research on self esteem over the lifespan indicates the following

Life-span development of self-esteem and its effects on important life outcomes

Affiliation.

• 1 Department of Psychology, University of Basel, Missionsstrasse 62, 4055 Basel, Switzerland. [email protected]
• PMID: 21942279
• DOI: 10.1037/a0025558

We examined the life-span development of self-esteem and tested whether self-esteem influences the development of important life outcomes, including relationship satisfaction, job satisfaction, occupational status, salary, positive and negative affect, depression, and physical health. Data came from the Longitudinal Study of Generations. Analyses were based on 5 assessments across a 12-year period of a sample of 1,824 individuals ages 16 to 97 years. First, growth curve analyses indicated that self-esteem increases from adolescence to middle adulthood, reaches a peak at about age 50 years, and then decreases in old age. Second, cross-lagged regression analyses indicated that self-esteem is best modeled as a cause rather than a consequence of life outcomes. Third, growth curve analyses, with self-esteem as a time-varying covariate, suggested that self-esteem has medium-sized effects on life-span trajectories of affect and depression, small to medium-sized effects on trajectories of relationship and job satisfaction, a very small effect on the trajectory of health, and no effect on the trajectory of occupational status. These findings replicated across 4 generations of participants--children, parents, grandparents, and their great-grandparents. Together, the results suggest that self-esteem has a significant prospective impact on real-world life experiences and that high and low self-esteem are not mere epiphenomena of success and failure in important life domains.

2012 APA, all rights reserved

Publication types

• Research Support, Non-U.S. Gov't
• Achievement*
• Age Factors
• Aged, 80 and over
• Depression / psychology
• Employment / psychology
• Health Status
• Job Satisfaction
• Life Change Events*
• Longitudinal Studies
• Middle Aged
• Personal Satisfaction*
• Self Concept*
• Young Adult

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Front Public Health

The Developmental Trajectory of Self-Esteem Across the Life Span in Japan: Age Differences in Scores on the Rosenberg Self-Esteem Scale From Adolescence to Old Age

Yuji ogihara.

1 Division of Cognitive Psychology in Education, Graduate School of Education, Kyoto University, Kyoto, Japan

2 Faculty of Science Division II, Tokyo University of Science, Tokyo, Japan

Takashi Kusumi

Associated data.

The datasets presented in this article are not readily available because participants were not asked to provide permission to disclose individual data at the time of data collection. Requests to access the minimal datasets (aggregate level) should be directed to Yuji Ogihara ( pj.ca.sut.sr@arahigoy ).

We examined age differences in global self-esteem in Japan from adolescents aged 16 to the elderly aged 88. Previous research has shown that levels of self-liking (one component of self-esteem) are high for elementary school students, low among middle and high school students, but then continues to become higher among adults by the 60s. However, it did not measure both aspects of self-esteem (self-competence and self-liking) or examine the elderly over the age of 70. To fully understand the developmental trajectory of self-esteem in Japan, we analyzed six independent cross-sectional surveys. These surveys administered the Rosenberg Self-Esteem Scale, which measured both self-competence and self-liking, on a large and diverse sample ( N = 6,113) that included the elderly in the 70s and 80s. Results indicated that, consistent with previous research, for both self-competence and self-liking, the average level of self-esteem was low in adolescence, but continued to become higher from adulthood to old age. However, a drop of self-esteem was not found over the age of 50, which was inconsistent with prior research in European American cultures. Our research demonstrated that the developmental trajectory of self-esteem may differ across cultures.

Introduction

Self-esteem, which is the positivity of a person's global evaluations of the self [e.g., Baumeister et al. ( 1 )], is one of the most famous indicators of mental health. To maintain good mental health, it is important to have a positive view of the self to some extent.

The average level of self-esteem changes across the life span along with changes in one's capacities (e.g., social, cognitive) and surrounding environments (e.g., social, economic). Uncovering the developmental trajectory of self-esteem across the life span is important at least for two reasons. First, it is crucial to reveal the effects of basic demographic variable on self-esteem. Age is one of the most frequently examined demographic variables. Thus, how age influences self-esteem should be investigated. Furthermore, when researchers are interested in the effects of other variables (e.g., socio-economic status, interpersonal relationships) on self-esteem, they should control for basic demographic variables that might confound these effects (e.g., age, gender). To statistically control for the effect of age, it is imperative to know in advance how age is associated with self-esteem. Second, investigating the developmental pattern across the lifetime contributes to understanding how self-esteem is formed, maintained, and influenced by changes in one's capacities and surrounding environments. For instance, finding two periods when self-esteem declines can estimate that a consistent factor common to the two periods might decrease self-esteem.

Revealing the developmental trajectory of self-esteem is also important for public health at least for two reasons. Such knowledge contributes to promoting public mental health by providing empirical evidence about the developmental pattern of self-esteem. First, knowing when self-evaluation tends to become negative over the life span facilitate effective prevention and provision. For example, parents and teachers can pay more attention to and provide more resource to people in the periods at higher risk. Second, knowing the period when self-evaluation is at high risk for turning negative can facilitate effective interventions and responses. Interventions and responses that are necessary depend on age categories (e.g., adolescence, old age). Moreover, people that are in the periods of lower self-esteem might feel relieved if they understand that they are not special and it is rather natural to have relatively low self-esteem during specific developmental stages.

Previous research especially in European American cultures has provided empirical evidence about the developmental trajectory of self-esteem over the life course. However, is this developmental pattern of self-esteem consistent across cultures?

Age Differences in Self-Esteem in European American Cultures

A large amount of research has investigated the developmental trajectory of self-esteem in European American cultures (especially in the U.S.; for reviews, see Orth and Robins ( 2 ); Robins and Trzensniewski ( 3 ). Robins et al. ( 4 ) investigated age differences in self-esteem from a broad range of population aged 9 to 90 years old in the U.S. They found that self-esteem is high in childhood, low in adolescence, but then continues to become higher in adulthood. Then, self-esteem peaks around the mid-60s, and shows a drop afterward. Moreover, Orth et al. ( 5 ) explored the developmental trajectory of self-esteem from young adults aged 25 to the elderly aged 104 by analyzing longitudinal data in the U.S. They showed that self-esteem increases from young adulthood through middle age, but then decreases from around the age of 60. In addition, Orth et al. ( 6 ) investigated the life-span development of self-esteem from adolescents aged 16 to the elderly aged 97 by examining other longitudinal data in the U.S. They demonstrated that self-esteem rises from adolescence to middle adulthood, peaks at about age 50, and declines in old age. This pattern of developmental change has been found not only in the U.S., but also in Germany. Orth et al. ( 7 ) examined the development of self-esteem from adolescents aged 14 to the elderly aged 89 by analyzing a longitudinal study. Results indicated that self-esteem increases from adolescence to middle adulthood, reaches a peak at about age 60 years, and decreases in old age in Germany.

Studies have shown that self-esteem reaches a peak in one's 50s or 60s, and then sharply drops in old age ( 4 – 7 ). This is a characteristic change, so it is important to reveal about when self-esteem peaks across the life span. This drop is thought to occur mainly for two reasons [e.g., Robins et al. ( 4 ); Robins and Tresniewski ( 3 )]. The first is the loss of things that are important to one's evaluation of oneself. These include the loss of socioeconomic positions or roles due to retirement, loss of close others (e.g., spouse, romantic partner), and a reduction in one's abilities (e.g., physical, cognitive). The second is a change in attitudes toward oneself. The elderly come to accept their limitations and faults, leading them to have more humble, modest, and balanced perspectives toward themselves.

Age Differences in Self-Esteem in Japan

Previous research has shown that self-esteem is profoundly affected by culture [e.g., Heine et al. ( 8 ); Schmitt and Allik ( 9 )], leading to the possibility that the developmental trajectory of self-esteem may differ across cultures. Thus, it is important to investigate the developmental change in self-esteem in cultures other than America and Europe 1 .

Prior research examined age differences in self-esteem from elementary school students aged 10 to the elderly in their 60s by analyzing cross-sectional data from a large, representative and diverse sample in Japan ( 12 ). It showed that levels of self-esteem were high for elementary school students, low among middle and high school students, but then gradually continued to become higher among adults, consistent with the pattern obtained in European American cultures ( 2 , 3 ). Moreover, previous research has indicated the same pattern of age differences in self-esteem from middle school students to the elderly in their 60s by analyzing another independent and large-sample survey ( 13 ).

However, previous research had two limitations. First, it did not directly examine age differences in global self-esteem in Japan. Prior research investigated age differences in self-esteem by focusing on one component of self-esteem: self-liking [“our affective judgment of ourselves, our approval or disapproval of ourselves, in line with internalized social values” (( 14 ), p. 325); also see, Tafarodi and Milne ( 15 ); Tafarodi and Swann ( 16 )]. It has been shown that self-esteem consists of self-liking and self-competence (“the overall sense of oneself as capable, effective, and in control”; ( 14 ), p. 325), which are strongly correlated with each other and construct self-esteem. Thus, it is strongly predicted that age differences in self-esteem would be consistent with those in self-liking. However, this has not been examined empirically. Although we do not have strong evidence, it is possible that patterns of age differences in self-competence are different from patterns of age differences in self-liking. To reveal the developmental trajectory of global self-esteem, it is desirable to directly investigate age differences in self-esteem by capturing both of its aspects simultaneously.

Second, previous research did not sufficiently investigate age differences in self-esteem in the elderly over the age of 70, leaving the developmental trajectory of self-esteem after the age of 70 in Japan unclear. Previous research in European American cultures has indicated that the average level of self-esteem drops sharply in the elderly period ( 2 , 3 ). To capture the whole picture of the developmental trajectory of self-esteem in Japan, it is necessary to investigate whether this sharp drop is also found among the elderly in Japan. Many studies have shown that people in Japan have more humble, modest and balanced attitudes toward themselves compared to people in European American cultures [e.g., Heine et al. ( 8 ); Heine and Hamamura ( 17 )]. Given that the sharp decline in self-esteem observed in European American cultures may be caused by increases in such attitudes in old age, it is possible that a decline may be absent or less sharp in Japanese older adults. Indeed, one prior study did not find a drop in self-liking between the ages of 50 and 69, which implies that a decline may be absent or found later in old age in Japan ( 18 ). Thus, the developmental pattern of self-esteem may differ across cultures, which should be investigated empirically.

Present Research

To overcome the first limitation of previous research, we measured global self-esteem by administering the Rosenberg Self-Esteem Scale [RSES; ( 11 )]. This scale is one of the most frequently used measures of global self-esteem. We predicted that the developmental trajectory of self-esteem would be consistent with that of self-liking: levels of self-esteem were low in adolescence, but then continued to become higher among adults. Here, we also empirically examined whether self-competence and self-liking were closely related to each other. We expected that self-competence and self-liking would be highly correlated with each other, and the developmental pattern of self-competence would be consistent with that of self-liking. To overcome the second limitation, we collected data covering a more diverse sample that included the elderly over the age of 70. We predicted that a drop of self-esteem would be absent or less sharp in Japanese older adults. In sum, in the current research, we investigated age differences in global self-esteem among a broader range of the population in Japan by using the RSES.

Prior research has shown that the pattern of age differences in self-esteem is similar between males and females in the U.S. [for a review see, Orth and Robins ( 2 )] and Germany ( 7 ). Although the patterns are consistent between gender, in some cases, small differences were found with females showing larger age differences than males ( 4 , 5 ). This was also the case in Japan ( 19 ). Thus, we also investigated whether age differences in self-esteem are moderated by gender. We predicted an absence of the moderating effect of gender, but if there were any differences, they would be small differences, which would be larger in females than in males.

We analyzed six independent web surveys administered to a large and diverse sample in Japan.

Each survey was conducted independently in 2009, 2011, 2012, 2013, 2017, and 2018. The data in 2009 was collected by Kyoto University Global COE Program (Revitalizing Education for Dynamic Hearts and Minds). The data for this secondary analysis, “International Comparative Research on Sense of Happiness, 2009–2011” was provided by the Social Science Japan Data Archive (Center for Social Research and Data Archives, Institute of Social Science, The University of Tokyo). The data from 2011, 2012, 2013, 2017, and 2018 were collected by us [e.g., Ogihara et al. ( 20 )]. We recruited participants from every prefecture in Japan on the internet via research firms. The research firms had their own large pools of participants. In each survey, a designated number of participants were assigned to each cell by age category and gender. Participants were rewarded after answering the survey. The summary of each survey is shown in Table 1 .

Summary of surveys.

A: the Japanese translation of the Rosenberg Self-Esteem Scale used in previous research [e.g., Heine et al. ( 8 ); Uchida et al. ( 21 )], B: Yamamoto et al. ( 22 ) .

Sample sizes by gender and generation are indicated in Table 2 . The sample sizes ranged from 763 to 1,331 and the total sample size was 6,113.

Sample sizes by gender and generation.

Self-Esteem

The Rosenberg Self-Esteem Scale, a 10-item measure of global self-esteem [e.g., “I feel that I have a number of good qualities,” “On the whole, I am satisfied with myself.”; ( 11 )], was administered.

In the 2017 and 2018 surveys, the Japanese translation of the RSES from Yamamoto et al. ( 22 ) was used (5-point scale; 1: Not applicable−5: Applicable). In the other surveys, a different translation of the RSES that has been used in previous research [e.g., Heine et al. ( 8 ); Uchida et al. ( 21 )] was administered (7-point scale; 1: Strongly disagree−7: Strongly agree) 2 . Reliabilities of the RSES in the six surveys were sufficiently high (αs > 0.86; Table 1 ).

The average scores for self-competence (e.g., “I feel that I have a number of good qualities”; SC1 3 ) and self-liking (e.g., “On the whole, I am satisfied with myself.”; SL1 4 ) were calculated by averaging the five items of the RSES, as was done in previous research ( 15 ). The reliabilities for self-competence (αs > 0.77; Table 1 ) and self-liking (αs > 0.69; Table 1 ) were sufficiently high.

First, we confirmed whether the Self-Esteem Scale measured the same concepts between genders and age-groups (i.e., measurement invariance/equivalence). We divided the participants into subgroups and conducted multi-group confirmatory factor analysis (CFA). For this analysis, we split the participants into three age groups: younger adults (10s 5 , 20s, 30s), middle-aged adults (40s, 50s), and older adults (60s, 70s, 80s) 6 . Following previous research ( 14 , 15 ), we made a two-factor model in which self-competence (measured by five items) and self-liking (measured by five items) constituted self-esteem ( Figure 1 ) 7 . For the multi-group CFA, we used IBM SPSS Amos (ver. 26).

A two-factor model in which self-competence and self-liking constitute self-esteem.

Then, to check whether self-competence and self-liking are closely related to each other, we calculated their correlation coefficients at the individual level and at the age level.

Next, we conducted hierarchical multiple linear regression analyses on each dataset for predicting self-esteem from age and gender 8 . The independent variables we entered in Step 1 were gender (male = 0, female = 1), the age, and their interaction (age × gender), and in Step 2, the age squared and its interaction with gender (age 2 × gender), and in Step 3 the age cubed and its interaction with gender (age 3 × gender). If the interaction effect was significant, we conducted hierarchical multiple linear regression analyses separately by gender for predicting self-esteem from the age (Step 1), the age squared (Step 2), and the age cubed (Step 3). In these analyses, we centered each age variable and weighted sample sizes to estimate the developmental trajectory of self-esteem more precisely.

Finally, we looked at whether these developmental patterns were found in each component (i.e., self-competence and self-liking) by conducting a series of hierarchical multiple linear regression analyses. The dependent variable was the average score for each component (self-competence or self-liking; not global self-esteem). In Step 1, the independent variables were age, the age-squared, and the type of component (categorical variable: self-competence = 0, self-liking = 1). In Step 2, the interaction terms were added (i.e., the age × the component, the age-squared × the component). We used IBM SPSS (ver. 25) for the analyses.

Measurement Invariance

We tested measurement invariance in two steps. First, we conducted a confirmatory factor analysis for each subgroup separately (single-group CFA) in each dataset. Second, we conducted confirmatory factor analysis by including the subgroups (multi-group CFA) at four successive levels in each dataset. These results are summarized in Table 3 .

Fit indices for the confirmatory factor analyses.

df, degree of freedom; SRMR, standardized root mean square residual; RMSEA, root mean square error of approximation; CI, confidence interval; CFI, comparative fit index .

Single-Group CFA

The model fits were acceptable to adequate in all the datasets ( Table 3 ), showing that the two-factor model successfully described the construction of self-esteem in each subgroup.

Multi-Group CFA

To evaluate the fitness of the model at four hierarchical levels, we used changes in CFI (ΔCFI) index. Specifically, if ΔCFI was smaller than 0.010, the successive model that constrained more equality was accepted ( 27 ). We did not use Δχ 2 for this evaluation because χ 2 is sensitive to sample size ( 27 ).

First, configural invariance was tested. Configural invariance indicates that structure of latent factors and observed variables (e.g., number of latent factors, same associations between each factor and observed items) are consistent across groups. Second, metric invariance (weak factorial invariance) was investigated. Metric invariance means that factor loadings are comparable across groups, indicating that participants in different groups respond to each item in a similar way. We constrained each factor loading to be equal across groups. Third, scalar invariance (strong factorial invariance) was tested. Scalar invariance indicates that factor loadings and intercepts of items are comparable across groups, showing that latent factor scores lead to observed scores in the same way across groups. We constrained each factor loading and intercept of each observed variable to be equal across groups. Finally, structural invariance (factor variance/covariance invariance) was examined. Structural invariance means that latent factors are distributed and associated similarly across groups. We constrained the variance of each factor and covariance of the two factors across groups.

Regarding gender, five out of the six datasets (2009, 2011, 2012, 2017, 2018 datasets) demonstrated scalar and structural invariance, and one (2013 dataset) showed partial scalar 9 and structural invariance. In the 2013 dataset, ΔCFI between the metric invariance model and the scalar invariance model was 0.011, which was slightly above the conventional criterion of 0.010 ( 27 ). This criterion is not a golden rule, and excluding only one constraint of item intercept cleared the criterion (partial scalar invariance). In all of the datasets, at least partial scalar and structural invariance were supported.

Regarding age-group, one out of the six datasets (2011 dataset) showed scalar and structural invariance, four (2009, 2012, 2013, 2018 datasets) showed partial scalar 10 and structural invariance, one (2017 dataset) showed metric and factorial invariance 11 . In the 2017 dataset, ΔCFI between the metric invariance model and the partial scalar invariance model was 0.012, which was slightly above the conventional criterion of 0.010 ( 27 ). Overall, in the most datasets, at least partial scalar and structural invariance were supported.

The Relationship Between Self-Competence and Self-Liking

We calculated correlation coefficients between self-competence and self-liking at the individual level and at the age level in each of the six studies ( Table 4 ). At the individual level, they were highly correlated for the total population ( r s > 0.72, p s < 0.001), for males ( r s > 0.68, p s < 0.001), and for females ( r s > 0.72, p s < 0.001). These strong relationships were also found within sub-populations (each age group for both males and females). Relatively lower coefficients for some sub-populations ( r = 0.42 for males in their 70s in 2018, r = 0.27 for females in their teens in 2018) were due to the small sample sizes ( n = 44 for males in their 70s in 2018, n = 5 for females in their teens in 2018; see Table 2 ). Similarly, at the age level, the correlations were large both for males ( r s > 0.71, p s < 0.001) and females ( r s > 0.79, p s < 0.001).

Correlation coefficients between self-competence and self-liking.

Age Differences in Global Self-Esteem

A summary of the regression models predicting self-esteem from age and gender is indicated in Table 5 .

Summary of regression models predicting self-esteem from age and gender.

Gender was coded as male = 0, female = 1 .

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender significantly increased the coefficient of determination (Step 2; Table 5 ). The addition of the age cubed and its interaction with gender did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted. We conducted a consistent analysis separately by gender because the interaction between the age squared and gender was significant.

The age significantly predicted self-esteem (Step 1), and the addition of the age squared term significantly increased the coefficient of determination (Step 2; Table 6 ). The addition of the age cubed term did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted ( Figure 2 ). Self-esteem showed a slight downward trend from the teens to the mid-30s (the lowest predicted score was the age of 32 12 ), but then it continued to become higher to the 80s 13 .

Summary of regression models predicting self-esteem from age by gender (in the 2009, 2011, and 2013 surveys).

Average and predicted self-esteem scores across ages in Japan (2009 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 2 ). Self-esteem continued to become higher from the teens to the 80s ( d = 0.97 14 ) 15 .

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. We conducted a consistent analysis separately by gender because the interaction between the age and gender was significant.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 3 ). Self-esteem continued to become higher from the 20s to the 50s ( d = 0.51).

Average and predicted self-esteem scores across ages in Japan (2011 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 3 ). Self-esteem continued to become higher from the 20s to the 50s ( d = 1.05). The slope for females ( B = 0.03, p < 0.001) was larger than that for males ( B = 0.01, p < 0.05).

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. The interaction between the age and gender was not significant, showing that self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 4 ) 16 .

Predicted self-esteem scores across ages in Japan (2012 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 5 ). Self-esteem continued to become higher from the teens to the 60s ( d = 1.00).

Average and predicted self-esteem scores across ages in Japan (2013 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 5 ). Self-esteem continued to become higher from the teens to the 60s ( d = 1.42). The slope for females ( B = 0.02, p < 0.001) was larger than that for males ( B = 0.01, p < 0.001).

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender significantly increased the coefficient of determination (Step 2; Table 5 ). The addition of the age cubed and its interaction with gender did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted. The interaction between the age squared and gender was not significant. Self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 6 ) 17 .

Predicted self-esteem scores across ages in Japan (2017 survey). Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. The interaction between the age and gender was not significant, showing that self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 7 ) 18 .

Predicted self-esteem scores across ages in Japan (2018 survey). Note. Error bars represent 95% confidence intervals.

Age Differences in Self-Competence and Self-Liking (The Two Components of Global Self-Esteem)

The results of the hierarchical multiple regression analyses are summarized in Table 7 . Except for females in 2009, the additions of their interaction terms did not significantly increase the coefficient of determination. Neither of the interaction terms significantly predicted the scores. These results consistently suggest that the developmental patterns for self-competence and self-liking were not different.

Hierarchical multiple linear regression predicting components of self-esteem from age, component type and their interactions.

For females in 2009, Step 2 significantly increased the coefficient of determination. The interaction term between age and the component significantly predicted the score. Thus, we conducted the same hierarchical multiple regression analyses on each component as we did for the global self-esteem ( Table 8 ). Although in both components age significantly explained the scores, the slope of the increase was higher for self-liking ( B = 0.02, p < 0.001) than for self-competence ( B = 0.01, p < 0.001; Figure 8 ). Yet, both patterns consistently indicated a continuous upward trend in self-esteem.

Regression models predicting each sub-component of self-esteem from age for females in the 2009 study.

Average and predicted self-esteem scores by component across ages among females in Japan (2009 survey). Note. Error bars represent 95% confidence intervals.

Due to the consistent relationship between self-competence and self-liking, the age patterns for each component were same as those for global self-esteem. Specifically, the quadratic increases were found among males in 2009 and both genders in 2017, and the linear increases were found in the other subgroups.

Summary of the Results and Implications

We investigated age differences in self-esteem in Japan from adolescents aged 16 to the elderly aged 88 by using the RSES. Previous cross-sectional research has investigated age differences in self-esteem in Japan ( 12 , 13 , 17 ). However, it had two limitations: (1) it did not directly examine age differences in global self-esteem and (2) it did not investigate self-esteem among the elderly over the age of 70. These limitations had to be overcome to fully understand the developmental trajectory in self-esteem across the life span in Japan. Therefore, we examined age differences in self-esteem by conducting the RSES on more diverse sample that included the elderly over the age of 70.

First, as predicted, we found a pattern to the developmental trajectory of global self-esteem that is consistent with previous research on self-liking ( 12 , 13 , 18 ). We had predicted that the developmental pattern of self-esteem would be consistent with that of self-liking, but we had not had empirical evidence to support it. In this research, we empirically confirmed that self-competence and self-liking are closely associated with each other and have a consistent developmental pattern in Japan. Specifically, across the six cross-sectional surveys, the average level of self-esteem was low in adolescence, but then gradually continued to become higher from young adulthood to late adulthood. This trajectory was consistent with findings in previous research in European American cultures ( 2 , 3 ).

Second, as expected, analyses showed that the average level of self-esteem continued to indicate an upward trend beyond the age of 50 in Japan. All of the six independent cross-sectional datasets consistently showed that self-esteem continued to become higher from adulthood to old age both for males and females. This finding was inconsistent with previous research that showed a drop in self-esteem over the age of 50 in European American cultures ( 2 , 3 ). With old age comes a more humble, modest, and balanced perspective toward oneself, which leads to a decline in self-esteem in old age in European American cultures [e.g., Robins et al. ( 4 ); Robins and Tresniewski ( 3 )]. Previous research has indicated that, compared to people in European American cultures, people in Japan have more humble and balanced attitudes toward themselves, not just in old age [e.g., Heine et al. ( 8 ); Heine and Hamamura ( 17 )], which may account for the absence of a drop of self-esteem among the Japanese people over the age of 50. Thus, this research demonstrates that different developmental patterns can emerge in different social/cultural environments.

One may wonder whether the absence of a sharp drop in self-esteem in Japan is caused by the fact that participants answered the questionnaire on the internet. Elderly people who use the internet may differ from the elderly population in general (e.g., they may be wealthier and healthier). However, this was also the case in previous research that observed a clear drop in self-esteem among the elderly in the U.S. (e.g., ( 4 )). Thus, this explanation is insufficient to account for the cultural difference in the developmental trajectory of self-esteem among the elderly. Still, it is desirable to collect more representative data especially from the elderly and see if the result is consistent with the present research.

Three datasets showed that gender differences in the pattern of age differences were absent. The other three datasets indicated that there were slight differences between gender: slopes for females were a little larger than those for males. In sum, the pattern of age differences in self-esteem was similar between gender, if any small differences. These results were consistent with previous research ( 2 , 7 , 19 ).

We also confirmed the measurement invariance of the Rosenberg's Self-Esteem Scale ( 11 ) across gender and age-groups in Japan. Regarding gender, five out of the six datasets demonstrated scalar and structural invariance, and one showed partial scalar and structural invariance. Thus, in all of the datasets, at least partial scalar and structural invariance were supported. Regarding age-group, one out of the six datasets showed scalar and structural invariance, four showed partial scalar and structural invariance, and one showed metric and structural invariance. Overall, in the most datasets, at least partial scalar and structural invariance were supported. These results showed that the model structure and the adequacy of the measure were invariant across gender and age-groups.

Limitations and Future Directions

This research investigated age differences in self-esteem from adolescents in their teens to the elderly in their 80s by analyzing six cross-sectional datasets from a large and diverse sample. But, in cross-sectional data, age differences involve cohort differences. This research is a reasonable first step to understand the developmental trajectory of self-esteem across the life span in Japan. In fact, the absence of a drop in self-esteem in Japan might be caused by the cohort effect. Specifically, older cohorts might have higher self-esteem and younger cohorts might have lower self-esteem ( 23 – 26 ), which might obscure the drop of self-esteem in Japan. Thus, in the future, it is necessary to analyze longitudinal data which can distinguish between age differences and cohort differences.

Another limitation is that, although the sample sizes were relatively large for teens ( n = 211), adults ( n 20s = 1,161, n 30s = 970, n 40s = 1,382, n 50s = 1,056), and the elderly in their 60s ( n = 1,031) and 70s ( n = 253), the sample size for the elderly in their 80s was small ( n = 49). Thus, the results for this age group may be unreliable. It would be desirable to examine this point further by collecting more representative data.

Data Availability Statement

Ethics statement.

This study was carried out in accordance with the recommendations of the Declaration of Helsinki and the Japanese Psychological Association with written informed consent from all subjects. Ethical approval was not required for this study in accordance with the local legislation and institutional requirements. This study asked participants to answer one of the most frequently used questionnaires (Rosenberg Self-Esteem Scale) and this questionnaire did not include any items that may harm participants. Thus, the protocol was not submitted to an ethics committee. Both individual researchers in charge of each survey and their respective research firms confirmed that all surveys were without any ethical concerns. All subjects gave written informed consent via the online questionnaire in accordance with the Declaration of Helsinki.

Author Contributions

YO contributed conception and design of the study, performed the statistical analysis, and wrote the first draft of the manuscript. TK collected the data and provided the critical comments on the manuscript. All authors contributed to manuscript revision and approved the final manuscript.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer ST declared a shared affiliation, with no collaboration, with the authors to the handling editor at the time of review.

Acknowledgments

We thank Pamela Taylor for her helpful comments on our previous versions of the manuscript.

1 Bleidorn et al. ( 10 ) examined cultural variation in age and gender differences in self-esteem across 48 nations including Japan. However, with regards to cultural differences in the developmental trajectory of self-esteem, it had at least three limitations. First, it investigated age differences in self-esteem between the ages of 16 and 45, leaving it unclear how average levels of self-esteem change after early middle age. Second, it indicated that self-esteem in females showed an upward trend from early adolescence to middle adulthood in Japan whereas self-esteem in males showed a downward trend during this period. Among 48 nations, this age pattern in males was found only in Japan, but the reason for this was not discussed. Third, as mentioned as a limitation by the authors, it used only one item “I see myself as someone who has high self-esteem” (a 5-point scale ranging from 1: disagree strongly to 5: agree strongly) instead of adequately constituted scale [e.g., Rosenberg Self-Esteem Scale; Rosenberg ( 11 )]. Although it was stated that the validity and reliability had been confirmed in the U.S. (although the item and its anchors differed from the original research), it is unclear whether the validity and reliability were confirmed in other nations including Japan. The word “self-esteem” is abstract and conceptual, so it may be interpreted differently across cultures and/or within each culture. This may have contributed to the exceptional pattern found in Japan (i.e., the continuous downward trend in males and the continuous upward trend in females).

2 The translations differed between years is because these studies were conducted as an omnibus survey with other researchers.

3 The other four items assessing self-competence were “I am able to do things as well as most other people.” (SC2), “I feel that I'm a person of worth, at least on an equal plane with others.” (SC3), “I feel I do not have much to be proud of.” (SC4), and “All in all, I am inclined to feel that I am a failure.” (SC5).

4 The other four items assessing self-liking were “I take a positive attitude toward myself.” (SL2), “I certainly feel useless at times.” (SL3), “At times I think I am no good at all.” (SL4), and “I wish I could have more respect for myself.” (SL5).

5 Although teenagers are usually not regarded as younger adults, because the size of this sample was relatively small, teenagers were conventionally included in the younger adults group in this analysis.

6 Because the datasets in 2011 and 2012 did not include participants aged over 60 years, we split the participants into two age groups: younger adults (20s, 30s) and middle-aged adults (40s, 50s).

7 Because the covariances between error terms of SC1, SC2, SC3, and error terms of SL1 and SL2 were large, we included them in the model (Model 1). These high associations may be due to the nature of the items: the items were affirmative, and the remaining (i.e., SC4, SC5, SL3, SL4, SL5) were reversed items. Even if we did not include these covariances in the model, the patterns of the results of multi-group CFA were consistent.

8 We did not examine whether the developmental pattern of self-esteem (e.g., the shape of changes, the magnitude of the slope) differed among the six datasets. Naturally, the developmental pattern is different because the time periods covered in the surveys are significantly different ( Tables 1 , ​ ,2). 2 ). Moreover, previous research has shown that the average levels of self-esteem decreased over time in Japan [( 23 – 25 ); for a review of historical changes in self-esteem in Japan see Ogihara ( 26 )]. Thus, the self-esteem score should not be simply averaged across the surveys that were conducted over nine years. Even if the self-esteem scores are z-transformed in each survey, the periods covered by the surveys are significantly different, and they should not be aggregated.

9 In the partial scalar invariance model, item intercepts that were constrained did not include SL5 (i.e., excluding the constraint of SL5's item intercept from the full scalar model).

10 In the partial scalar invariance models, item intercepts that were constrained were as follows. 2009: SC1, SC2, SC3, SL2; 2012: other than SL3; 2013: SC1, SC3, SL2, SL5; 2017: SC4, SL5; 2018: SC2, SL5.

11 The reason for these differences in the levels of measurement invariance could be due to differences in factors such as total sample sizes, proportions of age-group sample size, and type of scale translations. However, we do not have sufficient data to empirically detect the reason(s). Thus, in the future, it would be important to investigate the possibility that people across a broad range of age might respond to RSES items differently.

12 The lowest predicted score was at the age of 32 when raw data were unavailable. Thus, effect sizes were not calculated.

13 We describe developmental patterns of self-esteem based on age differences in self-esteem. However, it should be noted that our data were cross-sectional and not cross-temporal. Age differences include not only developmental changes that an individual experiences over the life course, but also cohort differences. Thus, these age differences do not necessarily mean developmental changes. Although findings obtained from cross-sectional research on self-esteem are consistent with those obtained from cross-temporal research [for a review, see Orth and Robins ( 2 ), this should be kept in mind (also see, the Limitations and Future Directions section of the Discussion below).

14 Effect sizes were calculated by using predicted average scores, standard deviations and sample sizes in three years (a given age, and 1 year before and after the age) to secure larger sample sizes and avoid unstable results.

15 Comparing the slopes in linear models (Step 1) between gender, the slope for females ( B = 0.011, p < 0.001) was slightly larger than that for males ( B = 0.009, p < 0.001).

16 We have also shown the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S1 ).

17 We have also shown the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S2 ).

18 We have also showed the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S3 ).

Funding. This work was supported by the Japanese Group Dynamics Association.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2020.00132/full#supplementary-material

Self-Esteem

How self-esteem changes over the lifespan, self-esteem builds over the lifespan and peaks at age 60..

Posted September 6, 2018 | Reviewed by Lybi Ma

• What Is Self-Esteem?
• Find a therapist near me

Positive self-regard varies from person to person, but research shows that this psychological resource rises and falls in systematic ways across the lifespan.

Scientists recently combed through numerous studies of self-esteem to chart the average changes that occur from childhood to old age. The trajectory they observed challenges ideas about how self-esteem develops and deepens our understanding of a trait thought to influence relationships, health, education , and professional success.

“This is the first time researchers have charted out, across studies, the trajectory of self-esteem,” says Brent Donnellan , a professor of psychology at Michigan State University who was not involved with the research. “It’s a massive contribution to understanding self-esteem across the lifespan.”

The team analyzed 331 studies that assessed self-esteem, collectively covering more than 164,000 people between 4 and 94 years old. Self-esteem is measured with questionnaires in which respondents state to what extent they agree with statements such as “I feel that I'm a person of worth, at least on an equal basis with others” or “I wish I could have more respect for myself.”

The investigators discovered that self-esteem tended to rise slightly from ages 4 to 11, remain stagnant from 11 to 15, increase markedly from 15 to 30, and subtly improve until peaking at 60. It stayed constant from 60 to 70 years old, declined slightly from ages 70 to 90, and dropped sharply from 90 to 94. (Fewer studies addressed the oldest and youngest age groups—just a couple each for the 4 to 6 range and 90 to 94 range—so the evidence is weaker for the tail ends of the spectrum.) The results were published in the journal Psychological Bulletin .

“The trajectory is much more positive than previously thought,” says Ulrich Orth , the lead author of the study and a developmental psychologist at the University of Bern. “Most people experience positive changes in self-esteem as they go through life, and only in very old age does the trend reverse.”

Every individual has a unique set of experiences; the trends observed only chart the average changes that occur. Still, the overall growth in self-esteem between ages 4 and 60 represents substantial change. “The cumulative increase in self-esteem going from childhood to young adulthood to midlife was much larger than I expected,” says Richard Robins , a psychology professor at the University of California, Davis, who was not involved in the research but has worked closely with Orth in the past. For example, the jump in self-esteem was greater in magnitude than the difference between men and women in body weight, Robins explains.

The findings challenge assumptions scientists previously held about certain age groups, Orth says. Past evidence suggested that children experience a decrease in self-esteem between 7 and 9 years old. It was thought that youngsters initially develop an inflated sense of self, which they revise when cognitive advancements allow them to distinguish between the real and ideal self—leading to a dip in self-esteem. However, self-worth turned out to increase slightly during this time window.

Another previous assumption was that adolescents experience a sharp drop in self-esteem thanks to factors such as challenging academic environments, social comparison, and the physiological changes brought on by puberty. Nevertheless, the review demonstrates that on average, self-esteem held steady. The finding doesn’t necessarily imply that everyone maintains the status quo, Robins notes. Changes that take place in adolescence likely lead some to grow and others to struggle, combining to display no overall change.

Past studies also suggested that older folks experience a notable drop in self-esteem, Orth says. The review demonstrated a more benign decrease through age 70 and a stark change only at age 90. Despite the challenges of aging, such as retirement , physical health problems, reduced social mobility, and loss of family and friends, the elderly can maintain relatively high levels of self-esteem. “It’s important to take care of the elderly and help them maintain high self-esteem,” Orth says. “It would be desirable for everyone to be satisfied with themselves when they look back on their life.” Evidence suggests that low self-esteem is a risk factor for developing depression , he adds, so addressing self-esteem could potentially help improve health and well-being for seniors.

Robins has observed the relationship between age and self-esteem firsthand. His father, Al, had worked for the government for years as a psychologist in human resources. When he was in his 80s, he moved to an assisted living home. Al struggled with feeling helpless, useless, and unable to leverage his professional skills, Robins says. He was also extremely frustrated by the institution’s inefficiencies. So Robins suggested that his father meet with the director and share a few ideas about how to improve operations and employee management . The two ended up meeting regularly. “It feels great to use all the knowledge I’ve acquired over the course of my life,” Robins recalls his father saying. “And, secondly, I’m finally getting this place into shape.”

Abigail Fagan is a Senior Associate Editor at Psychology Today .

• Find a Therapist
• Find a Treatment Center
• Find a Psychiatrist
• Find a Support Group
• Find Online Therapy
• United States
• Brooklyn, NY
• Chicago, IL
• Houston, TX
• Los Angeles, CA
• New York, NY
• Portland, OR
• San Diego, CA
• San Francisco, CA
• Seattle, WA
• Washington, DC
• Asperger's
• Bipolar Disorder
• Chronic Pain
• Eating Disorders
• Passive Aggression
• Personality
• Goal Setting
• Positive Psychology
• Stopping Smoking
• Low Sexual Desire
• Relationships
• Child Development
• Self Tests NEW
• Therapy Center
• Diagnosis Dictionary
• Types of Therapy

At any moment, someone’s aggravating behavior or our own bad luck can set us off on an emotional spiral that threatens to derail our entire day. Here’s how we can face our triggers with less reactivity so that we can get on with our lives.

• Emotional Intelligence
• Gaslighting
• Affective Forecasting
• Neuroscience

ORIGINAL RESEARCH article

The developmental trajectory of self-esteem across the life span in japan: age differences in scores on the rosenberg self-esteem scale from adolescence to old age.

• 1 Division of Cognitive Psychology in Education, Graduate School of Education, Kyoto University, Kyoto, Japan
• 2 Faculty of Science Division II, Tokyo University of Science, Tokyo, Japan

We examined age differences in global self-esteem in Japan from adolescents aged 16 to the elderly aged 88. Previous research has shown that levels of self-liking (one component of self-esteem) are high for elementary school students, low among middle and high school students, but then continues to become higher among adults by the 60s. However, it did not measure both aspects of self-esteem (self-competence and self-liking) or examine the elderly over the age of 70. To fully understand the developmental trajectory of self-esteem in Japan, we analyzed six independent cross-sectional surveys. These surveys administered the Rosenberg Self-Esteem Scale, which measured both self-competence and self-liking, on a large and diverse sample ( N = 6,113) that included the elderly in the 70s and 80s. Results indicated that, consistent with previous research, for both self-competence and self-liking, the average level of self-esteem was low in adolescence, but continued to become higher from adulthood to old age. However, a drop of self-esteem was not found over the age of 50, which was inconsistent with prior research in European American cultures. Our research demonstrated that the developmental trajectory of self-esteem may differ across cultures.

Introduction

Self-esteem, which is the positivity of a person's global evaluations of the self [e.g., Baumeister et al. ( 1 )], is one of the most famous indicators of mental health. To maintain good mental health, it is important to have a positive view of the self to some extent.

The average level of self-esteem changes across the life span along with changes in one's capacities (e.g., social, cognitive) and surrounding environments (e.g., social, economic). Uncovering the developmental trajectory of self-esteem across the life span is important at least for two reasons. First, it is crucial to reveal the effects of basic demographic variable on self-esteem. Age is one of the most frequently examined demographic variables. Thus, how age influences self-esteem should be investigated. Furthermore, when researchers are interested in the effects of other variables (e.g., socio-economic status, interpersonal relationships) on self-esteem, they should control for basic demographic variables that might confound these effects (e.g., age, gender). To statistically control for the effect of age, it is imperative to know in advance how age is associated with self-esteem. Second, investigating the developmental pattern across the lifetime contributes to understanding how self-esteem is formed, maintained, and influenced by changes in one's capacities and surrounding environments. For instance, finding two periods when self-esteem declines can estimate that a consistent factor common to the two periods might decrease self-esteem.

Revealing the developmental trajectory of self-esteem is also important for public health at least for two reasons. Such knowledge contributes to promoting public mental health by providing empirical evidence about the developmental pattern of self-esteem. First, knowing when self-evaluation tends to become negative over the life span facilitate effective prevention and provision. For example, parents and teachers can pay more attention to and provide more resource to people in the periods at higher risk. Second, knowing the period when self-evaluation is at high risk for turning negative can facilitate effective interventions and responses. Interventions and responses that are necessary depend on age categories (e.g., adolescence, old age). Moreover, people that are in the periods of lower self-esteem might feel relieved if they understand that they are not special and it is rather natural to have relatively low self-esteem during specific developmental stages.

Previous research especially in European American cultures has provided empirical evidence about the developmental trajectory of self-esteem over the life course. However, is this developmental pattern of self-esteem consistent across cultures?

Age Differences in Self-Esteem in European American Cultures

A large amount of research has investigated the developmental trajectory of self-esteem in European American cultures (especially in the U.S.; for reviews, see Orth and Robins ( 2 ); Robins and Trzensniewski ( 3 ). Robins et al. ( 4 ) investigated age differences in self-esteem from a broad range of population aged 9 to 90 years old in the U.S. They found that self-esteem is high in childhood, low in adolescence, but then continues to become higher in adulthood. Then, self-esteem peaks around the mid-60s, and shows a drop afterward. Moreover, Orth et al. ( 5 ) explored the developmental trajectory of self-esteem from young adults aged 25 to the elderly aged 104 by analyzing longitudinal data in the U.S. They showed that self-esteem increases from young adulthood through middle age, but then decreases from around the age of 60. In addition, Orth et al. ( 6 ) investigated the life-span development of self-esteem from adolescents aged 16 to the elderly aged 97 by examining other longitudinal data in the U.S. They demonstrated that self-esteem rises from adolescence to middle adulthood, peaks at about age 50, and declines in old age. This pattern of developmental change has been found not only in the U.S., but also in Germany. Orth et al. ( 7 ) examined the development of self-esteem from adolescents aged 14 to the elderly aged 89 by analyzing a longitudinal study. Results indicated that self-esteem increases from adolescence to middle adulthood, reaches a peak at about age 60 years, and decreases in old age in Germany.

Studies have shown that self-esteem reaches a peak in one's 50s or 60s, and then sharply drops in old age ( 4 – 7 ). This is a characteristic change, so it is important to reveal about when self-esteem peaks across the life span. This drop is thought to occur mainly for two reasons [e.g., Robins et al. ( 4 ); Robins and Tresniewski ( 3 )]. The first is the loss of things that are important to one's evaluation of oneself. These include the loss of socioeconomic positions or roles due to retirement, loss of close others (e.g., spouse, romantic partner), and a reduction in one's abilities (e.g., physical, cognitive). The second is a change in attitudes toward oneself. The elderly come to accept their limitations and faults, leading them to have more humble, modest, and balanced perspectives toward themselves.

Age Differences in Self-Esteem in Japan

Previous research has shown that self-esteem is profoundly affected by culture [e.g., Heine et al. ( 8 ); Schmitt and Allik ( 9 )], leading to the possibility that the developmental trajectory of self-esteem may differ across cultures. Thus, it is important to investigate the developmental change in self-esteem in cultures other than America and Europe 1 .

Prior research examined age differences in self-esteem from elementary school students aged 10 to the elderly in their 60s by analyzing cross-sectional data from a large, representative and diverse sample in Japan ( 12 ). It showed that levels of self-esteem were high for elementary school students, low among middle and high school students, but then gradually continued to become higher among adults, consistent with the pattern obtained in European American cultures ( 2 , 3 ). Moreover, previous research has indicated the same pattern of age differences in self-esteem from middle school students to the elderly in their 60s by analyzing another independent and large-sample survey ( 13 ).

However, previous research had two limitations. First, it did not directly examine age differences in global self-esteem in Japan. Prior research investigated age differences in self-esteem by focusing on one component of self-esteem: self-liking [“our affective judgment of ourselves, our approval or disapproval of ourselves, in line with internalized social values” (( 14 ), p. 325); also see, Tafarodi and Milne ( 15 ); Tafarodi and Swann ( 16 )]. It has been shown that self-esteem consists of self-liking and self-competence (“the overall sense of oneself as capable, effective, and in control”; ( 14 ), p. 325), which are strongly correlated with each other and construct self-esteem. Thus, it is strongly predicted that age differences in self-esteem would be consistent with those in self-liking. However, this has not been examined empirically. Although we do not have strong evidence, it is possible that patterns of age differences in self-competence are different from patterns of age differences in self-liking. To reveal the developmental trajectory of global self-esteem, it is desirable to directly investigate age differences in self-esteem by capturing both of its aspects simultaneously.

Second, previous research did not sufficiently investigate age differences in self-esteem in the elderly over the age of 70, leaving the developmental trajectory of self-esteem after the age of 70 in Japan unclear. Previous research in European American cultures has indicated that the average level of self-esteem drops sharply in the elderly period ( 2 , 3 ). To capture the whole picture of the developmental trajectory of self-esteem in Japan, it is necessary to investigate whether this sharp drop is also found among the elderly in Japan. Many studies have shown that people in Japan have more humble, modest and balanced attitudes toward themselves compared to people in European American cultures [e.g., Heine et al. ( 8 ); Heine and Hamamura ( 17 )]. Given that the sharp decline in self-esteem observed in European American cultures may be caused by increases in such attitudes in old age, it is possible that a decline may be absent or less sharp in Japanese older adults. Indeed, one prior study did not find a drop in self-liking between the ages of 50 and 69, which implies that a decline may be absent or found later in old age in Japan ( 18 ). Thus, the developmental pattern of self-esteem may differ across cultures, which should be investigated empirically.

Present Research

To overcome the first limitation of previous research, we measured global self-esteem by administering the Rosenberg Self-Esteem Scale [RSES; ( 11 )]. This scale is one of the most frequently used measures of global self-esteem. We predicted that the developmental trajectory of self-esteem would be consistent with that of self-liking: levels of self-esteem were low in adolescence, but then continued to become higher among adults. Here, we also empirically examined whether self-competence and self-liking were closely related to each other. We expected that self-competence and self-liking would be highly correlated with each other, and the developmental pattern of self-competence would be consistent with that of self-liking. To overcome the second limitation, we collected data covering a more diverse sample that included the elderly over the age of 70. We predicted that a drop of self-esteem would be absent or less sharp in Japanese older adults. In sum, in the current research, we investigated age differences in global self-esteem among a broader range of the population in Japan by using the RSES.

Prior research has shown that the pattern of age differences in self-esteem is similar between males and females in the U.S. [for a review see, Orth and Robins ( 2 )] and Germany ( 7 ). Although the patterns are consistent between gender, in some cases, small differences were found with females showing larger age differences than males ( 4 , 5 ). This was also the case in Japan ( 19 ). Thus, we also investigated whether age differences in self-esteem are moderated by gender. We predicted an absence of the moderating effect of gender, but if there were any differences, they would be small differences, which would be larger in females than in males.

We analyzed six independent web surveys administered to a large and diverse sample in Japan.

Each survey was conducted independently in 2009, 2011, 2012, 2013, 2017, and 2018. The data in 2009 was collected by Kyoto University Global COE Program (Revitalizing Education for Dynamic Hearts and Minds). The data for this secondary analysis, “International Comparative Research on Sense of Happiness, 2009–2011” was provided by the Social Science Japan Data Archive (Center for Social Research and Data Archives, Institute of Social Science, The University of Tokyo). The data from 2011, 2012, 2013, 2017, and 2018 were collected by us [e.g., Ogihara et al. ( 20 )]. We recruited participants from every prefecture in Japan on the internet via research firms. The research firms had their own large pools of participants. In each survey, a designated number of participants were assigned to each cell by age category and gender. Participants were rewarded after answering the survey. The summary of each survey is shown in Table 1 .

Table 1 . Summary of surveys.

Sample sizes by gender and generation are indicated in Table 2 . The sample sizes ranged from 763 to 1,331 and the total sample size was 6,113.

Table 2 . Sample sizes by gender and generation.

Self-Esteem

The Rosenberg Self-Esteem Scale, a 10-item measure of global self-esteem [e.g., “I feel that I have a number of good qualities,” “On the whole, I am satisfied with myself.”; ( 11 )], was administered.

In the 2017 and 2018 surveys, the Japanese translation of the RSES from Yamamoto et al. ( 22 ) was used (5-point scale; 1: Not applicable−5: Applicable). In the other surveys, a different translation of the RSES that has been used in previous research [e.g., Heine et al. ( 8 ); Uchida et al. ( 21 )] was administered (7-point scale; 1: Strongly disagree−7: Strongly agree) 2 . Reliabilities of the RSES in the six surveys were sufficiently high (αs > 0.86; Table 1 ).

The average scores for self-competence (e.g., “I feel that I have a number of good qualities”; SC1 3 ) and self-liking (e.g., “On the whole, I am satisfied with myself.”; SL1 4 ) were calculated by averaging the five items of the RSES, as was done in previous research ( 15 ). The reliabilities for self-competence (αs > 0.77; Table 1 ) and self-liking (αs > 0.69; Table 1 ) were sufficiently high.

First, we confirmed whether the Self-Esteem Scale measured the same concepts between genders and age-groups (i.e., measurement invariance/equivalence). We divided the participants into subgroups and conducted multi-group confirmatory factor analysis (CFA). For this analysis, we split the participants into three age groups: younger adults (10s 5 , 20s, 30s), middle-aged adults (40s, 50s), and older adults (60s, 70s, 80s) 6 . Following previous research ( 14 , 15 ), we made a two-factor model in which self-competence (measured by five items) and self-liking (measured by five items) constituted self-esteem ( Figure 1 ) 7 . For the multi-group CFA, we used IBM SPSS Amos (ver. 26).

Figure 1 . A two-factor model in which self-competence and self-liking constitute self-esteem.

Then, to check whether self-competence and self-liking are closely related to each other, we calculated their correlation coefficients at the individual level and at the age level.

Next, we conducted hierarchical multiple linear regression analyses on each dataset for predicting self-esteem from age and gender 8 . The independent variables we entered in Step 1 were gender (male = 0, female = 1), the age, and their interaction (age × gender), and in Step 2, the age squared and its interaction with gender (age 2 × gender), and in Step 3 the age cubed and its interaction with gender (age 3 × gender). If the interaction effect was significant, we conducted hierarchical multiple linear regression analyses separately by gender for predicting self-esteem from the age (Step 1), the age squared (Step 2), and the age cubed (Step 3). In these analyses, we centered each age variable and weighted sample sizes to estimate the developmental trajectory of self-esteem more precisely.

Finally, we looked at whether these developmental patterns were found in each component (i.e., self-competence and self-liking) by conducting a series of hierarchical multiple linear regression analyses. The dependent variable was the average score for each component (self-competence or self-liking; not global self-esteem). In Step 1, the independent variables were age, the age-squared, and the type of component (categorical variable: self-competence = 0, self-liking = 1). In Step 2, the interaction terms were added (i.e., the age × the component, the age-squared × the component). We used IBM SPSS (ver. 25) for the analyses.

Measurement Invariance

We tested measurement invariance in two steps. First, we conducted a confirmatory factor analysis for each subgroup separately (single-group CFA) in each dataset. Second, we conducted confirmatory factor analysis by including the subgroups (multi-group CFA) at four successive levels in each dataset. These results are summarized in Table 3 .

Table 3 . Fit indices for the confirmatory factor analyses.

Single-Group CFA

The model fits were acceptable to adequate in all the datasets ( Table 3 ), showing that the two-factor model successfully described the construction of self-esteem in each subgroup.

Multi-Group CFA

To evaluate the fitness of the model at four hierarchical levels, we used changes in CFI (ΔCFI) index. Specifically, if ΔCFI was smaller than 0.010, the successive model that constrained more equality was accepted ( 27 ). We did not use Δχ 2 for this evaluation because χ 2 is sensitive to sample size ( 27 ).

First, configural invariance was tested. Configural invariance indicates that structure of latent factors and observed variables (e.g., number of latent factors, same associations between each factor and observed items) are consistent across groups. Second, metric invariance (weak factorial invariance) was investigated. Metric invariance means that factor loadings are comparable across groups, indicating that participants in different groups respond to each item in a similar way. We constrained each factor loading to be equal across groups. Third, scalar invariance (strong factorial invariance) was tested. Scalar invariance indicates that factor loadings and intercepts of items are comparable across groups, showing that latent factor scores lead to observed scores in the same way across groups. We constrained each factor loading and intercept of each observed variable to be equal across groups. Finally, structural invariance (factor variance/covariance invariance) was examined. Structural invariance means that latent factors are distributed and associated similarly across groups. We constrained the variance of each factor and covariance of the two factors across groups.

Regarding gender, five out of the six datasets (2009, 2011, 2012, 2017, 2018 datasets) demonstrated scalar and structural invariance, and one (2013 dataset) showed partial scalar 9 and structural invariance. In the 2013 dataset, ΔCFI between the metric invariance model and the scalar invariance model was 0.011, which was slightly above the conventional criterion of 0.010 ( 27 ). This criterion is not a golden rule, and excluding only one constraint of item intercept cleared the criterion (partial scalar invariance). In all of the datasets, at least partial scalar and structural invariance were supported.

Regarding age-group, one out of the six datasets (2011 dataset) showed scalar and structural invariance, four (2009, 2012, 2013, 2018 datasets) showed partial scalar 10 and structural invariance, one (2017 dataset) showed metric and factorial invariance 11 . In the 2017 dataset, ΔCFI between the metric invariance model and the partial scalar invariance model was 0.012, which was slightly above the conventional criterion of 0.010 ( 27 ). Overall, in the most datasets, at least partial scalar and structural invariance were supported.

The Relationship Between Self-Competence and Self-Liking

We calculated correlation coefficients between self-competence and self-liking at the individual level and at the age level in each of the six studies ( Table 4 ). At the individual level, they were highly correlated for the total population ( r s > 0.72, p s < 0.001), for males ( r s > 0.68, p s < 0.001), and for females ( r s > 0.72, p s < 0.001). These strong relationships were also found within sub-populations (each age group for both males and females). Relatively lower coefficients for some sub-populations ( r = 0.42 for males in their 70s in 2018, r = 0.27 for females in their teens in 2018) were due to the small sample sizes ( n = 44 for males in their 70s in 2018, n = 5 for females in their teens in 2018; see Table 2 ). Similarly, at the age level, the correlations were large both for males ( r s > 0.71, p s < 0.001) and females ( r s > 0.79, p s < 0.001).

Table 4 . Correlation coefficients between self-competence and self-liking.

Age Differences in Global Self-Esteem

A summary of the regression models predicting self-esteem from age and gender is indicated in Table 5 .

Table 5 . Summary of regression models predicting self-esteem from age and gender.

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender significantly increased the coefficient of determination (Step 2; Table 5 ). The addition of the age cubed and its interaction with gender did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted. We conducted a consistent analysis separately by gender because the interaction between the age squared and gender was significant.

The age significantly predicted self-esteem (Step 1), and the addition of the age squared term significantly increased the coefficient of determination (Step 2; Table 6 ). The addition of the age cubed term did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted ( Figure 2 ). Self-esteem showed a slight downward trend from the teens to the mid-30s (the lowest predicted score was the age of 32 12 ), but then it continued to become higher to the 80s 13 .

Table 6 . Summary of regression models predicting self-esteem from age by gender (in the 2009, 2011, and 2013 surveys).

Figure 2 . Average and predicted self-esteem scores across ages in Japan (2009 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 2 ). Self-esteem continued to become higher from the teens to the 80s ( d = 0.97 14 ) 15 .

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. We conducted a consistent analysis separately by gender because the interaction between the age and gender was significant.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 3 ). Self-esteem continued to become higher from the 20s to the 50s ( d = 0.51).

Figure 3 . Average and predicted self-esteem scores across ages in Japan (2011 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 3 ). Self-esteem continued to become higher from the 20s to the 50s ( d = 1.05). The slope for females ( B = 0.03, p < 0.001) was larger than that for males ( B = 0.01, p < 0.05).

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. The interaction between the age and gender was not significant, showing that self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 4 ) 16 .

Figure 4 . Predicted self-esteem scores across ages in Japan (2012 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 5 ). Self-esteem continued to become higher from the teens to the 60s ( d = 1.00).

Figure 5 . Average and predicted self-esteem scores across ages in Japan (2013 survey). Note. Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared term did not significantly increase the coefficient of determination (Step 2; Table 6 ). Thus, the linear model (Step 1) was accepted ( Figure 5 ). Self-esteem continued to become higher from the teens to the 60s ( d = 1.42). The slope for females ( B = 0.02, p < 0.001) was larger than that for males ( B = 0.01, p < 0.001).

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender significantly increased the coefficient of determination (Step 2; Table 5 ). The addition of the age cubed and its interaction with gender did not significantly increase the coefficient of determination (Step 3). Thus, the quadratic model (Step 2) was accepted. The interaction between the age squared and gender was not significant. Self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 6 ) 17 .

Figure 6 . Predicted self-esteem scores across ages in Japan (2017 survey). Error bars represent 95% confidence intervals.

The model in Step 1 was significant, and the addition of the age squared and its interaction with gender did not significantly increase the coefficient of determination (Step 2; Table 5 ). Thus, the linear model (Step 1) was accepted. The interaction between the age and gender was not significant, showing that self-esteem continued to become higher from the 20s to the 50s both for males and females ( Figure 7 ) 18 .

Figure 7 . Predicted self-esteem scores across ages in Japan (2018 survey). Note. Error bars represent 95% confidence intervals.

Age Differences in Self-Competence and Self-Liking (The Two Components of Global Self-Esteem)

The results of the hierarchical multiple regression analyses are summarized in Table 7 . Except for females in 2009, the additions of their interaction terms did not significantly increase the coefficient of determination. Neither of the interaction terms significantly predicted the scores. These results consistently suggest that the developmental patterns for self-competence and self-liking were not different.

Table 7 . Hierarchical multiple linear regression predicting components of self-esteem from age, component type and their interactions.

For females in 2009, Step 2 significantly increased the coefficient of determination. The interaction term between age and the component significantly predicted the score. Thus, we conducted the same hierarchical multiple regression analyses on each component as we did for the global self-esteem ( Table 8 ). Although in both components age significantly explained the scores, the slope of the increase was higher for self-liking ( B = 0.02, p < 0.001) than for self-competence ( B = 0.01, p < 0.001; Figure 8 ). Yet, both patterns consistently indicated a continuous upward trend in self-esteem.

Table 8 . Regression models predicting each sub-component of self-esteem from age for females in the 2009 study.

Figure 8 . Average and predicted self-esteem scores by component across ages among females in Japan (2009 survey). Note. Error bars represent 95% confidence intervals.

Due to the consistent relationship between self-competence and self-liking, the age patterns for each component were same as those for global self-esteem. Specifically, the quadratic increases were found among males in 2009 and both genders in 2017, and the linear increases were found in the other subgroups.

Summary of the Results and Implications

We investigated age differences in self-esteem in Japan from adolescents aged 16 to the elderly aged 88 by using the RSES. Previous cross-sectional research has investigated age differences in self-esteem in Japan ( 12 , 13 , 17 ). However, it had two limitations: (1) it did not directly examine age differences in global self-esteem and (2) it did not investigate self-esteem among the elderly over the age of 70. These limitations had to be overcome to fully understand the developmental trajectory in self-esteem across the life span in Japan. Therefore, we examined age differences in self-esteem by conducting the RSES on more diverse sample that included the elderly over the age of 70.

First, as predicted, we found a pattern to the developmental trajectory of global self-esteem that is consistent with previous research on self-liking ( 12 , 13 , 18 ). We had predicted that the developmental pattern of self-esteem would be consistent with that of self-liking, but we had not had empirical evidence to support it. In this research, we empirically confirmed that self-competence and self-liking are closely associated with each other and have a consistent developmental pattern in Japan. Specifically, across the six cross-sectional surveys, the average level of self-esteem was low in adolescence, but then gradually continued to become higher from young adulthood to late adulthood. This trajectory was consistent with findings in previous research in European American cultures ( 2 , 3 ).

Second, as expected, analyses showed that the average level of self-esteem continued to indicate an upward trend beyond the age of 50 in Japan. All of the six independent cross-sectional datasets consistently showed that self-esteem continued to become higher from adulthood to old age both for males and females. This finding was inconsistent with previous research that showed a drop in self-esteem over the age of 50 in European American cultures ( 2 , 3 ). With old age comes a more humble, modest, and balanced perspective toward oneself, which leads to a decline in self-esteem in old age in European American cultures [e.g., Robins et al. ( 4 ); Robins and Tresniewski ( 3 )]. Previous research has indicated that, compared to people in European American cultures, people in Japan have more humble and balanced attitudes toward themselves, not just in old age [e.g., Heine et al. ( 8 ); Heine and Hamamura ( 17 )], which may account for the absence of a drop of self-esteem among the Japanese people over the age of 50. Thus, this research demonstrates that different developmental patterns can emerge in different social/cultural environments.

One may wonder whether the absence of a sharp drop in self-esteem in Japan is caused by the fact that participants answered the questionnaire on the internet. Elderly people who use the internet may differ from the elderly population in general (e.g., they may be wealthier and healthier). However, this was also the case in previous research that observed a clear drop in self-esteem among the elderly in the U.S. (e.g., ( 4 )). Thus, this explanation is insufficient to account for the cultural difference in the developmental trajectory of self-esteem among the elderly. Still, it is desirable to collect more representative data especially from the elderly and see if the result is consistent with the present research.

Three datasets showed that gender differences in the pattern of age differences were absent. The other three datasets indicated that there were slight differences between gender: slopes for females were a little larger than those for males. In sum, the pattern of age differences in self-esteem was similar between gender, if any small differences. These results were consistent with previous research ( 2 , 7 , 19 ).

We also confirmed the measurement invariance of the Rosenberg's Self-Esteem Scale ( 11 ) across gender and age-groups in Japan. Regarding gender, five out of the six datasets demonstrated scalar and structural invariance, and one showed partial scalar and structural invariance. Thus, in all of the datasets, at least partial scalar and structural invariance were supported. Regarding age-group, one out of the six datasets showed scalar and structural invariance, four showed partial scalar and structural invariance, and one showed metric and structural invariance. Overall, in the most datasets, at least partial scalar and structural invariance were supported. These results showed that the model structure and the adequacy of the measure were invariant across gender and age-groups.

Limitations and Future Directions

This research investigated age differences in self-esteem from adolescents in their teens to the elderly in their 80s by analyzing six cross-sectional datasets from a large and diverse sample. But, in cross-sectional data, age differences involve cohort differences. This research is a reasonable first step to understand the developmental trajectory of self-esteem across the life span in Japan. In fact, the absence of a drop in self-esteem in Japan might be caused by the cohort effect. Specifically, older cohorts might have higher self-esteem and younger cohorts might have lower self-esteem ( 23 – 26 ), which might obscure the drop of self-esteem in Japan. Thus, in the future, it is necessary to analyze longitudinal data which can distinguish between age differences and cohort differences.

Another limitation is that, although the sample sizes were relatively large for teens ( n = 211), adults ( n 20s = 1,161, n 30s = 970, n 40s = 1,382, n 50s = 1,056), and the elderly in their 60s ( n = 1,031) and 70s ( n = 253), the sample size for the elderly in their 80s was small ( n = 49). Thus, the results for this age group may be unreliable. It would be desirable to examine this point further by collecting more representative data.

Data Availability Statement

The datasets presented in this article are not readily available because participants were not asked to provide permission to disclose individual data at the time of data collection. Requests to access the minimal datasets (aggregate level) should be directed to Yuji Ogihara ( yogihara@rs.tus.ac.jp ).

Ethics Statement

This study was carried out in accordance with the recommendations of the Declaration of Helsinki and the Japanese Psychological Association with written informed consent from all subjects. Ethical approval was not required for this study in accordance with the local legislation and institutional requirements. This study asked participants to answer one of the most frequently used questionnaires (Rosenberg Self-Esteem Scale) and this questionnaire did not include any items that may harm participants. Thus, the protocol was not submitted to an ethics committee. Both individual researchers in charge of each survey and their respective research firms confirmed that all surveys were without any ethical concerns. All subjects gave written informed consent via the online questionnaire in accordance with the Declaration of Helsinki.

Author Contributions

YO contributed conception and design of the study, performed the statistical analysis, and wrote the first draft of the manuscript. TK collected the data and provided the critical comments on the manuscript. All authors contributed to manuscript revision and approved the final manuscript.

This work was supported by the Japanese Group Dynamics Association.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer ST declared a shared affiliation, with no collaboration, with the authors to the handling editor at the time of review.

Acknowledgments

We thank Pamela Taylor for her helpful comments on our previous versions of the manuscript.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2020.00132/full#supplementary-material

1. ^ Bleidorn et al. ( 10 ) examined cultural variation in age and gender differences in self-esteem across 48 nations including Japan. However, with regards to cultural differences in the developmental trajectory of self-esteem, it had at least three limitations. First, it investigated age differences in self-esteem between the ages of 16 and 45, leaving it unclear how average levels of self-esteem change after early middle age. Second, it indicated that self-esteem in females showed an upward trend from early adolescence to middle adulthood in Japan whereas self-esteem in males showed a downward trend during this period. Among 48 nations, this age pattern in males was found only in Japan, but the reason for this was not discussed. Third, as mentioned as a limitation by the authors, it used only one item “I see myself as someone who has high self-esteem” (a 5-point scale ranging from 1: disagree strongly to 5: agree strongly) instead of adequately constituted scale [e.g., Rosenberg Self-Esteem Scale; Rosenberg ( 11 )]. Although it was stated that the validity and reliability had been confirmed in the U.S. (although the item and its anchors differed from the original research), it is unclear whether the validity and reliability were confirmed in other nations including Japan. The word “self-esteem” is abstract and conceptual, so it may be interpreted differently across cultures and/or within each culture. This may have contributed to the exceptional pattern found in Japan (i.e., the continuous downward trend in males and the continuous upward trend in females).

2. ^ The translations differed between years is because these studies were conducted as an omnibus survey with other researchers.

3. ^ The other four items assessing self-competence were “I am able to do things as well as most other people.” (SC2), “I feel that I'm a person of worth, at least on an equal plane with others.” (SC3), “I feel I do not have much to be proud of.” (SC4), and “All in all, I am inclined to feel that I am a failure.” (SC5).

4. ^ The other four items assessing self-liking were “I take a positive attitude toward myself.” (SL2), “I certainly feel useless at times.” (SL3), “At times I think I am no good at all.” (SL4), and “I wish I could have more respect for myself.” (SL5).

5. ^ Although teenagers are usually not regarded as younger adults, because the size of this sample was relatively small, teenagers were conventionally included in the younger adults group in this analysis.

6. ^ Because the datasets in 2011 and 2012 did not include participants aged over 60 years, we split the participants into two age groups: younger adults (20s, 30s) and middle-aged adults (40s, 50s).

7. ^ Because the covariances between error terms of SC1, SC2, SC3, and error terms of SL1 and SL2 were large, we included them in the model (Model 1). These high associations may be due to the nature of the items: the items were affirmative, and the remaining (i.e., SC4, SC5, SL3, SL4, SL5) were reversed items. Even if we did not include these covariances in the model, the patterns of the results of multi-group CFA were consistent.

8. ^ We did not examine whether the developmental pattern of self-esteem (e.g., the shape of changes, the magnitude of the slope) differed among the six datasets. Naturally, the developmental pattern is different because the time periods covered in the surveys are significantly different ( Tables 1 , 2 ). Moreover, previous research has shown that the average levels of self-esteem decreased over time in Japan [( 23 – 25 ); for a review of historical changes in self-esteem in Japan see Ogihara ( 26 )]. Thus, the self-esteem score should not be simply averaged across the surveys that were conducted over nine years. Even if the self-esteem scores are z-transformed in each survey, the periods covered by the surveys are significantly different, and they should not be aggregated.

9. ^ In the partial scalar invariance model, item intercepts that were constrained did not include SL5 (i.e., excluding the constraint of SL5's item intercept from the full scalar model).

10. ^ In the partial scalar invariance models, item intercepts that were constrained were as follows. 2009: SC1, SC2, SC3, SL2; 2012: other than SL3; 2013: SC1, SC3, SL2, SL5; 2017: SC4, SL5; 2018: SC2, SL5.

11. ^ The reason for these differences in the levels of measurement invariance could be due to differences in factors such as total sample sizes, proportions of age-group sample size, and type of scale translations. However, we do not have sufficient data to empirically detect the reason(s). Thus, in the future, it would be important to investigate the possibility that people across a broad range of age might respond to RSES items differently.

12. ^ The lowest predicted score was at the age of 32 when raw data were unavailable. Thus, effect sizes were not calculated.

13. ^ We describe developmental patterns of self-esteem based on age differences in self-esteem. However, it should be noted that our data were cross-sectional and not cross-temporal. Age differences include not only developmental changes that an individual experiences over the life course, but also cohort differences. Thus, these age differences do not necessarily mean developmental changes. Although findings obtained from cross-sectional research on self-esteem are consistent with those obtained from cross-temporal research [for a review, see Orth and Robins ( 2 ), this should be kept in mind (also see, the Limitations and Future Directions section of the Discussion below).

14. ^ Effect sizes were calculated by using predicted average scores, standard deviations and sample sizes in three years (a given age, and 1 year before and after the age) to secure larger sample sizes and avoid unstable results.

15. ^ Comparing the slopes in linear models (Step 1) between gender, the slope for females ( B = 0.011, p < 0.001) was slightly larger than that for males ( B = 0.009, p < 0.001).

16. ^ We have also shown the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S1 ).

17. ^ We have also shown the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S2 ).

18. ^ We have also showed the results of the analysis conducted separately by gender to facilitate comparisons with other surveys ( Supplementary Material ; Table S1 ; Figure S3 ).

1. Baumeister RF, Campbell JD, Krueger JI, Vohs KD. Does high self-esteem cause better performance, interpersonal success, happiness, or healthier lifestyles? Psychol Sci Public Interest. (2003) 4:1–44. doi: 10.1111/1529-1006.01431

CrossRef Full Text | Google Scholar

2. Orth U, Robins RW. The development of self-esteem. Curr Dir Psychol Sci. (2014) 23:381–7. doi: 10.1177/0963721414547414

3. Robins RW, Trzesniewski KH. Self-esteem development across the lifespan. Curr Dir Psychol Sci. (2005) 14:158–62. doi: 10.1111/j.0963-7214.2005.00353.x

4. Robins RW, Trzesniewski KH, Tracy JL, Gosling SD, Potter J. Global self-esteem across the life span. Psychol Aging. (2002) 17:423–34. doi: 10.1037/0882-7974.17.3.423

PubMed Abstract | CrossRef Full Text | Google Scholar

5. Orth U, Trzesniewski KH, Robins RW. Self-esteem development from young adulthood to old age: a cohort-sequential longitudinal study. J Pers Soc Psychol. (2010) 98:645–58. doi: 10.1037/a0018769

6. Orth U, Robins RW, Widaman KF. Life-span development of self-esteem and its effects on important life outcomes. J Pers Soc Psychol. (2012) 102:1271–88. doi: 10.1037/a0025558

7. Orth U, Maes J, Schmitt M. Self-esteem development across the life span: a longitudinal study with a large sample from Germany. Dev Psychol. (2015) 51:248–59. doi: 10.1037/a0038481

8. Heine SJ, Lehman DR, Markus HR, Kitayama S. Is there a universal need for positive self-regard? Psychol Rev. (1999) 106:766–94. doi: 10.1037/0033-295X.106.4.766

9. Schmitt DP, Allik J. Simultaneous administration of the rosenberg self-esteem scale in 53 nations: exploring the universal and culture-specific features of global self-esteem. J Pers Soc Psychol. (2005) 89:623–42. doi: 10.1037/0022-3514.89.4.623

10. Bleidorn W, Arslan RC, Denissen JJ, Rentfrow PJ, Gebauer JE, Potter J, et al. Age and gender differences in self-esteem—A cross-cultural window. J Pers Soc Psychol. (2016) 111:396–410. doi: 10.1037/pspp0000078

11. Rosenberg M. Society and the Adolescent Self-Image . Princeton, NJ: Princeton University Press (1965).

Google Scholar

12. Ogihara Y. Age differences in self-liking in Japan: the developmental trajectory of self-esteem from elementary school to old age. Lett Evol Behav Sci. (2016) 7:33–6. doi: 10.5178/lebs.2016.48

13. Ogihara Y. The developmental trajectory of self-esteem in Japan: age differences in self-liking from middle school students to the elderly. JPN J Interpers Soc Psychol . (2018) 18:133–43. doi: 10.18910/70551

CrossRef Full Text

14. Tafarodi RW and Swann Jr WB. Self-linking and self-competence as dimensions of global self-esteem: initial validation of a measure. J Pers Assess. (1995) 65:322–42. doi: 10.1207/s15327752jpa6502_8

15. Tafarodi RW, Milne AB. Decomposing global self-esteem. J Pers. (2002) 70:443–84. doi: 10.1111/1467-6494.05017

16. Tafarodi RW, Swann WB. Two-dimensional self-esteem: theory and measurement. Pers Individ Dif. (2001) 31:653–73. doi: 10.1016/S0191-8869(00)00169-0

17. Heine SJ, Hamamura T. In search of East Asian self-enhancement. Pers Soc Psychol Rev. (2007) 11:4–27. doi: 10.1177/1088868306294587

18. Ogihara Y. A decline in self-esteem in adults over 50 is not found in Japan: age differences in self-esteem from young adulthood to old age. BMC Res Notes . (2019) 12:274. doi: 10.1186/s13104-019-4289-x

19. Ogihara Y. The pattern of age differences in self-esteem is similar between males and females in Japan: Gender differences in developmental trajectories of self-esteem from childhood to old age. Cogent Psychol. (2020) 7:1756147. doi: 10.1080/23311908.2020.1756147

20. Ogihara Y, Uchida Y, Kusumi T. How do Japanese perceive individualism? Examination of the meaning of individualism in Japan. Psychologia . (2014) 57:213–23. doi: 10.2117/psysoc.2014.213

21. Uchida Y, Kitayama S, Mesquita B, Reyes JAS, Morling B. Is perceived emotional support beneficial? Well-being and health in independent and interdependent cultures. Pers Soc Psychol Bull. (2008) 34:741–54. doi: 10.1177/0146167208315157

22. Yamamoto M, Matsui Y, Yamanari Y. The structure of perceived aspects of self. Jpn J Edu Psychol. (1982) 30:64–8. doi: 10.4992/jjpsy.53.106

23. Ogihara Y. The change in self-esteem among middle school students in Japan, 1989-2002. Psychology. (2016) 7:1343–51. doi: 10.4236/psych.2016.711136

24. Ogihara Y, Uchida Y, Kusumi T. Losing confidence over time: temporal changes in self-esteem among older children and early adolescents in Japan, 1999-2006. SAGE Open. (2016) 6:1–8. doi: 10.1177/2158244016666606

25. Oshio A, Okada R, Mogaki M, Namikawa T, Wakita T. Age and survey-year effects on self-esteem in Japan: a cross-temporal meta-analysis of scores on rosenberg's self-esteem scale. JPN J Edu Psychol. (2014) 62:273–82. doi: 10.5926/jjep.62.273

26. Ogihara Y. Temporal changes in individualism and their ramification in Japan: rising individualism and conflicts with persisting collectivism. Front Psychol. (2017) 8:695. doi: 10.3389/fpsyg.2017.00695

27. Cheung GW, Rensvold RB. Evaluating goodness of- fit indexes for testing measurement invariance. Struct Equ Model. (2002) 9:233–55. doi: 10.1207/S15328007SEM0902_5

Keywords: self-esteem, age difference, developmental trajectory, culture, self-competence, self-liking, Japan, Rosenberg self-esteem scale

Citation: Ogihara Y and Kusumi T (2020) The Developmental Trajectory of Self-Esteem Across the Life Span in Japan: Age Differences in Scores on the Rosenberg Self-Esteem Scale From Adolescence to Old Age. Front. Public Health 8:132. doi: 10.3389/fpubh.2020.00132

Received: 09 April 2019; Accepted: 01 April 2020; Published: 06 August 2020.

Reviewed by:

Copyright © 2020 Ogihara and Kusumi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Yuji Ogihara, yogihara@rs.tus.ac.jp

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 11 July 2022

Low self-esteem and the formation of global self-performance estimates in emerging adulthood

• Marion Rouault   ORCID: orcid.org/0000-0001-6586-3788 1 , 2   na1 ,
• Geert-Jan Will 3   na1 ,
• Stephen M. Fleming   ORCID: orcid.org/0000-0003-0233-4891 4 , 5 , 6 &
• Raymond J. Dolan   ORCID: orcid.org/0000-0001-9356-761X 4 , 5  

Translational Psychiatry volume  12 , Article number:  272 ( 2022 ) Cite this article

11k Accesses

6 Citations

4 Altmetric

Metrics details

• Human behaviour

High self-esteem, an overall positive evaluation of self-worth, is a cornerstone of mental health. Previously we showed that people with low self-esteem differentially construct beliefs about momentary self-worth derived from social feedback. However, it remains unknown whether these anomalies extend to constructing beliefs about self-performance in a non-social context, in the absence of external feedback. Here, we examined this question using a novel behavioral paradigm probing subjects’ self-performance estimates with or without external feedback. We analyzed data from young adults ( N  = 57) who were selected from a larger community sample ( N  = 2402) on the basis of occupying the bottom or top 10% of a reported self-esteem distribution. Participants performed a series of short blocks involving two perceptual decision-making tasks with varying degrees of difficulty, with or without feedback. At the end of each block, they had to decide on which task they thought they performed best, and gave subjective task ratings, providing two measures of self-performance estimates. We found no robust evidence of differences in objective performance between high and low self-esteem participants. Nevertheless, low self-esteem participants consistently underestimated their performance as expressed in lower subjective task ratings relative to high self-esteem participants. These results provide an initial window onto how cognitive processes underpinning the construction of self-performance estimates across different contexts map on to global dispositions relevant to mental health such as self-esteem.

Similar content being viewed by others

Ideological differences in the expanse of the moral circle

Microdosing with psilocybin mushrooms: a double-blind placebo-controlled study

Determinants of behaviour and their efficacy as targets of behavioural change interventions

Introduction.

A positive view of the self is a crucial determinant of mental health [ 1 , 2 ]. Low self-esteem has been associated with a number of psychiatric conditions, particularly those of an anxious and depressive nature [ 1 , 2 ]. People form beliefs about themselves and their abilities (“self-beliefs”) across many levels of abstraction, ranging from local confidence in individual decisions to estimates of performance on an entire task, up to global estimates about their own worth as expressed in reports of self-esteem. Having positive beliefs about self-worth (i.e., high self-esteem) is associated with a stronger ability to successfully deal with prospective situations, including how one deals with day-to-day challenges [ 3 ]. For instance, people with low self-esteem are often faster to disengage from a task in response to failure than those with high self-esteem [ 4 ]. Despite the recognized importance of self-beliefs for mental health, surprisingly little is known about the precise cognitive building blocks of self-beliefs, and their relationship with self-esteem [ 5 ].

Recent work examining the construction of momentary self-worth from social feedback [ 6 , 7 ] has started to uncover the formation of self-beliefs in a social context. Here low self-esteem participants were slower to update beliefs about how much others liked them, and faster to update momentary feelings of self-worth in response to social feedback. These findings provide initial evidence of a differential construction of self-beliefs being tied to a more global, stable construct such as self-esteem [ 6 , 7 ]. However, it remains unclear whether this is a specific idiosyncrasy of how low self-esteem individuals construct self-worth from social feedback or, alternatively, whether low self-esteem individuals have a domain-general bias when forming appropriate self-beliefs that extend to other non-social contexts. One possibility is that individuals with low self-esteem may maintain a negative self-view by consistently underestimating their abilities despite performing as well as those with high self-esteem, indicating a disconnection between a “local” self-evaluation on a given task and a “global” self-evaluation such as self-esteem.

Here we examined the formation of subjective self-performance estimates in participants with high and low self-esteem, in contexts with and without explicit feedback about performance. We leveraged a recently developed behavioral paradigm probing the formation of subjective self-performance estimates [ 8 , 9 ]. The main finding from this previous work is that decision confidence is a key factor contributing to the formation of self-performance estimates in the absence of feedback, a situation that echoes many real-life settings. We observed that decision difficulty, fluctuations in decision accuracy, and whether participants received feedback about their decisions all impacted their self-performance estimates. The present study employed this protocol to ask whether such subjective self-performance estimates, formed over the scale of minutes, relate to self-esteem. We previously proposed a hierarchical framework of metacognitive evaluation in which self-esteem may act as a global prior for generating self-performance estimates on a given task [ 10 ]. Specifically, under such a hierarchical framework of metacognitive evaluation – spanning decision confidence formed at a local level to self-esteem at a global level—we would expect self-esteem to provide a global context or prior for how self-performance estimates are formed on a given task [ 5 ]. We assume that self-esteem is a global estimate formed across longer timescales of months or years, whereas self-performance estimates are formed more rapidly, over the course of a few minutes of engaging in a task. Characterizing how these two constructs intersect is important to identify neurocognitive building blocks underpinning constructs relevant to mental health, such as self-esteem, and in turn facilitate novel interventions for disorders that are linked to altered self-esteem, a canonical example being depression [ 11 , 12 ].

To address these questions, we capitalized on a large dataset from a well-characterized community sample of adolescents and emerging adults ( N  = 2402; aged 14 to 24 at first measurement) who reported on their self-esteem across 1–3 timepoints spanning 4.5 years. We selected low and high self-esteem participants (aged 18–25) from the larger sample as individuals who scored within the bottom, or top, 10% of a self-esteem distribution so as to maximize power for detecting individual differences due to self-esteem [ 7 ]. A comparison between high and low self-esteem individuals was motivated by well-established findings that individuals with high self-esteem rates are among the healthiest in terms of low levels of depression and high levels of well-being in the population [ 13 ], providing a strong contrast against those with low self-esteem who experience substantial problems. Participants performed short blocks of two interleaved perceptual tasks and at the end of each block, they then selected the task on which they considered they had performed best and provided a subjective ability rating about each task. These two measures enable a window onto subjective self-performance estimates [ 8 ].

Consistent with previous findings, we found that participants underestimated their own performance in the absence of feedback, despite performing equivalently in situations with and without feedback. Participants with low self-esteem rated their performance lower compared to those with high self-esteem, despite task performance being similar in the two groups. We discuss the findings within a framework in which local metacognitive variables, such as decision confidence, influence the construction and maintenance of global self-esteem across longer times-scales.

Materials and methods

Participants.

We tested 62 human participants from the Neuroscience in Psychiatry Network (NSPN) cohort ( N  = 2402) who reported on their mental health, including measures of self-esteem, across 4.5 years for 1–3 measurements [ 7 ]. The NSPN 2400 Cohort is a general population sample of adolescents and emerging adults ( N  = 2402; aged 14–24 years at baseline) originally established to investigate a developmental change in mental health, cognition, and the brain (see ref. [ 14 ] for an in-depth cohort profile). Participants from Cambridgeshire and Greater London reported on sociodemographic characteristics and a range of mental health indices across multiple timepoints. A subsample ( N  = 785) nested within the larger cohort participated in detailed behavioral assessments of cognition using computerized tasks, clinical assessments, and IQ tests (see e.g., ref. [ 15 ]). A subset of this latter sample ( N  = 318) additionally underwent measures of brain structure and function using MRI (see e.g., ref. [ 16 ]).

For recruitment based on self-esteem, we used scores on the Rosenberg self-esteem scale (RSES) [ 17 ]. Mean RSES score of the large sample was 19.7 (on a scale of 0–30; SD = 5.62). We invited 184 participants with average RSES scores within the bottom decile (0–12) and top decile (27–30) of the large sample for further study and tested 53 participants (29 with low self-esteem; 24 with high self-esteem). To reach our target sample size of 30 participants in each group, we invited a further 51 participants whose recent RSES score was within the bottom or top decile of RSES scores and tested an additional ten participants. Five low self-esteem participants reported being in remission from a mental health problem for at least 3 years at the moment of testing. Participants were originally recruited for an fMRI study reported in (Will et al., 2020). The sample size was set to surpass the sample sizes of prior fMRI studies examining inter-individual differences in self-esteem (ten studies; median N  = 26; range = 17–48 [ 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ]). We further increased our power to detect individual differences by employing a targeted recruitment approach focusing on the extremes of a reported self-esteem distribution. After taking a break following MRI scanning, participants completed the self-performance estimate task reported here.

We matched groups based on gender and age, but not for subclinical symptoms of depression and anxiety [ 7 , 14 ]. As expected from the known associations between self-esteem and depression [ 2 ], we found strong correlations between the Rosenberg self-esteem score and the MFQ depression score (ρ(55) = −0.86, p  = 1.77 × 10 −17 ), and between the Rosenberg self-esteem score and the Trait Anxiety score (ρ(55) = −0.86, p  = 2.03 × 10 −17 ). When comparing the two groups to the large cohort ( N  = 2402), we observed that the low self-esteem group is at the 78.5 (±33.3) percentile in terms of depressive symptoms, while the high self-esteem group is at the 16.4 (± 6.1) percentile. In terms of well-being the low self-esteem group is at the 16.4 ± (0.04) percentile, while the high self-esteem group is at the 85 ± (0.04) percentile. These observations suggest that to characterize self-esteem-related problems, it is informative to contrast those who manifest such problems (i.e., low self-esteem participants) with those who have few such problems (i.e., high self-esteem participants).

Other inclusion criteria were applied: no current neurological or psychiatric disease, an address in London, no color-blindness and no contraindications to MRI (as the participants also underwent MRI scanning [ 7 ]). Five participants were excluded for responding at chance level (two participants, both high self-esteem), always selecting the same rating (one participant, low self-esteem), or failing the comprehension test of the rating scale during the practice (two participants, both high self-esteem), leaving N  = 57 participants for data analysis. The final sample consisted of 29 low self-esteem participants (mean age = 21.2, SD = 2.2; 18 women) and 28 high self-esteem participants (mean age = 21.1, SD = 2.3; 14 women). Participants were paid 8 GBP per hour for their participation and compensated for travel expenses. They provided written informed consent according to procedures approved by the London – Westminster NHS Research Ethics Committee (15/LO/1361).

Experimental design

Learning blocks.

Participants performed short learning blocks that randomly interleaved two “tasks” identified by two arbitrary color cues (Fig. 1a ). Participants were incentivized to learn about their own performance on each of the two tasks over the course of a block. Each block contained 2, 4, 6, 8, or 10 trials per task (which we refer to as “ learning duration ”), giving 30 blocks (=360 trials) per participant, presented in a pseudo-random order. We varied the learning duration to examine whether and how the number of decisions made by participants within each block impacted the construction of self-performance estimates.

a Participants performed short learning blocks of randomly alternating trials from two tasks (between 2 and 10 trials per task). At the end of each block, participants were asked to select the task on which they thought they had performed best (Task choice), as well as rate their overall ability at each task (Task rating). A new block ensued with two new color cues indicating two new tasks. b Each task required perceptual choices as to which of two boxes contained more dots. Trials were either easy or difficult according to the numerical dot difference between the left and right boxes. Following their response participants either received veridical feedback (correct, incorrect) about their perceptual judgment, or no-feedback. These four conditions resulted in six possible task pairings as displayed in a. c Each trial consisted in a perceptual judgment as to which of two boxes contained a higher number of dots, followed or not by provision of feedback.

Each task required a perceptual judgment as to which of the two boxes contained more dots (Fig. 1a ). The difficulty level of the judgment was controlled by the difference in dot number between boxes. Any given task (as indicated by the color cue) was either easy or difficult and provided either veridical feedback or no-feedback (Fig. 1b ). Importantly the color cues allowed participants to identify the two tasks but provided no information about task difficulty. These four task features provided six possible pairings of tasks in learning blocks. The order of blocks was randomized for each participant.

Two measures of self-performance estimates

Task choice. At the end of each block, participants were asked to choose the task for which they thought they performed best (Fig. 1a ). Specifically, they were asked to report which task they would like to perform in a short subsequent “test block” in order to gain a bonus. This procedure aimed to reveal self-performance estimates, because participants should choose the task they expect to be more successful at in the test block in order to gain maximum reward. To indicate their task choice, participants responded with two response keys that differed from those assigned to perceptual decisions to avoid any carry-over effects. The subsequent test block contained six trials from the chosen task (not shown in Fig. 1 ). No-feedback was provided during test blocks.

Task ratings. After the test block, participants were asked to rate their overall performance on each of the two tasks on a rating scale ranging from 50% (“chance level”) to 100% (“perfect”) to obtain explicit, parametric reports of self-performance estimates (Fig. 1a ). Ratings were made with the mouse cursor and could be given anywhere on the continuous scale. Intermediate ticks for percentages 60, 70, 80, and 90% correct were indicated on the scale but without verbal labels. There was no time limit on perceptual choices, task choices, and task ratings. After each block, participants were offered a break and could resume at any time, with the next learning block featuring two new tasks cued by two new colors. The present design is a modified version of the protocol from our original paper [ 8 ].

Trial structure . Each block featured two tasks, with each trial starting with a color cue presented for 1200 ms, indicating which of the two tasks will be performed in the current trial (Fig. 1 ). The stimuli were black boxes filled with white dots randomly positioned and presented for 300 ms, during which time participants were unable to respond. We used two difficulty levels characterized by a constant dot difference, but the spatial configuration of the dots inside a box varied randomly from trial-to-trial. One box was always half-filled (313 dots out of 625 positions), whereas the other contained 313 + 24 dots (difficult conditions) or 313 + 60 dots (easy conditions). The position of the box with the most dots was randomized across trials (half of the trials on the left, half of the trials on the right). Participants were asked to judge which box (left or right) contained more dots and the chosen box was then highlighted for 300 ms. Afterward, a colored rectangle (cueing the color of the current task) was presented for 1500 ms. The rectangle was either empty (on no-feedback trials) or contained the word “Correct” or “Incorrect” (on feedback trials), followed by an ITI of 600 ms.

Statistical analyses

To examine the influence of our experimental factors on self-performance estimates, we carried out three 2 × 2 × 2 repeated-measures ANOVAs on (1) objective performance (Table S1 ), (2) task choice (Table S2 ), and (3) task ratings (Table S3 ). Our factors were Feedback (present vs. absent), Difficulty (easy vs. difficult) as within-subject factors, and self-esteem level (high vs. low) as a between-subject factor. Because task choice frequencies are proportions, they were transformed using a classic arcsine square-root transformation before being entered into the ANOVA. Note that we reproduced these analyses based on past self-esteem status (as per recruitment) instead of current self-esteem status (on the testing session) and found virtually identical results (Tables S4 – S6 ).

Since objective performance naturally fluctuates even for a fixed difficulty level due to noise, we examined whether participants had some insight into these fluctuations. We additionally examined whether participants’ self-performance estimates reflected fluctuations in objective performance on a given learning block over and above variations in difficulty level. For each of the six pairings, we analyzed task choice and task ratings as a function of the absolute difference in performance between tasks for each participant (as in [ 8 ]) (Fig. 3 ). To quantify these effects, we performed a logistic (resp. linear) regression to further quantify the influence of fluctuations in objective performance on task choice (resp. task ratings), entering objective performance as block-wise regressors. We further introduced individual self-esteem (Rosenberg score) and its interaction with the difference in performance as additional regressors to examine if these could explain additional variance in task choice or task ratings. Regressors were z-scored to ensure comparability of regression coefficients. Each model was specified as Task Choice ~ β 0  +  β 1  × Difference in Performance +  β 2  × Self-esteem +  β 3  × Difference in Performance × Self-esteem, and participants were treated as a fixed effect in the regressions (due to few blocks per pairing per participant).

Finally, to visualize whether there were any effects of learning duration (the number of trials per task in each block) on self-performance estimates, task choice frequencies were averaged across participants for each of the six possible pairings and the five possible learning durations (Fig. 4 ). To investigate whether learning duration had a significant influence on task choice, separate logistic regressions were performed on each of the six task pairings. Each model was specified as Task Choice ~ β 0  +  β 1  × Learning Duration +  β 2  × Self-esteem +  β 3  × Learning Duration × Self-esteem (we continue to model the main effects of self-esteem on each individual task pairing but do not further test for the significance of these terms, as this effect is evaluated in the more powerful ANOVA approach above that collapses over task pairings). Similarly, we examined whether learning duration influenced task ratings with similar models as for task choices, but with linear regression models instead of logistic regressions, because the dependent variable was continuous rather than dichotomous. Our dependent variable was the difference in task ratings between the two tasks of a block. The use of a fixed-effects approach naturally limits the extent to which our findings can be generalized to the population level.

An experimental protocol probing the formation of self-performance estimates

To investigate the impact of self-esteem on self-performance estimates, participants ( N  = 57) engaged in 30 short learning blocks (4 to 20 trials) of two randomly interleaved visual discrimination tasks signaled by two arbitrary color cues (Fig. 1c ). We varied learning duration (the number of trials per task in each block) to examine whether participants differentially formed self-performance estimates depending on how much experience they had with each task. Each task required a perceptual discrimination judgment as to which of the two boxes contained a higher number of dots (Fig. 1c ). Two factors controlled task characteristics: task difficulty (either easy or difficult according to dot difference between boxes), and receipt of either veridical feedback (correct, incorrect) or no-feedback about performance on each perceptual choice (Fig. 1b ). This factorial design resulted in six possible task pairings for learning blocks (Fig. 1a ). For example, an Easy-Feedback task could be paired with a Difficult-Feedback task, or a Difficult-Feedback task could be paired with a Difficult-No-Feedback task, and so forth. At the end of each block, participants selected the task on which they believed they performed better (Task choice) and were rewarded on the basis of their performance on the chosen task (see Methods). They additionally provided a subjective rating of self-performance on each of the two tasks on a continuous scale (Task ratings) (Fig. 1a ). A short break ensued before the next learning block started when two new color cues indicated two new tasks. The two end-of-block measures, namely task choices and task ratings, provided proxies for self-performance estimates. In this way the design allowed us to compare self-performance estimates in participants with high or low self-esteem.

Self-esteem and self-performance estimates

We first examined whether high and low self-esteem participants differed in terms of objective performance on the perceptual tasks, conditional on the provision of feedback or not, and on task difficulty. We performed a 2 × 2 × 2 repeated-measures ANOVA on objective performance with two within-subject factors (Feedback and Difficulty), and with the self-esteem group as a between-subject factor (see Methods). First, we replicated our previous findings showing that participants performed better when tasks were easier (the main effect of Difficulty, F (1, 56) = 472.7, p  = 1.1 × 10 −28 ), but without a difference in performance in the presence or absence of feedback (Fig. S1 ) (main effect of Feedback, F (1, 56) = 0.622, p  = 0.434). High ( N  = 28) and low ( N  = 29) self-esteem participants did not differ in performance (main effect of Self-Esteem, F (1, 56) = 1.675, p  = 0.201). We found no significant interactions, except for interaction between Difficulty and Self-Esteem ( F (1,56) = 5.174, p  = 0.027), driven by slightly worse performance on easy tasks in the low self-esteem group (Table S1 ). Pairwise comparisons between each of the four experimental conditions showed no significant difference in performance in the easy conditions ( t 55  = 1.82, p  = 0.07 for feedback trials and t 55  = 1.55, p  = 0.12 for no-feedback trials). There was also no statistically significant difference between groups in the difficult condition with feedback ( t 55  = 0.75, p  = 0.45) nor in the difficult condition without feedback ( t 55  = −0.04, p  = 0.96). Together with a lack of the main effect of self-esteem on performance, these results suggest that any difference in self-performance estimates between self-esteem groups is likely to arise at a metacognitive level, rather than being driven by systematic differences in objective performance between groups across all experimental conditions of the design.

Next, we examined the construction of self-performance estimates in our perceptual tasks. We again applied the same 2 × 2 × 2 repeated-measures ANOVA, this time to task choices and task ratings. For our first measure of self-performance estimates, task choice, we replicated our prior work showing participants selected easy tasks as compared to difficult tasks more often at the end of blocks (main effect of Difficulty F (1, 56) = 108.8, p  = 1.2 × 10 −14 ). Participants were also more likely to select tasks that provided feedback, compared to those that did not (main effect of Feedback F (1, 56) = 93.8, p  = 1.7 × 10 −13 ). There was a trend-level interaction between Difficulty and Feedback ( F (1, 56) = 3.81, p  = 0.056), in accordance with previous findings showing an interaction in a subset of previous datasets (Fig. S1 ) [ 8 ]. This we assume reflects variability in how sensitive participants are to difficulty relative to feedback receipt. We found no main effect of Self-Esteem on task choice ( F (1, 56) = 0.295, p  = 0.59) and no significant interactions between Self-Esteem and other experimental factors (all p  > 0.33), meaning that task choices were most likely driven by experimentally manipulated factors as opposed to (task-unrelated) self-esteem. We also note that task choices can be insensitive to overall shifts in self-performance estimates across both tasks, which can cancel out when participants have to choose between pairs of tasks. This might be the case in low compared to high self-esteem individuals, for instance. To test for such effects, we next turned to our second measure of self-performance estimates, task ratings.

Finally, we analyzed our second measure of self-performance estimates: subjective task ability ratings. Consistent with previous findings [ 8 ], we found a main effect of Difficulty ( F (1, 56) = 211.7, p  = 1.7 × 10 −20 ) and of Feedback ( F (1, 56) = 139.9, p  = 9.7 × 10 −17 ) on task ratings, together with a significant interaction between these factors ( F (1, 56) = 35.6, p  = 1.8 × 10 −7 ). These results indicate that participants rated their self-performance lower in the absence of feedback, an effect exacerbated for easy as compared to difficult tasks (Fig. 2 and S1 ). Crucially we observed a main effect of Self-Esteem on task ratings ( F (1, 56) = 5.92, p  = 0.018), reflecting the fact that participants with low self-esteem reported lower self-performance estimates for both difficult and easy tasks as well as tasks with and without feedback, despite their objective task performance being equivalent to participants with high self-esteem (Table S3 ).

a A 2 × 2 × 2 repeated-measures ANOVA revealed a main effect of self-esteem status on task ratings, indicating lower self-performance estimates in participants with low self-esteem. Difficulty (Easy, Diff) and Feedback (Feedback, No–Feedback) were within-subject factors and self-esteem was a between-subject factor (see Methods and Results). Circles and error bars represent mean and SEM across participants ( N  = 28 with high self-esteem and N  = 29 with low self-esteem), and dots indicate individual data points. b Average objective performance across all conditions for high and low self-esteem groups separately. Bars and error bars indicate mean and SEM across participants and dots indicate individual data points. n.s. not significant (two-sample t -tes t , t 55  = 1.29, p  = 0.20).

Study participants were recruited on the basis of their self-esteem score at the time of inclusion in the database (“past” self-esteem level). We also assessed their self-esteem level at the time they performed the perceptual learning tasks (“current” self-esteem level) in order to create the groupings for the analyses reported above. Past self-esteem scores at the time of recruitment correlated with current self-esteem scores at the time of testing (ρ(55) = 0.72, p  = 2.6 × 10 −12 ). Nevertheless, to examine the robustness of our findings, we reproduced all our analyses but now based on past self-esteem groupings instead of current self-esteem groupings. Critically we found virtually identical results (Tables S4 – S6 ), indicating that our behavioral task interacts with self-esteem status in a stable manner. In particular, participants with lower self-esteem at the time of recruitment continued to provide lower subjective task ratings at the time of testing, despite objective performance being unaffected (Table S6 ).

Characterization of the influence of task factors on self-performance estimates

Having shown that self-esteem is linked to overall self-performance estimates in our task, we next characterized how participants’ self-performance estimates are influenced by block-to-block fluctuations in learning duration and performance and asked how the influence of these factors may interact with self-esteem. Building on our previous study [ 8 ], our experimental design with variable block lengths allowed us to characterize how experimental factors explain variation in subjective task ratings and examine if other previous findings replicate. In a first analysis, we reasoned that, even for a fixed difficulty level, there would be fluctuations in objective performance from block to block due to variability inherent to perceptual decision-making. To investigate whether participants were sensitive to such fluctuations when they provided self-performance estimates, we performed regression analyses predicting task choices and task ratings from the difference in objective performance between tasks (Fig. 3 ; see Methods). In a second analysis, we examined the influence of learning duration on the expression of self-performance estimates (Fig. 4 ; see Methods). In both these sets of analyses, we included self-esteem as an additional between-participant predictor and asked how it interacted with (i) the difference in objective performance between tasks and (ii) learning duration.

Self-performance estimates were measured as a task choice and b task ability ratings, each visualized here as a function of the absolute difference in performance between tasks, for a small, average, and large absolute difference in performance between tasks. Green (resp. orange) indicates easy (resp. difficult) tasks. Dotted lines (resp. full lines) indicate tasks without feedback (resp. with feedback). Error bars indicate SEM across participants ( N  = 57). Dots indicate individual data points; note that for task choices, task choice frequency took discrete values due to a limited number of data points per participant (see Methods). Significant effects of the difference in performance between tasks were found for end-of-block task choices (*** p  < .000001), except for when an easy-no-feedback task was paired with a difficult-no-feedback task (n.s.). For task ratings, there was a significant effect of the difference in performance between tasks in all blocks (*** p  < 0.0023).

Self-performance estimates measured as end-of-block task choices ( a ) or task ratings ( b ) as a function of learning duration (number of trials per task in each block) for the six possible task pairings. Error bars indicate SEM across participants ( N  = 57). Significant effects of learning duration on end-of-block self-performance estimates are indicated (* p  < 0.05, ** p  < 0.01) (see Methods).

First, in the total sample ( N  = 57), we found a significant effect of a difference in performance between tasks on end-of-block task choices (all task pairings β  = 1.09, all p  = 1.56 × 10 −7 ), except for when an Easy-No-Feedback task was paired with a Difficult-No-Feedback task ( β  = 0.083, p  = 0.49) (Fig. 3a ). These differences in performance did not interact with self-esteem, demonstrating that performance fluctuations continue to influence self-performance estimates irrespective of self-esteem level (interaction between self-esteem and difference in performance; all task pairings β  < 0.25, all p  > 0.21).

Using a similar approach, we uncovered a significant effect of differences in performance between tasks on end-of-block task ratings (Fig. 3b ) (all task pairings β  > 0.016, p  < 0.0023), meaning that the larger the difference in objective performance between tasks, the larger the difference in task ratings (irrespective of self-esteem). When examining interactions with self-esteem, we found that effects of fluctuations in performance did not differ as a function of self-esteem level for the majority (five out of six) of task pairings (interaction between self-esteem and difference in performance; all β   <  −0.0067, p  > 0.38). An exception was when an Easy-No-Feedback task was paired with a Difficult-Feedback task, for which the interaction between self-esteem and performance difference was significant ( β  = 0.016, p  = 0.022), without an effect of self-esteem itself ( β  = −0.011, p  = 0.088). This interaction indicates that participants with high self-esteem showed a greater influence of performance difference for this task pairing. Taken together, these findings indicate that participants’ end-of-block self-performance estimates were sensitive to fluctuations in objective difficulty, the presence of feedback, and fluctuations in task performance, with limited effects of self-esteem on these relationships.

Second, we examined the impact of learning duration (the number of decisions per task in each block) on self-performance estimates. Consistent with previous findings [ 8 ], regression analyses confirmed no significant effect of learning duration (number of trials per block) on end-of-block task choices for four out of six of the task pairings (all β  < 0.21, all p  > 0.099) (Fig. 4a ). An exception was when a Difficult-Feedback task was paired with a Difficult-No-Feedback task, with learning duration leading task choices to become less sensitive over time ( β  = −1.49, p  = 0.013). Similarly, when an Easy-No-Feedback task was paired with a Difficult-No-Feedback task, we found a significant effect of learning duration ( β  = −1.6, p  = 0.009) which interacted with self-esteem ( β  = −1.04, p  = 0.026) on task choices. For all other five out of six task pairings, there were no significant interactions with self-esteem (all β  < 0.12, all p  > 0.34).

Finally, we found no effect of learning duration on end-of-block task ratings for five out of six of the task pairings (Fig. 4b ) (all β  < 0.012, all p  > 0.16), with the exception of when an Easy-Feedback task was paired with an Easy-No-Feedback task ( β  = −0.018, p  = 0.033). We also found no interactions between self-esteem and learning duration (abs( β ) < 0.012, all p  > 0.15) on task ratings. Together these findings indicate that participants’ self-performance estimates were mostly insensitive to task duration, suggesting participants rapidly form an estimate of their expectations of success at the beginning of each block of trials and that the manner in which they do so is relatively insensitive to self-esteem level.

Humans construct beliefs about themselves and their abilities across many levels of abstraction, encompassing not only global constructs such as self-esteem but also self-performance estimates on a given task [ 5 ]. Having a favorable appraisal of oneself is a key component of mental well-being [ 1 , 2 ]. We previously proposed a hierarchical framework in which self-esteem acts as a global prior to self-performance estimates for a given task [ 10 ]. Here, we sought behavioral evidence that bears on this framework by relating self-esteem, a global construct, to subjective self-performance estimates created during tasks performed over a shorter temporal duration. To do this we leveraged a perceptual task for which we previously characterized how participants provide self-performance estimates [ 8 ].

We replicated our previous findings showing participants’ self-performance estimates are sensitive to task difficulty, feedback, and fluctuations in objective task performance. We further showed that participants with low self-esteem provide lower subjective task ratings than those with high self-esteem, in the absence of a main effect of self-esteem on objective performance. We compared a low self-esteem group with substantial problems to a healthy group of high self-esteem participants. Low self-esteem subjects’ propensity to consistently rate their performance as worse relative to those with high self-esteem—despite not performing any worse on objective measures—represents a candidate correlates of poor mental health. This disconnect between objective performance and its subjective evaluation may therefore be relevant for a better understanding of psychiatric disorders characterized by distortions in self-evaluation, as we further discuss below.

An important feature of our results is the absence of systematically lower performance in participants with low self-esteem across all experimental conditions of the design. This indicates a selective and consistent link between self-esteem and biases in confidence, uncontaminated by differences in performance. However, we did find a small, but significant, interaction between Difficulty and Self-Esteem in first-order task performance. One possible explanation for this effect is that lower expectations of self-performance may lead participants to engage less effort in the task, and thus display worse perceptual performance. In turn, such a cycle could become reinforcing, with lower perceptual performance further decreasing subjective task ratings. However, given the absence of systematic differences in performance between each of the four conditions of the design, we consider this alternative hypothesis less likely in light of our entire set of findings. More generally, this decoupling indicates that differences in global self-performance estimate stemmed from a metacognitive bias as opposed to a rational updating of confidence as a function of objectively lowered performance. This is a key insight as while previous reports have indicated that low self-esteem individuals also underestimate their performance on familiar tasks, it has remained unclear whether this is a consequence of negative self-beliefs, or due to negative experiences with the task at hand (for a review, see ref. [ 28 ]).

In the present study, a lack of a clear influence of low self-esteem on performance may reflect participants’ having limited experience with the perceptual task. We leveraged the fact that presumably, nobody had encountered the current perceptual task before in order to minimize prior beliefs about expected performance, thereby allowing us to isolate a ‘pure’ effect of self-esteem. Instead, had it been a memory task, for instance, participants might have retrieved and relied on general prior beliefs about their memory abilities [ 29 ]. The type of perceptual task we exploit is also likely to preclude influences seen in other cognitive domains, such as mathematics anxiety [ 30 ] or pervasive social effects such as stereotype threat (a perceived risk of confirming negative stereotypes about abilities associated with one’s social group) that are thought to influence subsequent performance [ 31 , 32 ]. Therefore, it is possible that relationships between self-esteem and task self-performance estimates may become even tighter in real-life metacognitive evaluations.

A key advantage of the current metacognitive task is that this difference can be interpreted through the lens of differential contributions to self-performance estimate formation. Although we cannot draw strong conclusions from non-significant findings, the lack of systematic statistical interactions between self-esteem and experimental factors (feedback presence, difficulty level) on self-performance estimates indicates participants with low self-esteem were not impaired in building self-performance estimates from task-specific factors. Specifically, they were also able to update self-performance estimates in the absence of feedback, indicating that they preserve an ability to track fluctuations in local confidence. Instead, participants with low self-esteem displayed a general underestimation of their performance as seen in subjective task ratings, independently of feedback condition and difficulty level. This result is consistent with a recent study showing that overall confidence in a perceptual task was associated with self-esteem score, in the absence of a relationship between self-esteem and metacognitive sensitivity (i.e., how well confidence tracks performance in the absence of feedback) [ 33 ]. Other studies have reported that self-esteem affects sensitivity to feedback, suggesting that high self-esteem may act as a ‘buffer’ against negative feedback [ 34 ]. Low self-esteem participants were found to provide self-worth ratings that are more sensitive to social evaluative feedback [ 7 ] or achievement feedback [ 34 ], although the type of feedback and task scope were substantially different from those of the current study. More generally, our results show that low and high self-esteem individuals continue to form global confidence estimates in a similar manner despite continuing to differ in their overall evaluation. This result is non-trivial and helps to delineate the source of confidence biases in self-esteem (as a generalized bias that appears to go beyond the influence of local task factors).

Many clinical and subclinical psychiatric symptoms are associated with alterations to various aspects of metacognition [ 35 , 36 ]. Low self-esteem is a robust predictor of concurrent and future mental health disorders, particularly those associated with negative cognitions and affect as expressed in anxious and depressive symptoms [ 1 ]. Importantly, the participants in our sample did not have a formal mental health diagnosis, providing some evidence that the observed associations between self-esteem and lower subjective performance ratings are likely to be explained by low self-esteem alone, rather than factors associated with patient status, such as stigma, the impact of therapy or medication. Notably, a previous study reported that self-esteem predicted overall confidence on a perceptual task in an online general population sample, even after controlling for depressive symptoms [ 33 ]. As is typically the case, here and in our previous study [ 7 ], self-esteem groups differed on trait anxiety, state anxiety, depression, and social anxiety, reflecting existing associations between low self-esteem and these symptoms. Indeed, in the DSM-V, a lowered sense of self-worth is one of the core diagnostic criteria for major depressive disorder. Likewise, self-esteem and self-efficacy are typically strongly decreased in depression and anxiety disorders. The ecological validity of our sample, therefore, does not allow us to distinguish a specific impact of self-esteem from unique shifts in co-morbid anxiety or depression levels, which are known to be tightly linked in longitudinal studies [ 2 ]. In another study using a dimensional approach, we identified lower levels of trial-by-trial decision confidence in subclinical participants who displayed higher scores on an “anxious-depression” transdiagnostic dimension [ 37 ]. It remains to be explored whether this alteration in decision confidence might generalize to more global aspects of metacognition, such as the self-performance estimates measured here [ 38 ]. However, to the extent that self-esteem is related to negative affective symptoms, the present results showing a link between low self-esteem and lower subjective task ratings provides initial evidence this may indeed be the case. Furthermore, in previous work, we have shown that global SPEs in a similar task are sensitive to trial-to-trial fluctuations in decision confidence [ 8 , 9 ]—suggesting factors that influence baseline decision confidence are also likely to influence global metacognition.

Similarly, previous work has provided evidence that other aspects of metacognition are shifted in the context of negative affective symptoms. A previous study of social anxiety documented a lack of a positivity bias—a tendency to overweight positive as compared to negative feedback—when processing feedback from a social task involving giving a speech in front of judges [ 39 ]. To the extent that social anxiety and low self-esteem are linked, these results suggest a similar lack of a positivity bias in learning may be linked to low self-esteem. Finally, in a face discrimination task participants with high anxiety manifest different feedback-related negativity correlates in EEG recordings following evaluative feedback, as compared to participants with low anxiety [ 40 ]. This indicates that anxiety might disrupt an evaluative component of performance monitoring, which we expect would extend to low self-esteem to the extent that anxiety and lowered self-esteem overlap. Another previous study provided empirical evidence that participants with depression differed in their cognitive reappraisal of positive information, suggested to be underpinned by a reduced integration of positive prediction errors [ 41 ]. In light of unexpected positive feedback about their own performance on a test, healthy participants positively updated their beliefs, whereas participants with depression did not change these task expectations [ 42 ]. To the extent that low self-esteem and depression overlap, a similar mechanism could partly explain our findings: participants with low self-esteem may not update their self-performance estimates following positive feedback as much as those with high self-esteem. However, we note that the impact of task factors on the formation of global estimates did not differ between self-esteem groups, so this hypothesis remains to be tested.

Replicating ours and others’ previous studies we found that task choices were sensitive to fluctuations in performance [ 8 , 9 , 43 ], an effect that remained when controlling for self-esteem level. We note that, unlike task ratings, the need to make binary task choices forces participants to separate between higher and lower self-performance estimates, even if self-performance on both tasks are close at the end of a block. This implies that any baseline shift in self-performance estimates that is common to both tasks may not manifest in task choices—possibly explaining why only ratings, but not choices, were sensitive to self-esteem level. We also replicate our previous result that learning duration did not systematically affect task choices [ 8 ] and extend this finding to the case of task ratings (Fig. 4 ). Although here we did not measure participants’ precision or confidence in their subjective task ratings, it is possible that uncertainty around expected performance decreases with learning duration, as participants have more samples to inform their self-performance estimates. Our analyses compared participants with high and low self-esteem levels. It would be interesting to examine whether there are any non-linear relationships between task factors, self-performance estimates, and self-esteem in a sample consisting of low, average, and high self-esteem participants.

While here we investigated the formation of global self-performance estimates over the course of short learning blocks, future work is needed, particularly using longitudinal measurements, to examine how global self-performance estimates develop over longer timescales [ 2 ] and impact subsequent metacognitive judgments [ 44 ]. This can provide a window onto the formation and maintenance of global dispositions that evolve across weeks or months, such as self-esteem itself [ 5 ]. In the present study, participants spanned a limited age range and it remains unknown how the formation of experimental self-performance estimates mirrors the maintenance and update of self-esteem across the lifespan. Some of these effects may be specific to adolescence, as previous work has shown that perceptual metacognitive sensitivity continues to mature in the 11-17 years old range [ 45 ]. In our sample, only a few people (<10%) shifted their reported self-esteem sufficiently to move between the self-esteem groups over the course of a couple of years. Nevertheless, it remains to be established how malleable such self-constructs are, though the lack of interactions between self-esteem and feedback in our experiment suggests a certain degree of stability or rigidity. Under a hierarchical framework, it is plausible that higher, more global, levels are more temporally stable whereas lower levels such as local decision confidence or self-performance estimates on individual tasks may be more malleable [ 5 , 46 ].

Metacognition operates across many levels of abstraction, from local confidence in individual decisions to self-performance estimates on a particular task, to global self-evaluations such as self-esteem. However, the relationships among these levels remain to be characterized. Our approach was to recruit participants from a community sample and use a task for which participants had no prior experience, academic stakes, or relevance, enabling us to isolate an effect of self-esteem on self-performance estimates that was distinct from other factors typically present in patient studies. Our study, therefore, connects two of these levels of metacognition in a simple lab-based task, disconnected from real-life evaluations, and finds that low self-esteem is associated with lower subjective performance estimates.

Data availability

Participants’ group-level data for statistical analyses are available at https://www.github.com/marionrouault/RouaultWillFlemingDolan/. Participants did not provide written consent regarding the posting of their anonymized data on public repositories; however, the raw datasets are available from the corresponding author upon reasonable request.

Code availability

MATLAB code for statistical analyses are available at https://www.github.com/marionrouault/RouaultWillFlemingDolan/ .

Orth U, Robins RW, Roberts BW. Low self-esteem prospectively predicts depression in adolescence and young adulthood. J Personal Soc Psychol. 2008;95:695–708.

Article   Google Scholar  

Sowislo JF, Orth U. Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies. Psychological Bull. 2013;139:213.

Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychological Rev. 1977;84:191.

Article   CAS   Google Scholar  

Di Paula A, Campbell JD. Self-esteem and persistence in the face of failure. J Personal Soc Psychol. 2002;83:711.

Seow TX, Rouault M, Gillan CM, Fleming SM. How local and global metacognition shape mental health. Biol Psychiatry. 2021;90:436–46.

Will G-J, Rutledge RB, Moutoussis M, Dolan RJ. Neural and computational processes underlying dynamic changes in self-esteem. eLife. 2017;6:e28098.

Article   PubMed   PubMed Central   Google Scholar  

Will G-J, Moutoussis M, Womack PM, Bullmore ET, Goodyer IM, Fonagy P, et al. Neurocomputational mechanisms underpinning aberrant social learning in young adults with low self-esteem. Transl Psychiatry. 2020;10:1–14.

Rouault M, Dayan P, Fleming SM. Forming global estimates of self-performance from local confidence. Nat Commun. 2019;10:1141.

Rouault M, Fleming SM. Formation of global self-beliefs in the human brain. Proc Natl Acad Sci USA. 2020;117:27268–76.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Rouault M, McWilliams A, Allen MG, Fleming SM. Human metacognition across domains: insights from individual differences and neuroimaging. Personal Neurosci. 2018;1:e17–13.

Stephan KE, Manjaly ZM, Mathys CD, Weber LA, Paliwal S, Gard T, et al. Allostatic self-efficacy: a metacognitive theory of dyshomeostasis-induced fatigue and depression. Front Hum Neurosci. 2016;10:550.

Vinckier F, Rigoux L, Oudiette D, Pessiglione M. Neuro-computational account of how mood fluctuations arise and affect decision making. Nat. Commun . 2018;9:1–12.

Cheng H, Furnham A. Perceived parental rearing style, self-esteem and self-criticism as predictors of happiness. J Happiness Stud. 2004;5:1–21.

Kiddle B, Inkster B, Prabhu G, Moutoussis M, Whitaker KJ, Bullmore ET, et al. Cohort profile: the NSPN 2400 cohort: a developmental sample supporting the Wellcome Trust NeuroScience in Psychiatry Network. Int J Epidemiol. 2018;47:18–19g.

Article   PubMed   Google Scholar  

Hula A, Moutoussis M, Will G-J, Kokorikou D, Reiter AM, Ziegler G, et al. Multi-round trust game quantifies inter-individual differences in social exchange from adolescence to adulthood. Computational Psychiatry. 2021;5:102–18.

Ziegler G, Hauser TU, Moutoussis M, Bullmore ET, Goodyer IM, Fonagy P, et al. Compulsivity and impulsivity traits linked to attenuated developmental frontostriatal myelination trajectories. Nat Neurosci. 2019;22:992–9.

Rosenberg M. Rosenberg self-esteem scale (RSE). Acceptance Commit Ther Measures package. 1965;61:18.

Google Scholar  

Somerville LH, Kelley WM, Heatherton TF. Self-esteem modulates medial prefrontal cortical responses to evaluative social feedback. Cereb Cortex. 2010;20:3005–13.

Yang J, Xu X, Chen Y, Shi Z, Han S. Trait self-esteem and neural activities related to self-evaluation and social feedback. Sci Rep. 2016;6:1–10.

Yang J, Dedovic K, Chen W, Zhang Q. Self-esteem modulates dorsal anterior cingulate cortical response in self-referential processing. Neuropsychologia. 2012;50:1267–70.

Van Schie CC, Chiu C-D, Rombouts SA, Heiser WJ, Elzinga BM. When compliments do not hit but critiques do: an fMRI study into self-esteem and self-knowledge in processing social feedback. Soc Cogn Affect Neurosci. 2018;13:404–17.

Onoda K, Okamoto Y, Nakashima K, Nittono H, Yoshimura S, Yamawaki S, et al. Does low self-esteem enhance social pain? The relationship between trait self-esteem and anterior cingulate cortex activation induced by ostracism. Soc Cogn Affect Neurosci. 2010;5:385–91.

Kashdan TB, DeWall CN, Masten CL, Pond RS Jr, Powell C, Combs D, et al. Who is most vulnerable to social rejection? The toxic combination of low self-esteem and lack of negative emotion differentiation on neural responses to rejection. PLoS ONE. 2014;9:e90651.

Gyurak A, Hooker CI, Miyakawa A, Verosky S, Luerssen A, Ayduk ÖN. Individual differences in neural responses to social rejection: the joint effect of self-esteem and attentional control. Soc Cogn Affect Neurosci. 2012;7:322–31.

Frewen PA, Lundberg E, Brimson-Théberge M, Théberge J. Neuroimaging self-esteem: a fMRI study of individual differences in women. Soc Cogn Affect Neurosci. 2013;8:546–55.

Chavez RS, Heatherton TF. Structural integrity of frontostriatal connections predicts longitudinal changes in self-esteem. Soc Neurosci. 2017;12:280–6.

Chavez RS, Heatherton TF. Multimodal frontostriatal connectivity underlies individual differences in self-esteem. Soc Cogn Affect Neurosci. 2015;10:364–70.

Baumeister RF, Campbell JD, Krueger JI, Vohs KD. Does high self-esteem cause better performance, interpersonal success, happiness, or healthier lifestyles? Psychological Sci Public Interest. 2003;4:1–44.

Hu X, Zheng J, Su N, Fan T, Yang C, Yin Y, et al . A Bayesian inference model for metamemory. Psychol Rev. 2021;128:824–85.

Desender K, Sasanguie D. Math anxiety relates positively to metacognitive insight into mathematical decision making. Psychol Res. 2021;86:1001–13.

Huguet P, Régner I. Counter-stereotypic beliefs in math do not protect school girls from stereotype threat. J Exp Soc Psychol. 2009;45:1024–7.

Gajdos T, Régner I, Huguet P, Hainguerlot M, Vergnaud J-C, Sackur J, et al. Does social context impact metacognition? Evidence from stereotype threat in a visual search task. PLoS ONE. 2019;14:e0215050.

Moses-Payne ME, Rollwage M, Fleming SM, Roiser JP. Postdecision evidence integration and depressive symptoms. Front Psychiatry. 2019;10:639.

Brown JD. High self-esteem buffers negative feedback: once more with feeling. Cognition Emot. 2010;24:1389–404.

David AS, Bedford N, Wiffen B, Gilleen J. Failures of metacognition and lack of insight in neuropsychiatric disorders. Philos Trans R Soc B Biol Sci. 2012;367:1379–90.

Hoven M, Lebreton M, Engelmann JB, Denys D, Luigjes J, van Holst RJ. Abnormalities of confidence in psychiatry: an overview and future perspectives. Transl Psychiatry. 2019;9:1–18.

Rouault M, Seow T, Gillan CM, Fleming SM. Psychiatric symptom dimensions are associated with dissociable shifts in metacognition but not task performance. Biol Psychiatry. 2018;84:443–51.

Hoven M, Denys D, Rouault M, Luigjes J, van Holst R. How do confidence and self-beliefs relate in psychopathology: a transdiagnostic approach. [Preprint]. 2022.

Koban L, Schneider R, Ashar YK, Andrews-Hanna JR, Landy L, Moscovitch DA, et al. Social anxiety is characterized by biased learning about performance and the self. Emotion. 2017;17:1144.

Aarts K, Pourtois G. Anxiety disrupts the evaluative component of performance monitoring: an ERP study. Neuropsychologia. 2012;50:1286–96.

Kube T, Schwarting R, Rozenkrantz L, Glombiewski JA, Rief W. Distorted cognitive processes in major depression: a predictive processing perspective. Biol Psychiatry. 2020;87:388–98.

Kube T, Rief W, Gollwitzer M, Gärtner T, Glombiewski JA. Why dysfunctional expectations in depression persist–Results from two experimental studies investigating cognitive immunization. Psychol Med. 2019;49:1532–44.

Lee AL, de Gardelle V, Mamassian P. Global visual confidence. Psychon Bull Rev. 2021;28:1233–42.

Sherman M, Seth A. Effects of expected task difficulty on metacognitive confidence and multitasking. [Preprint]. 2021.

Weil LG, Fleming SM, Dumontheil I, Kilford EJ, Weil RS, Rees G, et al. The development of metacognitive ability in adolescence. Conscious Cognition. 2013;22:264–71.

Kiebel SJ, Daunizeau J, Friston KJ. A hierarchy of time-scales and the brain. PLoS Comput Biol. 2008;4:e1000209.

Download references

Acknowledgements

For the purpose of Open Access, the authors have applied a CC-BY public copyright license to any author-accepted manuscript version arising from this submission. MR is the beneficiary of a postdoctoral fellowship from the AXA Research Fund. MR’s work is also supported by a department-wide grant from the Agence Nationale de la Recherche (ANR-17-EURE-0017, EUR FrontCog). This work has received support under the program «Investissements d’Avenir» launched by the French Government and implemented by ANR (ANR-10-IDEX-0001-02 PSL). G-JW was funded by the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement (No 707404) and the Sara van Dam z.l. Foundation, Royal Netherlands Academy of Arts & Sciences. SMF is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and Royal Society (206648/Z/17/Z). The Wellcome Centre for Human Neuroimaging is supported by core funding from the Wellcome Trust (203147/Z/16/Z). The Max Planck UCL Centre is a joint initiative supported by UCL and the Max Planck Society. The authors would like to thank Aislinn Bowler, Alexandra Hopkins, and Palee Womack for their assistance during recruitment and data collection.

Author information

These authors contributed equally: Marion Rouault, Geert-Jan Will.

Authors and Affiliations

Institut Jean Nicod, Département d’études cognitives, ENS, EHESS, CNRS, PSL University, 75005, Paris, France

Marion Rouault

Laboratoire de neurosciences cognitives et computationnelles, Département d’études cognitives, ENS, INSERM, PSL University, 75005, Paris, France

Department of Clinical Psychology, Utrecht University, Utrecht, The Netherlands

Geert-Jan Will

Wellcome Centre for Human Neuroimaging, University College London, London, UK

Stephen M. Fleming & Raymond J. Dolan

Max Planck UCL Centre for Computational Psychiatry and Ageing Research, University College London, London, UK

Department of Experimental Psychology, University College London, 26 Bedford Way, London, WC1H 0AP, UK

Stephen M. Fleming

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization: MR, G-JW, and SMF; Data collection: G-JW, Methodology: MR, G-JW, and SMF; Formal analysis: MR; Investigation: MR and G-JW; Writing—original draft: MR; Writing—review & editing: G-JW, SMF, and RJD; Funding acquisition: SMF and RJD.

Corresponding author

Correspondence to Marion Rouault .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary material, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Rouault, M., Will, GJ., Fleming, S.M. et al. Low self-esteem and the formation of global self-performance estimates in emerging adulthood. Transl Psychiatry 12 , 272 (2022). https://doi.org/10.1038/s41398-022-02031-8

Download citation

Received : 29 September 2021

Revised : 15 June 2022

Accepted : 17 June 2022

Published : 11 July 2022

DOI : https://doi.org/10.1038/s41398-022-02031-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

R425 first year student nurses ‘experience of encounters with death of a patient during clinical placement.

• Lebogang Molefe

BMC Nursing (2024)

Neural and computational underpinnings of biased confidence in human reinforcement learning

• Chih-Chung Ting
• Nahuel Salem-Garcia
• Maël Lebreton

Nature Communications (2023)

How do confidence and self-beliefs relate in psychopathology: a transdiagnostic approach

• Monja Hoven
• Judy Luigjes
• Ruth J. van Holst

Nature Mental Health (2023)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Self-Esteem

• Reference work entry
• First Online: 01 January 2022
• pp 4418–4421
• Cite this reference work entry

• Ulrich Orth 3  

201 Accesses

Self-esteem is defined as an individual’s subjective evaluation of his or her worth as a person (Donnellan et al. 2011 ). High self-esteem is characterized by feelings of self-acceptance and self-respect, whereas low self-esteem involves self-doubts and feelings of being a failure.

During the past one or two decades, research has made considerable progress in understanding the long-term stability of individual differences in self-esteem and in identifying the normative pattern of self-esteem change across the lifespan. These topics are important because research suggests that self-esteem has consequences for a broad set of outcomes, including social relationships, work, and health (Orth and Robins 2019 ).

Key Research Findings

Self-esteem is a relatively stable characteristic of individuals. Although people’s self-esteem may show some degree of fluctuation across short periods, individual differences in self-esteem (i.e., differences between people...

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Institutional subscriptions

Bleidorn W, Arslan RC, Denissen JJA et al (2016) Age and gender differences in self-esteem: a cross-cultural window. J Pers Soc Psychol 111:396–410. https://doi.org/10.1037/pspp0000078

Article   Google Scholar  

Donnellan MB, Trzesniewski KH, Robins RW (2011) Self-esteem: enduring issues and controversies. In: Chamorro-Premuzic T, von Stumm S, Furnham A (eds) The Wiley-Blackwell handbook of individual differences. Wiley-Blackwell, Chichester, pp 718–746

Google Scholar  

Donnellan MB, Kenny DA, Trzesniewski KH et al (2012) Using trait-state models to evaluate the longitudinal consistency of global self-esteem from adolescence to adulthood. J Res Pers 46:634–645. https://doi.org/10.1016/j.jrp.2012.07.005

Kuster F, Orth U (2013) The long-term stability of self-esteem: its time-dependent decay and nonzero asymptote. Personal Soc Psychol Bull 39:677–690. https://doi.org/10.1177/0146167213480189

Kuster F, Orth U, Meier LL (2013) High self-esteem prospectively predicts better work conditions and outcomes. Soc Psychol Personal Sci 4:668–675. https://doi.org/10.1177/1948550613479806

McGuire S, Manke B, Saudino KJ et al (1999) Perceived competence and self-worth during adolescence: a longitudinal behavioral genetic study. Child Dev 70:1283–1296. https://doi.org/10.1111/1467-8624.00094

Orth U, Robins RW (2019) Development of self-esteem across the lifespan. In: McAdams DP, Shiner RL, Tackett JL (eds) Handbook of personality development. Guilford, New York, pp 328–344

Orth U, Trzesniewski KH, Robins RW (2010) Self-esteem development from young adulthood to old age: a cohort-sequential longitudinal study. J Pers Soc Psychol 98:645–658. https://doi.org/10.1037/a0018769

Orth U, Robins RW, Widaman KF (2012) Life-span development of self-esteem and its effects on important life outcomes. J Pers Soc Psychol 102:1271–1288. https://doi.org/10.1037/a0025558

Orth U, Erol RY, Luciano EC (2018) Development of self-esteem from age 4 to 94 years: a meta-analysis of longitudinal studies. Psychol Bull 144:1045–1080. https://doi.org/10.1037/bul0000161

Sowislo JF, Orth U (2013) Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies. Psychol Bull 139:213–240. https://doi.org/10.1037/a0028931

Stinson DA, Logel C, Zanna MP et al (2008) The cost of lower self-esteem: testing a self- and social-bonds model of health. J Pers Soc Psychol 94:412–428. https://doi.org/10.1037/0022-3514.94.3.412

Trzesniewski KH, Donnellan MB, Robins RW (2003) Stability of self-esteem across the life span. J Pers Soc Psychol 84:205–220. https://doi.org/10.1037/0022-3514.84.1.205

Trzesniewski KH, Donnellan MB, Moffitt TE et al (2006) Low self-esteem during adolescence predicts poor health, criminal behavior, and limited economic prospects during adulthood. Dev Psychol 42:381–390. https://doi.org/10.1037/0012-1649.42.2.381

von Soest T, Wagner J, Hansen T et al (2018) Self-esteem across the second half of life: the role of socioeconomic status, physical health, social relationships, and personality factors. J Pers Soc Psychol 114:945–958. https://doi.org/10.1037/pspp0000123

Wagner J, Gerstorf D, Hoppmann C et al (2013) The nature and correlates of self-esteem trajectories in late life. J Pers Soc Psychol 105:139–153. https://doi.org/10.1037/a0032279

Wagner J, Hoppmann C, Ram N et al (2015) Self-esteem is relatively stable late in life: the role of resources in the health, self-regulation, and social domains. Dev Psychol 51:136–149. https://doi.org/10.1037/a0038338

Wagner J, Lüdtke O, Trautwein U (2016) Self-esteem is mostly stable across young adulthood: evidence from latent STARTS models. J Pers 84:523–535. https://doi.org/10.1111/jopy.12178

Download references

Author information

Authors and affiliations.

University of Bern, Bern, Switzerland

Ulrich Orth

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Ulrich Orth .

Editor information

Editors and affiliations.

Population Division, Department of Economics and Social Affairs, United Nations, New York, NY, USA

Department of Population Health Sciences, Department of Sociology, Duke University, Durham, NC, USA

Matthew E. Dupre

Section Editor information

Department of Population Health Sciences, Center for the Study of Aging and Human Development, Duke University, Durham, NC, USA

Heather R. Farmer

Center for the Study of Aging and Human Development, Duke University, Durham, NC, USA

Rights and permissions

Reprints and permissions

Copyright information

© 2021 Springer Nature Switzerland AG

About this entry

Cite this entry.

Orth, U. (2021). Self-Esteem. In: Gu, D., Dupre, M.E. (eds) Encyclopedia of Gerontology and Population Aging. Springer, Cham. https://doi.org/10.1007/978-3-030-22009-9_1093

Download citation

DOI : https://doi.org/10.1007/978-3-030-22009-9_1093

Published : 24 May 2022

Publisher Name : Springer, Cham

Print ISBN : 978-3-030-22008-2

Online ISBN : 978-3-030-22009-9

eBook Packages : Social Sciences Reference Module Humanities and Social Sciences Reference Module Business, Economics and Social Sciences

Share this entry

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

Low Self Esteem: What Does it Mean to Lack Self-Esteem?

Saul Mcleod, PhD

Editor-in-Chief for Simply Psychology

BSc (Hons) Psychology, MRes, PhD, University of Manchester

Saul Mcleod, PhD., is a qualified psychology teacher with over 18 years of experience in further and higher education. He has been published in peer-reviewed journals, including the Journal of Clinical Psychology.

Learn about our Editorial Process

Olivia Guy-Evans, MSc

Associate Editor for Simply Psychology

BSc (Hons) Psychology, MSc Psychology of Education

Olivia Guy-Evans is a writer and associate editor for Simply Psychology. She has previously worked in healthcare and educational sectors.

On This Page:

Self-esteem should be viewed as a continuum and can be high, medium, or low, and it is often quantified as a number in empirical research.

When considering self-esteem, it is important to note that both high and low levels can be emotionally and socially harmful to the individual. Indeed it is thought an optimum level of self-esteem lies in the middle of the continuum. Individuals operating within this range are thought to be more socially dominant within relationships.

Empirical Research

Research has shown key differences between individuals with high and low self-esteem. For example, people with high self-esteem focus on growth and improvement, whereas people with low self-esteem focus on not making mistakes in life.

Low self-esteem has been shown to be correlated with several negative outcomes, such as depression (Silverstone & Salsali, 2003).

Rosenberg and Owen (2001) offer the following description of low self-esteem people based on empirical research. People with low self-esteem are more troubled by failure and tend to exaggerate events as being negative.

For example, they often interpret non critical comments as critical. They are more likely to experience social anxiety and low levels of interpersonal confidence.

This in turn makes social interaction with others difficult as they feel awkward, shy, conspicuous, and unable to adequately express themselves when interacting with others (p. 409). Furthermore, low self-esteem individuals tend to be pessimistic towards people and groups within society.

Research has also shown that low self-esteem has to linked to an increased risk of teenage pregnancy.

Guindon (2002) asked school counsellors to list five characteristics that best describe students with low self-esteem. Over 1000 words were used and the most common are listed below:

• Withdrawn/shy/quiet
• Underachieving
• Negative (attitude)
• Socially inept
• Angry/hostile
• Unmotivated
• Dependent/follower
• Poor self-image
• Non-risk-taker
• Lacks self-confidence
• Poor communication

Low Self-Esteem in Children

It should be noted that, on average, self-esteem during childhood is found to be relatively high. However, there are individual differences, and some children are unfortunate to experience feelings of low self-esteem.

Low self-esteem in children tends to be related to physical punishment and the withholding of love and affection by parents. Carl Rogers would describe this as conditional positive regard, whereby individuals only receive positive attention from significant others (such as parents) when they act in a certain way. This reinforces to the child that they are only a person of value when they act a certain way (e.g., achieving A grades on a test).

Children with low self-esteem rely on coping strategies that are counterproductive such as bullying, quitting, cheating, avoiding, etc. Although all children will display some of these behaviors at times, low self-esteem is strongly indicated when these behaviors appear with regularity.

Socially children with low self-esteem can be withdrawn or shy and find it difficult to have fun. Although they may have a wide circle of friends, they are more likely to yield to group pressure and more vulnerable to bullying. At school, they avoid trying new things (for fear of failure) and will give up easily.

Low Self-Esteem in Teenagers

Self-esteem continues to decline during adolescence (particularly for girls). Researchers have explained this decline to body image and other problems associated with puberty.

Although boys and girls report similar levels of self-esteem during childhood, a gender gap emerges by adolescence in that adolescent boys have higher self-esteem than adolescent girls (Robins et al., 2002).

Girls with low self-esteem appear to be more vulnerable to perceptions of the ideal body image perpetuated in western media (through methods such as airbrushing models on magazine covers).

Abraham, T. (1988). Toward a Self-Evaluation Maintenance Model of Social behavior. In L. Berkowitz (Ed), Advances in Experimental Social Psychology (pp. 181–227).Academic Press.

Coopersmith, S. (1967). The Antecedents of Self-esteem . Freeman.

Harter, S. 1993. Causes and Consequences of Low Self-esteem in Children and Adolescents. In Baumeister, R.F. (Ed.) Self-Esteem: The Puzzle of Low Self-regard (pp. 87-116).

Mruk, C. (1995). Self-Esteem: Research, Theory, and Practice . Springer.

Guindon, M. H. (2002). Toward Accountability in the Use of the Self‐Esteem Construct. Journal of Counseling & Development, 80(2) , 204-214.

Robins, R.W., Trzesniewski, K.H., Tracy, J.L., Gosling, S.D., & Potter, J. (2002). Global self-esteem across the lifespan. Psychology and Aging , 17, 423-434.

Rosenberg, M. (1976). Beyond Self-Esteem: The Neglected Issues in Self-concept Research . Paper presented at the annual meetings of the ASA.

Rosenberg, M. (1979). Conceiving the Self . Basic Books.

Rosenberg, M., & Owens, T.J. (2001). Low self-esteem people: A collective portrait. In T.J. Owens. S. Stryker, & N. Goodmanm (Eds.), Extending self-esteem theory and research (pp. 400-436). New York: Cambridge University Press.

Silverstone, P. H., & Salsali, M. (2003). Low self-esteem and psychiatric patients: Part I–The relationship between low self-esteem and psychiatric diagnosis. Annals of General Psychiatry, 2(1) , 2.

Viktor, G. (1982). The Self-Concept. Annual Review of Sociology , 8:1–33.

Viktor, G., & Schwalbe, M.L. (1983). Beyond the Looking-glass Self: Social Structure and Efficacy-Based Self-Esteem. Social Psychology Quarterly , 46:77–88.

Related Articles

Relationships

How To Fix a Relationship With Bad Communication

Relationships , Dating Tips

10 Warning Signs of Bad Communication in a Relationship

Dating Tips , Relationships

My Boyfriend Is Boring: What Should I Do?

Why Do Relationships Fail?

Boundaries with Adult Children

Boundaries vs Expectations