Self-esteem and depression in a Taiwanese population: a meta-analysis.
Beck (1987) argued that, according to the cognitive theory of depression, negative self-evaluation is a proximal factor in individual depression. Moreover, Rosenberg (1979) said that self-esteem refers to self-perception and self-worth. Thus, self-esteem is a proximal factor in individual depression. The claim made by Leary, Schreindorfer, and Haupt (1995) that self-esteem has a strong correlation with depression has been confirmed in many studies. For example, Conley, Haines, Hilt, and Metalsky (2001) reported a significant negative correlation between self-esteem and depression (r = -.45) in a study of 8-year-olds. Burwell and Shirk (2006) similarly found a significant negative correlation between self-esteem and depression (r = -.63) in a study of 14-year-olds. The magnitude of the correlation, which ranged from r = -.45 to r = -.63, apparently correlated with age. Moreover, in a study of secondary school students Campbell (1997) found that the correlation was stronger in female students (r = -.62) than in male students (r = -.35). Similarly, Schafer, Wickrama, and Keith (1998) reported a stronger correlation coefficient between self-esteem and depression in married women (r = -.64) compared to that in married men (r = -.40). Dixon and Robinson Kurpius (2008) found a significant negative correlation between self-esteem and depression (r = -.58). In a study targeting adults aged from 58 to 64 years, Fernandez, Mutran, and Reitzs (1998) found a significant negative correlation between self-esteem and depression (r = -.45). Researchers who have focused on specific groups include Courtney, Gamboz, and Johnson (2008), who studied the correlation between self-esteem and depression in 197 teenagers with eating disorders. They found that low self-esteem had a significant positive correlation with depression (r = .20). Crockenberg and Leerkes (2003) addressed the same issue but focused specifically on a population of pregnant women. Again, self-esteem had a significant negative correlation with depression (r = -.65). Because of the widely varying reports regarding the strength of the
correlation between self-esteem and depression, we performed a meta-analysis to clarify the correlation.
In a study of junior high school students conducted in Taiwan, Wang (2009) reported a significant negative correlation between self-esteem and depression (r = -.45). Gao and Lin (2008) similarly reported a negative correlation (r = -.60) between self-esteem and depression in a study of high school students. In other studies of high school or college students conducted in Taiwan similar findings have been reported (Su, 2005; Yang, 1996). For example, Wu, Lu, and Tung (2009) claimed that college students' self-esteem had a direct impact on their depressive mood. In a large-scale study Chuo, Hong, Su, and Chen (2009) found that the association between self-esteem and depression increases with age. Wu and Huang (2010) conducted a 3-year longitudinal study and pointed out that, among all the dynamic factors on the trajectories of early adolescent depressive symptoms, low self-esteem had a considerable impact on the initial stage and on the trajectories of depression. In summary, self-esteem negatively correlates with depression at all age levels ranging from elementary school students to adults and in both general and specific populations. And the trend seems to be clear: the higher the level of one's self-esteem, the less likely one will suffer from depression.
Previously conducted meta-analyses of the relationship between self-esteem and depression include those by Sowislo and Orth (2013), who analyzed 77 studies published in English and found a strong correlation between self-esteem and depression and a large effect size ([bar.[gamma]] = -.57), and Cai, Wu, and Brown (2009), who reported that, in studies performed in the People's Republic of China, a moderate to high correlation between self-esteem and depression has been reported. Our meta-analysis is the first in which this relationship has been studied specifically in a Taiwanese population.
In this study, self-esteem was defined as the individual evaluation of the self. Depression refers to a depressive affect rather than to clinical syndromes. To identify studies potentially relevant to this study, we performed database searches in Taiwan's National Digital Library of Theses and Dissertations; the Periodical Information Center of National Central Library; and the Abstracts of Conference Paper Database. Keywords we used were a combination of the terms of self-esteem (self-esteem or self-concept) and depression (depressive, depressed, or depression).
The electronic search retrieved 225 studies. Eight additional studies were identified by checking the reference lists of all relevant articles. The inclusion criteria were as follows: First, studies should have been published in traditional Chinese characters. Second, the samples were taken from the normal population; clinical participants were excluded. Third, studies should provide sample size and the information to compute the correlation between self-esteem and depression.
Coding the Studies
The effect size calculated in the present study was measured using Pearson product moment correlation. The following variables were coded for each study: (a) the proportion of males in the sample; (b) the age range of the sample (if the study reported grade but not age, age was estimated by adding six to the grade level, and if a range was given, the median was used as an estimate of the mean); (c) birth cohort, which was obtained by deducting the mean age of the samples from the year in which the surveys were distributed; (d) sample type, categorizing samples into four types (i.e., general students, students with special needs, general population other than students, and physically ill people); (e) publication status, classifying publications as journal articles, theses, dissertations, or conference papers; and (f) the specified measures of self-esteem and depression. The coding was performed by the first author on two occasions: from October 2011 to March 2012 and from March 2012 to April 2012. The stability of coding estimated by correlation coefficients was computed for continuous variables such as sample size and sample age. For categorical variables such as gender, publication status, self-esteem and depression measures, and sample type, the percentages of agreement were computed. The results showed that the consistency percentages for categorical moderating variables ranged from 98.31% to 100% and the correlation coefficients for continuous moderating variables ranged from .95 to 1.00.
MetaWin Version 2, developed by Rosenberg, Adams, and Gurevitch (2000), was used for statistical analyses. In the fixed-effects model, sampling errors are assumed to be the only source of variation in effect sizes. In the random-effects model, both sampling errors and variation in true effect size are the sources to explain the variation in effect sizes (Rosenberg et al., 2000). Because of the implausibility of the assumptions in the fixed-effects model, the random-effects model was adopted. The variance of [bar.[gamma]]1 = [[([1-[y.sup.2.sub.1]).sup.2]]/[N-2]] computing was used as the weight when the weighted mean effect size [bar.[gamma]]
Descriptive Statistics of the Studies
After applying the above criteria, the meta-analysis included 32,005 participants from 59 samples in 50 studies. The studies were published between 1984 and 2011, and the sample size ranged from 59 to 1,688 participants. Of the 59 reported effect sizes, 47 (79.66%) were based on both male and female samples, four were based on only male samples, and eight were based on only female samples. Mean age in the samples ranged from 8.50 to 68.56 years. Publication outlets included journal articles, theses or dissertations, and conference papers. In the birth cohorts date of birth (DOB) ranged from the 1930s to the 1990s. Thirteen effect sizes were obtained for the birth cohort of samples in which DOB was in the 1980s (37.14%), and 12 were obtained for the birth cohort of samples in which DOB was in the 1990s (34.19%).
Magnitude of Overall Effect Size
The weighted mean effect size [bar.[gamma]] was -.48. A 95% confidence interval of -.52 to -.45, between which 0 was not located, confirming that the weighted mean effect size for the correlation between self-esteem and depression was significantly lower than 0. According to Cohen (1992), the weighted mean effect size was large, indicating a high correlation between self-esteem and depression.
A test for heterogeneity indicated that effect sizes were heterogeneous (Q = 79.92, df = 58, p < .05). That is, the variance in effect size was exceeded by sampling errors. Hence, the effects of moderators including gender, age, birth cohort, sample type, publication status, self-esteem measure, and depression measure were examined.
Publication status. Table 1 shows that the weighted mean effect sizes were negative for both journal articles ([bar.[gamma]] = -.48) and unpublished theses and dissertations ([bar.[gamma]] = -.48), and significantly differed from 0. The result of [Q.sub.b] = .01 (df = 1, p > .05) indicated that the weighted mean effect sizes for different publication statuses did not differ significantly. Restated, publication status did not moderate the correlation between self-esteem and depression.
According to the Rothenthal fail-safe number, 16,228 missing studies with a mean correlation of zero were needed to reduce the mean correlation from statistical significance to nonsignificance. This result is consistent with the effect of publication status, lending support for no publication bias.
Gender. The relationship between the effect sizes and the proportion of male samples was analyzed using regression analysis. The regression analysis obtained a slope of .00 (p > .05), which indicated that the proportion of male and female
samples did not affect the correlation between self-esteem and depression. When the moderating effect of gender was examined, the proportion of male samples (continuous data) was used as the predictive variable in the regression analysis. The proportion was obtained by dividing the total number of male samples by the total number of male and female samples. For effect sizes based on only male samples, the predicting variable was coded 100%. For effect sizes based on only female samples, the predicting variable was coded 0%. Therefore, the analysis was based on both male and female samples.
Age. Similarly, regression analysis was performed to evaluate age effect. The slope of -.00 (p > .05) obtained in the regression analysis results indicated that the mean age of the samples did not significantly affect the correlation between self-esteem and depression. Sample age was therefore not associated with the correlation between self-esteem and depression.
Birth cohort. The birth cohort of the samples ranged between 1930s and 1990s. Birth cohort was used as a predictive variable in the regression analysis for the correlation between self-esteem and depression. The slope of [Q.sub.b] = -.00 (p > .05) obtained in the regression analysis indicated that birth cohort was not a significant predictor of the relationship between self-esteem and depression.
Sample type. Table 1 contains a summary of the analytical results for the different sample types. The weighted mean effect sizes for different sample types varied across studies, [Q.sub.b] = 11.04 (df = 3, p < .05), which indicated that sample type moderated the correlation between self-esteem and depression.
Self-esteem measure. Researchers using the Rosenberg Self-Esteem Inventory (RSE; 1979) or a modified version reported a weighted mean effect size of [bar.[gamma]] = -.55 (k = 21). Researchers using other self-esteem measures reported a weighted mean effect size of [bar.[gamma]] = 46 (k = 38). Both values were negative and significantly different from 0. The result of [Q.sub.b] = 9.72 (df = 2, p < .01) indicates that the mean correlation coefficients for various self-esteem measures were different. As the 95% confidences intervals did not overlap, those who had used the RSE recorded a significantly higher mean correlation compared to those who had used other self-esteem measures.
Depression measure. Some primary researchers used multiple self-depression measures. To investigate the effect of depression measure, multiple effect sizes from the same sample were considered independently. Coding multiple effect sizes from the same sample yielded 62 effect sizes. As only one effect size was calculated using the Zung Depression Scale (Zung, 1965), this effect size was excluded from the calculation of the mean effect size. The final analysis included 61 effect sizes. All the weighted mean effect sizes were negative and significantly different from 0. The weighted mean effect size was [bar.[gamma]] = -.57 for studies in which an original or modified version of either the first or second edition of the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, & Erbaugh, 1961; BDI-II,
Beck, Steer, & Brown, 1996) had been used. The weighted mean effect size was [bar.[gamma]] = -.58 for studies in which the Center for Epidemiological Studies Depression Scale (Radloff, 1977) or a modified version of it had been used. The result of [Q.sub.b] = 24.32 (df = 5,p < .001) indicates that the depression measure was related to the correlation between self-esteem and depression. At 95% confidence intervals, in which the mean effect sizes for these two measures ([bar.[gamma]] = -.42) exceeded those of studies where other depression measures had been used.
Chentsova-Dutton and Tsai (2009) suggested that in Western cultures it is crucial to have a good feeling toward oneself. When one does not have good self-esteem, one is likely to feel depressed. In contrast, in Eastern cultures self-criticism is sometimes viewed as a good driving force for self-improvement.
Accordingly, Chentsova-Dutton and Tsai claimed that the correlation between self-esteem and depression in Eastern cultures is not as strong as it is in Western cultures. However, Chentsova-Dutton and Tsai's argument was not supported when one compares the results in the present study with the results gained in previous meta-analyses. According to the guidelines proposed by Cohen (1992), our study had a large mean effect size. An effect size of [bar.[gamma]] = -.57 was similarly considered large in the meta-analysis conducted by Sowislo and Orth (2013). The effect size obtained by Cai et al. (2009) was also large and close to the effect size reported in the present study. Most of the studies included in the meta-analysis by Sowislo and Orth (2013) were conducted in Western societies. In contrast, the studies included in the research conducted by Cai et al. (2009) had been performed in China, whereas those analyzed in the present study had been performed in Taiwan. The fact that the mean effect sizes reported in these three meta-analyses were comparable seems to contradict the argument by Chentsova-Dutton and Tsai that people from different cultural backgrounds have different perspectives about the importance of self-esteem.
In the present study we found that sample types were associated with the correlation between self-esteem and depression. The effect size was substantially larger for physically ill people. We should pay special attention to, and provide special caring for, this group of people. Medical care providers and/or the families of these people should not only assist in providing for the physical needs of those who are ill but they should also pay attention to their self-esteem. In practice, hospitals can provide counseling services to physically ill patients or arrange religious services to provide spiritual comfort. Such services can help prevent people being overly focused on their inability to deal with everyday, routine work because of their illness. Hopefully, doing so can help physically ill people maintain an appropriate self-esteem level and can reduce the incidence of tragic events such as suicide.
The results in this study revealed that demographic variables had no significant moderating effects. Specifically, gender was not significantly related to the correlation between self-esteem and depression, which is consistent with results gained in the earlier study by Sowislo and Orth (2013). Results of our analysis showed that there was no cultural difference in terms of the impact of gender on this relation between self-esteem and depression. Moreover, sample age was unrelated to the correlation between self-esteem and depression. That is, the correlation between self-esteem and depression did not significantly differ between children and adults. Therefore, parents and teachers should be aware that the importance of self-esteem for children's psychological well-being is no less than it is for adults. When disciplining children, parents and teachers should bear in mind that inappropriate language can easily reduce the self-esteem of children.
In everyday life, parents and teachers should pay attention to the words they use and should try to use positive language when disciplining children.
There are some limitations in this study. First, researchers in all studies included in this meta-analysis used self-report inventories. Hence, our study results may not be generalizable to studies in which responses obtained by other means (i.e., nonself-report) were used. Future researchers should examine the moderating effect of the type of reporting. This information could be useful to understand if peer, teacher, or parent reports are good sources for providing information on the level of self-esteem and depression. Second, there was a possibility of shared method effect between self-esteem and depression self-report inventories. Therefore, the mean correlations obtained in this study might be larger than the true correlations. Lastly, as the studies we used for our meta-analysis were correlational, the causal relationship between self-esteem and depression cannot be established.
Ayuso-Mateos, J. L. (2000). Global burden of unipolar depressive disorders in the year 2000. Retrieved from http://www.who.int/healthinfo/statistics/bod depression.pdf
Beck, A. T. (1987). Cognitive models of depression. Journal of Cognitive Psychotherapy: International Quarterly, 1, 5-37.
Beck, A. T., Steer, R. A., & Brown, G. (1996). Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation.
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.
Burwell, R. A., & Shirk, S. R. (2006). Self-processes in adolescent depression: The role of self-worth contingencies. Journal of Research on Adolescence, 16, 479-490. http://doi.org/fkdr3s
Cai, H., Wu, Q., & Brown, J. D. (2009). Is self-esteem a universal need? Evidence from the People's Republic of China [In Chinese]. Asian Journal of Social Psychology, 12, 104-120. http://doi. org/d3rhvf
Campbell, T. L. (1997). Understanding the association between self-concept, daily hassles, and depressive and anxiety symptoms among adolescents. Unpublished doctoral dissertation, University of Ottawa, Canada.
Chentsova-Dutton, Y. E., & Tsai, J. L. (2009).Understanding depression across cultures. In I. Gotlib & C. Hammen (Eds.), Handbook of depression (2nd ed., pp. 363-403). New York: Guilford.
Chuo, S. L., Hong, L. Y., Su, C. L., & Chen, H. C. (2009). A research of the Chinese version Beck Youth Inventories [In Chinese]. Psychological Testing, 56, 639-669.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159. http://doi.org/as7
Conley, C. S., Haines, B. A., Hilt, L. M., & Metalsky, G. I. (2001). The Children's Attributional Style Interview: Developmental tests of cognitive diathesis-stress theories of depression. Journal of Abnormal Child Psychology, 29, 445-463. http://doi.org/fpbngn
Courtney, E. A., Gamboz, J., & Johnson, J. G. (2008). Problematic eating behaviors in adolescents with low self-esteem and elevated depressive symptoms. Eating Behaviors, 9, 408-414. http:// doi.org/bzzwv5
Crockenberg, S. C., & Leerkes, E. M. (2003). Parental acceptance, postpartum depression, and maternal sensitivity: Mediating and moderating processes. Journal of Family Psychology, 17, 80-93. http://doi.org/ctxzqb
Derogatis, L. R. (1983). SCL-90-R: Administration, scoring and procedures manual-II for the revised version. Baltimore, MD: Clinical Psychometric Research.
Dixon, S. K., & Robinson Kurpius, S. E. (2008). Depression and college stress among university undergraduates: Do mattering and self-esteem make a difference? Journal of College Student Development, 49, 412-424. http://doi.org/fmrhf9
Fernandez, M. E., Mutran, E. J., & Reitzs, D. C. (1998). Moderating the effects of stress on depressive symptoms. Research on Aging, 20, 163-182. http://doi.org/c8qpvk
Gao, M. K., & Lin, L. W. (2008). Exploring the relations between explanatory style, life stress, and depression in high school students: From cognitive diathesis-stress model perspective [In Chinese]. The Archive of Guidance & Counseling, 30, 41-59.
John Tung Foundation. (2002). Adolescent Self-Check List for Depressive Emotions. Retrieved from http://www.jtf.org.tw/overblue/young/
John Tung Foundation. (2011, December 6). A survey study on the relationship between physical activity, source of pressure, and depression among junior, senior and vocational high school students in five major cities of Taiwan. Retrieved from http://jtf5123.wiwe.com.tw/
Leary, M. R., Schreindorfer, L. S., & Haupt, A. L. (1995). The role of low self-esteem in emotional and behavioral problems: Why is low self-esteem dysfunctional? Journal of Social Clinical Psychology, 14, 297-314. http://doi.org/dwrc77
Lee, Y. (1999). Epidemiological survey of depressive disorders in Kaohsiung metropolis: Development of a culture-relevant "Taiwanese Depression Screening Questionnaire". Retrieved from Taiwan National Science Council database (NSC88-2413-H182A-001).
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401. http://doi.org/b4z
Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
Rosenberg, M., Adams, D., & Gurevitch, J. (2000). MetaWin Version 2: Statistical software for meta-analysis. Sunderland, MA: Sinauer.
Schafer, R. B., Wickrama, K., & Keith, P. (1998). Stress in marital interaction and change in depression: A longitudinal analysis. Journal of Family Issues, 19, 578-594. http://doi.org/c9387q
Sowislo, J. F., & Orth, U. (2013). Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies. Psychological Bulletin, 139, 213-240. http://doi.org/kh8
Su, H. Y. (2005). A study on the relationship among parent-child relationship, perfectionism, self-esteem and depressive tendency of adolescents. Unpublished master's thesis, National Kaohsiung Normal University, Taiwan, ROC.
Wang, C. C. (2009). A study of the relationship among self-concept, parentification and depression of single-parent family's students in junior high school. Unpublished master's thesis, National Chiayi University, Taiwan, ROC.
World Health Organization. (2001). The world health report 2001. Mental health: New understanding, new hope. Geneva: Author.
Wu, C. T., & Huang, Y. T. (2010). The impact of dynamic factors on trajectories of early adolescent depressive symptom [In Chinese]. Formosa Journal of Mental Health, 23, 535-562.
Wu, P. H., Lu, W. M., & Tung, H. Y. (2009). The effects of self-concept, proactive coping, and depressive mood on the self-identity of college students in Taiwan: Using structural equation modeling. Journal of Education and Psychology, 32, 55-78.
Yang, S. N. (1996). Integration of depression theory-cognitive approach. Unpublished master's thesis, National Chengchi University, Taiwan, ROC.
Zung, W. W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70. http://doi.org/cd3gnc
National Changhua University of Education
National Changhua University of Education
Shu-Jiuan Chen, Graduate Institute of Education, National Changhua University of Education;
Chia-Hui Chiu, Department of Foreign Languages and Literature, Tunghai University; Chiungjung Huang, Graduate Institute of Education, National Changhua University of Education.
Shu-Jiuan Chen is now an independent scholar in Taichung City, Taiwan, ROC.
Correspondence concerning this article should be addressed to: Shu-Jiuan Chen, 6 F1. -1, No. 110,
Sec. 2, Hankou Road, Situn District, Taichung City 407, Taiwan, ROC. Email: firstname.lastname@example.org
Table 1. The Effects of Moderators on the Relationship Between Self-Esteem and Depression 95% CI Moderator k mean Lower Upper [Q.sub.b] r limit limit Participant group 11.04 * General students 40 -.47 -.52 -.43 Students with special 3 -.42 -.75 -.09 needs General population 9 -.42 -.53 -.32 other than students Physically ill people 7 -.64 -.76 -.51 Publication status 0.01 Journal papers 18 -.48 -.53 -.44 Unpublished papers 41 -.48 -.55 -.41 Self-esteem measures 9.72** RSE 21 -.55 -.60 -.49 Miscellaneous 38 -.46 -.49 -.41 Depression measures 24.32 *** BDI 15 -.57 -.64 -.50 CES-D 11 -.58 -.67 -.50 ASCLDE 5 -.39 -.54 -.23 TDSQ 3 -.50 -.82 -.17 SCL-90-R 3 -.40 -.73 -.07 Miscellaneous 24 -.42 -.47 -.36 Note. RSE = Rosenberg Self-Esteem Inventory and Modified Rosenberg Self-Esteem Inventory; BDI = Beck Depression Inventory and Modified Beck Depression Inventory; CES-D = Center for Epidemiological Studies Depression Scale and Modified Center for Epidemiological Studies Depression Scale; ASCLDE = Adolescent Self-Check List for Depressive Emotions and Modified Adolescent Self-Check List for Depressive Emotions (John Tung Foundation, 2002); TDSQ = Taiwanese Depression Screening Questionnaire and Modified Taiwanese Depression Screening Questionnaire (Lee, 1999); SCL-90-R = Symptom Checklist-90-Revised and Modified Symptom Checklist-90-Revised (Derogatis, 1983).
|Printer friendly Cite/link Email Feedback|
|Author:||Chen, Shu-Jiuan; Chiu, Chia-Hui; Huang, Chiungjung|
|Publication:||Social Behavior and Personality: An International Journal|
|Date:||May 1, 2013|
|Previous Article:||Gender differences in buyer-seller negotiations: emotion regulation strategies.|
|Next Article:||Effects of trust and perceived risk on user acceptance of a new technology service.|