Social inequalities in adolescent depression: The role of parental social support and optimism

Abstract

Aims:

Interpersonal theory suggests relationships between socio-economic status (SES) and adolescent psychopathology mediated by negative parenting. This study examines the role of perceived parental social support and optimism in understanding adolescents’ depression and self-rated health among a sample of Hungarian youth.

Methods:

Using a self-administered questionnaire, data (N = 881) were collected from high-school students (14–20 years old) in Szeged, Hungary (a regional centre in the southeastern region, near to the Serbian border, with a population of 170,000 inhabitants). To analyse the overall structure of the relationship between objective/subjective SES, parental support, optimism and health outcomes (depression, self-perceived health), structural equation modelling (SEM) was employed.

Results:

Findings suggest the following: (1) SES variables generate social inequalities in adolescent depression through parental social support, particularly maternal support; and (2) parents provide youths with different levels of social support that in turn may strengthen or weaken optimism during the socialization process.

Conclusions:

In addressing depression prevention and treatment, we may want to take into account socio-economic differences in social networks and levels of optimism, which may influence youths’ psychosocial adjustment and development of psychopathology.

Keywords

Social inequalities adolescent depression self-perceived health social support optimism

Introduction

While a growing body of empirical research has demonstrated the importance of examining the relationship between social inequalities and physical and mental health (Sekine et al., 2011), there are few significant socio-economic differences in adolescents’ health in terms of mortality and morbidity (Halldórsson, Kunst, Köhler & Mackenbach, 2000). This relative ‘socio-economic status (SES) equality’ might be explained by features of socialization (e.g. a growing independence from parental influences) (West & Sweeting, 2004). Despite this equalization, there may be differences in those factors preceding socio-economic health differences in adult life, such as stressful life events and mental health problems (Piko & Fitzpatrick, 2007). Moreover, childhood socio-economic conditions may initiate lifelong cumulative disadvantages (Hatch, 2005). Particularly subjective SES, parental education and parents’ unemployment play an important role in determining depression in youths (Kaltiala-Heino, Rimpalä, Rantanen & Laippala, 2001).

Depression is the most common clinical disorder during adolescence; in addition, between 20% and 50% of adolescents report significant, subsyndromal levels of depression (Kessler, Avenevoli & Merikangas, 2001) that also contribute to lower self-perceived health among adolescents (Piko, 2007). Since adolescence is a sensitive developmental period of rapid change and adjustment to socialization tasks, we need to know more about SES-based inequalities in depression in light of these processes (Wickrama, Noh & Elder, 2009).

As a recent review paper argues, the different explanations for social inequalities in health are related to each other through the mechanism of socialization that involves the intergenerational transfer of attitudes, beliefs and behaviours relevant to health (Siahpush & Singh, 2000). Adolescents’ family relationships may be both a risk and protective factor for the development of depression. Low level of family cohesion, parental commitments and social support, and frequent conflicts are associated with adolescent depression (Kaltiala-Heino et al., 2001). Social support (particularly perceived support) as a developmental asset can act as a strong protective factor against depression, and can increase well-being and resilience (Hankin, 2006; Piko, 1998; Piko, Kovacs & Fitzpatrick, 2009).

Social support, however, is not independent of SES (Kerr, Preuss & King, 2006); for example, parents from a higher social class engage more in conversation with their children (Kroenke, 2008) or they are more likely to provide secure attachment (Sund & Wichstrøm, 2002). As a review paper suggests, SES predicts negative parenting, which in turn predicts depression in a linear fashion (Grant et al., 2003). That is, the link between SES and adolescent psychopathology may be mediated by negative parenting. Among the possible mediators between social support and adolescent depression, control beliefs and optimism have been noted (Herman-Stahl & Petersen, 1999) as potentially important to contributing to social inequalities in health (Taylor & Seeman, 1999). Lower optimism, on the other hand, is a strong predictor of depressive symptoms over time among youths (Wong & Lim, 2009). Adolescents’ beliefs about their ability to handle situations and maintain their expectations for positive outcomes (that is, dispositional optimism) are also formed during the process of socialization (Herman-Stahl & Petersen, 1999). Fewer studies have explored the optimism–health link in adolescence or its relationship to health disparities, although youths with lower optimism have been shown to be less optimistic (Khullar, Oreskovic, Perrin & Goodman, 2011). Overall, as a meta-analysis suggests, path analysis would be useful for developing conceptual models in which parenting mediates between parents’ SES and adolescents’ psychological symptoms through cognitive attributions, such as optimism (Grant et al., 2003).

Based on related previous literature, one of the central goals of the present paper is to examine the mediating role of perceived parental social support and optimism in the relationship between SES and adolescent depression and self-perceived health among a sample of Hungarian high-school students. Overall, we hypothesize that social inequalities in adolescent health, in part, stem from differences in social support provided by the parents that can either strengthen or weaken their child’s optimistic attitudes by acting as a mediator. We anticipate that subjective SES plays a direct role in adolescents’ social support and health, whereas objective SES indicators play indirect roles through subjective SES (that is, SES self-assessment) in the final path model.

Method

Data were collected in the spring of 2008. Five schools were chosen randomly from a list of all high schools in Szeged. These high-school classes were chosen randomly from a sample of all classes in the population of all high schools. This sample, based on a stratified data collection, consisted of 881 students (14–20 years of age) attending high school (grades 1–5) in Szeged (a regional centre in the southeastern region of Hungary near to the Serbian border, with a population of 170,000 inhabitants).¹ Of the sampled students, 44.6% were female and the median age of the sample was 16 years (M = 16.6, SD = 1.3). Of the 900 questionnaires distributed, 881 were returned, without incentives to participate in the study (response rate was 97.8%). Students completed the questionnaires themselves during the class period (which took about 40 minutes to complete). Those students who did not participate were either absent or youths whose parents did not give them permission to participate in the study. Student participation was voluntary and confidentiality was emphasized, noting that the data were being collected for research purposes only. Trained graduate students distributed the questionnaires in each class using a standardized procedure. Parents were informed about the study and their consent was obtained prior to data collection.

Self-perceived health, as a global health indicator, was measured by asking students how they compared their health status to that of their peers. The responses to the question were: (1) poor; (2) fair; (3) good; and (4) excellent (Piko & Fitzpatrick, 2007).

Depressive symptomatology was measured by a validated shortened version of the original 27-item Children’s Depression Inventory (CDI), which is a self-rated depressive symptom scale for children adapted from the Beck Depression Inventory for adults (Kovacs, 1992). Each item assesses a single symptom, such as sadness, and was coded from 0 to 2. The scale, based on the current data, was reliable with a Cronbach’s α of .74. We weighted the shortened CDI by a factor of 3.375 (number of original CDI items 27/shortened version items 8 = 3.375) for comparison purpose (Piko et al., 2009).

In applying SES in the study, we focused on the multidimensionality of SES. This means that both ‘objective’ and ‘subjective’ social status measures were used giving priority to the latter (Piko & Fitzpatrick, 2007). The objective social status measures were based on employment (occupational) status and the educational level (schooling) of the student’s parents. Since social structure is based primarily on a dual-earning system, both father’s and mother’s educational level and employment status were measured. Employment status was divided into five categories: (1) professional and managerial; (2) skilled non-manual; (3) skilled manual; (4) unskilled manual; and (5) unemployed (including housewife for mothers). The last category was added based on the consideration of previous studies: these suggested that the employment status was linearly related to health outcomes including unemployed/housewife categories (Federico, Falese & Capelli, 2009; Kaltiala-Heino et al., 2001). A four-level classification of education was used to measure father’s and mother’s schooling: (1) primary education; (2) apprenticeship; (3) General Certificate of Education (i.e. high-school level); and (4) university or college degree. In additon, the subjective social status (SES self-assessment) was measured by the following question: ‘How would you rate your family’s socioeconomic status?’ The response categories were: (1) lower; (2) lower-middle; (3) middle; (4) upper-middle; and (5) upper class. As previous studies indicated, SES self-assessment seemed to be the most significant and consistent factor in generating social inequalities in adolescent health (Piko & Fitzpatrick, 2007; Siahpush & Singh, 2000).

The Measures of Perceived Social Support were used to assess the level of satisfaction with the support experienced by youths as provided by their parents. Each sub-scale (one for social support from mother and one for social support from father) contained six items, for example ‘I often feel really appreciated by my father/mother’. Responses were based on the following categories: 1 = not at all like my experience; 2 = somewhat like my experience; 3 = much like my experience; and 4 = very much like my experience. The final perceived social support scales were coded from 6 to 24 and reliable with Cronbach’s α coefficients of .92 (father support) and .91 (mother support) (Turner & Marino, 1994).

Dispositional optimism was measured using the eight items of the Hungarian version of the Life Orientation Test (Scheier & Carver, 1985). Students were asked to indicate their degree of agreement with statements such as ‘In uncertain times, I usually expect the best’ or ‘I hardly ever expect things to go my way’ (reverse item). A five-point response scale was used ranging from 0 = strongly disagree to 4 = strongly agree (except for four reverse-coded items). This scale was reliable with a Cronbach’s α of .70.

The analysis begins with an examination of the descriptive statistics. To analyse the overall structure of the relationship between objective/subjective SES, parental support, optimism and health outcomes, structural equation modelling (SEM) with bootstrapping, maximum likelihood estimation was employed. A detailed description of this analysis can be found in Appendix 1.

Results

Table 1 presents the detailed socio-economic, psychosocial and health profiles of Hungarian high-school students in the sample. The mean score of depressive symptomatology for the sample was 8.1 (SD = 8.0, range: 0–47), that of optimism was 28.1 (SD = 5.5, range: 9–40). According to the social support scales, students received more support from mothers (M = 19.8, SD = 4.3) compared to support from their fathers (M = 16.7, SD = 5.2) (p < .001). Nearly half of them perceived their own health as good (45.5%). Most of the students considered themselves middle class (63.1%), 2.8% reported being lower class and only 1.7% belonged to the upper (elite) class. Most of the students’ parents were skilled manual labourers and reported having completed some apprenticeship or high school.

Table 1.

Descriptive statistics for study variables.

Variables	n (%) M (SD)
Depressive symptomatology (CDI) (score: 0–47)	− 8.1 (8.0)
Optimism (LOT) (score: 9–40)	− 28.1 (5.5)
Social support from father (score: 6–24)	− 16.7 (5.2)
Social support from mother (score: 6–24)	− 19.8 (4.3)
Self-perceived health
Excellent	180 (20.4)
Good	401 (45.5)
Fair	254 (28.8)
Poor	33 (3.7)
Father’s schooling
Primary schooling	53 (6.0)
Apprenticeship	410 (46.5)
High-school level	242 (27.5)
College/university degree	146 (16.6)
Mother’s schooling
Primary schooling	79 (9.0)
Apprenticeship	337 (38.3)
High-school level	266 (30.2)
College/university degree	190 (21.6)
Father’s employment status
Professional/managerial	140 (15.9)
Other non-manual	90 (11.1)
Skilled manual	500 (56.8)
Unskilled manual	64 (7.3)
Unemployed	64 (7.3)
Mother’s employment status
Professional/managerial	212 (13.7)
Other non-manual	179 (20.3)
Skilled manual	310 (35.2)
Unskilled manual	41 (4.6)
Unemployed	216 (24.5)
SES self-assessment
Lower class	25 (2.8)
Lower-middle class	131 (14.9)
Middle class	556 (63.1)
Upper-middle class	143 (16.2)
Upper class	15 (1.7)

Note: Percentages do not add up to 100% in most cases due to missing values.

Confirmatory factor analysis (CFA) models of indicator variables representing social resources (SES variables, SES self-assessment and social support) and psychological health (depression and self-rated health) was conducted to test the validity of the measurement model. In both cases the latent variables representing respective constructs were allowed to correlate, whereas measurement errors were not related. Questionnaire items (as discussed in the Method section) loaded only one respective latent variable. CFA for social resources showed acceptable model-data fit, with χ² (d.f. = 60, N = 848) = 356.62, p < .001, NFI = .93, TLI = .92, CFI = .94, RMSEA = .07 (90% CI = .06–.08). Factor loadings for observed indicators of latent variables were: .85, .82, .85, .78 for maternal support; .73, .75, .86, .83 for paternal support; .58, .63, .66, .76 for objective SES (p < .001 for all loadings). All latent variables were significantly related (correlation coefficients ranging from .13 to .26; p < .001 for all), except for the association among maternal support and objective SES (.06, p = .127). CFA for psychosocial health variables indicated that data fitted the model well, with χ² (d.f. = 5, N = 848) = 7.17, p = .208, NFI = .97, TLI = .98, CFI = .99, RMSEA = .02 (90% CI = .00–.05). Factor loadings for observed indicators of latent variables were .40, .46, .55 and .40 for depression (p < .001 for all loadings). Both latent variables were significantly related (correlation coefficient: −.25, p < .001). Since a comparison of the hypothesized model (no constrained paths) with a model assuming that the paths from SES self-assessment to maternal support are constrained and equal in both gender groups showed that the model assuming paths of equal values for both groups did not differ significantly from the hypothesized model, it was assumed that the gender effects are negligible and further analyses were conducted with combining the gender groups.

SEM for the hypothesized model resulted in obtaining good fit indices, χ² (d.f. = 195, N = 848) = 569.04, p < .001, NFI = .90, TLI = .92, CFI = .93, RMSEA = .047 (90% CI = .04–.05). In all cases observed variables loaded significantly and the latent variables had the following factor loading ranges: .58–.76 for objective SES; .78–.85 for maternal support; .74–.86 for paternal support; .40–.54 for optimism;.40–.54 for depression. Objective SES was related to subjective SES (.37, p < .001) and paternal support (.09, p = .05), but unrelated to maternal support (.03, p =.575). Subjective SES was in turn related to support from father (.15, p < .001) and from mother (.11, p = .006), as well as optimism (.09, p = .049). Paternal support was related to higher optimism (.30, p < .001) and depression (−.12, p = .044), but unrelated to self-perceived health (.07, p = .083). Maternal support was related to higher optimism (.26, p < .001) and to lower depression (−.18, p = .002), but unrelated to self-perceived health (−.01, p =.877). Finally, optimism was related to better self-perceived health (.39, p < .001) and lower depression (−.53, p < .001). Residuals of parental and maternal support covaried significantly (.22, p < .001); significant covariance was also found for depression and self-perceived health (−.16, p = .025). Fit indices and results of comparisons between the hypothesized model (assuming that there are effects of the predictors on the outcome variables) and alternative models are presented in Table 2.

Table 2.

Fit indices and results of comparisons between the hypothesized model (assuming an effect of the predictor on the outcome variable) and alternative models.

Model	Fit indices					Comparison with the hypothesized model: Δ χ² (d.f., n) p ^a
	χ² (d.f., n) p	NFI	TLI	CFI	RMSEA
Alternative Model 1: Direct effect of objective SES on parental support constrained to zero	572.49 (197, 848) p < .001	.90	.93	.93	.047	3.45 (2, 881) p = 0.178
Alternative Model 2: Direct effect of subjective SES on optimism constrained to zero	572.89 (196, 881) p < .001	.90	.93	.93	.047	3.86 (1, 881) p = .049
Alternative Model 3: Effects of parental support on depression and on self-perceived health constrained to zero	582.75 (199, 881) p < .001	.90	.93	.93	.047	13.71 (4, 881) p = .008

A significant χ² indicates that the alternative model presents a significantly worse fit than the hypothesized model.

Among the alternative models only one had good fit indices and did not differ significantly from the hypothesized model (Figure 1). This model, with direct effects of objective SES on parental support constrained to zero, should be accepted as the final model. As indicated in Table 2 this model is more parsimonious and fits the data as well as the more complex, hypothesized model. In particular, fit indices for this final model were χ² (d.f. = 197, N = 848) = 572.49, p < .001, NFI = .90, TLI = .93, CFI = .93, RMSEA = .047 (90% CI = .04–.05). The path coefficients for the final model are presented in Figure 2.

Figure 1.

Hypothesized model: Relationships between objective SES, subjective SES, parental support, optimism, depression and self-perceived health.

Figure 2.

Results of structural equation modelling for the final model: Relationships between objective SES, subjective SES, parental support, optimism, depression and self-perceived health.

Higher objective SES was related to higher perceived SES, which in turn predicted higher paternal and maternal support. Higher parental support indices were related to higher optimism. Higher maternal (and also paternal, but in smaller degree) support was also related to lower depression. Higher optimism was related to better self-perceived health and lower depression. In all cases observed variables loaded respective latent variables significantly with the following ranges of factor loadings: .58–.76 for objective SES; .78–.85 for maternal support; .74–.86 for paternal support; .40–.50 for optimism;.40–.54 for depression. Standardized indirect effects of objective SES were .03 for self-perceived health and −.02 for depression. Standardized indirect effects for subjective SES were .09 for self-perceived effects and –.06 for depression. The variables in the model predicted 45% of the variation in depressive symptoms and 18% of the variation in self-perceived health (Figure 2).

Additional analyses, aimed at testing the mediation effects, were performed using the Sobel test. Subjective SES was found to mediate the relationship between objective SES and paternal support (Z = 4.31, p < .001) and the relationship between objective SES and maternal support (Z = 3.11, p < .001). Paternal support mediated the relationship between subjective SES and optimism (Z = 3.46, p < .001). Maternal support was also a significant mediator between subjective SES and optimism (Z = 2.66, p = .004). Optimism was a significant mediator in the associations between paternal support and depression (Z = −3.45, p < .001) and self-reported health (Z = 2.19, p = .014). Finally, optimism mediated the relationship between maternal support and depression (Z = −3.30, p < .001) and the relationship between maternal support and self-reported health (Z = 2.15, p = .016). In conclusion, objective SES directly predicted subjective SES. Subjective SES was related to social support, which in turn mediated the effects of subjective SES on optimism. Social support from mother and father was indirectly related to adolescents’ depression and perceived health, and its effects were mediated by optimism.

Discussion

The focus of the analyses was to attempt to explore the role of perceived parental social support and optimism in explicating the relationship between adolescent depression and self-rated health as health outcomes and SES among a sample of Hungarian high-school students. We were aware that parental social support, as a developmental asset, might serve as a protective factor for adolescent health and psychosocial adjustment (Hankin, 2006; Kaltiala-Heino et al., 2001; Wong & Lim, 2009). Likewise, optimism may also serve as a protection for maintaining adolescent resilience (Wong & Lim, 2009). However, previous studies also suggested that there were SES differences in levels of social support received (Kerr et al., 2006). In addition, optimism is also influenced by parenting style and parents’ SES (Herman-Stahl & Petersen, 1999). Therefore, we proposed a path analysis applying a possible conceptual model of these interrelationships based on a previous meta-analysis (Grant et al., 2003).

Subjective social status (which is strongly influenced by objective SES variables) was found to be associated with both paternal and maternal support. In addition, parental social support was (negatively) related to depression. This may be explained, in part, by previous findings that parents from higher social classes tend to talk/interact more with their children and also provide more rational/material support (Kroenke, 2008). As retained SEM indicated, maternal support (which may be more closely linked to emotional support) was more implicitly related to depression than paternal support, as a previous study had also indicated (Piko, 1998).

As the results suggest, those who evaluated themselves as belonging to higher social classes also received more social support from parents, which contributed to lower levels of depressive symptoms and a better perception of health, through optimism. Optimism is a learned attitude that develops during socialization (Herman-Stahl & Petersen, 1999) and is also associated with SES and therefore may generate social inequalities in health (Taylor & Seeman, 1999). As part of healthy personality development, youths must learn effective coping skills and ego-protective attitudes, such as optimism; however, the parents’ resources related to these may reflect their socio-economic background. The final model also suggested that whereas parental support was not directly related to self-perceived health, optimism (positively) and depression (negatively) was associated with self-perceived health. This may be explained by the fact that self-perceived health is primarily influenced by youths’ well-being (Piko, 2007).

Despite any limitations (e.g. the cross-sectional nature of analyses, the use of only self-report data without parental information or medical verifications, the lack of information on household income), we believe these findings make a unique contribution to the literature on social inequalities in adolescent mental health. In this study, we followed the path model of a meta-analysis that had suggested a path from SES through parental behaviour to adolescent psychopathology (Grant et al., 2003). Therefore, we applied a mediational model including parental social support and optimism. We believe this is a major strength of the paper. Moreover, although SES and health relationships among adolescents have been analysed in a number of countries, these patterns have not been examined among samples of Hungarian high-school students.

Overall, our results indicate that during adolescence parental social support and optimism may serve as important mediators in the SES–health relationship. More precisely, our findings suggest that: (1) SES variables may generate social inequalities in adolescent depression through parental social support, particularly maternal support; and (2) social inequalities in adolescents’ psychological health may be explained by differences in parents’ resources that provide youths with different levels of social support, which in turn may strengthen or weaken optimism during the socialization process.

Conclusions

Results support a concept of SES inequalities in depression and perceived health that may be mediated by parental social support and optimism. These findings have important clinical and health policy implications since these protective factors have enormous potential for interventions. These may include efforts to improve adolescents’ resilience and protective factors. For clinical implications, our paper highlights the notion of optimism, with the pattern of associations assumed in the cognitive behavioural treatment for adolescent depression. One of the main goals of cognitive behavioural therapy (CBT) is to address the cognitive, behavioural and social factors that underlie and maintain distress (Reinecke & Ginsburg, 2008). Cognitive therapists endeavour to provide children and their parents with strategies for coping with life’s problems, and a sense of optimism. That is, to empower both the children and their parents. Recently, in developing prevention programmes of anxiety and depression, optimistic thinking skills have received considerable attention; some interventions address this issue among children from disadvantaged schools (Roberts et al., 2010). Such programmes could potentially contribute to developing or strengthening certain attitudes (such as optimism) when otherwise parents from lower social SES backgrounds may lack the necessary material and psychosocial resources.

Footnotes

Appendix 1 Acknowledgements

This study was supported by the ETT 012-08/2009 research grant of the Ministry of Health Care (Hungary) and in the frame of the following personal research grant: TÁMOP 4.2.1.-B/09/0/KONV-2010-005.

Notes

References

Arbuckle

J. L.

(2006). Amos 6.0.1. Chicago, IL: Smallwaters Corporation.

Bandalos

D. L.

Finney

S. J.

(2001). Item parceling issues in structural equation modeling. In Marcoulides

G. A.

Schumacker

R. E.

(Eds), New developments and techniques in structural equation modeling (pp. 269–296). Hillsdale, NJ: Lawrence Erlbaum Associates.

Byrne

B. M.

(2009). Structural equation modeling with AMOS. Basic concepts, applications, and programming, 2nd edn. New York: Routledge.

Federico

Falese

Capelli

(2009). Socio-economic inequalities in physical activity practice among Italian children and adolescents: A cross-sectional study. Journal of Public Health, 17, 377–384.

Grant

K. E.

Compas

B. E.

Stuhlmacher

A. F.

Thurm

A. E.

McMahon

S. D.

Halpert

J. A.

(2003). Stressors and child and adolescent psychopathology: Moving from markers to mechanisms of risk. Psychological Bulletin, 129, 447–466.

Halldórsson

Kunst

A. E.

Köhler

Mackenbach

J. P.

(2000). Socioeconomic inequalities in the health of children and adolescents: A comparative study of the five Nordic countries. European Journal of Public Health, 10, 281–288.

Hankin

B. L.

(2006). Adolescent depression: Description, causes, and interventions. Epilepsy & Behavior, 8, 102–114.

Hatch

S. L.

(2005). Conceptualizing and identifying cumulative adversity and protective resources: Implications for understanding health inequalities. Journal of Gerontology, 60B, 130–134.

Herman-Stahl

Petersen

A. C.

(1999). Depressive symptoms during adolescence: Direct and stress-buffering effects of coping, control beliefs, and family relationships. Journal of Applied Developmental Psychology, 20, 45–62.

10.

Kaltiala-Heino

Rimpalä

Rantanen

Laippala

(2001). Adolescent depression: The role of discontinuities in life course and social support. Journal of Affective Disorders, 64, 155–166.

11.

Kerr

D. C. R.

Preuss

L. J.

King

C. A.

(2006). Suicidal adolescents’ social support from family and peers: Gender-specific associations with psychopathology. Journal of Abnormal Child Psychology, 34, 103–114.

12.

Kessler

R. C.

Avenevoli

Merikangas

K. R.

(2001). Mood disorders in children and adolescents: An epidemiologic perspective. Biological Psychiatry, 49, 1002–1014.

13.

Khullar

Oreskovic

N. M.

Perrin

J. M.

Goodman

(2011). Optimism and the socioeconomic status gradient in adolescent adiposity. Journal of Adolescent Health, 49, 553–555.

14.

Kovacs

(1992). Children’s Depression Inventory (CDI). Toronto: Multi-Health Systems.

15.

Kroenke

(2008). Socioeconomic status and health: Youth development and neomaterialist and psychosocial mechanisms. Social Science & Medicine, 66, 31–42.

16.

Landis

R. S.

Beal

Tesluk

P. E.

(2000). A comparison of approaches to forming composite measures in structural equation models. Organizational Research Methods, 3, 186–207.

17.

MacKinnon

D. P.

(2008). Path analysis mediation models. In MacKinnon

D. P.

(Ed.), Introduction to statistical mediation analysis (pp. 127–171). Mahwah, NJ: Lawrence Erlbaum Associates.

18.

Piko

(1998). Social support and health in adolescence: A factor analytical study. British Journal of Health Psychology, 3, 333–344.

19.

Piko

(2007). Self-perceived health among adolescents: The role of gender and psychosocial factors. European Journal of Pediatrics, 166, 701–708.

20.

Piko

B. F.

Fitzpatrick

K. M.

(2007). Socioeconomic status, psychosocial health and health behaviours among Hungarian adolescents. European Journal of Public Health, 17, 353–360.

21.

Piko

B. F.

Kovacs

Fitzpatrick

K. M.

(2009). Parental, school-related and individual protective factors for adolescent depressive symptomatology. European Child and Adolescent Psychiatry, 18, 617–624.

22.

Reinecke

M. A.

Ginsburg

G. S.

(2008). Cognitive-behavioral treatment of depression during childhood and adolescence. In Abela

J. R. Z.

Hankin

B. L.

(Eds), Handbook of depression in children and adolescents (pp. 179–206). New York: Guilford Press.

23.

Roberts

C. M.

Kane

Bishop

Cross

Fenton

Hart

(2010). The prevention of anxiety and depression in children from disadvantaged schools. Behavior Research & Therapy, 48, 68–73.

24.

Scheier

M. F.

Carver

C. S.

(1985). Optimism, coping and health: Assessment and implications of generalized outcome experiences. Health Psychology, 5, 219–247.

25.

Sekine

Tatsuse

Kagamimori

Chandola

Cable

Marmot

… Lahelma

(2011). Sex inequalities in physical and mental functioning of British, Finnish, and Japanese civil servants: Role of job demand, control and work hours. Social Science & Medicine, 73, 595–603.

26.

Siahpush

Singh

G. K.

(2000). A multivariate analysis of the association between social class of origin and current social class with self-rated general health and psychological health among 16-year-old Australians. Australian and New Zealand Journal of Medicine, 30, 653–659.

27.

Sund

A. M.

Wichstrøm

(2002). Insecure attachment as a risk factor for future depressive symptoms in early adolescents. Journal of American Child & Adolescent Psychiatry, 41, 1478–1485.

28.

Taylor

S. E.

Seeman

T. E.

(1999). Psychosocial resources and SES–health relationship. Annals of the New York Academy of Sciences, 896, 210–225.

29.

Thompson

Hickey

(2005). Society in focus. Boston, MA: Pearson.

30.

Turner

R. J.

Marino

(1994). Social support and social structure: A descriptive epidemiology. Journal of Health and Social Behavior, 35, 513–519.

31.

West

Sweeting

(2004). Evidence on equalisation in health in youth from the West of Scotland. Social Science & Medicine, 59, 13–27.

32.

Wickrama

K. A. S.

Noh

Elder

G. H.

(2009). An investigation of family SES-based inequalities in depressive symptoms from early adolescence to emerging adulthood. Advances in Life Course Research, 14, 147–161.

33.

Wong

S. S.

Lim

(2009). Hope versus optimism in Singaporean adolescents: Contributions to depression and life satisfaction. Personality and Individual Differences, 46, 648–652.