changing.As change accumulates, retest correlations will eventually asymp-tote toward zero. For example, the finding that retest correlationsfor self-evaluations changed at the rate of 12% per year (i.e., 88%is stable year-to-year) suggests that after 20 years only 8% (.8820)

of the between-person variance in self-evaluations is retained,whereas 92% of variance (100 – 8) has changed. This prediction isinconsistent with actual retest correlations of life satisfaction inmore recent panel studies that remain above .30 over retest inter-vals of 20 years or more (Lucas & Donnellan, 2007; Lucas &Donnellan, 2012; Schimmack, Krause, Wagner, & Schupp, 2010;Schimmack & Lucas, 2010).

The high retest stability of individual differences over retestintervals longer than two decades suggests that some factors thatproduce variation in personality across individuals are stable.Similar conclusions can be drawn from findings of longitudinalbehavioral genetics studies that have found substantial geneticinfluences on personality traits and well-being across the life span(Briley & Tucker-Drob, 2014; Kandler et al., 2010; Nes, Røysamb,Tambs, Harris, & Reichborn-Kjennerud, 2006). However, stableinfluences are not solely based in our biology. Evidence is accu-mulating that certain features of the environment can also producestability in personality (Bleidorn, Kandler, & Caspi, 2014; Briley& Tucker-Drob, 2014).

To allow for the possibility that stable influences on personalityexist, Kenny and Zautra (1995) introduced the trait-state-errormodel (TSE). The state and error factors in the TSE model areequivalent to Conley’s (1984) model, with state factor reflectingchanging influences on personality. The major advantage of theTSE model is that it includes a stable trait factor. The stable traitfactor in the model essentially allows for an above-zero asymptotein retest correlations. For example, if retest correlations overperiods of 20 years no longer decrease but hover around r ! .30,the data suggest that 30% of between-person variance in a person-ality characteristic is influenced by a stable factor that neverchanges. Because this factor has a persistent influence on person-ality at every time point, all measures share a minimum of 30% ofthe variance independent of the time interval between measures.

At present, only relatively few studies have used the TSE modelto quantify stability and change of personality characteristics. Themain reason is that the model requires a minimum of four occa-sions of measurement, relatively long time periods, and largesamples to provide meaningful parameter estimates; that is, pa-rameter estimates with relatively narrow confidence intervals (CIs)around them. For example, Anusic, Lucas, and Donnellan (2012)conducted a short-term panel study that included measures of theBig Five and life satisfaction. A total of 237 students completedeight waves of measurement over a 2-month interval. The studyrevealed that personality traits were more stable than life satisfac-tion judgments, which were more stable than pleasant and unpleas-ant affect. However, the 2-month interval was too short to reliablyestimate the trait factor.

So far, the model mainly has been used for panel studies of lifesatisfaction where it could be fit to raw data (Lucas & Donnellan,2007; Lucas & Donnellan, 2012; Schimmack & Lucas, 2010;Schimmack, Krause, Wagner, & Schupp, 2010). The main findingis that somewhere between 30% and 50% of the between-personvariation in life satisfaction is attributable to a stable trait factor.Donnellan, Kenny, Trzesniewski, Lucas, and Conger (2012) andKuster and Orth (2013) obtained similar results for self-esteem.Based on Conley’s finding that personality traits are more stablethan self-evaluations (see also Anusic et al., 2012), it is likely thata stable factor accounts for even more variation in personalitytraits that exists at the between-person level. However, this pre-

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

767META-ANALYSIS OF STABILITY AND CHANGE

diction has not been tested because there are no equivalent panelstudies of personality.

In sum, the existing evidence suggests that personality traits aremore stable than self-concepts and self-evaluative judgments suchas judgments of self-esteem and life satisfaction. However, theevidence regarding the relative contribution of stable and changinginfluences on these constructs is inconclusive because previousstudies have been limited by their design. The main limitation isthat original studies often do not include all measures, sufficientmeasurement points, or a sufficient time period to observe signif-icant change in individual differences.

In the present study, we were able to compare the stability andchange of personality, well-being, and self-esteem measures bydeveloping a Meta-Analytic Stability and Change model (MASC)for retest correlations. MASC can be regarded as an extension ofConley’s meta-analytic model and as a meta-analytic version ofKenny and Zautra’s (1995) TSE model for original data. The maincontribution of our article is to introduce MASC as a model formeta-analyses of retest correlations and to apply the model toretest correlations of personality traits (Roberts & DelVecchio,2000), life satisfaction ratings (Schimmack & Oishi, 2005), self-esteem ratings (Trzesniewski, Donnellan, & Robins, 2003), andratings of affect. This model can be simultaneously applied tovarious individual differences and provide quantitative tests ofdifferences in stability and change among them.

Statistical Model

MASC can be considered a meta-analytic version of Kenny andZautra’s (1995) TSE model for raw data. TSE is a structuralequation model, and structural equation models are most mean-ingful when they are interpreted as realistic models of a set ofcausal processes that could have produced an observed pattern ofdata (Borsboom, Mellenbergh, & Van Heerden, 2003). Our modelof stability and change is illustrated in Figure 1, where the boxesrepresent observed variances in a measure of individual differ-ences (e.g., IQ tests, trait ratings, life satisfaction ratings). Thereare two boxes because the measure was administered twice. The

model parameters predict the observed retest correlation betweenthe two observed measures.

The model in Figure 1 distinguishes two sources of variance inresponses at the between-person level: reliable variance and ran-dom error variance. The assumption that random measurementerror influences personality measures is not controversial. It iscommon practice to estimate the amount of reliable variance in apersonality measure and to report this information. However,random measurement error is often ignored in the interpretation ofresearch findings (Schmidt & Hunter, 1996). For the meta-analyticintegration of retest correlations, it is essential to distinguish be-tween random measurement error and true change (Conley, 1984;Charles, 2005). The reason is that differences in reliability candistort comparisons of the true stability and change of the constructthat is being measured. For example, a single life satisfactionrating is less reliable than a multiple item life satisfaction scale ora 20-item personality scale (Schimmack & Oishi, 2005). Withoutcorrection for unreliability, it is impossible to compare the stabilityof personality traits and life satisfaction. Thus, MASC corrects forunreliability and examines the stability and change of the reliablefactor in Figure 1.

There has been some debate about the ability of the TSE todistinguish between random measurement error and true change(Lucas & Donnellan, 2012; Schimmack et al., 2010). The reason isthat the estimate of the error parameter depends on the length ofthe shortest retest interval. To illustrate, in the German Socio-Economic Panel (e.g., see Lucas & Donnellan, 2012), life satis-faction is measured in annual intervals. A 1-year retest intervalimplies that some events can produce changes in life satisfactionthat are only observed on a single occasion. For example, Petramay rate her life satisfaction a 7 in February 2001. She loses herjob in October 2001. Her life satisfaction rating in February 2002is a 4. However, in December 2002 she gets a new job, and inFebruary 2003 she rates her life satisfaction a 7. The TSE modelcannot distinguish the temporary shift in the life satisfaction ratingin 2002 from random measurement error (e.g., Petra was in a badmood when she rated her life satisfaction in 2002), although anactual life event produced a real change in life satisfaction thatlasted from October 2001 to December 2002, a period of "1 year.

This is not a problem for MASC because a meta-analysisincludes studies with different retest intervals. For example, retestintervals for life satisfaction judgments range from a few minuteswhen the same item was administered twice in a single survey tomore than 20 years (Schimmack & Oishi, 2005). By includingretest correlations over shorter time periods, the model does abetter job of distinguishing random measurement error from truechange. Random measurement error in this case also includestransient influences, such as mood effects and other factors thatmay influence response styles but do not necessarily reflectchanges in one’s true standing on a psychological characteristic(Chmielewski & Watson, 2009).

Figure 1 shows that two factors contribute to reliable variationacross individuals at any moment in time. One factor representsstable causes. Kenny and Zautra (1995) called this factor trait (i.e.,the T in the TSE). We think that this terminology can createconfusion because the term trait is also used to refer to personalityconstructs like extraversion. We use the term trait to refer to theBig Five and similar constructs, and the term stable factor to referto stable causes of individual differences. For example, a study

Figure 1. The Meta-Analytic Stability and Change model (MASC),shown fitted to a variable measured at two occasions.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

768 ANUSIC AND SCHIMMACK

might report retest correlations of weight in an adult sample. Onestable factor that produces individual differences in weight isheight. Thus, height would be one stable cause of individualdifferences in weight. If height would explain 25% of the between-person variance in weight, the path coefficient from height toweight on any single occasion would be .5 (.52 ! 25% explainedvariance). The 25% shared variance between height and weight attwo occasions would produce a retest correlation of r ! .25 inweight measures. The retest correlation could no longer asymptoteto zero because the stable factor height will produce a minimumcorrelation of r ! .25. The square of the coefficient S in Figure 1can be called the stability coefficient (S2), and it reflects the extentto which individual differences are caused by stable factors, justlike the heritability coefficient in twin studies refers to the amountof variation across individuals that can be attributed to the effect ofgenetic variation.

In MASC, the nature of stable factors that contribute to retestcorrelations is unknown. MASC can only estimate the proportionof reliable variance on a single occasion that can be attributed toinfluences of stable factors. This includes direct stable influences(e.g., genes, stable aspects of the environment) but also influencesof events that remain even after the event itself ends. For example,an adult person is no longer in her childhood environment, but herchildhood environment may have long lasting influences on herthoughts, feelings, and behaviors, and thus would be considered astable cause of adult personality (i.e., one’s childhood environmentcannot change in adulthood).

The other factor that contributes to retest correlation is thechange factor. Kenny and Zautra (1995) referred to this factor asstate. We think it is confusing to refer to this factor as state in amodel of stability and change of personality traits because statetypically refers to occasion-specific states (e.g., mood). For thisreason, we use the term change factor. As with the stable factor,the change factor reflects combined effects of several influences.These influences are grouped together because they share theproperty that they can change over time. As such, changing influ-ences and stable influences are mutually exclusive. To use theexample of weight again, changing factors that can contribute toweight are diet and exercise. When individuals change their diet orexercise habits, their weight changes, and these changes willinfluence the rank-order of individuals. For example, Michael mayweigh 200 pounds and lose 50 pounds; Natasha may weigh 160pounds and gain 10 pounds. As a result of these changes, Michaeland Natasha switch places in the rank-order. These changes willlower retest correlations. Over longer periods of time, morechanges can accumulate resulting in smaller retest correlationsover longer retest intervals than over shorter retest intervals.

Figure 1 illustrates that the stable factor and the changing factorare two mutually exclusive factors that determine the reliablecomponent of measured individual differences. This implies thatthe total reliable between-person variance is the sum of the amountof between-person variance explained by the stable factor (S2) andthe amount of between-person variance explained by the changingfactor (C2). Because retest correlations are based on standardizedvariables with variances of 1, it follows that

S2 ! C2 " 1 and that S2 " 1 # C2 (1)

Thus, although S2 and C2 are conceptually distinct factors, theyare captured by a single parameter in a model with standardized

coefficients. As a result, they reflect relative proportions of stableand changing variance. Original data would be needed to estimatethe absolute amount of stable variance when the same measure isused. Once more, this is also true for behavioral genetic models,where environmental factors account for all of the variance notexplained by genetic factors.

The final parameter in MASC models the rate of change ofchanging factors. Figure 1 shows that changing factors can changeduring the time interval between two retests. It is important that theamount of change that occurs during a retest interval depends onthe length of the retest interval. Over a very brief interval, there isvery little time for any changing factor to change. Over a very longinterval, many changing factors will have changed and after infi-nitely long interval all changing factors will have changed. Factorsthat never change, and factors whose influence never changes evenif the factor itself changes (e.g., a life event that continues to exerta constant influence on personality even after the event is over),are by definition stable factors and are captured in the stable factor.

In structural equation modeling, it is common to estimate therate of change with the path coefficient from changing factors atTime 1 to changing factors at Time 2. This approach is useful instudies with equally spaced retest intervals (e.g., daily ratings).However, this approach is less useful for comparisons of retestintervals that vary in length because coefficients vary as a functionof the retest interval. One solution to this problem is to convertcoefficients for different time intervals into a common time inter-val. For example, Conley (1984) expressed the rate of change interms of the retest correlation over a 1-year period. We decided tofollow the same procedure and estimate a 1-year stability of thechanging factors. In Figure 1, the quantity of change and stabilityof changing factors is quantified by the amount of variance in thechanging factor at Time 2 that has not yet changed versus theamount of variance in changing factors that has changed.

To use a concrete example, we plotted a hypothetical pattern ofretest correlations in Figure 2. The intercept in this figure is .80.This means that without any real changes, retest correlations areonly r ! .80. The remaining 20% of the variance is measurementerror. Thus, the reliability of the hypothetical test scores in thisstudy is 80%. Figure 2 also shows that retest correlations decreaseover time, but then hit an asymptote at r ! .50. This finding showsthat 50% of the variance in test scores never changes. Thus, 50%of the total variance is determined by stable factors. Adjusted forunreliability, this implies that 62.5% (.50/.80 ! .625) of thereliable variance is caused by stable factors (S2 ! .625). Changingfactors account for the remaining 30% of the total variance in testscores (80% reliable variance # 50% stable variance ! 30%changing variance). Changes in these factors lead to decreasingcorrelations over longer retest intervals. For example, as the retestinterval increases from zero to 1 year, retest correlations decreasefrom .80 to .65. This drop implies that the stability coefficient inFigure 1 is .50. The reason is that changing factors contribute 30%of the variance, and stability of .50 implies that the total effect ofchanging factors is .15 over a 1-year retest interval because .30 !

.50 ! .15. Adding to this the effect of stable factors, r ! .50,produces the 1-year retest correlation of r ! .65 that is observedfor a retest interval of one year. A stability coefficient of .50 overa 1-year interval implies that changing factors at Time 1 predictonly 25% of the variation in changing factors at Time 2. As Figure2 shows, a stability coefficient of .50 leads to a rather rapid decay

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

769META-ANALYSIS OF STABILITY AND CHANGE

in retest correlations and predictability over time. After 5 years,99% of between-person variance in the changing factors haschanged, whereas only 1% is left unchanged.

MASC uses nonlinear regression analyses to estimate threeregression coefficients that maximize predictions of retest corre-lations (rt1#t2) from the length of the retest interval (TIME). Thenonlinear equation is

rt1#t2 " REL2 ! !(1 # C2) ! (C2) ! StChTIME" (2)

where REL2 is the amount of reliable variance that exists at thebetween-person level, C2 is the proportion of this reliable varianceexplained by the change factor, (1 # C2) is the proportion of thereliable variance that is explained by the stable factor, and StCh isthe stability of the changing factors over 1 unit of TIME (e.g., 1year).

Current Study

Equation 2 is the basic form of MASC. It can be extended asnecessary to evaluate the impact of moderator variables on differ-ent components of the model. Moderator variables are specified asproduct terms of the basic parameters and moderator variables. Weinclude these product terms to examine whether model parametersdiffer significantly between studies of personality traits, self-esteem, life satisfaction, and affect. Based on existing literature,we expect the largest influence of stable factors on personalitytraits (Conley, 1984; Roberts & DelVecchio, 2000). Life satisfac-tion and self-esteem should show less stability over longer periodsof time, yet they should show evidence of notable influence ofstable factors as well (Anusic et al., 2012; Donnellan et al., 2012;Kuster & Orth, 2013; Lucas & Donnellan, 2012; Schimmack &Lucas, 2010).

It is more difficult to make predictions about where affect wouldfit on this continuum. Existing longitudinal studies of affect areless prevalent than those that focus on other psychological con-structs. Some studies have suggested that affect is influenced bystable personality traits to a greater degree than life satisfaction(e.g., Schimmack, Radhakrishnan, Oishi, Dzokoto, & Ahadi,2002), suggesting that affect would show greater influences ofstable factors. On the other hand, there is some longitudinalevidence that, at least in the short-term, ratings of affect are lessstable than those of life satisfaction (Anusic et al., 2012). Onepossible explanation for these inconsistent findings is that affect ismore strongly influenced by stable factors than life satisfactionjudgments, but that changing factors in affect decay more quickly.This would lead to lower stability for affect measures in theshort-term and higher stability of affect measures in the long-term.Our study provides the first test of this hypothesis.

We also evaluate additional moderators of stability andchange. The most prominent moderator in the literature is age.There is widespread consensus that stability of personality traitsincreases from childhood into adulthood (Roberts & DelVec-chio, 2000; Specht et al., 2011; Wortman, Lucas, & Donnellan,2012). Recent evidence suggests that stability decreases againin old age. A study by Lucas and Donnellan (2007) found thatstability in life satisfaction also increases throughout the lifespan, with greatest stability occurring in the sixth decade of life,and a similar trend has been found for self-esteem (Donnellanet al., 2012; Kuster & Orth, 2013; Trzesniewski et al., 2003).However, the processes that lead to higher stability over the lifespan are still unclear. In this study, we evaluate for the first timeto our knowledge the extent to which increasing stability isattributable to increasing reliability, increasing influence ofstable factors, or increasing stability of changing factors. Forexample, if personality traits become increasingly more influ-enced by stable factors, and personality traits influence well-being (Diener & Lucas, 2000), we would expect that well-beingalso becomes more stable with age because of the greaterinfluence of stable factors. However, it is also possible thatwell-being becomes more stable if changing factors such asincome and social relationships become more stable with age.

Method

Data Selection

We relied on Trzesniewski et al.’s (2003) meta-analysis ofself-esteem to obtain studies that reported retest correlations forself-esteem. We used Roberts and DelVecchio’s (2000) meta-analysis to obtain a list of studies containing retest correlationsfor personality traits. We used Schimmack and Oishi’s (2005)meta-analysis to obtain studies containing retest correlations forlife satisfaction. These sources provided the bulk of our data.We conducted additional searches using PsycInfo and WebOf-Science databases with keywords for individual differences(i.e., personality, affect, feelings, mood, life satisfaction, hap-piness) paired with each of the following keywords: consis-tency, stability, longitudinal, trait, change. We defined each ofthe four types of individual differences (personality traits, lifesatisfaction, affect, and self-esteem) broadly and retained anystudies that explicitly referenced one of these four psycholog-

Figure 2. Example trajectory of retest correlations (y-axis) taken overdifferent retest intervals (x-axis). Reliability ! 1 – Error. In this example,Error ! .20 (Reliability ! .80). Stability ! .50, that is, .50/.80 ! .625 ofreliable variance. Change ! .30, that is, .30/.80 ! .375 of reliable variance.Stability of the Change component defines the shape of the curve and is .50in this example.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

770 ANUSIC AND SCHIMMACK

ical characteristics. For example, we did not limit our search forpersonality traits to the Big Five, but included any studies thatmeasured what is typically considered a personality variable(e.g., constraint or alienation, as assessed by the Multidimen-sional Personality Questionnaire; Tellegen, 1982). This exten-sive search included all studies published prior to 2009. Wescanned articles obtained in this manner for retest correlations.All studies that reported retest correlations were included in themeta-analysis. Next, we searched for any additional relevantpapers that may have been missed by our preliminary search byreviewing the references of the selected articles and searchingfor any newer papers that cited them. Any studies found in thisway that contained information about retest correlations werealso included in the analyses.

For our analyses, we defined the unit of analysis as the smallestsample within the study for which retest information was provided.For example, if a study contained separate retest correlations foryounger and older females and males, we coded these as fourdifferent samples (i.e., younger females, younger males, olderfemales, older males). This allowed us to more accurately examinethe role of age and gender as potential moderators of stability andchange of individual differences.

We computed additional retest correlations from several pub-licly available panel studies. We obtained life satisfaction datafrom the British Household Panel Study, the Swiss HouseholdPanel, the German Socioeconomic Panel, the National Survey ofFamilies and Households, the Americans’ Changing Lives study,and the National Health and Nutrition Examination Survey. TheGerman Socioeconomic Panel also provided retest correlations forpersonality for years 2005–2009. Additional self-esteem data wereobtained from the Americans’ Changing Lives study, the NationalEducational Longitudinal Study, and the National LongitudinalStudy. For these datasets, we obtained separate retest correlationsover all possible retest intervals separately for both genders and forthe following age groups (age was coded at the first wave of datacollection): under 15 years, 15 to 30 years, 30 to 60 years, and over60 years.

For each retest correlation reported in the articles or com-puted from panel studies, we recorded the following informa-tion: construct (personality traits, life satisfaction, affect, orself-esteem), age, gender, scale length, retest interval, and retestcorrelation. For age, we coded the mean age at the start of theretest interval as reported in the article. If age range wasreported instead of the mean, we took the mean of the twoendpoints of the range. Some articles also reported informationon sample sizes of specific age groups (e.g., 10 participantsaged 12–15 years, 48 participants aged 16 –20 years)—in thesecases we estimated age as the weighted mean. We centered theage variable around the overall mean (32.4 years), and trans-formed it by dividing it by 10. Thus, the moderating effects ofage can be interpreted as change over a 10-year period. Wecoded gender as the proportion of female participants in thesample. Scale length was coded as the number of items in thescale for which the retest correlation was reported. If thisinformation was not explicitly given in the article, we tried tofind the information about the scale elsewhere (e.g., otherarticles that used the same scale). Because the associationbetween scale length and reliability is not linear (e.g., thereliability difference between a single-item and a five-item

scale is larger than the difference between 25- and 30-itemmeasures), for our analyses we transformed the scale lengthvariable by taking the natural logarithm of the actual number ofitems. We recorded retest interval in years. Finally, raw retestcorrelations were recorded as reported in the article or com-puted from the panel data. We included only retest correlationsof self-reported measures, and omitted any retest correlationsbased on informant reports, observer ratings, or projective tests.

Some articles did not provide information about age, genderdistribution, or scale length. In such cases, we imputed valuesfor those variables using the mean for that construct. Forexample, if age information was missing in a study that reporteda retest correlation for life satisfaction, we used the average ageacross all life satisfaction studies (40.83 years) as the age forthat study. In total, 13 cases were missing age information (2from personality, 4 from affect, and 7 from life satisfactionsubsamples), 116 were missing information on gender distribu-tion in their sample (30 form personality, 32 from affect, 22from self-esteem, and 32 for life satisfaction subsamples), and15 cases were missing information about scale length (5 frompersonality, 5 from affect, 1 from self-esteem, and 4 from lifesatisfaction subsamples).

We next limited our overall sample to include only retestcorrelations taken over retest intervals of 15 years or less. Thereason for doing so was that retest correlations over longerretest intervals simply were not available for certain constructs(i.e., affect and self-esteem), which complicates comparisons ofdifferent constructs. As an important goal of our study was tocompare stability and change of different individual differenceconstructs, we included only the range of retest intervals thatwas common across the constructs.

Next, we aggregated all information across cases that wereobtained from the same sample over the same retest intervallength. For example, a three-wave study in which the data werecollected annually could provide two retest correlations over a1-year retest interval (Time 1 to Time 2, and Time 2 to Time 3).In this case, we averaged all information (age, gender compo-sition, retest correlation, etc.) across these two retest periods inorder to obtain only a single retest correlation for each retestinterval length in each sample. If an article reported retestcorrelations for multiple measures or multiple domains (e.g.,different personality subscales or domains), we averaged acrossthese in the same manner. The data used in our meta-analysisare available online as supplemental material.

MASC

We used the nls function of the stats package of the RStatistical Software for our analyses (R Development CoreTeam, 2010). We fit three models to the data. The first modelestimated overall reliability, relative effects of the stable andthe changing factors, and the stability of the changing factorsfrom all available retest correlations. The second model in-cluded moderating effects of construct. This model was used totest whether there were differences in the model parametersacross personality, life satisfaction, self-esteem, and affect.This was estimated by adding product terms between the threedummy variables (that distinguished between the four individ-ual differences of interest) and the model parameters. The third

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

771META-ANALYSIS OF STABILITY AND CHANGE

model included all moderating effects. This allowed us to testwhether, in addition to differences between construct, there wasany evidence of the effects of scale length, gender, and age (andage squared) on model parameters. The complete script for ouranalyses can be found in the Online Supplement.

Results

Data Characteristics

Our final dataset contained 984 retest correlations: 243 forpersonality (179 studies), 106 for affect (59 studies), 206 forself-esteem (100 studies), and 429 for life satisfaction (104 studies;see the Online Supplement for details). The modal number of retestcorrelations per study was 1. Multiwave studies provided multipleretest correlations. For example, a study with three measurementoccasions provides three retest correlations over the interval fromT1 to T2, T2 to T3, and T1 to T3. The relatively high number ofretest correlations for life satisfaction stems is because of theinclusion of a few panel studies with many retest correlations.Table 1 provides descriptive statistics about the samples. As can beseen Table 1, there was substantial variability in scale length, age,and retest interval among the studies.

Overall Model

We first fit a model without moderators. In this model, retestinterval explained 28.8% of the variance in retest correlations.The next model added three dummy variables that distinguishedbetween retest correlations of personality traits and self-esteem(Dummy 1), life satisfaction (Dummy 2), and affect (Dummy3). Personality traits were used as the reference group. Theamount of explained variance increased substantially to 48.3%,$R2 ! .195, F(9, 972) ! 40.8, p % .0001. We then addedgender, age, age2, and scale length as moderators. Scale lengthwas only included as a moderator of the reliability parameter.Inclusion of these moderator variables further increased theamount of explained variance by 10% to 58.3%, $R2 ! .100,F(10, 962) ! 23.0, p % .0001.

Table 2 shows the estimates of reliability, relative influencesof stable and changing factors, and 1-year and 10-year stabili-ties of the changing factors from the final model. Full modelresults that include effects of moderators and 95% CIs areavailable in the Online Supplement. The results are also de-picted in Figure 3, which shows the observed retest correlationsand the model estimates for personality traits, life satisfaction,self-esteem, and affect for different age groups.

The reliability estimate for personality traits was .72, 95%CI ! (.65, .78). The results suggested that affect measures wereless reliable than measures of other constructs (difference be-tween measures of personality traits and affect was #.14, 95%CI ! (#.21, #.06)). One plausible reason for this finding is thataffect measures sometimes have shorter time frames (days,weeks). Thus, affect ratings are more likely to change soquickly that it becomes impossible to distinguish real changesfrom random measurement error. Unfortunately, we were notable to test this hypothesis because many studies failed toprovide adequate information about the time frame of affectmeasures. However, for the purpose of comparisons with otherconstructs, the distinction between short-term fluctuations inmood and random measurement error is not important.

The second and third columns of Table 2 show the contribu-tion of stable and changing factors to individual differencesacross different constructs. The results show that stable factorsaccounted for 83% of individual differences in personalitytraits; the remaining 17% of the reliable variance was explainedby changing factors. As predicted, changing factors explainedsubstantially more variance in affect (difference of 41%), lifesatisfaction (difference of 31%), and self-esteem (difference of27%) than personality traits. The results for life satisfaction andself-esteem are very similar to each other, whereas affect ap-pears to be even more strongly influenced by changing factors.However, the CI around this estimate is large and overlapssubstantially with those for life satisfaction and self-esteem (seeTable 1 in the online supplemental material). Before more databecome available, our main conclusion is that personality traitsare more stable than well-being and self-esteem.

The fourth column shows the estimates of the 1-year stabilityof the changing factors for personality traits, life satisfaction,self-esteem, and affect. We found significant but small stabilityof changing factors that influence personality (.25). With thissmall amount of stability, retest correlations reach the asymp-tote set by stable factors very quickly (Figure 3). Indeed, the1-year stability of .25 implies that after 2 years only 0.004% ofvariance in changing factors remains ((.252)2). We found strongevidence that the changing factors that influence well-being andself-esteem are considerably more stable. Our estimate of the1-year stability of the changing factors for life satisfaction (.78)is consistent with previous estimates of about .80. Parameterestimates for affect and self-esteem are similar and CIs overlap.However, it is important to note that even 1-year stabilities ashigh as .80 would lead to large changes over longer periods oftime because small changes accumulate over time. To illustrate

Table 1Means (and SDs) for Age, Gender Distribution, Scale Length, Retest Interval, and RetestCorrelation for the Four Types of Individual Differences and the Overall Sample

Personality Affect Self-esteem Life satisfaction Overall

Age 29.1 (16.5) 33.2 (18.6) 18.5 (12.5) 40.8 (21.4) 32.4 (20.2)Female (%) 46.5 54.9 48.7 50.7 49.7Scale length 22.7 (17.0) 9.5 (5.8) 12.4 (16.4) 2.0 (2.9) 10.1 (14.3)Retest interval 4.5 (3.6) 3.0 (3.2) 3.2 (3.1) 6.1 (4.1) 4.8 (3.9)Retest correlations 243 106 206 429 984

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

772 ANUSIC AND SCHIMMACK

this we also show 10-year stability estimates in the fifth columnof Table 2. As this column shows, virtually no variance inchanging factors of life satisfaction and self-esteem remainsafter 10-years (e.g., .092 ! .01).

In sum, the most important result was that the between-personvariance in personality traits is notably more stable than variance

in well-being and self-esteem, reflecting greater influences ofstable factors on personality traits. Moreover, the changing factorsthat influence personality traits tend to change more rapidly thanthose for well-being and self-esteem. Because our analyses con-trolled for potential differences in reliability, this finding cannot beattributed to measurement artifacts.

Table 2Estimates of the Model Parameters From Nonlinear Regression With Construct, Scale Length,Age, and Gender as Moderators of Retest Correlations

Reliability(1 # Error)

Stability(1 # Change) Change

1-year stability of theChange component

10-year stability of theChange component

Personality .72 .83 .17 .25 .00Affect .58 .42 .58 .88 .28Self-esteem .71 .56 .44 .79 .09Life satisfaction .67 .52 .48 .78 .08

Note. See online supplemental material for complete results with 95% confidence intervals and estimates ofmoderating effects of scale length, gender, and age.

Figure 3. Observed retest correlations as a function of retest interval. x ! under 20 years of age, dot ! between20 and 40 years of age, triangle ! older than 60. Regression lines illustrate model predictions for average age(solid gray line), 15-year-olds (dotted line), 30-year-olds (solid black line), and 60-year-olds (dashed lines).

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

773META-ANALYSIS OF STABILITY AND CHANGE

Additional Moderators

As predicted by psychometric theory, scale length was a signif-icant moderator of reliability, estimate ! .04, 95% CI ! (.03, .05).It is interesting that life satisfaction judgments were as reliable aspersonality measures after controlling for the effect of scale length.This finding undermines the claims that life satisfaction judgmentsare highly unreliable and susceptible to influence of mood andpriming effects (Schwarz & Strack, 1999). The results are ratherconsistent with the view that life satisfaction judgments are basedon chronically accessible information and provide reliable infor-mation about individuals’ well-being (Schimmack, Diener, & Oi-shi, 2002; Schimmack & Oishi, 2005).

Gender was not a significant moderator of any of the modelparameters. Meta-analyses have relatively low power to detectgender effects because there is relatively little variation in thegender composition of samples. However, original data also pro-duced similar results for women and men (Schimmack & Lucas,2010). Thus, our results are likely to generalize across genders.

Age had statistically significant linear and quadratic effects onthe effect of stable versus changing factors. The coefficient esti-mates can be interpreted as differences in parameters betweenpeople who are 10 years apart in age. The linear estimate for themoderating effect of age on the extent to which stable factorsinfluence personality is .10, 95% CI ! (.07, .02), and the quadraticestimate is #.02, 95% CI ! (#.03, #.02). This suggests that theinfluence of stable factors on personality increases with age, butthe increase is greater early on than later in life. For example,according to the model, stable factors contribute 83% of reliablebetween-person variance in personality traits for 32-year-olds (av-erage age), 55% for 12-year-olds, and 95% for 52-year-olds.1 Thisfinding is consistent with the findings from Roberts and DelVec-chio’s (2000) meta-analysis that personality stability increaseswith age, particularly during adolescence and early adulthood(Figure 3).

Discussion

The stability of individual difference constructs has been one ofthe most debated issues in personality psychology. In this article,we used a novel meta-analytic method, MASC, to estimate thecontribution of stable influences to retest correlations of person-ality traits, life satisfaction, affect, and self-esteem. Our findingssupport the hypothesis that stable factors have a stronger influenceon personality traits than on the other individual differences. Thisdifference cannot be attributed to differences in the reliability ofpersonality and well-being measures because our model controlsfor differences in reliability.

We also were able for the first time to compare the stability ofthe changing factors across different individual differences. Wefound that the changing factors that influence personality traits arenot very stable. In contrast, we replicated the finding of highstability for the changing factors that influence life satisfaction.We also found high stability for the changing factors that influenceself-esteem, consistent with previous findings by Donnellan et al.(2012) and Kuster and Orth (2013). Our finding that stabilitydiffers across constructs has important implications for identifyingpredictors of stability and change in these constructs. For example,if personality traits were influenced by the same environmental

factors as well-being, the changing factors of personality should beas stable as the changing factors of well-being.

There have been few direct comparisons of environmental in-fluences on personality and well-being. Schimmack, Schupp, andWagner (2008) examined the influence of unemployment in anational representative sample of Germans. Consistent with lon-gitudinal studies, unemployment predicted lower life satisfactionand a lower hedonic balance (lower positive affect, higher negativeaffect), but unemployment did not predict personality traits. Thisfinding suggests that unemployment is one of the changing factorsthat can produce changes in well-being that can last several years(Lucas, Clark, Georgellis, & Diener, 2004) without producingsimilar changes in personality traits.

Our results are also consistent with a recent study of 14,000participants that examined predictors of changes in personalityover a 4-year period. Despite the large sample, the study found fewsignificant effects of major life events such as marriage, childbirth,and widowhood on personality traits (Specht et al., 2011). It isimportant that the same sample has been used to demonstrate thatthese environmental factors influence life satisfaction (Lucas,2007).

Nevertheless, our findings suggest that changing factors have aninfluence on personality and refute the “plaster model” that saysthat personality traits are fixed after age 30. Our model merelysuggests that these factors account for a relatively smaller portionof the variance compared with stable factors. Indeed, individuallife events may have rather small effects on personality, but takentogether changing circumstances can contribute to changes inpersonality. Consistent with this idea, several studies have linkedenvironmental events to personality changes (Bleidorn, 2012;Jackson, Thoemmes, Jonkmann, Lüdtke, & Trautwein, 2012; Zim-mermann & Neyer, 2013).

Our study provides additional support for the hypothesis thatstability increases with age. It is interesting that our results indi-cated that this increase is not because of increasing stability of thechanging factors, which would suggest that influences of thechanging factors that contribute to personality become longerlasting with age (e.g., people may hold on to their jobs for longerin older adulthood than in early adulthood). Rather, we foundrelatively stronger influences of stable factors with age. One ex-planation for this finding is that internal biological factors or theenvironmental factors (or both) become more stable with age. Forexample, higher environmental stability later in life (e.g., stableincome and housing) may contribute to increasing contribution ofthe stable factor. That is, the influence of increased income fol-lowing graduation may have permanent effects on life satisfactionand would be considered a new stable influence. People may alsoactively shape their environments to fit their existing predisposi-tions, leading to increased stability of personality over the life span(Fraley & Roberts, 2005). However, results based on retest corre-lations are difficult to interpret because standardized coefficientsonly provide information about the relative proportions of varianceattributed to each model component. Thus, an alternative expla-nation is that stable factors remain constant throughout the lifespan, but there is less influence of changing factors later in life. For

1 .83 & (#2) ! .10 & (#2)2 ! (#.02) ! .55; .83 & (2) ! .10 & (2)2 !(#.02) ! .95.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

774 ANUSIC AND SCHIMMACK

this reason, it is important to examine stability and change ofpersonality and well-being in original studies using structuralequation modeling of covariances (Kenny & Zautra, 1995; Schim-mack & Lucas, 2010). Moreover, more longitudinal twin studiesare needed to examine the contributions of genes and environmentto stability and change. Existing evidence suggests that stabilityincreases with age because environmental factors increasinglycontribute to stability of personality in adulthood (Bleidorn et al.,2014; Briley & Tucker-Drob, 2014).

Our results also provide new evidence regarding the influence ofpersonality on well-being. Schimmack et al. (2002) proposed thatpersonality traits have a direct influence on affect because the BigFive traits like extraversion and neuroticism influence individuals’typical mood levels. In contrast, the influence of personality traitson life satisfaction is indirect; personality only influences lifesatisfaction because people rely on their typical mood levels tojudge their lives. That is, independent of actual life events, peoplewho are typically in a good mood will judge their lives morepositively. This model predicts that personality traits are strongerpredictors of hedonic balance (positive—negative affect) than oflife satisfaction (Schimmack et al., 2008). If personality is morestable, the model also predicts that affect should be more stablethan life satisfaction. Our results provided no evidence of greaterstability of affect over life satisfaction. However, the wide CIsshow that the existing data are inconclusive. In the future, it will beimportant to test this hypothesis in larger samples, using originalpanel data that measured personality, life satisfaction, and affect.

Another limitation of retest correlations is that correlation co-efficients do not take mean level changes into account. Thus, ourresults do not contradict findings that average levels of personalitychange with age (e.g., Roberts et al., 2008). Mean level changeshave been demonstrated and may explain some of the changingvariance in our model. The reason is that individuals are unlikelyto change at the very same moment to exactly the same degree. Forexample, the average level of conscientiousness increases consid-erably between ages 20 and 30. However, if Janice’s conscien-tiousness increases considerably from age 22 to 24 and Jason’sconscientiousness increases considerably from age 25 to age 28,our model would show some short-term changes in the rank-orderof conscientiousness. To examine this issue further, it would benecessary to model mean-level and rank-order stability and changewith original data.

The finding that the stability of personality traits is higher thanthe stability of other psychological constructs can have interestingimplications for clinical applications. Specifically, personalitytraits in adulthood appear to be highly stable. Indeed, cliniciansoften report difficulties in treating conditions such as personalitydisorders. However, life satisfaction and affect are less stable.Thus, although it may be difficult to increase quality of life bychanging one’s personality, it would be possible to design inter-ventions to increase life satisfaction, affective experiences, andself-esteem. This being said, it is also important to recognize thatthe stable factors in our model are only stable given the naturallyoccurring circumstances in which the studies were conducted.Future research needs to uncover the actual stable factors, and abetter understanding of these factors may suggest novel ways tochange them. Still, we believe our results can give clinicians someidea about the constraints on change and the efforts that may berequired to induce lasting changes in psychological characteristics.

Limitations and Future Directions

In this article, we demonstrated how a novel meta-analyticapproach can be applied to study the stability and change ofindividual differences. The main advantage of this approach is thatit treats both stability and change of constructs as theoreticallyimportant. Furthermore, this method is able to separate transientinfluences that lead to changes over short periods of time(Chmielewski & Watson, 2009) from longer lasting changes thatare more likely related to important changes in life circumstances.

An important limitation of this approach, as with other trait-stateapproaches, is that the retest correlations must asymptote duringthe duration of the study in order for the model parameters to beestimated correctly (e.g., Anusic et al., 2012). If the asymptote isnot reached over the study interval, the model may have difficultyseparating the influences of the stable factors from those of thechanging factors. Another limitation particular to the nonlinearregression approach to modeling stability and change is that theapproach presented in this paper can be used only for descriptivepurposes (e.g., to estimate relative proportions of source variance).It does not offer an explanation of the particular mechanismsresponsible for stability and change of any particular construct.However, it may be possible to extend this model to includebivariate correlations (i.e., correlations of one variable at one timewith another variable at another time). This may provide usefulinsight in the dynamics of covariation between variables over time.

Finally, it is important to note that some studies in this meta-analysis provided multiple retest correlations. Ideally, nonlinearmultilevel models should be used in this case to account fornonindependence among the data. However, nonlinear multilevelmodels tend to be difficult to estimate when within-study data arescarce. In this case there were many studies that provided only oneretest correlation, which resulted in difficulties with model con-vergence. A reasonable approximation is to use a simple nonlinearmodel. As a result, CIs should be interpreted with caution and arelikely to be even wider than suggested in this article. Thus, it iseven more important to conduct longitudinal studies that canprovide more conclusive evidence about stability and change ofimportant individual differences. At present, our meta-analysisprovides the best scientific evidence on this important issue.

Despite these limitations, we believe that the meta-analyticapproach presented in this paper is advantageous to other modelsin literature. Such an approach provides more informative esti-mates of the sources of stability. Moreover, this approach can beused to analyze archival data for which only retest correlations areavailable. Further research needs to consider conditions underwhich this model can and cannot be identified in order to furtherfacilitate feasibility of this useful statistical method. Another im-portant direction for future research is to use MASC to examinethe stability and change of other personality characteristics. Forexample, Conley (1984) proposed a hierarchy of consistency withintelligence as the most stable construct. MASC makes it possibleto compare parameter estimates for different constructs and tobuild a comprehensive hierarchy of stability of individual differ-ences.

References

Adams, S. H. (1994). Role of hostility in women’s health during midlife:A longitudinal study. Health Psychology, 13, 488–495.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

775META-ANALYSIS OF STABILITY AND CHANGE

Adams, V. H., III. (1997). A paradox in African American quality of life.Social Indicators Research, 42, 205–219.

Akse, J., Hale, W. W., III, Engels, R. C. M. E., Raaijmakers, Q. A. W., &Meeus, W. H. J. (2007). Stability and change in personality type mem-bership and anxiety in adolescence. Journal of Adolescence, 30, 813–834.

Alfonso, V. C., Allison, D. B., Rader, D. E., & Gorman, B. S. (1996). Theextended satisfaction with life scale: Development and psychometricproperties. Social Indicators Research, 38, 275–301.

Allemand, M., Zimprich, D., & Hertzog, C. (2007). Cross-sectional agedifferences and longitudinal age changes of personality in middle adult-hood and old age. Journal of Personality, 75, 323–358.

Allen, J. P., Hauser, S. T., Bell, K. L., & O’Connor, T. G. (1994).Longitudinal assessment of autonomy and relatedness in adolescent-family interactions as predictors of adolescent ego development andself-esteem. Child Development, 65, 179–194.

Alsaker, F. D., & Olweus, D. (1992). Stability of global self-evaluations inearly adolescence: A cohort longitudinal study. Journal of Research onAdolescence, 2, 123–145.

Anusic, I., Lucas, R. E., & Donnellan, M. B. (2012). Dependability ofpersonality, life satisfaction, and affect in short-term longitudinal data.Journal of Personality, 80, 33–58. http://dx.doi.org/10.1111/j.1467-6494.2011.00714.x

Arnold, J. (1998). Do self-ideal discrepancies and global self-esteemamount to the same thing? An empirical investigation. British Journal ofGuidance and Counselling, 16, 190–202.

Atkinson, T. (1982). The stability and validity of quality of life measures.Social Indicators Research, 10, 113–132.

Baird, B. M., Le, K., & Lucas, R. E. (2006). On the nature of intraindi-vidual personality variability: Reliability, validity, and associations withwell-being. Journal of Personality and Social Psychology, 90, 512–527.

Baltes, P. B., & Nesselroade, J. R. (1972). Cultural change and adolescentpersonality development: An application of longitudinal sequences. De-velopmental Psychology, 7, 244–256.

Bates, M. E., & Pandina, R. J. (1989). Individual differences in the stabilityof personality needs: Relations to stress and substance use during ado-lescence. Personality and Individual Differences, 10, 1151–1157.

Baur, P. A., & Okun, M. A. (1983). Stability of life satisfaction in late life.The Gerontologist, 23, 261–265.

Blais, M. R., Vallerand, R. J., Pelletier, L. G., & Brière, N. M. (1989).L’Échelle de satisfaction de vie: Validation canadienne-française du“Satisfaction With Life Scale” [The Life Satisfaction Scale: Validationof the French Canadian “Satisfaction with Life Scale”]. Canadian Jour-nal of Behavioural Science/Revue canadienne des sciences du comporte-ment, 21, 210–223.

Bleidorn, W. (2012). Hitting the road to adulthood: Short-term personalitydevelopment during a major life transition. Personality and SocialPsychology Bulletin, 38, 1594 –1608. http://dx.doi.org/10.1177/0146167212456707

Bleidorn, W., Kandler, C., & Caspi, A. (2014). The behavioural genetics ofpersonality development in adulthood: Classic, contemporary, and futuretrends. European Journal of Personality, 28, 244–255. http://dx.doi.org/10.1002/per.1957

Bleidorn, W., Kandler, C., Riemann, R., Angleitner, A., & Spinath, F. M.(2009). Patterns and sources of adult personality development: GrowthCurve analyses of the NEO PI-R scales in a longitudinal twin study.Journal of Personality and Social Psychology, 97, 142–155.

Block, J., & Robins, R. W. (1993). A longitudinal study of consistency andchange in self-esteem from early adolescence to early adulthood. ChildDevelopment, 64, 909–923.

Bolton, B. (1979). Longitudinal stability of the primary and secondarydimensions of the 16PF-E. Multivariate Experimental Clinical Re-search, 4, 67–71.

Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2003). The theo-retical status of latent variables. Psychological Review, 110, 203–219.http://dx.doi.org/10.1037/0033-295X.110.2.203

Bowling, A., & Farquhar, M. (1996). Associations with changes in lifesatisfaction among three samples of elderly people living at home.International Journal of Geriatric Psychiatry, 11, 1077–1087.

Brief, A. P., Butcher, A. H., George, J. M., & Link, K. E. (1993).Integrating bottom-up and top-down theories of subjective well-being:The case of health. Journal of Personality and Social Psychology, 64,646–653.

Briley, D. A., & Tucker-Drob, E. M. (2014). Genetic and environmentalcontinuity in personality development: A meta-analysis. PsychologicalBulletin, 140, 1303–1331. http://dx.doi.org/10.1037/a0037091

Bromberger, J. T., & Matthews, K. A. (1996). A “feminine” model ofvulnerability to depressive symptoms: A longitudinal investigation ofmiddle-aged women. Journal of Personality and Social Psychology, 70,591–598.

Busseri, M. A., Sadava, S. W., & Decourville, N. (2007). A hybrid modelfor research on subjective well-being: Examining common- andcomponent-specific sources of variance in life satisfaction, positiveaffect, and negative affect. Social Indicators Research, 83, 413–445.

Cairns, E., McWhirter, L., Duffy, U., & Barry, R. (1990). The stability ofself-concept in late adolescence: Gender and situational effects. Person-ality and Individual Differences, 11, 937–944.

Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality ofAmerican life: Perceptions, evaluations, and satisfactions. New York,NY: Russell Sage Foundation.

Cantoni, L. J. (1955). A study in emotional adjustment: The correlation ofstudent and adult forms of the Bell Adjustment Inventory over a periodof thirteen years. Educational and Psychological Measurement, 15,137–143.

Caputo, D. V., Psathas, G., & Plapp, J. M. (1966). Test–retest reliability ofthe EPPS. Educational and Psychological Measurement, 26, 883–886.

Carmelli, D., Rosenman, R. H., & Chesney, M. A. (1987). Stability of theType A structured interview and related questionnaires in a 10-yearfollow up of an adult cohort of twins. Journal of Behavioral Medicine,10, 513–525.

Carter, S. D. (2004). Re-examining the temporal aspects of affect: Rela-tionships between repeatedly measured affective state, subjective well-being, and affective disposition. Personality and Individual Differences,36, 381–391.

Caspi, A., & Roberts, B. W. (2001). Personality development across thelife course: The argument for change and continuity. PsychologicalInquiry, 12, 49–66. http://dx.doi.org/10.1207/S15327965PLI1202_01

Chamberlain, K., & Zika, S. (1992). Stability and change in subjectivewell-being over short time periods. Social Indicators Research, 26,101–117.

Charles, E. P. (2005). The correction for attenuation due to measurementerror: Clarifying concepts and creating confidence sets. PsychologicalMethods, 10, 206–226. http://dx.doi.org/10.1037/1082-989X.10.2.206

Charman, T. (1994). The stability of depressed mood in young adolescents:A school-based survey. Journal of Affective Disorders, 30, 109–116.

Chiriboga, D. A. (1982). Consistency in adult functioning: The influence ofsocial stress. Ageing and Society, 2, 7–29.

Chmielewski, M., & Watson, D. (2009). What is being assessed and whyit matters: The impact of transient error on trait research. Journal ofPersonality and Social Psychology, 97, 186–202. http://dx.doi.org/10.1037/a0015618

Coleman, P. G., Ivani-Chalian, C., & Robinson, M. (1993). Self-esteemand its sources: Stability and change in later life. Ageing and Society, 13,171–192.

Conley, J. J. (1984). The hierarchy of consistency: A review and model oflongitudinal findings on adult individual differences in intelligence,personality and self-opinion.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

776 ANUSIC AND SCHIMMACK

Costa, P. T. Jr., Herbst, J. H., McCrae, R. R., & Siegler, I. C. (2000).Personality at midlife: Stability, intrinsic maturation, and response to lifeevents. Assessment, 7, 365–378.

Costa, P. T., Jr., & McCrae, R. R. (1978). Age differences in personalitystructure revisited: Studies in validity, stability, and change. Interna-tional Journal of Aging and Human Development, 8, 261–275.

Costa, P. T., Jr., & McCrae, R. R. (1988). Personality in adulthood: Asix-year longitudinal study of self-reports and spouse ratings on the NEOpersonality inventory. Journal of Personality and Social Psychology, 54,853–863.

Costa, P. T., Jr., & McCrae, R. R. (1992). Multiple uses for longitudinalpersonality data. European Journal of Personality, 6, 85–102.

Costa, P. T., Jr., McCrae, R. R., & Arenberg, D. (1980). Enduring dispo-sitions in adult males. Journal of Personality and Social Psychology, 38,793–800.

Costa, P. T., Jr., McCrae, R. R., & Norris, A. H. (1981). Personal adjust-ment to aging: Longitudinal prediction from neuroticism and extraver-sion. Journal of Gerontology, 36, 78–85.

Costa, P. T., Jr., McCrae, R. R., & Zonderman, A. B. (1987). Environ-mental and dispositional influences on well-being: Longitudinalfollow-up of an American national sample. British Journal of Psychol-ogy, 78, 299–306.

Cramer, D. (1984). Life and job satisfaction: A two-wave panel study.Journal of Psychology, 129, 261–267.

Crook, M. N. (1943). A retest with the Thurstone Personality Scheduleafter six and one-half years. Journal of General Psychology, 28, 111–120.

Cumming, G. (2014). The new statistics: Why and how. PsychologicalScience, 25, 7–29. http://dx.doi.org/10.1177/0956797613504966

Daleiden, E., Chorpita, B. F., & Lu, W. (2000). Assessment of tripartitefactors of emotion in children and adolescents II: Concurrent validity ofthe affect and arousal scales for children. Journal of Psychopathologyand Behavioral Assessment, 22, 161–182.

Darley, J. G. (1938). Changes in measured attitudes and adjustments.Journal of Social Psychology, 9, 189–199.

Davis, M. H., & Franzoi, S. L. (1991). Stability and change in adolescentself-consciousness and empathy. Journal of Research in Personality, 25,70–87.

Davis, T. N., & Satterly, D. J. (1968). Personality profiles of studentteachers. British Journal of Educational Psychology, 39, 183–187.

Deihl, L. M., Vicary, J. R., & Deike, R. C. (1997). Longitudinal trajectoriesof self-esteem from early to middle adolescence and related psychoso-cial variables among rural adolescents. Journal of Research on Adoles-cence, 7, 393–411.

Diener, E., Emmons, R. A., Larsen, R. J., & Griffen, S. (1985). Thesatisfaction with life scale. Journal of Personality Assessment, 49, 71–75.

Diener, E., & Lucas, R. E. (2000). Explaining differences in societal levelsof happiness: Relative standards, need fulfillment, culture, and evalua-tion theory. Journal of Happiness Studies, 1, 41–78. http://dx.doi.org/10.1023/A:1010076127199

Donnellan, M. B., Conger, R. D., & Burzette, B. G. (2005). Criterion-related validity, self-other agreement, and longitudinal analyses for theIowa personality questionnaire: A short alternative to the MPQ. Journalof Research in Personality, 39, 458–485.

Donnellan, M. B., Conger, R. D., & Burzette, R. G. (2007). Personalitydevelopment from late adolescence to young adulthood: Differentialstability, normative maturity, and evidence for the maturity-stabilityhypothesis. Journal of Personality, 75, 237–263.

Donnellan, M. B., Kenny, D. A., Trzesniewski, K. H., Lucas, R. E., &Conger, R. D. (2012). Using trait-state models to evaluate the longitu-dinal consistency of global self-esteem from adolescence to adulthood.Journal of Research in Personality, 46, 634–645. http://dx.doi.org/10.1016/j.jrp.2012.07.005

Duske, J. B., & Flaherty, J. F. (1981). The development of the self-conceptduring the adolescent years. Monographs of the Society for Research inChild Development, 46 (Serial No. 191).

Eccles, J. S., Wigfield, A., Flanagan, C. A., Miller, C., Reuman, D. A., &Yee, D. (1989). Self-concepts, domain values, and self-esteem: Rela-tions and changes at early adolescence. Journal of Personality, 57,283–310.

Eells, L. W. (1985). The effect of role change on physical health, mentalhealth and general life satisfaction: A panel analysis (Doctoral disser-tation). Available from ProQuest Dissertations and Theses database.(UMI No. 303396505)

Elliott, M. (1996). Impact of work, family, and welfare receipt on women’sself-esteem in young adulthood. Social Psychology Quarterly, 59, 80–95.

Engel, M. (1959). The stability of the self-concept in adolescence. Journalof Abnormal and Social Psychology, 58, 211–217.

Farnsworth, P. R. (1938). A genetic study of the Bernreuter PersonalityInventory. Pedagogical Seminar and Journal of Genetic Psychology, 52,3–13.

Felson, R. B., & Zielinski, M. A. (1989). Children’s self-esteem andparental support. Journal of Marriage and Family, 51, 727–735.

Fenzel, L. M. (2000). Prospective study of changes in global self-worth andstrain during the transition to middle school. Journal of Early Adoles-cence, 20, 93–116.

Ferguson, C. J. (2010). A meta-analysis of normal and disordered person-ality across the life span. Journal of Personality and Social Psychology,98, 659–667. http://dx.doi.org/10.1037/a0018770

Fincham, F. D., & Bradbury, T. N. (1993). Marital satisfaction, depression,and attributions: A longitudinal analysis. Journal of Personality andSocial Psychology, 64, 442–452.

Fraley, R. C., & Roberts, B. W. (2005). Patterns of continuity: A dynamicmodel for conceptualizing the stability of individual differences inpsychological constructs across the life course. Psychological Review,112, 60–74. http://dx.doi.org/10.1037/0033-295X.112.1.60

Gerrard, M., Gibbons, F. X., Reis-Bergan, M., & Russell, D. W. (2000).Self-esteem, self-serving cognitions, and health risk behavior. Journal ofPersonality, 68, 1177–1201.

Glatzer, W. (1991). Quality of life in advanced industrialized countries:The case of West Germany. In F. Strack, M. Argyle, & N. Schwarz(Eds.), Subjective well-being: An interdisciplinary perspective (pp. 261–79). Oxford, England: Pergamon Press.

Gold, D., & Andres, D. (1985). Personality test reliability: Correlates forolder subjects. Journal of Personality Assessment, 49, 530–532.

Gold, S. R., & Henderson, B. B. (1990). Daydreaming and curiosity:Stability and change in gifted children and adolescents. Adolescence, 25,701–708.

Granleese, J., & Joseph, S. (1994). Reliability of the Harter Self-PerceptionProfile for Children and predictors of global self-worth. The Journal ofGenetic Psychology, 155, 487–492.

Grigoriadis, S., & Fekken, G. C. (1992). Person reliability of the MinnesotaMultiphasic Personality Inventory. Personality and Individual Differ-ences, 13, 491–500.

Hampson, S. E., & Goldberg, L. R. (2006). A first large cohort study ofpersonality trait stability over the 40 years between elementary schooland midlife. Journal of Personality and Social Psychology, 91, 763–779.

Hare, B. R. (1985). Stability and change in self-perception and achieve-ment among black adolescents: A longitudinal study. The Journal ofBlack Psychology, 11, 29–42.

Hawkins, D. N., & Booth, A. (2005). Unhappily ever after: Effects oflong-term, low-quality marriages on well-being. Social Forces, 84,446–465.

Headey, B., & Wearing, A. (1989). Personality, life events, and subjectivewell-being: Toward a dynamic equilibrium model. Journal of Person-ality and Social Psychology, 57, 731–739.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

777META-ANALYSIS OF STABILITY AND CHANGE

Heise, D. R. (1969). Separating reliability and stability in test–retestcorrelation. American Sociological Review, 34, 93–101. http://dx.doi.org/10.2307/2092790

Helson, R., & Moane, G. (1987). Personality change in women fromcollege to midlife. Journal of Personality and Social Psychology, 53,176–186.

Helson, R., & Wink, P. (1992). Personality change in women from theearly 40s to the early 50s. Psychology and Aging, 7, 46–55.

Hirsch, B. J., & Rapkin, B. D. (1987). The transition to junior high school:A longitudinal study of self-esteem, psychological symptomatology,school life, and social support. Child Development, 58, 1235–1243.

Hoffmann, H., & Wehler, R. (1978). Pre- and posttraining MMPI scores ofwomen alcoholism counselors. Journal of Studies on Alcohol, 39, 1952–1955.

Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiencespredicting changes in self-esteem of sixth- and seventh-grade students.Journal of Educational Psychology, 82, 117–127.

Holahan, C. K. (1981). Lifetime achievement patterns, retirement and lifesatisfaction of gifted aged women. Journal of Gerontology, 36, 741–749.

Holmlund, U. (1991). Change and stability of needs from middle adoles-cence to young adulthood in Swedish females. European Journal ofPersonality, 5, 379–385.

Holsen, I., Kraft, P., & Vittersø, J. (2000). Stability in depressed mood inadolescence: Results from a 60-year longitudinal panel study. Journal ofYouth and Adolescence, 29, 61–78.

Horley, J., & Lavery, J. J. (1991). The stability and sensitivity of subjectivewell-Being measures. Social Indicators Research, 24, 113–122.

Huebner, E. S., Funk III, B. A., & Gilman, R. (2000). Cross-sectional andlongitudinal psychosocial correlates of adolescent life satisfaction re-ports. Canadian Journal of School Psychology, 16, 53–64.

Hunter, S. M., Johnson, C. C., Vizelberg, I. A., Webber, L. S., & Berenson,G. S. (1991). Tracking Type A behavior in children and young adults:The Bogalusa heart study. Journal of Social Behavior and Personality,6, 71–84.

Jackson, J. J., Thoemmes, F., Jonkmann, K., Lüdtke, O., & Trautwein,U. (2012). Military training and personality trait development: Doesthe military make the man, or does the man make the military?Psychological Science, 23, 270 –277. http://dx.doi.org/10.1177/0956797611423545

Johnson, W., McGue, M., & Krueger, R. F. (2005). Personality stability inlate adulthood: A behavioral genetic analysis. Journal of Personality, 73,523–551.

Kandler, C., Bleidorn, W., Riemann, R., Spinath, F. M., Thiel, W., &Angleitner, A. (2010). Sources of cumulative continuity in personality:A longitudinal multiple-rater twin study. Journal of Personality andSocial Psychology, 98, 995–1008. http://dx.doi.org/10.1037/a0019558

Keller, M. C., Coventry, W. L., Heath, A. C., & Martin, N. G. (2005).Widespread evidence for non-additive genetic variation in Cloninger’sand Eysenck’s personality dimensions using a twin plus sibling design.Behavior Genetics, 35, 707–721.

Keltikangas-Järvinen, L. (1988). Stability of type A behavior during ado-lescence, young adulthood, and adulthood. Journal of Behavioral Med-icine, 12, 387–396.

Keltikangas-Järvinen, L. (1990). The stability of self-concept during ado-lescence and early adulthood: A six-year follow-up study. The Journalof General Psychology, 117, 361–368.

Kenny, D. A., & Zautra, A. (1995). The trait-state-error model for multi-wave data. Journal of Consulting and Clinical Psychology, 63, 52–59.http://dx.doi.org/10.1037/0022-006X.63.1.52

Kunzmann, U., Little, T. D., & Smith, J. (2000). Is age-related stability ofsubjective well-being a paradox? Cross-sectional and longitudinal evi-dence from the Berlin Aging Study. Psychology and Aging, 15, 511–526.

Kuster, F., & Orth, U. (2013). The long-term stability of self-esteem: Itstime-dependent decay and nonzero asymptote. Personality and SocialPsychology Bulletin, 39, 677– 690. http://dx.doi.org/10.1177/0146167213480189

Lang, F. R., & Heckhausen, J. (2001). Perceived control over developmentand subjective well-being: Differential benefits across adulthood. Jour-nal of Personality and Social Psychology, 81, 509–523.

Larson, R. W., Moneta, G., Richards, M. H., & Wilson, S. (2002). Conti-nuity, stability, and change in daily emotional experience across adoles-cence. Child Development, 73, 1151–1165.

Leon, G. R., Gillum, B., Gillum, R., & Gouze, M. (1979). Personalitystability and change over a 30-year period: Middle age to old age.Journal of Consulting and Clinical Psychology, 47, 517–524.

Lepper, H. S. (1998). Use of other-reports to validate subjective well-beingmeasures. Social Indicators Research, 44, 367–379.

Ligon, M. B. (2007). Improving life satisfaction of elders through oralhistory: The narrator’s perspective (Doctoral dissertation). Availablefrom ProQuest Dissertation and Theses database. (UMI No. 304705209)

Lintunen, T., Leskinen, E., Oinonen, M., Salinto, M., & Rahkila, P. (1995).Change, reliability, and stability in self-perceptions in early adolescence:A four-year follow-up study. International Journal of Behavioral De-velopment, 18, 351–364.

Lovibond, P. F. (1998). Long-term stability of depression, anxiety, andstress syndrome. Journal of Abnormal Psychology, 107, 520–526.

Lucas, R. E. (2007). Adaptation and the set-point model of subjectivewell-being: Does happiness change after major life events? CurrentDirections in Psychological Science, 16, 75–79. http://dx.doi.org/10.1111/j.1467-8721.2007.00479.x

Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2004). Unem-ployment alters the set point for life satisfaction. Psychological Science,15, 8–13. http://dx.doi.org/10.1111/j.0963-7214.2004.01501002.x

Lucas, R. E., Diener, E., & Suh, E. (1996). Discriminant validity ofwell-being measures. Journal of Personality and Social Psychology, 71,616–628.

Lucas, R. E., & Donnellan, M. B. (2007). How stable is happiness? Usingthe START S model to estimate the stability of life satisfaction. Journalof Research in Personality, 41, 1091–1098. http://dx.doi.org/10.1016/j.jrp.2006.11.005

Lucas, R. E., & Donnellan, M. B. (2012). Estimating reliability of single-item life satisfaction measures: Results from four national panel studies.Social Indicators Research, 105, 323–331. http://dx.doi.org/10.1007/s11205-011-9783-z

Lyubomirsky, S., & Lepper, H. S. (1999). A measure of subjective hap-piness: Preliminary reliability and construct validation. Social IndicatorsResearch, 46, 137–155.

Magnus, K., Diener, E., Fujita, F., & Pavot, W. (1993). Extraversion andneuroticism as predictors of objective life events: A longitudinal anal-ysis. Journal of Personality and Social Psychology, 65, 1046–1053.

Mann, J. H. (1958). Self-ratings and the EPPS. Journal of Applied Psy-chology, 42, 267–268.

Marks, G. N., & Fleming, N. (1999). Influences and consequences ofwell-being among Australian young people: 1980–1995. Social Indica-tors Research, 46, 301–323.

Marsh, H. W. (1996). Physical self-description questionnaire: Stability anddiscriminant validity. Research Quarterly for Exercise and Sport, 67,249–264.

Marsh, H. W., Craven, R., & Debus, R. (1998). Structure, stability, anddevelopment of young children’s self-concepts: A multicohort-multioccasion study. Child Development, 69, 1030–1053.

Marsh, H. W., & Grayson, D. (1994). Longitudinal stability of latent meansand individual differences: A unified approach. Structural EquationModeling, 1, 317–359.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

778 ANUSIC AND SCHIMMACK

Marsh, H. W., & Smith, I. D. (1982). Multitrait-multimethod analyses oftwo self-concept instruments. Journal of Educational Psychology, 74,430–440.

McCrae, R. R., & Costa, P. T., Jr. (1991). Adding liebe und arbeit: The fullfive-factor model and well-being. Personality and Social PsychologyBulletin, 17, 227–232.

McGue, M., Bacon, S., & Lykken, D. T. (1993). Personality stability andchange in early adulthood: A behavioral genetic analysis. DevelopmentalPsychology, 29, 96–109.

McGuire, S., Manke, B., Saudino, K. J., Reiss, D., Hetherington, E. M., &Plomin, R. (1999). Perceived competence and self-worth during adoles-cence: A longitudinal behavioral genetic study. Child Development, 70,1283–1296.

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 onthe big five dimensions? A longitudinal study from ages 5 to 7. Journalof Personality and Social Psychology, 89, 90–106.

Melamed, A. R., Silverman, M. S., & Lewis, G. J. (1974). Three yearfollow-up of women religious on the 16 personality factor questionnaire.Review of Religious Research, 15, 64–70.

Mendelson, B. K., White, D. R., & Mendelson, M. J. (1996). Self-esteemand body esteem: Effects of gender, age, and weight. Journal of AppliedDevelopmental Psychology, 17, 321–346.

Meyer, J. M., Heath, A. C., Eaves, L. J., Mosteller, M., & Schieken, R. M.(1988). The predictive power of Cattell’s Personality Questionnaires: Aneighteen-month prospective study. Personality and Individual Differ-ences, 9, 203–212.

Moneta, G. B., Schneider, B., & Csikszentmihalyi, M. (2001). A longitu-dinal study of the self-concept and experiential components of self-worth and affect across adolescence. Applied Developmental Science, 5,125–142.

Morizot, J., & Le Blanc, M. (2003). Continuity and change in personalitytraits from adolescence to midlife: A 25-year longitudinal study com-paring representative and adjudicated men. Journal of Personality, 71,705–756.

Mortimer, J. T., Finch, M. D., & Kumba, D. (1982). Persistence and changein development: The multidimensional self-concept. Life-Span Develop-ment and Behavior, 4, 263–313.

Mortimer, J. T., & Lorence, J. (1981). Self-concept stability and changefrom late adolescence to early adulthood. Research in Community andMental Health, 2, 5–42.

Mroczek, D. K., & Spiro, A., III. (2003). Modeling intraindividual changein personality traits: Findings from the normative aging study. TheJournals of Gerontology, 58B, 153–165.

Muldoon, O. T. (2000). Social group membership and self-perceptions inNorthern Irish children: A longitudinal study. British Journal of Devel-opmental Psychology, 18, 65–80.

Muntaner, C., Garcia-Sevilla, L., Fernandez, A., & Torrubia, R. (1988).Personality dimensions, schizotypal and borderline personality traits andpsychosis proneness. Personality and Individual Differences, 9, 257–268.

Nelson, G. (1990). Women’s life strains, social support, coping, andpositive and negative affect: Cross-sectional and longitudinal tests of thetwo-factor theory of emotional well-being. Journal of Community Psy-chology, 18, 239–263.

Nes, R. B., Røysamb, E., Tambs, K., Harris, J. R., & Reichborn-Kjennerud,T. (2006). Subjective well-being: Genetic and environmental contribu-tions to stability and change. Psychological Medicine, 36, 1033–1042.http://dx.doi.org/10.1017/S0033291706007409

Neyer, F. J., & Asendorpf, J. B. (2001). Personality-relationship transac-tion in young adulthood. Journal of Personality and Social Psychology,81, 1190–1204.

Nichols, R. C. (1967). Personality change and the college. AmericanEducational Research Journal, 4, 173–190.

Nolen-Hoeksema, S., Girgus, J. S., & Seligman, M. E. P. (1986). Learnedhelplessness in children: A longitudinal study of depression, achieve-ment, and explanatory style. Journal of Personality and Social Psychol-ogy, 51, 435–442.

Nottelmann, E. D. (1987). Competence and self-esteem during transitionfrom childhood to adolescence. Developmental Psychology, 23, 441–450.

O’Malley, P. M., & Bachman, J. G. (1983). Self-esteem: Change andstability between ages 12 and 23. Developmental Psychology, 19, 257–268.

Ormel, J., & Rijsdijk, F. V. (2000). Continuing change in neuroticismduring adulthood-structural modeling of a 16-year, 5-wave communitystudy. Personality and Individual Differences, 28, 461–478.

Ormel, J., & Schaufeli, W. B. (1991). Stability and change in psychologicaldistress and their relationship with self-esteem and locus of control: Adynamic equilibrium model. Journal of Personality and Social Psychol-ogy, 60, 288–299.

Ostir, G. V., Smith, P. M., Smith, D., & Ottenbacher, K. J. (2005).Reliability of the positive and negative affect schedule (PANAS) inmedical rehabilitation. Clinical Rehabilitation, 19, 767–769.

Pavot, W., & Diener, E. (1993a). The affective and cognitive context ofself-reported measures of subjective well-being. Social Indicators Re-search, 28, 1–20.

Pavot, W., & Diener, E. (1993b). Review of the satisfaction with life scale.Psychological Assessment, 5, 164–172.

Pavot, W., Diener, E., Colvin, C. R., & Sandvik, E. (1991). Furthervalidation of the satisfaction with life scale: Evidence for the cross-method convergence of well-being measures. Journal of PersonalityAssessment, 57, 149–161.

Pedersen, W. (1991). Mental health, sensation seeking and drug use pat-terns: A longitudinal study. British Journal of Addiction, 86, 195–204.

Plant, W. T., & Telford, C. W. (1966). Changes in personality for groupscompleting different amounts of college over two years. Genetic Psy-chology Monographs, 74, 3–36.

Popham, S. M., Holden, R. R. (1991). Psychometric properties of MMPIfactor scales. Personality and Individual Differences, 12, 513–517.

Pullmann, H., & Allik, J. (2000). The Rosenberg Self-Esteem Scale: Itsdimensionality, stability and personality correlates in Estonian. Person-ality and Individual Differences, 28, 701–715.

Pullmann, H., Raudsepp, L., & Allik, J. (2006). Stability and change inadolescents’ personality: A longitudinal study. European Journal ofPersonality, 20, 447–459.

Putnam, S. H., Kurtz, J. E., & Houts, D. C. (1996). Four-month test–retestreliability of the MMPI-2 with normal male clergy. Journal of Person-ality Assessment, 67, 341–353.

Rammstedt, B., & John, O. P. (2007). Measuring personality in one minuteor less: A 10-item short version of the Big Five Inventory in English andGerman. Journal of Research in Personality, 41, 203–212.

Rantanen, J., Metsäpelto, R. L., Feldt, T., Pulkkinen, L., & Kokko, K.(2007). Long-term stability in the Big Five personality traits in adult-hood. Scandinavian Journal of Psychology, 48, 511–518.

R Development Core Team. (2010). R: A language and environment forstatistical computing. Vienna, Austria: R Foundation for StatisticalComputing.

Reitzes, D. C., Mutran, E. J., & Fernandez, M. E. (1996). Does retirementhurt well-being? Factors influencing self-esteem and depression amongretirees and workers. The Gerontologist, 36, 649–656.

Reker, G. T., & Wong, P. T. P. (1984). Psychological and physicalwell-being in the elderly: The Perceived Well-Being Scale (PWB).Canadian Journal on Aging, 3, 23–32.

Rhee, H. (2000). Prevalence and predictors of headaches in U.S. adoles-cences. Headache, 40, 528–538.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

779META-ANALYSIS OF STABILITY AND CHANGE

Roberts, R. E. L., & Bengtson, V. L. (1996). Affective ties to parents inearly adulthood and self-esteem across 20 years. Social PsychologyQuarterly, 59, 96–106.

Roberts, B. W., Caspi, A., & Moffitt, T. E. (2001). The kids are alright:Growth and stability in personality development from adolescence toadulthood. Journal of Personality and Social Psychology, 81, 670–683.

Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistencyof personality traits from childhood to old age: A quantitative review oflongitudinal studies. Psychological Bulletin, 126, 3–25. http://dx.doi.org/10.1037/0033-2909.126.1.3

Roberts, B. W., Wood, D., & Caspi, A. (2008). The development ofpersonality traits in adulthood. In O. P. John, R. W. Robins, & L. A.Pervin (Eds.), Handbook of personality: Theory and research (3rd ed.,pp. 375–398). New York, NY: Guilford Press.

Robins, R. W., Fraley, R. C., Roberts, B. W., Trzesniewski, K. H. (2001).A longitudinal study of personality change in young adulthood. Journalof Personality, 69, 618–640.

Roeser, R. W., & Eccles, J. S. (1998). Adolescents’ perceptions of middleschool: Relation to longitudinal changes in academic and psychologicaladjustment. Journal of Research on Adolescence, 8, 123–158.

Rubin, R. A. (1978). Stability of self-esteem ratings and their relation toacademic achievement: A longitudinal study. Psychology in the Schools,15, 430–433.

Ryff, C. D. (1989). Happiness is everything, or is it? Explorations on themeaning of psychological well-being. Journal of Personality and SocialPsychology, 57, 1069–1081.

Salmela-Aro, K., & Nurmi, J. (1997). Goal contents, well-being, and lifecontext during transition to university: A longitudinal study. Interna-tional Journal of Behavioral Development, 20, 471–491.

Savin-Wiliams, R. C., & Demo, D. H. (1984). Developmental change andstability in adolescent self-concept. Developmental Psychology, 20,1100–1110.

Schimmack, U., Diener, E., & Oishi, S. (2002). Life-satisfaction is amomentary judgment and a stable personality characteristic: The use ofchronically accessible and stable sources. Journal of Personality, 70,345–384. http://dx.doi.org/10.1111/1467-6494.05008

Schimmack, U., Krause, P., Wagner, G. G., & Schupp, J. (2010). Stabilityand change of well-being: An experimentally enhanced latent state-trait-error analysis. Social Indicators Research, 95, 19–31. http://dx.doi.org/10.1007/s11205-009-9443-8

Schimmack, U., & Lucas, R. E. (2010). Environmental influences onwell-being: A dyadic latent panel analysis of spousal similarity. SocialIndicators Research, 98, 1–21. http://dx.doi.org/10.1007/s11205-009-9516-8

Schimmack, U., & Oishi, S. (2005). The influence of chronically andtemporarily accessible information on life satisfaction judgments. Jour-nal of Personality and Social Psychology, 89, 395–406. http://dx.doi.org/10.1037/0022-3514.89.3.395

Schimmack, U., Radhakrishnan, P., Oishi, S., Dzokoto, V., & Ahadi, S.(2002). Culture, personality, and subjective well-being: Integrating pro-cess models of life satisfaction. Journal of Personality and SocialPsychology, 82, 582–593. http://dx.doi.org/10.1037/0022-3514.82.4.582

Schimmack, U., Schupp, J., & Wagner, G. G. (2008). The influence ofenvironment and personality on the affective and cognitive componentof subjective well-being. Social Indicators Research, 89, 41–60. http://dx.doi.org/10.1007/s11205-007-9230-3

Schmidt, F. L., & Hunter, J. E. (1996). Measurement error in psychologicalresearch: Lessons from 26 research scenarios. Psychological Methods, 1,199–223. http://dx.doi.org/10.1037/1082-989X.1.2.199

Schofield, W. (1953). A study of medical students with the MMPI: II.Group and individual changes after two years. Journal of Psychology,36, 137–141.

Schwarz, N., & Strack, F. (1999). Reports of subjective well-being: Judg-

mental processes and their methodological implications. In D. Kahne-man, E. Diener, & N. Schwartz (Eds.), Well-being: The foundations ofhedonic psychology (pp. 61–84). New York, NY: Russell Sage Foun-dation.

Schyns, P. (2001). Income and satisfaction in Russia. Journal of HappinessStudies, 2, 173–204.

Scollon, C. N., & Diener, E. (2006). Love, work, and changes in extraver-sion and neuroticism over time. Journal of Personality and SocialPsychology, 91, 1152–1165.

Shek, D. T. L. (2007). A longitudinal study of perceived parental psycho-logical control and psychological well-being in Chinese adolescents inHong Kong. Journal of Clinical Psychology, 63, 1–22.

Simon, A., & Thomas, A. (1983). Notes and shorter communications:Means, standard deviations and stability coefficients on the EPI forFurther Education and College of Education students. Personality andIndividual Differences, 4, 95–96.

Small, B. J., Hertzog, C., Hultsch, D. F., & Dixon, R. A. (2003). Stabilityand change in adult personality over 6 years: Findings from the VictoriaLongitudinal Study. Journal of Gerontology, 58B, 3, 166–176.

Specht, J., Bleidorn, W., Denissen, J. J. A., Hennecke, M., Hutteman, R.,Kandler, C., . . . Zimmermann, J. (2014). What drives adult personalitydevelopment? A comparison of theoretical perspectives and empiricalevidence. European Journal of Personality, 28, 216–230. http://dx.doi.org/10.1002/per.1966

Specht, J., Egloff, B., & Schmukle, S. C. (2011). Stability and change ofpersonality across the life course: The impact of age and major lifeevents on mean-level and rank-order stability of the Big Five. Journal ofPersonality and Social Psychology, 101, 862–882. http://dx.doi.org/10.1037/a0024950

Stacey, C. A., & Gatz, M. (1991). Cross-sectional age differences andlongitudinal change on the Bradburn affect balance scale. Journal ofGerontology, 46, 76–78.

Stein, J. A., Newcomb, M. D., & Bentler, P. M. (1986). Stability andchange in personality: A longitudinal study from early adolescence toyoung adulthood. Journal of Research in Personality, 20, 276–291.

Stevens, D. P., & Truss, C. V. (1985). Stability and change in adultpersonality over 12 and 20 years. Developmental Psychology, 21, 568–584.

Stewart, L. H. (1964). Change in personality test scores during college.Journal of Counseling Psychology, 11, 211–220.

Stricker, L. J., & Ross, J. (1964). An assessment of some structuralproperties of the Jungian personality typology. Journal of Abnormal andSocial Psychology, 68, 62–71.

Suh, E., Diener, E., & Fujita, F. (1996). Events and subjective well-Being:Only recent events matter. Journal of Personality and Social Psychol-ogy, 70, 1091–1102.

Tellegen, A. (1982). Brief manual for the Differential Personality Ques-tionnaire. Unpublished manuscript, University of Minneapolis.

Tellegen, A., Lykken, D. T., Bouchard, T. J., Jr., Wilcox, K. J., Segal,N. L., & Rich, S. (1988). Personality similarity in twins reared apart andtogether. Journal of Personality and Social Psychology, 54, 1031–1039.http://dx.doi.org/10.1037/0022-3514.54.6.1031

Terracciano, A., Costa, P. T., Jr., & McCrea, R. R. (2006). Personalityplasticity after age 30. Personality and Social Psychology Bulletin, 32,999–1010.

Terracciano, A., McCrae, R. R., & Costa, P. T., Jr. (2010). Intra-individualchange in personality stability and age. Journal of Research in Person-ality, 44, 31–37. http://dx.doi.org/10.1016/j.jrp.2009.09.006

Thompson, E. R. (2007). Development and validation of an internationallyreliable short-form of the positive and negative affect schedule (PA-NAS). Journal of Cross-Cultural Psychology, 38, 227–242.

Trzesniewski, K. H., Donnellan, M. B., & Robins, R. W. (2003). Stabilityof self-esteem across the life span. Journal of Personality and SocialPsychology, 84, 205–220. http://dx.doi.org/10.1037/0022-3514.84.1.205

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

780 ANUSIC AND SCHIMMACK

Tuddenham, R. D. (1962). Constancy of personal morale over a fifteen-year interval. Child Development, 33, 663–673.

Vaidya, J. G., Gray, E. K., Haig, J., & Watson, D. (2002). On the temporalstability of personality: Evidence for differential stability and the role oflife experiences. Journal of Personality and Social Psychology, 83,1469–1484.

Van Aken, M. A. G., Denissen, J. A., Branje, J. T., Dubas, S. D., &Goossens, L. (2006). Midlife concerns and short-term personalitychange in middle adulthood. European Journal of Personality, 20,497–513.

Veenhoven, R. (1994). Is happiness a trait? Tests of theory that a bettersociety does not make people any happier. Social Indicators Research,32, 101–160.

Viken, R. J., Rose, R. J., Kaprio, J., & Koskenvuo, M. (1994). A devel-opmental genetic analysis of adult personality: Extraversion and neurot-icism from 18 to 59 years of age. Journal of Personality and SocialPsychology, 66, 722–730.

Vittersø, J. (2001). Personality traits and subjective well-being: Emotionalstability, not extraversion, is probably the important predictor. Person-ality and Individual Differences, 31, 903–914.

Wade, T. J., Thompson, V., Tashakkori, A., & Valente, E. (1989). Alongitudinal analysis of sex by race differences in predictors of adoles-cent self-esteem. Personality and Individual Differences, 10, 717–729.

Wallace, J. R., Cunningham, T. F., & Del Monte, V. (1984). Change andstability in self-esteem between late childhood and early adolescence.Journal of Early Adolescence, 4, 253–257.

Waltz, M. (1986). A longitudinal study on environmental and dispositionaldeterminants of life quality: Social support and coping with physicalillness. Social Indicators Research, 18, 71–93.

Watson, D. (2004). Stability versus change, dependability versus error:Issues in the assessment of personality over time. Journal of Research inPersonality, 38, 319–350.

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-dation of brief measures of positive and negative affect: The PANASscales. Journal of Personality and Social Psychology, 54, 1063–1070.

Watson, D., & Humrichouse, J. (2006). Personality development in emerg-ing adulthood: Integrating evidence from self-ratings and spouse ratings.Journal of Personality and Social Psychology, 91, 959–974.

Watson, D., & Walker, L. M. (1996). The long-term stability and predictivevalidity of trait measures of affect. Journal of Personality and SocialPsychology, 70, 567–577.

Weinstein, S. M., Mermelstein, R. J., Hankin, B. L., Hedeker, D., & Flay,B. R. (2007). Longitudinal patterns of daily affect and global moodduring adolescence. Journal of Research on Adolescence, 17, 587–600.

Wheeler, D. S., & Schwarz, J. C. (1989). Millon clinical multiaxial inven-tory (MCMI) scores with a collegiate sample: Long-term stability andself-other agreement. Journal of Psychopathology and Behavioral As-sessment, 11, 339–352.

Whitbourne, S. K., Zuschlag, M. K., Elliot, L. B., & Waterman, A. S.(1992). Psychosocial development in adulthood: A 22-year sequentialstudy. Journal of Personality and Social Psychology, 63, 260–271.

Wilhelm, K., & Parker, G. (1990). Reliability of the parental bondinginstrument and intimate bond measure scales. Australian and New Zea-land Journal of Psychiatry, 24, 199–202.

Wood, D., & Roberts, B. W. (2006). Cross-sectional and longitudinal testsof the personality and role identity structural model (PRISM). Journal ofPersonality, 74, 779–810.

Woodall, K. L., & Matthews, K. A. (1993). Changes in and stability ofhostile characteristics: Results from a 4-year longitudinal study of chil-dren. Journal of Personality and Social Psychology, 64, 491–499.

Wortman, J., Lucas, R. E., & Donnellan, M. B. (2012). Stability andchange in the Big Five personality domains: Evidence from a longitu-dinal study of Australians. Psychology and Aging, 27, 867–874. http://dx.doi.org/10.1037/a0029322

Yardley, J. K., & Rice, R. W. (1991). The relationship between mood andsubjective well-being. Social Indicators Research, 24, 101–111.

Yasuda, T., Lubin, B., Kim, J., & Van Witlock, R. (2003). The Japaneseversion of the Multiple Affect Adjective Checklist–Revised: Develop-ment and validation. Journal of Clinical Psychology, 59, 93–109.

Yonge, G. D., & Regan, M. C. (1975). A longitudinal study of personalityand choice of major. Journal of Vocational Behavior, 7, 41–65.

Zimmerman, M. A., Copeland, L. A., Shope, J. T., & Dielman, T. E.(1997). A longitudinal study of self-esteem: Implications for adolescentdevelopment. Journal of Youth and Adolescence, 26, 117–141.

Zimmermann, J., & Neyer, F. J. (2013). Do we become a different personwhen hitting the road? Personality development of sojourners. Journal ofPersonality and Social Psychology, 105, 515–530. http://dx.doi.org/10.1037/a0033019

Zuckerman, M., & O’Loughlin, R. E. (2006). Self-enhancement by socialcomparison: A prospective analysis. Personality and Social PsychologyBulletin, 32, 751–760.

Received October 10, 2014Revision received June 18, 2015

Accepted June 23, 2015 !

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

781META-ANALYSIS OF STABILITY AND CHANGE