You are on page 1of 21

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223440007

A three-wave longitudinal study of self-evaluations during


young adulthood

Article in Journal of Research in Personality · April 2007


DOI: 10.1016/j.jrp.2006.06.004

CITATIONS READS

56 614

4 authors, including:

Kali Trzesniewski Rand D Conger


University of California, Davis University of California, Davis
88 PUBLICATIONS 18,163 CITATIONS 372 PUBLICATIONS 44,884 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Kali Trzesniewski on 09 July 2018.

The user has requested enhancement of the downloaded file.


Journal of Research in Personality 41 (2007) 453–472
www.elsevier.com/locate/jrp

A three-wave longitudinal study of self-evaluations


during young adulthood 夽
M. Brent Donnellan a,¤, Kali H. Trzesniewski b, Katherine J. Conger c,
Rand D. Conger c
a
Department of Psychology, Michigan State University, East Lansing, MI 48823, USA
b
University of Western Ontario, Canada
c
University of California, Davis,USA

Available online 4 August 2006

Abstract

This investigation provides much needed data on stability and change in self-evaluations during
young adulthood. The Messer and Harter [Messer, B. & Harter, S. (1986). Manual for the adult self-
perception proWle. Unpublished manuscript. Department of Psychology, University of Denver] Adult
Self-Perception ProWle was given to a sample of over 400 young adults in 1999, 2001, and 2003. Sev-
eral conclusions emerged from this study. Generally speaking, domain-speciWc self-evaluations
showed rank-order consistency and very small normative changes during this interval. Nonetheless,
there were individual diVerences in change in self-evaluations that warrant additional theoretical and
empirical attention. In terms of the prediction of global self-esteem, physical self-esteem was one of
the strongest independent predictors of global self-esteem. Finally, weighting domain speciWc self-
evaluations by importance ratings did not consistently improve the prediction of global self-esteem.
The implications of these results for understanding the psychological nature of young adulthood and
for designing future studies using importance ratings are discussed.
© 2006 Elsevier Inc. All rights reserved.


This research is currently supported by grants from the National Institute of Child Health and Human Devel-
opment, the National Institute on Drug Abuse, and the National Institute of Mental Health (HD047573,
HD051746, and MH051361). Support for earlier years of the study also came from multiple sources, including the
National Institute of Mental Health (MH00567, MH19734, MH43270, MH59355, MH62989, and MH48165), the
National Institute on Drug Abuse (DA05347), the National Institute of Child Health and Human Development
(HD027724), the Bureau of Maternal and Child Health (MCJ-109572), and the MacArthur Foundation Research
Network on Successful Adolescent Development Among Youth in High-Risk Settings.
*
Corresponding author.
E-mail address: donnel59@msu.edu (M.B. Donnellan).

0092-6566/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jrp.2006.06.004
454 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

Keywords: Self-concept development; Harter adult self-perception scales; Early adulthood; Young adulthood;
Global self-esteem

1. Introduction

What happens to self-evaluations during young adulthood? This is an intriguing ques-


tion because the young adult years are a time of contextual change as individuals leave
their families of origin, Wnish school, enter the work force, and start families of their own
(Rindfuss, 1991). As such, young adulthood serves as a particularly important time for
investigating stability and change in individual characteristics; however, very few studies
have investigated the development of self-evaluations during this developmental period.
The main objective of this paper is to provide much needed data on self-evaluations during
the young adult years using three waves of longitudinal data.
Researchers who study the development of individual diVerences emphasize that
there are diVerent types of stability and change (e.g., Caspi, Roberts, & Shiner, 2005;
Roberts & Pomerantz, 2004). Thus, there are multiple ways to frame and answer ques-
tions about the development of self-evaluations during young adulthood. We focus on
three types of stability and change in this paper: diVerential (or rank-order), normative,
and individual diVerences. DiVerential stability concerns the degree to which individual
diVerences are maintained over time and is often examined with stability coeYcients.
Normative stability concerns whether average levels increase, decrease, or remain con-
stant over time. Individual diVerences in change concern whether or not individuals
diVer in rates of change and can be assessed through the presence of signiWcant variabil-
ity around change parameters in growth models (e.g., Singer & Willett, 2003) or using
indices of reliable individual change (e.g., Donnellan, Conger, & Burzette, in press;
Roberts, Caspi, & MoYtt, 2001; Robins, Fraley, Roberts, & Trzesniewski, 2001). A
focus on individual change is important because average trends may obscure personal-
ity changes that are evident at the level of the individual (e.g., Mroczek & Spiro, 2003).
Before describing the present study, we brieXy review the literature on stability and
change in self-esteem during adolescence and early adulthood based on recent longitu-
dinal studies and meta-analytic results.

1.1. A brief review of research on stability and change in self-evaluations

Self-concept researchers make distinctions between global self-esteem and domain-spe-


ciWc self-evaluations (e.g., see Byrne, 1996 for a review; Shavelson, Hubner, & Stanton,
1976). Measures of global self-esteem assess overall self-regard whereas domain-speciWc
self-evaluations assess self-regard in particular areas of functioning such as physical
appearance or academic competence. Rosenberg, Schooler, Schoenbach, and Rosneberg
(1995) argued that that it was important to make a distinction between these constructs
because they are not interchangeable. Therefore, we separately review literature on global
and domain-speciWc self-evaluations.
In terms of global self-esteem, there is appreciable diVerential stability in this construct
as assessed by retest correlations. Individuals who are higher (or lower) relative to their
peers in global self-esteem at one time point also tend to be higher (or lower) in self-esteem
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 455

relative to their peers at a future time point (Trzesniewski, Donnellan, & Robins, 2003).
For example, Neyer and Asendorpf (2001) reported a correlation of .49 for self-esteem
across a four-year interval for a sample of German young adults and Trzesniewski et al.’s
(2003) meta-analysis showed that retest correlations for self-esteem increased from adoles-
cence into adulthood, indicating that global self-evaluations become relatively more stable
in adulthood.
In terms of normative stability, there is converging evidence that mean-levels of global
self-esteem tend to increase from adolescence to young adulthood. For example, Robins,
Trzesniewski, Tracy, Gosling, and Potter (2002) found that individuals increase in positive
feelings about themselves during the transition to adulthood using a large internet survey
with a single-item measure of global self-esteem. Donnellan, Trzesniewski, and Conger
(2006) replicated this increase with growth curve modeling using the Rosenberg Self-
Esteem Scale (Rosenberg, 1965) as did Baldwin and HoVman (2002). These recent Wndings
are broadly consistent with previous work conducted in 1970s and 1980s (reviewed in Pas-
carella & Terenzini, 1991) indicating that normative levels of global self-esteem increase
from adolescence to young adulthood.
The literature on individual diVerences in self-esteem change is smaller than the litera-
ture on diVerential and normative stability. Nonetheless, researchers have found signiWcant
variability about slope parameters in growth models for self-esteem during adolescence
which indicates individual diVerence in rates of change (e.g., Donnellan et al., 2006; Bald-
win & HoVman, 2002). These individual diVerences are consistent with earlier studies that
used cluster analysis to identify groups of adolescents with diVerent patterns of intraindi-
vidual change in self-esteem (Deihl, Vicary, & Deike, 1997; Hirsch & DuBois, 1991; Zim-
merman, Copeland, Shope, & Dielman, 1997). Accordingly, there is evidence of individual
diVerences in change in self-esteem, a point that was recently emphasized by Harter and
Whitesell (2003).
Relative to global self-esteem, domain speciWc self-evaluations have received far less
attention, although these individual diVerences have been studied by some developmental
researchers (e.g., Cole et al., 2001; Harter, 1999, 2003; McGuire et al., 1999; Young &
Mroczek, 2003). This literature indicates that there is appreciable rank-order stability in
domain speciWc self-evaluations, especially by adolescence (e.g., Cole et al., 2001; Marsh,
1989; McGuire et al., 1999; Young & Mroczek, 2003). For example, Cole et al. (2001)
reported that median disattenuated 6-month stability coeYcients for the Harter scales
increased linearly from grades 3 to 11 ending in the .80–.90 range (see Fig. 2, p. 1734). Simi-
larly, Byrne (1996) reported that Marsh’s Self-Descriptive Questionnaire III (SDQ-III)
shows substantial levels of rank-order stability over an 18-month period for young adults
(median r D .74). Thus, domain speciWc self-evaluations appear to show an appreciable
amount of rank order consistency over time that may also increase with age much like
global self-esteem.
In terms of normative changes, the existing evidence suggests that mean-level changes
occur during adolescence. Cole et al. (2001) found that mean-levels increased during the
high school years for certain domains, such as behavioral conduct and physical appear-
ance (Fig. 3, p. 1735). Young and Mroczek (2003) found time-related increases in self-
perceptions of Job Competence, Romantic Appeal, and Physical Appearance in an
adolescent sample. However, an interesting Wnding was that age itself was negatively
related to the initial levels of Athletic Competence and Physical Appearance and
positively related to Job Competence and Romantic Appeal. Thus, younger adolescents
456 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

rated themselves higher on Athletic Competence and Physical Appearance than older
adolescents and lower on Job Competence and Romantic Appeal than older adolescents.
Marsh (1989) found increases in most domain-speciWc aspects of the self-concept from
late adolescence to early adulthood. Finally, in terms of individual diVerences in change,
Young and Mroczek (2003) found individual diVerences in rates of change for six of nine
self-concept domains.

1.2. Relations between domain speciWc self-evaluations and global self-esteem

James (1890) theorized that domain speciWc self-evaluations served as the building
blocks of global self-appraisals. Thus, from a practical point of view, there has been a long
standing interest in how well domain speciWc self-evaluations statistically predict global
self-evaluation. For example, Harter (2003, p. 627) reported that she consistently Wnds that
perceptions of physical attractiveness are the strongest predictors of global self-worth (e.g.,
r D .64) whereas perceptions of athletic competence are the weakest predictors (e.g., r D .31).
Thus, certain domain speciWc self-evaluations appear to be more predictive of global self-
esteem than others.
A theoretically signiWcant and controversial issue is whether or not self-appraisals of the
importance of speciWc domains are critical for understanding how domain speciWc self-
evaluations relate to global self-evaluations (e.g., Harter, 1999, 2003; Marsh, 1986, 1993,
1995; Pelham, 1995a, 1995b; Pelham & Swann, 1989). This issue can be traced directly to
the often cited formula proposed by James (1890) that self-esteem is a result of successes
divided by pretensions. The implication is that there are individual diVerences in the arenas
that contribute to levels of overall self-regard (i.e., individuals diVer in their pretensions).
For example, excellence at golf may be a strong determinant of the overall self-regard of
Tiger Woods but play a much less central role as a determinant of self-esteem for the
authors of this paper. Although this perspective is inXuential, it has proven diYcult to
unequivocally verify (e.g., Byrne, 1996; Marsh, 1995; Pelham, 1995a, 1995b). One aspect of
the debate centers on the use of importance ratings, typically single-items, to weight
domain-speciWc self-evaluations.
This debate over importance ratings raises a practical concern for longitudinal studies.
Quite simply, it strikes us as an open question as to whether or not it is worth participants’
time and eVort to complete importance ratings. If, for instance, importance ratings do not
improve the ability of domain speciWc self-evaluations to predict variability in global self-
esteem, then it becomes harder to justify the practice of continuing to ask participants to
complete these sorts of ratings when there are time constraints. Accordingly, we will exam-
ine importance ratings over time and critically evaluate their utility in predicting global
self-esteem.

2. The present study

The Wrst objective of this paper is to examine diVerential stability, normative stability,
and individual diVerences in change in self-evaluations during young adulthood with three
waves of longitudinal data. The Messer and Harter (1986) Adult Self-Perception ProWle
(ASPP; see also Byrne, 1996, pp. 189–197) was used to assess the multi-dimensional nature
of the adult self-concept. This instrument has “the potential to make a rich and valuable
contribution to self-concept assessment [and] deserves much greater attention than
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 457

it has apparently thus far been paid” (Byrne, 1996, p. 196). There are 12 ASPP scales:
Sociability assesses self-perceptions of competence in social interactions; Job Competence
assesses self-perceptions in the domains of work and career; Nurturance assesses self-per-
ceptions in the ability to care for others; Athletic Competence assesses self-perceptions of
sports and Wtness related abilities; Physical Appearance assesses self-evaluations of attrac-
tiveness; Adequacy as a Provider assesses self-evaluations in the ability to meet the material
needs of signiWcant others; Morality assesses self-evaluations of the ethics of personal
behavior; Household Management assesses self-evaluations in running an organized house-
hold; Intimate Relationships assesses self-evaluations in the ability to maintain close and
fulWlling personal relationships; Intelligence assesses self-evaluations of intellectual capa-
bilities; Sense of Humor assesses self-evaluations of the ability to Wnd humor in life; and
Global Self-Worth assesses global self-evaluations. The separate assessment of global self-
worth is a notable feature of the ASPP because it makes it possible to examine how well
domain-speciWc evaluations predict overall levels of self-esteem. This is the second broad
objective of this study.
We made four predictions for this investigation. First, we predicted that diVerential
stability will generally be substantial in young adulthood (i.e. retest correlations at or
above .50) given the previously reviewed literature. Nonetheless, we also expected to
Wnd relatively lower stability coeYcients for domain speciWc self-evaluations in aspects
of life that young adults may still be negotiating such as intimate relationships and
providing for a family. Second, in terms of normative stability, we predicted that
mean-level changes in self-concept will be slight but in the direction of increases over
time given the general trend toward greater psychological adjustment in young adult-
hood (e.g., Donnellan et al., in press; Roberts et al., 2001). Third, we expected to Wnd
individual diVerences in rates of change given the increasing recognition of variability
in rates of change for individual diVerences (e.g., Mroczek, Spiro, & GriYn, 2006) and
including self-evaluations (e.g., Young & Mroczek, 2003). Fourth, we expected that
physical self-esteem would be the strongest predictor of global self-esteem whereas ath-
letic competency would be the weakest predictor given previous results reported by
Harter (1999).
Although there is widespread scholarly interest in the topic of gender diVerences in
self-evaluations (e.g., Kling, Hyde, Showers, & Buswell, 1999; see also Hyde, 2005), much
of the literature examines mean-level gender diVerences. The general Wnding is that men
score higher than women on measures of global self-esteem (e.g., Kling et al., 1999) and
that there are some mean-level diVerences on certain domain speciWc self-evaluations
that fall along gender stereotypical lines (e.g., Cole et al., 2001; Harter, 1999, 2003;
Marsh, 1989; Young & Mroczek, 2003). For instance, women typically score lower on
self-evaluations of physical attractiveness compared to men whereas they typically score
higher than men on self-evaluations of interpersonal skills in close relationships. How-
ever, mean-level diVerences should not be interpreted as evidence for gender diVerences
in patterns of continuity and change (see also Rowe, Vazsonyi, & Flannery, 1994). Given
our primary focus, we also conducted exploratory analyses related to gender diVerences
in stability and change but did not make speciWc predictions in that regard. As Cole et al.
(2001) noted, there is “no theoretical basis for anticipating that [diVerential] stability ƒ
would vary by gender” (p. 1727). Thus, we pursued these analyses as a potentially impor-
tant extension of earlier research and in the spirit of discovery rather than theoretical
test.
458 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

3. Method

3.1. Sample and procedures

Data analyzed in this report are drawn from 409 participants (57.5% female) from the
Family Transitions Project (FTP; see Conger & Conger, 2002 for a broad overview) who
completed the Adult Self-Perception ProWle (ASPP; Messer & Harter, 1986) during Years
6 (primarily collected during calendar year 1999), 8 (primarily collected during calendar
year 2001), and 10 (primarily collected during calendar year 2003) of the FTP. We refer to
these as the Time 1, 2, and 3 assessments, respectively. The ASPP was not administered at
any other waves. The average age of participants was 23.26 years at Time 1 (SD D .47;
Median D 23; Min. D 22, Max. D 25), 25.25 years at Time 2 (SD D .47; Median D 25;
Min. D 24, Max. D 27), and 27.25 years at Time 3 (SD D .50; Median D 27; Min. D 26,
Max. D 29) based on answers to the question, “What was your age in years at your last
birthday?”. The ethnic/racial background of the FTP was predominately European Ameri-
can. Participants completed the ASPP when research assistants visited their places of resi-
dence and were paid approximately $10 per hour of participation.

3.2. Measures

3.2.1. Adult Self-Perception ProWle (ASPP; Messer & Harter, 1986)


The 50-item ASPP assesses domain-speciWc self-perceptions with 11 four-item scales and
assesses global self-worth with a six-item scale. Table 1 provides descriptive statistics for the
ASPP scales. All scales use a forced-choice response format where participants Wrst indicate
which of two statements is most self-descriptive. One statement reXects a positive self-evalu-
ation and the other statement reXects a negative self-evaluation. For example, a pair of
statements pertaining to global self-worth is “Some adults are dissatisWed with themselves”
and “Other adults are satisWed with themselves.” After selecting which description is most
self-descriptive, participants then indicate whether the statement they chose is “Sort of true

Table 1
Means and standard deviations for the adult self-perception proWle
Time 1 Time 2 Time 3
Mean SD Alpha Mean SD Alpha Mean SD Alpha
Sociability 3.09 0.64 .79 3.06 0.62 .80 3.03 0.64 .82
Job Competence 3.25 0.52 .70 3.29 0.51 .75 3.25 0.51 .68
Nurturance 3.19 0.54 .78 3.20 0.55 .80 3.21 0.56 .80
Athletic Competence 2.66 0.74 .83 2.64 0.76 .86 2.62 0.78 .87
Physical Appearance 2.79 0.64 .84 2.75 0.63 .84 2.75 0.63 .86
Adequacy as a Provider 3.08 0.57 .80 3.06 0.58 .85 3.06 0.58 .84
Morality 3.16 0.57 .78 3.22 0.54 .82 3.18 0.56 .84
Household Management 2.99 0.65 .82 2.97 0.67 .85 2.94 0.65 .83
Intimate Relationships 3.05 0.66 .83 2.98 0.65 .83 2.96 0.66 .83
Intelligence 3.11 0.58 .80 3.11 0.55 .80 3.09 0.58 .83
Sense of Humor 3.28 0.55 .76 3.26 0.53 .75 3.24 0.54 .77
Global Self-Worth 3.18 0.57 .87 3.16 0.56 .88 3.15 0.59 .89
Note. Time 1 self-evaluations were collected primarily in 1999, Time 2 self-evaluations were collected primarily in
2001, and Time 3 self-evaluations were collected primarily in 2003.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 459

Table 2
Ratings of importance for the domains assessed by the adult self-perception proWle
Domain Time 1 Time 2 Time 3 Time 1
to Time 3
Mean SD Percentage Mean SD Percentage Mean SD Percentage
retest
rating item rating item rating item
correlation
“Very “Very “Very
Important” Important” Important”
Sociability 3.47 .61 51.8 3.39 .61 45.2 3.38 .62 44.7 .43
Job Competence 3.77 .45 78.4 3.74 .47 75.6 3.73 .48 74.6 .31
Nurturance 3.44 .63 50.6 3.46 .63 53.1 3.48 .66 56.7 .42
Athletic 2.76 .88 21.8 2.67 .85 16.1 2.63 .86 16.4 .53
Competence
Physical 2.78 .76 14.4 2.74 .75 12.7 2.70 .77 13.0 .49
Appearance
Adequacy as a 3.52 .56 55.0 3.50 .57 54.3 3.57 .54 59.9 .32
Provider
Morality 3.55 .59 59.7 3.55 .58 59.2 3.56 .60 61.4 .40
Household 3.18 .71 34.8 3.17 .71 34.0 3.20 .70 34.7 .44
Management
Intimate 3.44 .70 54.8 3.41 .67 50.0 3.42 .64 49.4 .49
Relationships
Intelligence 3.46 .61 51.8 3.40 .61 46.0 3.41 .60 47.2 .50
Sense of Humor 3.60 .57 63.3 3.54 .57 57.9 3.54 .58 57.2 .36
Note. N D 409 except for Job Competence in 1999, Intimate Relationships in 2001, and Athletic Competence
in 2003, where N D 408.

for me” or “Really true for me.” Item responses were converted to a four-point rating scale
and scored so that higher scores reXect greater feelings of self-worth or self-competence.
Scores on the Global Self-Worth Scale at Time 1 were highly correlated with scores on the
Rosenberg Self-Esteem Scale for this sample (Rosenberg, 1965; r D .74, p < .05). The Rosen-
berg Self-Esteem scale was not administered at Time 2 or Time 3.

3.2.2. Importance ratings


Respondents also reported how important each one of the 11 domains assessed by the
ASPP were to them on a four-point scale (1 D “Very Important” to 4 D “Not Very Impor-
tant”). These questions were reverse scored so that higher scores reXected more impor-
tance. Table 2 displays descriptive data on the Importance Ratings. In the following data
analyses, we examine the main and moderating eVects of these ratings in predicting global
self-esteem from domain-speciWc self-evaluations.1

1
Messer and Harter (1986) detail how to compute a Discrepancy Score which is purported to more directly
capture the idea that self-esteem is inXuenced by evaluations of competency in areas that are of major importance
to the individual (James, 1890). The procedure is to Wrst select those areas where respondents indicated that the
domain was very important (i.e., he or she rated that area a 4) and then subtract the corresponding ASSP scale
score from that maximum possible score of 4. Consistent with predictions, Discrepancy Scores were negatively
correlated with Global Self-Esteem at all three assessments (r D ¡.50 in 1999, r D ¡.47 in 2001, and r D ¡.56 in
2003). Unfortunately, one limitation of this procedure is that discrepancies are only generated for those domains
that are very important to the individual (i.e., areas they rated as a 4 in terms of importance). Moreover, Discrep-
ancy Scores have been criticized on psychometric grounds including unreliability and diYculties in interpretation
(see Byrne, 1996 for a review of these issues). Given these concerns we did not employ this scoring procedure in
this paper. Any interested reader may contact the Wrst author for further details concerning these variables.
460 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

4. Results and discussion

4.1. Overview of longitudinal analyses

We analyzed the longitudinal data using a mixture of simple and more complex analy-
ses. DiVerential stability was assessed with retest correlations which were interpreted
using Cohen’s (1988) rule of thumb (e.g., an r of .5 or larger was considered large; see also
McCartney & Rosenthal, 2000). We should point out that an examination of correlations
is necessarily limited to a consideration of only one time interval. To examine normative
stability, an eVect size was calculated by dividing the mean diVerence of the ASPP scales
between any two waves by the standard deviation of the scale at the earlier wave, which
results in a d-metric eVect size similar to a Cohen’s d (see Kline, 2004). We interpreted
these eVect sizes using Cohen’s rule of thumb (e.g., a d of 冷.2冷 was considered small; see e.g.,
McCartney & Rosenthal, 2000). EVect sizes of this sort are also limited to the consider-
ation of only one time interval. As such, we used growth curve modeling (e.g., Singer &
Willett, 2003) to identify normative trends in change over time. To investigate whether or
not there were individual diVerences in change we used the Reliable Change Index
(Jacobson & Truax, 1991) and examined whether or not there was signiWcant variance in
the change parameter from the growth curve models. As a set, these analyses provide a
relatively comprehensive picture of stability and change in self-evaluations across a four-
year interval.

4.2. DiVerential stability and change

Table 3 displays the stability coeYcients for the ASPP as indexed by retest correla-
tions. As seen in Table 3, the correlations were generally large following Cohen’s rule of
thumb. The average four-year stability coeYcient for the 12 ASPP scales was .56
(SD D .11) and ranged from .38 for Job Competence to .77 for Athletic Competence. The
average two-year stability coeYcients were comparable over both intervals (1999–2001:
M D .62, SD D .08, Min. D .48, Max. D .79; 2001 to 2003: M D .63, SD D .10, Min. D .43,

Table 3
DiVerential stability for the adult self-perception proWle scores
Scale Stability from Stability from Stability from
Time 1 to Time 3 Time 1 to Time 2 Time 2 to Time 3
Sociability .65 .68 .74
Job Competence .38 .48 .43
Nurturance .59 .63 .65
Athletic Competence .77 .79 .80
Physical Appearance .63 .69 .69
Adequacy as a Provider .41 .49 .50
Morality .47 .56 .57
Household Management .59 .61 .65
Intimate Relationships .53 .62 .60
Intelligence .56 .65 .68
Sense of Humor .53 .64 .63
Global Self-Worth .61 .64 .64
Note. Correlations were used to assess rank-order stability and all were statistically signiWcant at p < .05.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 461

Max. D .80) and in each case, Job Competence had the smallest coeYcient and Athletic
Competence had the largest. Consistent with our predictions, the two scales with the low-
est stability coeYcients, Job Competence and Adequacy as a Provider are domains of life
that are still being negotiated by individuals in their mid-20s. Nonetheless, it appears that
self-evaluations are a fairly stable individual characteristic in young adulthood when con-
sidering retest coeYcients.

4.3. Normative stability and change

Mean-level diVerences in the ASPP scales were Wrst examined with d metric eVects sizes
computed such that positive scores indicated that later scores were higher than earlier
scores. EVect sizes were calculated for all three intervals (Time 1 to Time 3, Time 1 to Time
2, and Time 2 to Time 3). As seen in Table 4, none of these eVect sizes were greater than
absolute value .15 indicating that average changes over time were, at best, modest. This
simple inspection of the mean-level eVect sizes helped anticipate the growth modeling Wnd-
ings. We then used the Hierarchical Linear Modeling version 6.0 software package
(Raudenbush, Bryk, & Congdon, 2004) to conduct growth modeling analyses. Models were
estimated using restricted maximum likelihood estimation procedures.
We initially estimated an intercepts only model which can be used to calculate intraclass
correlations which quantify the ratio of between person variance to total variance (e.g.,
Singer & Willett, 2003, p. 96). These coeYcients are displayed in Table 4 and indicate that
the majority of variability in most ASPP domains lies between persons (e.g., the average
intraclass correlation was .60 with a range of .43–.78). We then estimated linear models for
each of the domains assessed by the ASPP. We used linear models given the suggestion by
Raudenbush and Bryk (2002) that this simple model is useful for modeling change in
studies that have only a small number of observations taken over a relatively short time
(p. 163). We ‘‘centered” age around 22 which is the earliest age in our sample. Thus, the
intercept represents the average self-esteem score at age 22. Intercepts and slopes from
Table 4
Normative stability for the adult self-perception proWle Scores
DiVerence scores in d Metric Growth model parameters
1999–2003 1999–2001 2001–2003 ICC Intercept Slope
Sociability ¡0.10 ¡0.04 ¡0.05 .69 3.11 ¡.02¤
Job Competence 0.00 0.07 ¡0.07 .43 3.27 .00
Nurturance 0.04 0.02 0.01 .62 3.19 .00
Athletic Competence ¡0.05 ¡0.02 ¡0.04 .78 2.67 ¡.01
Physical Appearance ¡0.06 ¡0.05 ¡0.01 .67 2.80 ¡.01
Adequacy as a Provider ¡0.03 ¡0.03 ¡0.01 .47 3.09 ¡.01
Morality 0.04 0.10 ¡0.06 .53 3.17 .00
Household Management ¡0.08 ¡0.04 ¡0.04 .62 3.02 ¡.02¤
Intimate Relationships ¡0.14 ¡0.11 ¡0.03 .58 3.08 ¡.03¤
Intelligence ¡0.03 0.00 ¡0.03 .63 3.12 ¡.01
Sense of Humor ¡0.07 ¡0.04 ¡0.04 .60 3.30 ¡.01
Global Self-Worth ¡0.05 ¡0.03 ¡0.03 .63 3.19 ¡.01
Note. EVect sizes in the d metric were used to assess normative stability by dividing the mean diVerence by the
standard deviation of the scale at the earlier wave. Positive ds reXect that the average score was higher at the later
wave whereas a negative d reXects that the average score was lower. See text for description of growth models.
¤
p < .05 for slope parameters.
462 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

these models are displayed in Table 4. When considering normative stability, our primary
focus was on whether or not there was evidence that the slope parameter was reliably
diVerent from zero. This provided evidence for age-related systematic growth or decline in
a particular domain and corresponds to existing research on mean-level trends in self-eval-
uation. We found only three cases where the slope was statistically signiWcant: Sociability,
Household Management, and Intimate Relationships. In these cases, there was an average
decline in self-evaluations with age in these domains. However, as was seen with the d met-
ric analyses, the point estimates for the slopes were not particularly large in relation to the
scale metric (1–4).2

4.4. Individual diVerences in change

Individual diVerences in change were Wrst examined with the Reliable Change Index
(RCI: Jacobson & Truax, 1991). The RCI is calculated by dividing the diVerence score by
the standard error of the diVerence using a formula reported in Jacobson and Truax (1991,
p. 14). The diVerence between Time 3 and Time 1 was used for these analyses. Thus, the
RCI “tells us whether change reXects more than the Xuctuations of an imprecise measuring
instrument” (Jacobson & Truax, 1991, p. 14). It is important to note that the Jacobson and
Truax (1991) formula accounts for measurement unreliability such that the size of diVer-
ence required for passing the threshold for reliable change increases as unreliability in the
measurement instrument increases. This raises the question as to what estimate of mea-
surement reliability should be used in calculations of the RCI. Ideally, a short-term retest
correlation would be used but unfortunately short-term test–retest information is not
available for the ASPP scales (Byrne, 1996); therefore we used the internal consistencies of
the ASPP scales in 1999 as our estimate of reliability.3
Following conventions, individuals were classiWed as reliably increasing in a particular
self-evaluation if their RCI was above 1.96 and classiWed as reliably decreasing in a partic-
ular self-evaluation if their RCI was below ¡1.96. Individuals were classiWed as staying
consistent if their RCI scores were between ¡1.96 and 1.96. These distributions are
reported for the whole sample in Table 5. We tested if the observed distribution of individ-
uals classiWed as increasing, consistent, and decreasing deviated from expected distribu-
tions if changes were assumed to occur at random. According to conventions based on
classical test theory, 2.5% of the sample should be classiWed as increasing (n D 10) and 2.5%

2
We were concerned that the overall age eVects were perhaps driven by the few individuals with extreme age
scores. For instance, only four individuals were 22 at the 1999 assessment and only Wve were 29 at the 2003 assess-
ment. Accordingly, we dropped these individuals from the analyses and repeated the growth curve analyses for
those cases where the slope was statistically signiWcant. The eVects were unchanged for Sociability and Intimate
Relationships; however, the eVect for Household Management became marginally signiWcant (Slope D .01,
p D .053). All told, the concern that a few cases were driving the overall eVect did not appear to be well founded,
except perhaps for Household Management.
3
We acknowledge that this solution may not be ideal and may result in biased estimates of reliable change. We
are also aware of dangers in blurring important distinctions between test–retest coeYcients and internal consis-
tency estimates. However, we assumed that the short-term retest coeYcients would be relatively close to the inter-
nal consistency estimates. We also experimented with lower reliability values to see how this might aVect the
results. In particular, if a reliability of .60 is used for all scales (an extremely low value given the 2-year retest co-
eYcients) then there is still signiWcant deviation from chance for 6 of the 12 dimensions: Job Competence, Nurtur-
ance, Adequacy as a Provider, Morality, Sense of Humor, and Global Self-Worth.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 463

Table 5
Individual diVerences in change for adult self-perception proWle scores
Scale Reliable change index distributions from Time 1 to Time 3 Growth model parameters
Decreasing Consistent Increasing 2(df D 2) Intercept SD Slope SD
Sociability 25 (6.1%) 368 (90.0%) 16 (3.9%) 27.23 .56 .06
Job Competence 27 (6.6%) 356 (87.0%) 26 (6.4%) 57.30 .43 .06
Nurturance 39 (9.5%) 323 (79.0%) 47 (11.5%) 232.20 .46 .06
Athletic 24 (5.9%) 371 (90.7%) 14 (3.4%) 22.03 .67 .06
Competence
Physical 60 (14.7%) 304 (74.3%) 45 (11.0%) 391.07 .57 .06
Appearance
Adequacy as a 62 (15.2%) 297 (72.6%) 50 (12.2%) 452.16 .44 .07
Provider
Morality 51 (12.5%) 298 (72.9%) 60 (14.7%) 439.39 .48 .08
Household 31 (7.6%) 353 (86.3%) 25 (6.1%) 69.93 .53 .05
Management
Intimate 79 (19.3%) 286 (69.9%) 44 (10.8%) 618.97 .57 .08
Relationships
Intelligence 52 (12.7%) 305 (74.6%) 52 (12.7%) 370.94 .53 .08
Sense of Humor 54 (13.2%) 314 (76.8%) 41 (10.0%) 304.16 .50 .07
Global Self-Worth 56 (13.7%) 314 (76.8%) 39 (9.5%) 310.16 .47 .05
Note. The 2 tests if the observed distribution was signiWcantly diVerent from 2.5% Decreasing (n D 10), 95%
Consistent (n D 389), and 2.5% Increasing (n D 10). All 2 test values, intercept standard deviations, and slope
standard deviations were statistically signiWcant at p < .05.

of the sample should be classiWed as decreasing (n D 10) and the remaining 95% should be
classiWed as consistent (n D 389) based on chance expectations. All 12 observations devi-
ated from this distribution indicating that reliable change was occurring for this sample at
the level of the individual.
We also calculated the number of domains of self-evaluation that demonstrated reliable
change for each member of the sample. The average number of changes was 1.37
(SD D 2.00) and we found that 179 sample members (43.8%) demonstrated no reliable
changes in any domain whereas 106 members changed in 1 domain (25.9%), 48 changed in
2 domains (11.7%), 38 changed in 3 domains (9.3%), 10 changed in 4 domains (2.4%), 6
changed in 5 domains (1.5%), and 22 changed in 6 or more domains (5.3%). Thus, the
modal pattern in this sample was for no reliable change on any ASPP scale.
Another way to identify the presence of individual diVerences in change is to test
whether or not there is signiWcant variance about the slope in a growth modeling context.
As seen in Table 5, these parameters were statistically signiWcant in all cases indicating that
there were individual diVerences in the rate of change for all self-evaluations. Also shown
in Table 5 is the presence of individual variability for the intercepts for all self-evaluations.
This result is largely unremarkable given the well-known observation that there are cross-
sectional diVerences in self-evaluations.

4.5. Gender diVerences

Gender diVerences in rank-order stability were formally tested by comparing correla-


tions for women and men using formula 2.8.11 in Cohen, Cohen, West, and Aiken (2003,
p. 49). To reduce the number of comparisons we only tested for gender diVerences in the
464 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

four-year stability coeYcients. There were Wve cases where the diVerence was statistically
signiWcant at p < .05: Job Competence (r D .47 for Women compared to r D .24 for Men,
Z D 2.69), Physical Appearance (r D .66 for Women compared to r D .51 for Men, Z D 2.37),
Household Management (r D .64 for Women compared to r D .50 for Men, Z D 2.09), Intel-
ligence (r D .63 for Women compared to r D .40 for Men, Z D 3.24), and Sense of Humor
(r D .60 for Women compared to r D .42 for Men, Z D 2.49). In all Wve cases, self-percep-
tions were more stable for women than for men. It is important to emphasize that we did
not expect to Wnd these diVerences in retest correlations and that the diVerences in magni-
tude between women and men were not large. Nonetheless, future research should pursue
replications of these intriguing diVerences.
Gender diVerences in normative stability and individual diVerences in change were
formally tested by including Gender (1 D Female, 0 D Male) as a level 2 predictor of the
intercepts and slopes in the growth models. Gender was a signiWcant predictor of slopes
for only one scale, Athletic Competence, and results indicated that women declined
more sharply than men (Unstandardized Beta for the Slope D ¡.03, SE D .01). However,
gender predicted the intercepts for Nurturance, Athletic Competence, Physical Appear-
ance, Morality, and Intimate Relationships which was akin to evidence of mean-level
gender diVerences in these domains. Women reported higher scores than men on Nur-
turance (Unstandardized Beta D .32, SE D .06), Intimate Relationships (Unstandard-
ized Beta D .16, SE D .07) and Morality (Unstandardized Beta D .18, SE D .06);
conversely women reported lower scores than men on Athletic Competence (Unstan-
dardized Beta D ¡.43, SE D .07) and Physical Appearance (Unstandardized
Beta D ¡.22, SE D .07). Finally, there were no gender diVerences in the RCI patterns
and there was no evidence for a gender diVerence in average number of RCI changes
(Women: M D 1.45, SD D 1.85; Men: M D 1.26, SD D 2.20; d D .10). Thus, there was evi-
dence for mean-level gender diVerences but much less evidence for gender diVerences in
individual change.

4.6. How well do domain-speciWc self-evaluations predict global self-evaluations?

Table 6 presents the cross-sectional correlations between domain-speciWc self-evalua-


tions and global self-evaluations.4 Table 6 also shows how well each domain independently
predicts global self-esteem in a regression equation where all domains are used as predic-
tors of global self-esteem. The 11 domains concurrently explained between 72% and 75% of
the variability in global self-esteem across all three waves. Moreover, Physical Appearance
appeared to be one of the best predictors of Global Self-Esteem at any occasion given that
it had the largest zero-order correlation and standardized regression coeYcient when pre-
dicting global self-esteem.
Finally, we examined whether particular domains could predict relative changes
in global self-esteem. Thus, we regressed Global Self-Esteem at Time 3 on speciWc
self-evaluations at Time 1 and Global Self-Esteem at Time 1. Physical Appearance,
Household Management, and Global Self-Esteem at Time 1 were statistically signiW-
cant predictors in this analysis ( D .18 for Physical Appearance;  D .13 for Household

4
We tested whether or not there were gender diVerences in these associations and none were statistically signiW-
cant at p < .05. Therefore, we did not pursue further tests of gender diVerences in this section.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 465

Table 6
Domain-speciWc evaluations as cross-sectional predictors of Global Self-Worth
Scale Time 1 Time 2 Time 3
Zero- Standardized Zero- Standardized Zero- Standardized
order regression order regression order regression
correlation coeYcient correlation coeYcient correlation coeYcient
Sociability .57¤ .06 .48¤ ¡.04 .58¤ .08¤
Job Competence .61¤ .09¤ .63¤ .20¤ .61¤ .18¤
Nurturance .40¤ .04 .33¤ .02 .36¤ ¡.03
Athletic Competence .37¤ .03 .33¤ .06 .30¤ .01
Physical Appearance .68¤ .34¤ .63¤ .29¤ .64¤ .28¤
Adequacy as a Provider .64¤ .24¤ .58¤ .20¤ .59¤ .18¤
Morality .53¤ .14¤ .56¤ .20¤ .57¤ .22¤
Household Management .39¤ .01 .37¤ ¡.04 .51¤ .07¤
Intimate Relationships .54¤ .05 .53¤ .14¤ .60¤ .11¤
Intelligence .62¤ .08¤ .61¤ .10¤ .65¤ .15¤
Sense of Humor .57¤ .14¤ .52¤ .12¤ .51¤ ¡.01
R2 .74¤ .72¤ .75¤
¤
p < .05.

Management,  D .47 for Global Self-Esteem, ps < .05). However, results for domains
other than Physical Appearance were not robust when we changed intervals. That is,
when we regressed Global Self-Esteem at Time 3 on speciWc self-evaluations at Time 2
and Global Self-Esteem at Time 2, only Physical Appearance and Global Self-Esteem
at Time 2 were signiWcant predictors ( D .14 for Physical Appearance and  D .39 for
Global Self-Esteem, ps < .05) and when we regressed Global Self-Esteem at Time 2 on
speciWc self-evaluations at Time 1 and Global Self-Esteem at Time 1, only Physical
Appearance, Intimate Relationships, and Global Self-Esteem at Time 2 were signiWcant
predictors ( D .16 for Physical Appearance,  D .15 for Intimate Relationships, and
 D .47 for Global Self-Esteem, ps < .05). Thus, there was robust evidence that positive
self-evaluations in the domain of physical appearance predicted relative increases in
Global Self-Esteem in young adulthood whereas the evidence for other domains was
much less convincing.

4.7. Importance ratings

As seen in Table 2, the absolute ratings were relatively high given the maximum possible
score of 4 on a particular domain. This is consistent with the contention that the ASPP
assesses self-perceptions in domains that are generally important for young adults. More-
over, the four-year retest coeYcients reported in Table 2 were high in light of the single-
item nature of the rating (e.g., the average correlation was .43, SD D .07). Age-related
changes in importance ratings were tested using growth modeling. We wish to acknowl-
edge at the outset that there are inherent limitations in single-item ratings in terms of
reliability and we viewed these analyses very cautiously. Complete tables describing these
results are available from the Wrst author. There was evidence at p < .05 that slope terms
were signiWcantly diVerent from zero for ratings of six of the eleven domains assessed by
the ASPP (Sociability D ¡.02; Job Competence D ¡.01; Athletic Competence D ¡.03;
466 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

Physical Appearance D ¡.02; Intelligence D ¡.02; Sense of Humor D ¡.02). In all cases
importance ratings in these domains declined with age.5
Following Marsh (1986, 1995), we conducted a series of hierarchical regression analy-
ses to test whether or not weighting self-evaluations by individual importance ratings
improved the prediction of global self-esteem. In the Wrst step, global self-esteem was
regressed on the 11 self-evaluations, individual importance ratings were added in the sec-
ond step, and the interactions between importance ratings and self-evaluations were
entered in the third step. We centered all domain-speciWc self-evaluations and impor-
tance ratings and calculated interaction terms using these centered scores. The self-eval-
uations accounted for the lion’s share of the variance in global self-esteem at each wave
as was reported in Table 6. Importance ratings accounted for a statistically signiWcant
amount of additional variance at all three occasions (R2 D .03, .02, .and .02 at Time 1, 2,
and 3, respectively). For all three years, importance ratings of physical appearance had a
statistically signiWcant (p < .05) independent eVect on global self-esteem ( D ¡.16, ¡.11,
and ¡.13 at Time 1, 2, and 3, respectively). The eVect was negative indicating that indi-
viduals who rated physical appearance more important than the sample average had
lower self-esteem. The only other signiWcant importance rating was for athletics at Time
3 ( D .07); however, we were not conWdent that this was a robust eVect given that it
failed to pass the statistical signiWcance threshold at Time 1 ( D .04, p D .20) and at Time
2 ( D .06, p D .07). Finally, interaction terms only accounted for a signiWcant amount of
additional variance at Time 3 (R2 D .02) and the only signiWcant independent eVect was
for physical appearance X importance rating ( D .12). Interaction terms did not
improve prediction of global self-esteem at Time 1 (R2 D .01, p D .47;  for physical
appearance X importance rating D .05, p D .08) or at Time 2 (R2 in 2001 D .01, p D .43;
 for physical appearance X importance rating D .07, p < .05). All told, these analyses
did not indicate overwhelmingly that weighting domain speciWc importance ratings
improved the prediction of global self-esteem.

5. General discussion

The broad goals of this investigation were to provide data on stability and change in
domain-speciWc self-evaluations during young adulthood and to examine how well domain
speciWc self-evaluation statistically predicted global self-esteem. Five major conclusions
emerged from our analyses. First, self-evaluations show rank-order consistency across a
four-year interval during young adulthood. Second, most self-evaluations demonstrate
very little normative changes during this interval. Third, although there is a good deal of
evidence of diVerential and absolute stability, there are individual diVerences in changes in

5
We used gender as a level 2 predictor to test whether or not gender inXuenced trajectories of importance rat-
ings. Gender predicted the intercepts for three domains. Women had higher intercepts than men for ratings of im-
portance of Nurturance (Unstandardized Beta D .20, SE D .07) and Intimate Relationships (Unstandardized
Beta D .16, SE D .08) but lower intercepts than men for ratings of importance for Athletic Competence (Unstan-
dardized Beta D ¡.62, SE D .10). Gender predicted slopes for ratings of Physical Appearance (Unstandardized
Beta D .04, SE D .02) and Adequacy as a Provider (Unstandardized Beta D .03, SE D .02). Graphical displays of
these results (available from the Wrst author) indicated that women declined less dramatically than men on ratings
of importance for Physical Appearance. In terms of the rating of the importance of Adequacy as a Provider,
women increased their ratings of the importance of Adequacy as a Provider with age much more sharply than
men who evinced, if anything, a slight decline on the rating of the importance of this dimension.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 467

self-evaluations that deserve additional empirical scrutiny. Fourth, domain speciWc self-
evaluations account for a very substantial amount of variance in global self-esteem and
physical self-esteem consistently appears to be one of the strongest independent predictors
of global self-esteem. Finally, weighting self-evaluations by importance ratings did not
consistently improve the prediction of global self-esteem. Prior to discussing some of our
Wndings in more detail, we Wrst consider why personality psychologists should care about
self-evaluations (see Robins, Norem, & Cheek, 1999).

5.1. Personality and self-evaluations

One general take home message from this study is that self-evaluations during young
adulthood are not ephemeral qualities that change dramatically from year to year. They
appear to behave like other individual diVerence constructs such as core personality traits.
This raises important questions about the connections between basic dispositions such as
those captured by personality trait inventories and self-esteem. Robins et al. (1999) pointed
out that basic personality dimensions inXuence how people think and feel about them-
selves. Thus, the relative stability of the ASPP scales in young adulthood may also stem
from the fact that core personality traits show diVerential stability in young adulthood
(e.g., Roberts & DelVecchio, 2000).
Indeed, individual diVerences in core personality traits might help explain the stability
of self-evaluations both directly through self-perceptual mechanisms and indirectly
through the inXuence that these characteristics exert on actual adaptation. For instance,
trait negative aVect may directly impact self-evaluations because individuals high in this
dimension of personality perceive most things negatively compared to individuals lower in
this trait (e.g., Watson & Clark, 1984). On the other hand, a trait like conscientiousness
may indirectly lead to higher self-evaluations in domains related to work and economic
well-being because this dimension of personality has been linked to extrinsic career success
even controlling for cognitive abilities (e.g., Judge, Higgins, Thoresen, & Barrick, 1999).
Future research is needed to test the pathways linking basic personality dispositions to self-
evaluations.
However, we do not believe that self-evaluations can be completely reduced to basic
traits. For instance, we do not think that self-evaluations are a perfect reXection of
underlying traits. For example, it is unlikely that the average woman in this sample was
actually less physically attractive than the average man in this sample although the
average self-perception of attractiveness was lower for women then men. Moreover, it
might not make psychological sense to talk about accuracy with regard to overall self-
regard. For instance, it isn’t clear whether or not there is an appropriate external stan-
dard for judging overall self-regard (e.g., Tangney & Leary, 2003) and we tend to agree
with the perspective that there is no such external yardstick for global self-esteem.
To be sure, self-evaluations can be appreciated as phenomenological constructs (e.g.,
Harter, 1999, 2003) that might help researchers better understand thoughts, feelings,
and behavior (e.g., Harter, 2003). In sum, although self-evaluations may reXect basic
dispositions, we believe that self-evaluations are nonetheless interesting in their own
right and worthy of study by personality researchers (e.g., Robins et al., 1999). All told,
we believe that personality research should encompass studies of self-concept, self-
esteem, life narratives, life goals, and motives as well as studies of the Big Three, Big
Five, or Big Seven traits.
468 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

5.2. The study of stability and change in self-evaluations provides a psychological lens on
young adulthood

The study of stability and change in self-evaluations sheds light on the phenomenology
of young adulthood. Namely, from the vantage point of many self-evaluations, young
adulthood appears to be a time of relatively stable self-regard and few normative changes.
That is, we found evidence of appreciable diVerential and absolute stability in the self-eval-
uations assessed by the ASPP. To be sure, there were three domains that showed slight sys-
tematic declines during the young adult years, Sociability, Household Management, and
Intimate Relationships; however, these drops were absolutely small (i.e., no d was larger
than ¡.14) and reXected less sanguine appraisals in these particular domains with age.
Nonetheless, it appears that the typical experiences of young adulthood do not appear to
create great disturbances or upheavals in the self-evaluations that we assessed.
At Wrst blush, Wnding evidence for stability in self-evaluations may appear counter-intu-
itive considering that young adulthood is considered by demographers to be a time of
dense and diverse change (Rindfuss, 1991). Nevertheless, these results are sensible in light
of current theorizing on the development of individual diVerences. Arnett (2000, 2004)
characterizes the years between 18 and 25 as “the volitional years of life” during which
individuals are relatively independent from social roles and expectations. This is a time of
relative freedom in terms of what individuals ought to do and how individuals ought to act.
Caspi and MoYtt (1993) argue that individuals are able to bring their characteristic ways
of thinking, feeling, and acting to the open situations, which in turn reinforces those very
characteristics. Thus, we would expect stability to predominant over instability during this
developmental period. Likewise, there was not a great deal of normative change in these
aspects of the self during the intervals we investigated, perhaps because there is an absence
of a clear press to think and feel in a particular way. These conditions would seem to pro-
mote continuity rather than change in self-evaluations when considering both diVerential
and normative stability.
Nonetheless, there was consistent evidence of individual diVerences in change. This Wnd-
ing indicates that young adulthood might be best characterized as a time of individual
diVerences in change in self-evaluations rather than as a time of normative change in self-
evaluations. The existence of these individual diVerences raises interesting questions about
the predictors of individual diVerences in change. Indeed, these Wndings suggest that
researchers should continue to focus on intraindividual diVerences in change in future
studies (e.g., Mroczek & Spiro, 2003; Young & Mroczek, 2003).

5.3. Importance ratings may not be that useful

An inspection of the results for the importance ratings revealed a few interesting
insights. First, the average importance ratings were relatively high for all of the domains
assessed by the Messer and Harter (1986) scales. This bolsters our conWdence in the
assertion that the ASPP taps self-perceptions in domains that are relevant to young
adults. Second, importance ratings for over half of the ASPP dimensions show a slight
decline in young adulthood. Third, importance ratings of physical appearance were con-
sistently negatively related to self-esteem controlling for all self-evaluations and the
other importance ratings in hierarchical regression analyses. This Wnding suggests that
there could be a cost to self-esteem in terms of placing too much importance on physical
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 469

appearance. Nonetheless, we have tentatively concluded that it is not worth the partici-
pants’ time and eVort to complete these measures. This argument is based on at least two
considerations.
First, weighting self-evaluations by the importance rating did not consistently enhance
the predictive validity of self-evaluations for explaining variability in global self-esteem.
Thus, if the goal is to use importance ratings to improve the eYciency of domain speciWc
predictors then researchers are likely to be unimpressed by the results. Importance ratings
themselves had predictive validity above and beyond self-evaluations for predicting global
self-esteem. However, it is hard to be impressed by a block of predictors that accounts for
an additional 2–3% of variability after accounting for 70% or so of the variability in global
self-esteem using the self-evaluations themselves.
Second, we are concerned that individuals may be unaware or unwilling to admit the
importance of each of these domains for their overall sense of self-worth. In short, a single
item rating may not adequately capture the actual psychological importance of the
domains being rated by the ASPP. For instance, physical appearance received the second
to the lowest average importance rating at Time 1 and 2 and the lowest rating at Time 3
despite the fact that it was highly correlated with global self-esteem scores at all waves. In
addition, self-evaluations of physical appearance predicted relative changes in global self-
esteem from Time 1 to time 3. In this case, the importance rating did not adequately reXect
the relatively strong link between self-perceptions of physical appearance and evaluations
of overall self-worth.
To be clear, we acknowledge that single item importance ratings may have value for
applications that we did not consider. Certain readers may reasonably quarrel with our sin-
gular focus on the prediction of global self-esteem. For instance, importance ratings may
provide descriptive information about what domains individuals value at diVerent points
in the life span. However, we suggest that researchers who wish to seriously study impor-
tance ratings will beneWt from constructing multi-item measures of importance rather than
relying on single-item rating scales (see also Marsh, 1995).

6. Limitations and conclusions

A few limitations and caveats should be noted. First, future research should replicate
and extend our results on more diverse samples. The FTP sample was predominately Euro-
pean American and drawn from one region of the United States. Second, although three-
wave studies are more informative than two-wave studies, even more waves of data will
aVord an opportunity to more precisely capture the function of the absolute development
of self-evaluations. Finally, we may have missed the developmental stage or stages where
changes in self-evaluations were most pronounced. For instance, more “action” may have
occurred during the 18–22 interval, a period when sample members were not administered
the ASPP.
Nonetheless, given the paucity of longitudinal data on domain-speciWc self-evaluations
in young adulthood, we believe that our investigation provides much needed data on a
neglected topic. These analyses provide useful psychometric data on the Harter Adult
scales and further indicate that domain-speciWc self-evaluations are fairly stable individual
diVerence constructs. In short, we found that self-evaluations show appreciable amounts of
rank-order and normative stability during a four-year interval in young adulthood. Sub-
stantively, our Wndings provide an important insight into the psychological aspects of
470 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

young adulthood; that is, most individuals in this study appeared to be moving through
their mid 20s without experiencing dramatic upheavals in their self-concepts.

References

Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. Amer-
ican Psychologist, 55, 469–480.
Arnett, J. J. (2004). Emerging adulthood: The winding road from the late teens through the twenties. New York:
Oxford University Press.
Baldwin, S. A., & HoVman, J. P. (2002). The dynamics of self-esteem: a growth-curve analysis. Journal of Youth
and Adolescence, 31, 101–113.
Byrne, B. M. (1996). Measuring self-concept across the life-span: issues and instrumentation. Washington, DC:
American Psychological Association.
Caspi, A., & MoYtt, T. E. (1993). When do individual diVerences matter? A paradoxical theory of personality
coherence. Psychological Inquiry, 4, 247–271.
Caspi, A., Roberts, B. W., & Shiner, R. L. (2005). Personality development: stability and change. Annual Review of
Psychology, 56, 453–484.
Cohen, J. (1988). Statistical power for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied multiple regression/correlation analysis for the behavioral
sciences (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.
Cole, D. A., Maxwell, S. E., Martin, J. M., Peeke, L. G., Seroczynski, A. D., Tram, J. M., et al. (2001). The develop-
ment of multiple domains of child and adolescent self-concept: a cohort sequential longitudinal design. Child
Development, 72, 1723–1746.
Conger, R. D., & Conger, K. J. (2002). Resilience in midwestern families: selected Wndings from the Wrst decade of
a prospective longitudinal study. Journal of Marriage and the Family, 64, 361–373.
Deihl, L. M., Vicary, J. R., & Deike, R. C. (1997). Longitudinal trajectories of self-esteem from early to middle
adolescence and related psychosocial variables among rural adolescents. Journal of Research on Adolescence,
7, 393–411.
Donnellan, M. B., Trzesniewski, K. H., & Conger, R. D. (2006). Personality correlates of self-esteem during the
transition from late adolescence to young adulthood. Unpublished manuscript. Department of Psychology,
Michigan State University.
Donnellan, M. B., Conger, R. D, & Burzette, R. G. (in press). Personality development from late adolescence to
young adulthood: rank-order stability, normative changes, and evidence for the maturity-stability hypothesis.
Journal of Personality.
Harter, S. (1999). The construction of the self. New York: Guilford.
Harter, S. (2003). Development of self-representations during childhood and adolescence. In M. R. Leary & J. P.
Tangney (Eds.), Handbook of self and identity (pp. 610–642). New York: Wiley.
Harter, S., & Whitesell, N. R. (2003). Beyond the debate: why some adolescents report stable self-worth over time
and situation, whereas others report changes in self-worth. Journal of Personality, 71, 1027–1058.
Hirsch, B. J., & DuBois, D. L. (1991). Self-esteem in early adolescence: the identiWcation and prediction of con-
trasting longitudinal trajectories. Journal of Youth and Adolescence, 20, 53–72.
Hyde, J. S. (2005). The gender similarities hypothesis. American Psychologist, 60, 581–592.
Jacobson, N. S., & Truax, P. (1991). Clinical signiWcance: a statistical approach to deWning meaningful change in
psychotherapy research. Journal of Clinical and Consulting Psychology, 59, 12–19.
James, W. (1890). The principles of psychology. New York: Henry Holt and Company.
Judge, T. A., Higgins, C. A., Thoresen, C. J., & Barrick, M. R. (1999). The big Wve personality traits, general mental
ability, and career success across the life span. Personnel Psychology, 52, 621–652.
Kline, R. (2004). Beyond signiWcance testing: Reforming data analysis methods in behavioral research. Washington
DC: American Psychological Association.
Kling, K. C., Hyde, J. S., Showers, C. J., & Buswell, B. N. (1999). Gender diVerences in self-esteem: a meta-analysis.
Psychological Bulletin, 125, 470–500.
Marsh, H. W. (1986). Global self-esteem: its relation to speciWc facets of self-concept and their importance. Jour-
nal of Personality and Social Psychology, 51, 1224–1236.
Marsh, H. W. (1989). Age and sex eVects in multiple dimensions of self-concept: preadolescence to early adult-
hood. Journal of Educational Psychology, 81, 417–430.
M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472 471

Marsh, H. W. (1993). Relations between global and speciWc domains of self: the importance of individual impor-
tance, certainty, and ideals. Journal of Personality and Social Psychology, 65, 975–992.
Marsh, H. W. (1995). A Jamesian model of self-investment and self-esteem: comment on Pelham (1995). Journal
of Personality and Social Psychology, 69, 1151–1160.
McCartney, K., & Rosenthal, R. (2000). EVect size, practical importance, and social policy for children. Child
Development, 71, 173–180.
McGuire, S., Manke, B., Saudino, K. J., Reiss, D., Hetherington, E. M., & Plomin, R. (1999). Perceived compe-
tence and self-worth during adolescence: a longitudinal behavior genetic study. Child Development, 6,
1283–1296.
Messer, B. & Harter, S. (1986). Manual for the adult self-perception proWle. Unpublished manuscript. Department
of Psychology, University of Denver.
Mroczek, D. K., & Spiro, A. (2003). Modeling intraindividual change in personality traits: Wnding from the Nor-
mative Aging Study. Journal of Gerontology: psychological Sciences, 58B, 153–165.
Mroczek, Spiro, A., & GriYn, P. W. (2006). Personality and aging. In J. E. Birren & K. W. Schaie (Eds.), Handbook
of the psychology of aging (sixth ed., pp. 363–377). New York: Academic Press.
Neyer, F. J., & Asendorpf, J. B. (2001). Personality-relationship transaction in young adulthood. Journal of Per-
sonality and Social Psychology, 81, 1190–1204.
Pascarella, E. T., & Terenzini, P. T. (1991). How college aVects students: Wndings and insights from 20 years of
research. San Francisco: Jossey-Bass.
Pelham, B. W. (1995a). Self-investment and self-esteem: evidence for a Jamesian model of self-worth. Journal
of Personality and Social Psychology, 69, 1141–1150.
Pelham, B. W. (1995b). Further evidence for a Jamesian model of self-worth: reply to Marsh (1995). Journal of
Personality and Social Psychology, 69, 1161–1165.
Pelham, B. W., & Swann, W. B. (1989). From self-conceptions of self-worth: on the sources and structure of global
self-esteem. Journal of Personality and Social Psychology, 57, 672–680.
Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models: applications and data analysis methods (2nd ed.).
Thousands Oaks: Sage.
Raudenbush, S., Bryk, A., & Congdon, R. (2004). Hierarchical Linear Modeling Version 6.0. ScientiWc Software
International.
Rindfuss, R. R. (1991). The young adult years: diversity, structural change, and fertility. Demography, 28,
493–512.
Roberts, B. W., Caspi, A., & MoYtt, T. E. (2001). The kids are alright: growth and stability in personality develop-
ment from adolescence to adulthood. Journal of Personality and Social Psychology, 81, 670–683.
Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency of personality from childhood to old age:
a quantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25.
Roberts, B. W., & Pomerantz, E. M. (2004). On traits, situations, and their integration: a developmental perspec-
tive. Personality and Social Psychology Review, 8, 402–416.
Robins, R. W., Fraley, R. C., Roberts, B. W., & Trzesniewski, K. H. (2001). A longitudinal study of personality
change in young adulthood. Journal of Personality, 69, 617–640.
Robins, R. W., Norem, J. K., & Cheek, J. M. (1999). Naturalizing the self. In L. A. Pervin & O. P. John (Eds.),
Handbook of personality: theory and research (2nd ed., pp. 443–4777). New York: Guilford press.
Robins, R. W., Trzesniewski, K. H., Tracy, J. L., Gosling, S. D., & Potter, J. (2002). Self-esteem across the lifespan.
Psychology and Aging, 17, 423–434.
Rosenberg, M. (1965). Society and adolescent self-image. Princeton, NJ: Princeton University.
Rosenberg, M., Schooler, C., Schoenbach, C., & Rosneberg, F. (1995). Global self- esteem and speciWc self-esteem:
diVerent concepts, diVerent outcomes. American Sociological Review, 60, 141–156.
Rowe, D. C., Vazsonyi, A. T., & Flannery, D. J. (1994). No more than skin deep: ethnic and racial similarity in
developmental process. Psychological Bulletin, 101, 396–413.
Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Validation of construct interpretations. Review of Educa-
tional Research, 46, 407–441.
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: modeling change and event occurrence. New
York: Oxford University Press.
Tangney, J. P., & Leary, M. R. (2003). The next generation of self research. In M. R. Leary & J. P. Tangney (Eds.),
Handbook of Self and Identity (pp. 667–674). New York: Wiley.
Trzesniewski, K. H., Donnellan, M. B., & Robins, R. W. (2003). Stability of self-esteem across the lifespan. Journal
of Personality and Social Psychology, 84, 205–220.
472 M.B. Donnellan et al. / Journal of Research in Personality 41 (2007) 453–472

Watson, D., & Clark, L. A. (1984). Negative aVectivity: the disposition to experience aversive emotional states.
Psychological Bulletin, 96, 465–490.
Young, J. F., & Mroczek, D. K. (2003). Predicting intraindividual self-concept trajectories during adolescence.
Journal of Adolescence, 26, 586–600.
Zimmerman, M. A., Copeland, L. A., Shope, J. T., & Dielman, T. E. (1997). A longitudinal study of self-esteem.
Implications for adolescent development.. Journal of Youth and Adolescence, 26, 117–141.

View publication stats

You might also like