The mediating role of self-esteem in the association between perceived social support and youth mental health difficulties.

The mediating role of self-esteem in the association between perceived social support and youth mental health difficulties

Background

This study examined whether self-esteem mediates the association between perceived social support and youth mental health difficulties in a heterogeneous sample of school-recruited youths and clinically referred inpatient and outpatient youths.

Methods

The study included 312 children and adolescents (mean age = 14.36 years, SD = 2.19, range = 9–19; 37.5% boys, 62.5% girls). The primary analysis was a covariate-adjusted mediation model in the full sample, with perceived social support (MSPSS) as the predictor, self-esteem (RSE) as the mediator, and Strengths and Difficulties Questionnaire (SDQ) Total Difficulties as the outcome. Hierarchical clustering based on SDQ, RSE, and MSPSS item profiles was conducted as a secondary, exploratory analysis.

Results

Higher perceived social support was associated with higher self-esteem (B = 0.26, 95% CI [0.20, 0.31], bootstrapped p < 0.001), and higher self-esteem was associated with fewer mental health difficulties (B = −0.43, 95% CI [−0.50, −0.35], bootstrapped p < 0.001) in the covariate-adjusted full-sample model. The indirect effect via self-esteem was statistically detectable (B = −0.11, 95% CI [−0.14, −0.08], bootstrapped p < 0.001), indicating partial mediation. Exploratory cluster-specific analyses indicated that the indirect effect was significant in the lower-difficulties/higher-resources cluster (B = −0.05, 95% CI [−0.09, −0.02], bootstrapped p = 0.003), but not in the higher-difficulties/lower-resources cluster (B = −0.01, 95% CI [−0.04, 0.00], bootstrapped p = 0.077).

Conclusion

Self-esteem appears to represent one pathway linking perceived social support to fewer mental health difficulties in youth. Cluster-specific findings should be interpreted as exploratory and suggest that the strength of this pathway may vary across psychosocial profiles.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13034-026-01141-1.

Introduction

Mental health difficulties emerging during childhood and adolescence may shape developmental trajectories with long-term consequences for psychosocial functioning and well-being [1]. Perceived social support and self-esteem are two factors that have repeatedly been linked to better mental health outcomes in young people, yet the way these factors are connected remains less clearly specified [2].

Perceived social support refers to the subjective sense that support is available from family, friends, and significant others [3]. There is a distinction between objective and subjective support, although the terms can overlap. Research in mental health usually investigates perceived support, which is subjectively experienced by the individual. Across developmental and clinical studies, perceived support is generally associated with fewer emotional and behavioral problems [4, 5].

Self-esteem, in turn, reflects a person’s global evaluation of self-worth and has consistently been associated with lower levels of depression, anxiety, and broader psychological distress [6].

A central unresolved question is whether self-esteem may partly explain how perceived social support relates to mental health. We modelled perceived social support as the predictor and self-esteem as the mediator for theory-based reasons: perceived social support may be viewed as an interpersonal resource external to the individual, whereas self-esteem reflects an internal self-evaluative process related to global self-worth [7]. This rationale is broadly consistent with the review by [3], which found that positive social support was linked to lower levels of depression, anxiety, stress, and behavioural problems, as well as to higher self-esteem and self-efficacy. In this framework, supportive relationships may foster more positive self-evaluations [8] and a greater sense of security, which in turn may be associated with fewer emotional and behavioural difficulties. At the same time, this directional specification should be understood as theory-informed rather than causal, because the present data are cross-sectional and perceived social support and self-esteem may influence each other bidirectionally [8, 9].

Beyond full-sample associations, young people may differ in their overall psychosocial profiles. A person-centered clustering approach can therefore be useful for describing psychological heterogeneity in combinations of symptoms, self-esteem, and perceived support. However, when clusters are created from the same variables later entered into mediation models, subgroup results require cautious interpretation due to reduced variance and circularity, which may influence the apparent pattern of effects.

In addition to global difficulties, domain-specific patterns also matter. Perceived social support is most consistently tied to fewer internalizing (emotional) symptoms in adolescent samples, with several studies indicating that this association is at least partly transmitted via higher self-esteem [10]. By contrast, links with externalizing domains (conduct, hyperactivity/inattention) are more heterogeneous; adolescents with elevated levels of hyperactivity and inattention symptoms frequently report lower self-esteem, but mediation of social support effects through self-esteem is less well established in these domains [11, 12]. Finally, higher prosocial tendencies are associated with greater self-worth and can be strengthened by supportive classroom climates; recognition of prosocial behaviour predicts lower subsequent depressive symptoms [13, 14]. More broadly, positive self-esteem has been linked to better psychological outcomes in adolescents, and both social support and self-esteem may function as protective factors that help maintain psychological balance and resilience in the face of mental health difficulties [15, 16]. These considerations motivate symptom group-level tests of the mediation model alongside analyses of total difficulties.

The present study had two aims. The primary aim was to test whether self-esteem mediates the association between perceived social support and mental health difficulties in the full sample, after adjustment for age, sex, and recruitment source (school vs. clinical). As a secondary exploratory extension, we used hierarchical clustering to characterize heterogeneity in overall psychosocial profiles and to examine whether this pattern appeared more or less detectable across the resulting profile groups. We did not treat clustering as an alternative to the primary mediation model, but rather as a person-centered descriptive approach to exploring heterogeneity.

We expected that greater perceived social support would be associated with fewer mental health difficulties partly through higher self-esteem. We examined possible profile-specific differences in this indirect pathway only as an exploratory secondary question and therefore did not formulate a confirmatory directional hypothesis regarding which profile would show a larger indirect effect.

Methods

Participants

The study was based on an original sample of 397 children and adolescents. The school group was enrolled from elementary, vocational, and high schools in Szeged, Hungary, whereas the clinical group comprised inpatients and outpatients of the Child and Adolescent Psychiatric Unit of the University of Szeged. No diagnosis-based restrictions were applied in the clinical group; youths with any psychiatric diagnosis were eligible, provided that they met the general inclusion criteria. Thus, the clinical sample was intended to reflect the diagnostically heterogeneous population of the unit rather than a disorder-specific subgroup. Inclusion criteria for both groups were being enrolled in fifth grade or above and having normal intelligence. Children below fifth grade or those who had already completed high school were not included. Participation further required parental/legal guardian consent and youth assent/consent. For the present analyses, participants with incomplete Multidimensional Scale of Perceived Social Support (MSPSS), Rosenberg Self-Esteem Scale (RSE), or Strengths and Difficulties Questionnaire (SDQ) data were excluded, resulting in a questionnaire-complete analytic sample of 312 youths (mean age = 14.36 years, SD = 2.19; range = 9–19 years).

Procedure

After approval had been obtained from school principals, parents and children in the school group were informed about the study in writing. Consent was provided by completing the consent form and by providing parental and/or child email addresses. The questionnaire link was then sent by email to those who agreed to participate. The online survey was administered using LimeSurvey, and data were stored on a protected server of the University of Szeged, Hungary. No personal information was collected in the survey. Psychiatric participants received information and consented in person, and completed questionnaires in paper-and-pencil format, whereas school participants completed them online because of COVID-19 restrictions. Questionnaire completion in the school group was supervised by the main researcher, while data collection in the clinical group was overseen by members of the research team. Only youth self-report questionnaire data were used in the present analyses. The study was approved by the Institutional Review Board of the University of Szeged, Hungary (No. 17/2019).

Measures

The MSPSS assesses perceived support from family, friends, and significant others [17]. The RSE measures global self-esteem [7]. The SDQ captures psychological and behavioral difficulties and prosocial strengths [18]. In line with conventional SDQ scoring, reverse-keyed items were rescored according to the manual, and the primary outcome was the 20-item Total Difficulties score, which excludes the Prosocial scale. The self-report version of the SDQ is primarily recommended for respondents aged approximately 11–17 years, depending on literacy and comprehension. In the present sample, the vast majority of participants fell within this age range; however, approximately 1% of the sample were aged 9–10 years. These cases were retained in the analyses due to their negligible proportion and because prior research suggests that SDQ self-reports from children aged 9–10 years are broadly comparable to those obtained from 11-year-olds [19]. Prosocial behavior was examined separately in exploratory subscale analyses. Internal consistencies in the present sample were excellent for MSPSS (alpha = 0.91), good for RSE (alpha = 0.86), acceptable for SDQ Total Difficulties (alpha = 0.78), and lower for the Prosocial scale (alpha = 0.64). Full descriptions are provided in the Supplementary Information.

Statistical analysis

Analyses were conducted in R (version 4.4.2). The primary analysis was a covariate-adjusted mediation model in the full sample, with perceived social support (MSPSS total) as the predictor, self-esteem (RSE total) as the mediator, and SDQ Total Difficulties as the outcome. Age, sex, and recruitment source (school vs. clinical) were included as covariates. Path coefficients and indirect effects were estimated using nonparametric bootstrapping with 5,000 resamples. Percentile 95% confidence intervals and empirical bootstrap p-values were derived from the bootstrap distribution. For the mediation models, interpretation focused primarily on the effect estimates and bootstrap confidence intervals, while bootstrap p-values are also reported for completeness. As a secondary, exploratory analysis, hierarchical clustering (Ward’s method with Euclidean distance) was applied to min–max normalized SDQ, RSE, and MSPSS item profiles [20]. This clustering step was used as an exploratory descriptive follow-up rather than as a pre-specified confirmatory analysis. Between-cluster differences in age and questionnaire total scores were examined using Wilcoxon rank-sum tests, whereas differences in categorical sample characteristics, including sex, recruitment source, presence of any diagnosis, and comorbidity, were examined using chi-square tests. The resulting clusters were labeled post hoc according to their overall psychosocial profiles: Cluster 1 was termed the higher-difficulties/lower-resources profile, whereas Cluster 2 was termed the lower-difficulties/higher-resources profile. We then repeated the same covariate-adjusted mediation model within each cluster, retaining age, sex, and recruitment source as covariates, and conducted exploratory adjusted subscale-level indirect-effect analyses for emotional symptoms, conduct problems, hyperactivity/inattention, peer problems, and prosocial behavior. For these exploratory subscale analyses, empirical bootstrap p-values were additionally adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure [21] within each sample-specific subscale family. Cluster quality was summarized with the average silhouette width, agglomerative coefficient, and cophenetic correlation.

Results

Sample characteristics

The questionnaire-complete analytic sample comprised 312 youths. Full-sample characteristics are summarized in Table 1. The mean age was 14.36 years (SD = 2.19; range = 9–19 years), 117 participants were boys (37.5%), and 195 were girls (62.5%).Table 1Full-sample and cluster-specific sample characteristics

Characteristic Full sample Cluster 1 Cluster 2 p (Cluster 1 vs 2)

N 312 108 204

Age, mean (SD) 14.36 (2.19) 14.68 (1.99) 14.19 (2.27) 0.068

Age range 9.0–19.0 11.0–18.6 9.0–19.0

Boys, n (%) 117 (37.5%) 29 (26.9%) 88 (43.1%) 0.007

Girls, n (%) 195 (62.5%) 79 (73.1%) 116 (56.9%)

School sample, n (%) 197 (63.1%) 34 (31.5%) 163 (79.9%)  < 0.001

Clinical sample, n (%) 115 (36.9%) 74 (68.5%) 41 (20.1%)

Clinical subsample—outpatient, n (%) 35 (30.4%) 19 (25.7%) 16 (39.0%) 0.145

Clinical subsample—inpatient, n (%) 80 (69.6%) 55 (74.3%) 25 (61.0%)

Any diagnosis, n (%) 113 (36.2%) 72 (66.7%) 41 (20.1%)  < 0.001

Comorbidity, n (%) 69 (22.1%) 51 (47.2%) 18 (8.8%)  < 0.001

Characteristics are shown for the full sample and for the two exploratory psychosocial profiles. Cluster 1 represents the higher-difficulties/lower-resources profile, whereas Cluster 2 represents the lower-difficulties/higher-resources profile. P-values refer to comparisons between Cluster 1 and Cluster 2. Percentages are based on the total number of participants in each column, except for outpatient/inpatient rows, where percentages are based on the clinical subsample within that column

Full-sample mediation analysis

The covariate-adjusted mediation results for the full sample are presented in Table 2, and the corresponding model is illustrated in Fig. 1. Greater perceived social support was associated with higher self-esteem (a path: B = 0.26, 95% CI [0.20, 0.31], bootstrapped p < 0.001), and higher self-esteem was associated with fewer mental health difficulties (b path: B = −0.43, 95% CI [−0.50, −0.35], bootstrapped p < 0.001). The direct effect of perceived social support on SDQ Total Difficulties remained statistically detectable (c′ path: B = −0.10, 95% CI [−0.14, −0.06], bootstrapped p < 0.001). The indirect effect through self-esteem was also statistically detectable (a × b: B = −0.11, 95% CI [−0.14, −0.08], bootstrapped p < 0.001), indicating partial mediation.Table 2Covariate-adjusted mediation models predicting SDQ Total Difficulties in the full sample

Path Estimate CI lower CI upper Bootstrap p label Standardized

a 0.2552 0.2002 0.308 0.0004 0.4572

b  − 0.4296  − 0.5046  − 0.3527 0.0004  − 0.5101

c′  − 0.1029  − 0.1433  − 0.0629 0.0004  − 0.219

a × b  − 0.1096  − 0.1431  − 0.0797 0.0004  − 0.2332

total  − 0.2126  − 0.2548  − 0.1729 0.0004  − 0.4522

Perceived social support (MSPSS total) was specified as the predictor, self-esteem (RSE total) as the mediator, and SDQ Total Difficulties as the outcome. All models were adjusted for age, sex, and recruitment source. Path coefficients, percentile 95% bootstrap confidence intervals, and empirical bootstrap p-values were estimated using 5,000 bootstrap resamples. Path labels are defined as follows: a = MSPSS → RSE, b = RSE → SDQ, c′ = direct MSPSS → SDQ effect, a × b   = indirect effect, and total = total effect

In exploratory subscale analyses conducted in the full sample, indirect effects were statistically detectable for emotional symptoms (B = −0.05, 95% CI [−0.07, −0.04], bootstrapped p < 0.001), conduct problems (B = −0.01, 95% CI [−0.02, −0.003], bootstrapped p = 0.005), hyperactivity/inattention (B = −0.03, 95% CI [−0.04, −0.02], bootstrapped p < 0.001), and peer problems (B = −0.02, 95% CI [−0.03, −0.01], bootstrapped p < 0.001), but not for prosocial behavior (B = 0.00, 95% CI [−0.004, 0.01], bootstrapped p = 0.293). Full-sample subscale-level indirect effects are reported in Supplementary Table 1.

Exploratory cluster analysis and cluster-specific results

As a secondary, exploratory analysis, hierarchical clustering of the questionnaire-complete sample (N = 312) identified two psychosocial profile groups: Cluster 1 (n = 108), characterized by higher difficulties and lower psychosocial resources, and Cluster 2 (n = 204), characterized by lower difficulties and higher psychosocial resources. The clustering results are shown in Fig. 2, and the demographic and clinical cluster characteristics are summarized in Table 1, with detailed ICD-10 primary diagnosis frequencies by cluster reported in Supplementary Table 2.

Relative to Cluster 2, Cluster 1 showed substantially higher SDQ Total Difficulties (Mean = 19.42, SD = 4.17 vs. Mean = 10.98, SD = 4.45; p < 0.001), lower self-esteem (RSE: Mean = 10.77, SD = 4.95 vs. Mean = 21.51, SD = 4.77; p < 0.001), and lower perceived social support (MSPSS: Mean = 59.69, SD = 13.83 vs. Mean = 74.24, SD = 8.29; p < 0.001). Cluster 1 contained more clinically referred youths and more diagnosed/comorbid cases, whereas Cluster 2 contained more school-recruited youths (all p-values for source, diagnosis, and comorbidity < 0.001). Sex also differed between clusters (p = 0.007), whereas age did not (p = 0.068). Cluster-quality indices suggested a modest but interpretable two-cluster partition (average silhouette width = 0.17; agglomerative coefficient = 0.91; cophenetic correlation = 0.55).

In the exploratory cluster-specific mediation models, the indirect effect was not statistically detectable in Cluster 1 (B = −0.01, 95% CI [−0.04, 0.00], bootstrapped p = 0.077), whereas it was statistically detectable in Cluster 2 (B = −0.05, 95% CI [−0.09, −0.02], bootstrapped p = 0.003). Detailed cluster-specific adjusted mediation results are presented in Supplementary Fig. 1 and Supplementary Table 3.

At the cluster level, exploratory subscale analyses showed that none of the Cluster 1 indirect effects remained statistically detectable after FDR correction, whereas all five indirect effects remained statistically detectable in Cluster 2. Subscale-level indirect effects across the full sample and clusters are reported in Supplementary Table 1 and illustrated in Fig. 3.

Discussion

The present study examined whether self-esteem mediates the association between perceived social support and youth mental health difficulties in a heterogeneous sample spanning both school-recruited and clinically referred youths. The primary finding was that, in the full sample and after adjustment for age, sex, and recruitment source, higher perceived social support was associated with fewer mental health difficulties partly through higher self-esteem. This pattern is consistent with the view that self-esteem may represent one pathway linking supportive social environments to better mental health in youth [10, 22]. By including both school-recruited and clinically referred youths, the present study may help address a gap in a literature that has often relied on higher-functioning samples.

The full-sample subscale analyses suggested that the indirect pathway through self-esteem was most evident for emotional symptoms, and to a lesser extent for hyperactivity/inattention, peer problems, and conduct problems. This pattern is broadly compatible with the idea that self-evaluative processes may be especially relevant for distress-related symptom domains [6, 10]. At the same time, the more modest findings for hyperactivity/inattention and conduct problems are in line with previous work suggesting that the role of self-esteem may be less uniform across externalizing-related domains [11, 12].

The cluster-specific findings are best understood in relation to two broad psychosocial profiles: a higher-difficulties/lower-resources profile (Cluster 1) and a lower-difficulties/higher-resources profile (Cluster 2). At the exploratory cluster level, the subscale pattern was more heterogeneous. After FDR correction, none of the Cluster 1 subscale indirect effects remained statistically detectable, whereas statistically detectable indirect effects were observed across all five subscales in Cluster 2. However, these subgroup findings should be interpreted cautiously. The clusters were derived from the same psychosocial variables that were later examined in the mediation models, therefore apparent between-cluster differences may partly reflect statistical properties of the clustering procedure, including reduced within-cluster variance and range restriction, rather than fully distinct mechanisms.

This interpretation is consistent with previous work suggesting that perceived social support and self-esteem are broadly protective, but that their associations with mental health may vary across symptom domains and levels of clinical complexity [3, 16, 23, 24]. In the lower-difficulties/higher-resources profile, the indirect pathway through self-esteem was more consistently detectable, which is in line with studies suggesting that supportive relationships may be linked to fewer mental health difficulties partly through more positive self-evaluations [10, 22]. By contrast, in the higher-difficulties/lower-resources profile, weaker or less consistent indirect effects may reflect reduced within-cluster variability, range restriction, and greater diagnostic and clinical heterogeneity. This is also compatible with the broader literature indicating that self-esteem and social support are relevant but non-specific protective factors, whose effects may be less uniform in youths with more complex or severe clinical presentations [3, 12, 15, 23]. In more symptomatic youths, self-esteem may therefore remain clinically relevant, but it is likely to operate alongside additional disorder-specific, family-level, neurodevelopmental, and symptom-focused mechanisms.

Clinically, these findings suggest that self-esteem may be one relevant pathway through which supportive relationships are associated with better mental health in youth. From a prevention and intervention perspective, this is important because both perceived support and self-esteem are potentially modifiable targets. Social support can be mobilized through family-, peer-, school-, mentoring-, and group-based approaches, and systematic review evidence suggests that such approaches may contribute to improved child and adolescent mental health when they enhance supportive relationships and connectedness [3, 25]. These implications are also consistent with classic stress-buffering and main-effect models of social support: supportive relationships may reduce the impact of stressful experiences, while also promoting well-being more generally, independent of current stress exposure [3]. In the higher-difficulties/lower-resources profile, this may be particularly relevant as part of a broader clinical approach, because strengthening family, peer, and school-based support may provide a relational foundation for more direct symptom-focused treatment. In the lower-difficulties/higher-resources profile, combined efforts to maintain supportive relationships and strengthen positive self-evaluations may have preventive value by reinforcing existing psychosocial resources. School-based social and emotional learning programmes may also improve social-emotional skills, attitudes, behaviour, and academic outcomes, providing a relevant preventive context for strengthening interpersonal resources [26]. Self-esteem-focused interventions have similarly shown beneficial effects in children and adolescents, particularly when self-esteem or self-concept is an explicit intervention target [27, 28]. In clinical settings, self-esteem may be addressed through psychoeducation, cognitive-behavioural techniques, structured group interventions, and broader mental health promotion approaches [23, 29]. However, the weaker exploratory cluster-specific findings in the higher-difficulties/lower-resources group suggest that self-esteem-focused work alone is unlikely to be sufficient for more symptomatic youths. For these individuals, interventions may need to combine relationship-based support, self-evaluative work, and more direct symptom-focused clinical treatment.

Limitations

This study has several limitations. First, the data were cross-sectional, and therefore, the mediation models cannot establish temporal ordering or causality. The tested direction from perceived social support to self-esteem to mental health difficulties was theory-informed rather than causal. Second, the cluster-specific mediation analyses were exploratory, were not intended as pre-specified confirmatory tests, and were vulnerable to circularity because the clusters were derived from the same variables later entered into the mediation models. Finally, the prosocial scale showed lower internal consistency than the other measures, which also argues for caution in interpreting prosocial findings.

Conclusion

In summary, self-esteem appears to be one pathway linking perceived social support to fewer mental health difficulties in youths from school and clinical settings. In the present study, the full-sample adjusted model provides the most robust evidence for this pathway, whereas the cluster-specific results are best understood as exploratory indications that its detectability may vary across psychosocial profiles.

Supplementary Information

References

1.Smith C Carlson BE Stress, coping, and resilience in children and youth Soc Serv Rev 1997 71 2 231 256 10.1086/604249 Smith C, Carlson BE. Stress, coping, and resilience in children and youth. Soc Serv Rev. 1997;71(2):231–56. 10.1086/604249 .

2.Kong F You X Loneliness and self-esteem as mediators between social support and life satisfaction in late adolescence Soc Indic Res 2013 110 1 271 279 10.1007/S11205-011-9930-6 Kong F, You X. Loneliness and self-esteem as mediators between social support and life satisfaction in late adolescence. Soc Indic Res. 2013;110(1):271–9. 10.1007/S11205-011-9930-6 .

3.Bauer A Stevens M Purtscheller D Knapp M Fonagy P Evans-Lacko S Paul J Mobilising social support to improve mental health for children and adolescents: a systematic review using principles of realist synthesis PLoS ONE 2021 10.1371/JOURNAL.PONE.0251750 34015021 PMC8136658 Bauer A, Stevens M, Purtscheller D, Knapp M, Fonagy P, Evans-Lacko S, et al. Mobilising social support to improve mental health for children and adolescents: a systematic review using principles of realist synthesis. PLoS ONE. 2021. 10.1371/JOURNAL.PONE.0251750 .PMC8136658 34015021

4.Ahmed W Minnaert A van der Werf G Kuyper H Perceived social support and early adolescents’ achievement: the mediational roles of motivational beliefs and emotions J Youth Adolesc 2010 39 1 36 46 10.1007/S10964-008-9367-7 20091215 PMC2796962 Ahmed W, Minnaert A, van der Werf G, Kuyper H. Perceived social support and early adolescents’ achievement: the mediational roles of motivational beliefs and emotions. J Youth Adolesc. 2010;39(1):36–46. 10.1007/S10964-008-9367-7 .PMC2796962 20091215

5.Teo AR Choi HJ Valenstein M Social relationships and depression: ten-year follow-up from a nationally representative study PLoS ONE 2013 10.1371/JOURNAL.PONE.0062396 23646128 PMC3640036 Teo AR, Choi HJ, Valenstein M. Social relationships and depression: ten-year follow-up from a nationally representative study. PLoS ONE. 2013. 10.1371/JOURNAL.PONE.0062396 .PMC3640036 23646128

6.Brown JD Marshall MA Self-esteem and emotion: some thoughts about feelings Pers Soc Psychol Bull 2001 27 5 575 584 10.1177/0146167201275006 Brown JD, Marshall MA. Self-esteem and emotion: some thoughts about feelings. Pers Soc Psychol Bull. 2001;27(5):575–84. 10.1177/0146167201275006 .

7.Rosenberg M Schooler C Schoenbach C Self-esteem and adolescent problems: modeling reciprocal effects Am Sociol Rev 1989 54 6 1004 10.2307/2095720 Rosenberg M, Schooler C, Schoenbach C. Self-esteem and adolescent problems: modeling reciprocal effects. Am Sociol Rev. 1989;54(6):1004. 10.2307/2095720 .

8.Pan Y Chan MM Liu G Li S Deng X The longitudinal association between self-esteem and social relationships among Chinese adolescents: roles of developmental stages and gender J Happiness Stud 2024 10.1007/S10902-024-00802-2 Pan Y, Chan MM, Liu G, Li S, Deng X. The longitudinal association between self-esteem and social relationships among Chinese adolescents: roles of developmental stages and gender. J Happiness Stud. 2024. 10.1007/S10902-024-00802-2 .

9.Wagner J Brandt ND Bien K Bombik M The longitudinal interplay of self-esteem, social relationships, and academic achievement during adolescence: theoretical notions and bivariate meta-analytic findings Z Erziehungswiss 2024 27 1 39 61 10.1007/S11618-023-01177-5/TABLES/2 Wagner J, Brandt ND, Bien K, Bombik M. The longitudinal interplay of self-esteem, social relationships, and academic achievement during adolescence: theoretical notions and bivariate meta-analytic findings. Z Erziehungswiss. 2024;27(1):39–61. 10.1007/S11618-023-01177-5/TABLES/2 .

10.Chang CW Yuan R Chen JK Social support and depression among Chinese adolescents: the mediating roles of self-esteem and self-efficacy Child Youth Serv Rev 2018 88 128 134 10.1016/J.CHILDYOUTH.2018.03.001 Chang CW, Yuan R, Chen JK. Social support and depression among Chinese adolescents: the mediating roles of self-esteem and self-efficacy. Child Youth Serv Rev. 2018;88:128–34. 10.1016/J.CHILDYOUTH.2018.03.001 .

11.Mazzone L Postorino V Reale L Guarnera M Mannino V Armando M Fatta L De Peppo L Vicari S Self-esteem evaluation in children and adolescents suffering from ADHD Clin Pract Epidemiol Ment Health 2013 9 1 96 102 10.2174/1745017901309010096 23878614 PMC3715757 Mazzone L, Postorino V, Reale L, Guarnera M, Mannino V, Armando M, et al. Self-esteem evaluation in children and adolescents suffering from ADHD. Clin Pract Epidemiol Ment Health. 2013;9(1):96–102. 10.2174/1745017901309010096 .PMC3715757 23878614

12.Wells AE Hunnikin LM Ash DP van Goozen SHM Low self-esteem and impairments in emotion recognition predict behavioural problems in children J Psychopathol Behav Assess 2020 42 4 693 701 10.1007/S10862-020-09814-7/TABLES/3 Wells AE, Hunnikin LM, Ash DP, van Goozen SHM. Low self-esteem and impairments in emotion recognition predict behavioural problems in children. J Psychopathol Behav Assess. 2020;42(4):693–701. 10.1007/S10862-020-09814-7/TABLES/3 .

13.Ma C Mastrotheodoros S Lan X Linking classmate autonomy support with prosocial behavior in Chinese left-behind adolescents: the moderating role of self-esteem and grit Pers Individ Dif 2022 195 111679 10.1016/J.PAID.2022.111679 Ma C, Mastrotheodoros S, Lan X. Linking classmate autonomy support with prosocial behavior in Chinese left-behind adolescents: the moderating role of self-esteem and grit. Pers Individ Dif. 2022;195:111679. 10.1016/J.PAID.2022.111679 .

14.Nagaoka D Tomoshige N Ando S Morita M Kiyono T Kanata S Fujikawa S Endo K Yamasaki S Fukuda M Nishida A Hiraiwa-Hasegawa M Kasai K Being praised for prosocial behaviors longitudinally reduces depressive symptoms in early adolescents: a population-based cohort study Front Psychiatry 2022 13 865907 10.3389/FPSYT.2022.865907 35656347 PMC9152118 Nagaoka D, Tomoshige N, Ando S, Morita M, Kiyono T, Kanata S, et al. Being praised for prosocial behaviors longitudinally reduces depressive symptoms in early adolescents: a population-based cohort study. Front Psychiatry. 2022;13:865907. 10.3389/FPSYT.2022.865907 .PMC9152118 35656347

15.Arsandaux J Galéra C Salamon R The association of self-esteem and psychosocial outcomes in young adults: a 10-year prospective study Child Adolesc Ment Health 2021 26 2 106 113 10.1111/CAMH.12392 32400105 Arsandaux J, Galéra C, Salamon R. The association of self-esteem and psychosocial outcomes in young adults: a 10-year prospective study. Child Adolesc Ment Health. 2021;26(2):106–13. 10.1111/CAMH.12392 .32400105

16.Liu Q Jiang M Li S Yang Y Social support, resilience, and self-esteem protect against common mental health problems in early adolescence: a nonrecursive analysis from a two-year longitudinal study Medicine (Baltimore) 2021 100 4 e24334 10.1097/MD.0000000000024334 33530225 PMC7850671 Liu Q, Jiang M, Li S, Yang Y. Social support, resilience, and self-esteem protect against common mental health problems in early adolescence: a nonrecursive analysis from a two-year longitudinal study. Medicine (Baltimore). 2021;100(4):e24334. 10.1097/MD.0000000000024334 .PMC7850671 33530225

17.Zimet GD Dahlem NW Zimet SG Farley GK The multidimensional scale of perceived social support J Pers Assess 1988 52 1 30 41 10.1207/S15327752JPA5201_2 2280326 Zimet GD, Dahlem NW, Zimet SG, Farley GK. The multidimensional scale of perceived social support. J Pers Assess. 1988;52(1):30–41. 10.1207/S15327752JPA5201_2 .2280326

18.Goodman R The strengths and difficulties questionnaire: a research note J Child Psychol Psychiatry Allied Discip 1997 38 5 581 586 10.1111/J.1469-7610.1997.TB01545.X 9255702 Goodman R. The strengths and difficulties questionnaire: a research note. J Child Psychol Psychiatry Allied Discip. 1997;38(5):581–6. 10.1111/J.1469-7610.1997.TB01545.X .9255702

19.Hobbs MJ Laurens KR Psychometric comparability of self-report by children aged 9–10 versus 11 years on the strengths and difficulties questionnaire (SDQ) Child Indic Res 2020 13 1 301 318 10.1007/S12187-019-09633-7/METRICS Hobbs MJ, Laurens KR. Psychometric comparability of self-report by children aged 9–10 versus 11 years on the strengths and difficulties questionnaire (SDQ). Child Indic Res. 2020;13(1):301–18. 10.1007/S12187-019-09633-7/METRICS .

20.Murtagh F Legendre P Ward’s hierarchical agglomerative clustering method: which algorithms implement ward’s criterion? J Classif 2014 31 3 274 295 10.1007/S00357-014-9161-Z/METRICS Murtagh F, Legendre P. Ward’s hierarchical agglomerative clustering method: which algorithms implement ward’s criterion? J Classif. 2014;31(3):274–95. 10.1007/S00357-014-9161-Z/METRICS .

21.Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing J Roy Stat Soc Ser B (Methodol) 1995 57 1 289 300 10.1111/J.2517-6161.1995.TB02031.X Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc Ser B (Methodol). 1995;57(1):289–300. 10.1111/J.2517-6161.1995.TB02031.X .

22.Poudel A Gurung B Khanal GP Perceived social support and psychological wellbeing among Nepalese adolescents: the mediating role of self-esteem BMC Psychol 2020 10.1186/S40359-020-00409-1 32357919 PMC7195718 Poudel A, Gurung B, Khanal GP. Perceived social support and psychological wellbeing among Nepalese adolescents: the mediating role of self-esteem. BMC Psychol. 2020. 10.1186/S40359-020-00409-1 .PMC7195718 32357919

23.Mann M Hosman CMH Schaalma HP De Vries NK Self-esteem in a broad-spectrum approach for mental health promotion Health Educ Res 2004 19 4 357 372 10.1093/HER/CYG041 15199011 Mann M, Hosman CMH, Schaalma HP, De Vries NK. Self-esteem in a broad-spectrum approach for mental health promotion. Health Educ Res. 2004;19(4):357–72. 10.1093/HER/CYG041 .15199011

24.Sowislo JF Orth U Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies Psychol Bull 2013 139 1 213 240 10.1037/A0028931 22730921 Sowislo JF, Orth U. Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies. Psychol Bull. 2013;139(1):213–40. 10.1037/A0028931 .22730921

25.Das JK Salam RA Lassi Z Khan MN Mahmood W Patel V Bhutta Z Interventions for adolescent mental health: an overview of systematic reviews J Adolesc Health 2016 39 9 754 762 10.1016/J.JADOHEALTH.2016.06.020 PMC5026677 27664596 Das JK, Salam RA, Lassi Z, Khan MN, Mahmood W, Patel V, et al. Interventions for adolescent mental health: an overview of systematic reviews. J Adolesc Health. 2016;39(9):754–62. 10.1016/J.JADOHEALTH.2016.06.020 .PMC5026677 27664596

26.Durlak JA Weissberg RP Dymnicki AB Taylor RD Schellinger KB The impact of enhancing students’ social and emotional learning: a meta-analysis of school-based universal interventions Child Dev 2011 82 1 405 432 10.1111/J.1467-8624.2010.01564.X 21291449 Durlak JA, Weissberg RP, Dymnicki AB, Taylor RD, Schellinger KB. The impact of enhancing students’ social and emotional learning: a meta-analysis of school-based universal interventions. Child Dev. 2011;82(1):405–32. 10.1111/J.1467-8624.2010.01564.X .21291449

27.Haney P Durlak JA Changing self-esteem in children and adolescents: a meta-analytic review J Clin Child Psychol 1998 27 4 423 433 10.1207/S15374424JCCP2704_6 9866079 Haney P, Durlak JA. Changing self-esteem in children and adolescents: a meta-analytic review. J Clin Child Psychol. 1998;27(4):423–33. 10.1207/S15374424JCCP2704_6 .9866079

28.O’Mara AJ Marsh HW Craven RG Debus RL Do self-concept interventions make a difference? A synergistic blend of construct validation and meta-analysis Educ Psychol 2006 41 3 181 206 10.1207/S15326985EP4103_4 O’Mara AJ, Marsh HW, Craven RG, Debus RL. Do self-concept interventions make a difference? A synergistic blend of construct validation and meta-analysis. Educ Psychol. 2006;41(3):181–206. 10.1207/S15326985EP4103_4 .

29.Biney H Giles E Hutt M Matthews R Lacey JH Self-esteem as a catalyst for change in adolescent inpatients with anorexia nervosa: a pilot randomised controlled trial Eat Weight Disord 2022 27 1 189 198 10.1007/S40519-021-01161-0 33713335 PMC8860801 Biney H, Giles E, Hutt M, Matthews R, Lacey JH. Self-esteem as a catalyst for change in adolescent inpatients with anorexia nervosa: a pilot randomised controlled trial. Eat Weight Disord. 2022;27(1):189–98. 10.1007/S40519-021-01161-0 .PMC8860801 33713335