Journal of Adolescent Health
Volume 44, Issue 4 , Pages 380-386, April 2009

Spirituality and Depressive Symptoms in a School-Based Sample of Adolescents: A Longitudinal Examination of Mediated and Moderated Effects

  • John E. Pérez, Ph.D.

      Affiliations

    • Department of Psychology, University of Massachusetts, Boston, Massachusetts
    • Corresponding Author InformationAddress correspondence to: John E. Pérez, Department of Psychology, University of Massachusetts, Boston, 100 Morrissey Blvd., Boston, MA 02125-3393.
    • ,
  • Todd D. Little, Ph.D.

      Affiliations

    • Department of Psychology, University of Kansas, Lawrence, Kansas
    • ,
  • Christopher C. Henrich, Ph.D.

      Affiliations

    • Department of Psychology, Georgia State University, Atlanta, Georgia

Received 29 May 2008; accepted 27 August 2008. published online 11 November 2008.

Article Outline

Abstract 

Purpose

To prospectively examine whether personal agency beliefs and direct coping mediate the association between spirituality and depressive symptoms in a school-based sample of adolescents, and whether gender, race, or grade level moderate this model.

Method

Students (N = 1096) from sixth through ninth grades in a northeastern public school system were administered self-report instruments in group format at baseline, 6-month follow-up, and 1-year follow-up. Demographic variables and constructs of spirituality, personal agency, direct coping, and depressive symptoms were assessed.

Results

Structural equation modeling was used to examine the longitudinal associations among the constructs while controlling for socioeconomic status and baseline depressive symptoms. For the total sample, the model predicted 21% of the variance in depressive symptoms over 1 year. The model was moderated by gender but not by race or grade level. The model explained 28% of the variance in depressive symptoms for girls and 16% of the variance in depressive symptoms for boys. Moreover, there was an indirect effect of spirituality on depressive symptoms for girls but not for boys.

Conclusion

These results suggest mechanisms by which spirituality may maintain lower levels of depressive symptoms among adolescent girls during periods of transition to middle and high school.

Keywords: Spirituality, Religion, Depression, Adolescent, Personal agency, Self-efficacy, Coping behavior

 

Adolescence represents a period of increased vulnerability to depression. Rates of major depression dramatically increase from childhood to adolescence, particularly for girls [1], [2]. Depression during adolescence is associated with a number of maladaptive outcomes including poor academic performance, substance abuse, high-risk sexual behavior, antisocial behavior, physical health problems, and suicidal behavior [3], [4]. Adolescents with subclinical levels of depressive symptoms also demonstrate significant impairment in social and academic functioning and are at high risk for developing major depressive episodes [5]. These data suggest that adolescence is a critical period for understanding the factors that buffer adolescents against depressive symptoms.

Studies with adults show that various dimensions of religiousness and spirituality are generally associated with lower levels of depressive symptoms [6]. However, little attention has been paid to this link among children and adolescents [7], [8]. In recent, nationally representative samples of adolescents, higher levels of religious beliefs and involvement were associated with lower levels of depressive symptoms [9], [10]. Controlling for demographic variables (i.e., gender, race, grade level, and socioeconomic status), one study found that perceived social support from adolescents’ religious community was associated with lower levels of depressive symptoms; conversely, perceived criticism and demands from their religious community were associated with higher levels of depressive symptoms [11].

Although religiousness and spirituality are usually associated with lower levels of depressive symptoms among adolescents, the potential mechanisms in this association have rarely been examined empirically, and few studies have used a theoretical framework to guide the research. Four potential psychosocial mechanisms have received some empirical attention in the religion–health literature: health practices, social support, belief structures, and psychosocial resources such as self-esteem and self-efficacy [12].

Bandura's social cognitive theory provides a framework for exploring potential mechanisms in the spirituality–depression connection. Social cognitive theory is grounded in a view of human agency that emphasizes intentionality and forethought, self-regulation, and self-reflectiveness in adaptation and change [13]. Human functioning is the product of a dynamic, reciprocal interplay of personal, behavioral, and environmental influences [13]. One potential mediator, drawn from social cognitive theory, is personal agency. Personal agency reflects the belief in one's ability to organize and execute the sources of action required to manage prospective situations. Religion provides a context for mastery experiences and social modeling that can foster personal agency. A relationship with a transcendental power (e.g., God) can empower people with the belief that they can achieve any goal. There is some evidence that religious beliefs are associated with personal agency beliefs and internal locus of control among both adults and adolescents [14], [15]. In addition, longitudinal studies with children and adolescents demonstrate that agency-related beliefs regarding social and academic outcomes are inversely related to depressive symptoms [16], [17].

Another possible mediator of the spirituality–depression link is coping behavior. Pargament [18] has built a strong program of research demonstrating the direct effects of both positive and negative religious coping on health. Less attention, however, has been paid to non-religious coping behaviors as potential mediators in the religion–health connection [12]. According to Lazarus and Folkman's transactional model of stress and coping, coping behaviors are dependent upon multiple factors, including the perceived controllability of the situation [19]. Although multiple coping behaviors are often employed in the context of stressful situations, problem-focused coping is employed when a stressor is perceived as controllable. Direct or approach-oriented coping may reflect cognitive attempts to change ways of thinking about the problem and behavioral attempts to resolve the problems by dealing with them directly [19]. Direct coping is linked to higher personal agency and lower negative affect among adolescents. [20], [21], [22]. Religion can frame the way one thinks about problems and may be linked to direct coping via personal agency. An intrinsically motivated relationship with God may lead people to appraise stressors as challenges rather than threats [18]; with God's support, humans are empowered to approach and overcome obstacles.

Religiousness and spirituality vary as a function of age, gender, and race. There is a general decline in religiousness during adolescence [23]. Compared to adolescent boys, adolescent girls report higher levels of general religiousness [9], [23], religious judgment [24], positive religious coping [25], and daily spiritual experiences [25]. In several studies, black adolescents show higher levels of religious involvement than do white adolescents [9], [23], [26]. Together with findings that depression varies as a function of age and gender [1], [2], these results suggest that demographic variables may moderate the association between spirituality, agency beliefs, coping, and depressive symptoms.

Using social cognitive theory and the transactional model of stress and coping to guide our hypotheses, we examined personal agency beliefs and direct coping as potential mechanisms that may mediate the relationship between spirituality and depressive symptoms in a longitudinal design with adolescents. We also examined whether grade level, gender, and race moderated this model. In this study, spirituality is a multidimensional construct defined as the frequency of an individual's spiritual or religious practices, personal conviction of God's existence based on experience, and a highly internalized relationship between God and the self [27]. Consistent with social cognitive theory, personal agency beliefs are domain specific; we assessed those situations that are specific to establishing and maintaining friendships. Similarly, direct coping was measured in relation to friendships. This domain was chosen because interpersonal stressors are salient predictors of depressive symptoms among adolescents, especially among girls [28].

Back to Article Outline

Hypotheses 

We hypothesized the following: first, we proposed that greater spirituality will predict decreases in depressive symptoms at 1-year follow-up. This association will be mediated by personal agency beliefs and direct coping. Greater spirituality will positively predict personal agency which, in turn, will positively predict direct coping. In the final link, higher levels of direct coping will predict lower depressive symptoms. Second, we hypothesized that the link between spirituality and depressive symptoms will be moderated by grade level, gender, and race. Specifically, the amount of variance explained in depressive symptoms will be greater for older students, girls, and blacks.

Back to Article Outline

Methods 

Participants 

Students (N = 1096, 50% female) from a northeastern public school system participated (80% active-consent participation rate) as they made the transition to grades 7–9 (baseline age range: 11–15 years). Nine elementary schools (grade 6), a middle school (grades 7 and 8), and the first year of high school (grade 9) were represented. The sample was ethnically and socio-economically diverse: 66.8% white, 16.6% black, 6.9% Hispanic/Latino, and 9.8% other. Approximately 19% of students in the district received free or reduced-priced meals.

Measures 

Demographic variables including gender, race, and grade level were provided by both student self-report and school records. Socioeconomic status (SES) was a dichotomous variable based on students’ eligibility for free or reduced lunch (obtained from school records), the criteria for which is family income of less than 1.85 times that of the federal poverty level.

Index of Core Spiritual Experience (INSPIRIT) [27]. The first six INSPIRIT items that measure the frequency of spiritual practices, experiences that convince a person that a higher power exists, and experiences that evoke feelings of closeness with one's higher power were used in the current study, using a four-point scale (not at all true to completely true). The INSPIRIT has shown validity associations with life purpose and satisfaction, health-promoting attitude, and decreased frequency of medical symptoms as well as positive validity correlations with intrinsic religious motivation (r = .61–.69) and a negative correlation with extrinsic religious motivation (r = −.26) [27], [29].

Multi-dimensional Control Agency Means-Ends protocol (Multi-CAM) [30]. Personal agency beliefs were measured with six items from the Multi-CAM using a four-point scale (ranging from not at all true to completely true). These items assessed students’ beliefs about their ability and effort to achieve their interpersonal goals (e.g., establishing and maintaining friendships). Multi-CAM personal agency beliefs have been associated with more social cooperation, more emotional support-seeking, less social exploitation, and less anxiety among children [22].

Behavioral Inventory of Strategic Control (BISC) [31]. Using a four-point scale (ranging from not at all true to completely true), direct coping was measured by three items from the BISC that assessed students’ attempts to actively solve interpersonal problems (e.g., making new friends). The BISC has been associated with positive social outcomes for children such as having many friends, good social engagement, less loneliness, and less anxiety [31].

Children's Depression Inventory (CDI) [32]. The CDI, a 27-item modification of the Beck Depression Inventory, is a widely used, validated measure of depressive symptoms in children. Each item consists of three statements representing levels of severity of a common symptom of depression, which are assigned a numerical value ranging from 0 to 2. High scores on the CDI indicate higher levels of depressive symptoms.

Procedure 

This study was part of a larger project exploring how various psychosocial factors affect adolescents’ academic and social development. Approval was obtained from the appropriate IRB, each participant's parent/guardian provided informed consent, and each participant provided informed assent. Data were collected in three waves, 6 months apart. At each wave, questionnaires were group administered (approximately 25 students per classroom) by trained staff during three 50-minute sessions over a 3-week period. Teachers were present during the testing session; however, they did not oversee the questionnaire administration. Half of the items from each scale were randomly assigned to two parallel-forms questionnaires, which were administered in random order. The CDI was administered as a complete measure with one of the forms. Student participants were given school supplies and the schools were paid for participation in the study.

Analytic procedures 

The sample of 1096 participated in at least two waves of measurement. At each wave, missing data (approximately 11%) were estimated using full-information regression estimation techniques [33]. Outliers (approximately 4%) were identified using the same regression estimation technique and adjusted to be within two standard deviations of the predicted value from the saturated regression equation. The suicidal ideation item of the CDI was omitted from the analyses due to limited variance. In addition, because depressive symptoms were positively skewed, a square-root transformation of the items was performed to normalize its distribution.

Structural equation modeling (SEM) with LISREL 8.80 [34] was used to examine the longitudinal associations between the latent constructs. All constructs were represented using three parceled indicators [35], each of which had acceptable internal consistency for SEM (see Table 1). The measurement model was first tested by specifying a longitudinal confirmatory factor model. Then, the fit of the hypothesized model was examined. The hypothesized model controlled for Time 1 depressive symptoms by regressing all latent variables in the model on it. Thus, the model examined how Time 1 spirituality, mediated by Time 2 personal agency and Time 2 direct coping, predicted changes in depressive symptoms from Time 1 to Time 3. Because socioeconomic status (SES) correlated significantly with depressive symptoms at Time 1 (r = −.08, p < .01) and depressive symptoms at Time 3 (r = −.11, p < .01), it was included as a covariate in all analyses. Group comparisons were made by gender, race, and grade level. In all models, the error variances for Time 1 depressive symptoms and Time 3 depressive symptoms were allowed to correlate, and the factor loadings were constrained to be equal across time. In all cross-group comparisons, loadings for the manifest variables were constrained to be equal across groups. Measurement equivalence of the factor loadings across groups was compared using change in CFI (a difference ≤.01 is acceptable). Similarity of latent regression paths across groups was tested using the χ2 difference test [36].

Table 1. LISREL estimates of measured parcels for the total sample in the freely estimated model (N = 1096)
ParcelλSEθSEαR2
Depressive Symptoms (Time 1)
V1.24.01.02.00.71.70
V2.26.01.01.00.78.87
V3.20.01.02.00.68.67
Spirituality (Time 1)
V4.61.02.23.01.60.61
V5.72.02.13.02.61.80
V6.55.02.27.01.75.53
Personal Agency (Time 2)
V7.62.02.14.01.59.73
V8.68.02.10.01.59.82
V9.61.02.17.01.56.69
Direct Coping (Time 2)
V10.69.02.28.02—.63
V11.69.02.27.02—.65
V12.56.02.32.02—.50
Depressive Symptoms (Time 3)
V13.24.01.02.00.75.77
V14.26.01.01.00.82.88
V15.20.01.02.00.73.73

Note. Lambda (λ) represents the unstandardized variance explained in each indicator (i.e., parcel) by the latent variable. Theta (θ) represents the unstandardized error variance. Standard errors of estimate (SE) are given for both lambda and theta. Alpha (α) represents Cronbach's alpha (internal consistency) for each of the measured parcels. Single-item indicators were used for the direct coping parcels. R2 represents the amount of variance explained by the latent variable in each of the measured parcels.

Back to Article Outline

Results 

The measurement model, which allowed all constructs to correlate with each other, showed an excellent fit (χ2 (92, N = 1096) = 169.08, p < .001, NNFI = .99, CFI = .99, RMSEA = .028). The hypothesized path model shown in Figure 1 also showed excellent fit (χ2 (95, N = 1096) = 172.26, p < .001, NNFI = .99, CFI = .99, RMSEA = .027). A χ2 difference test showed no significant drop in fit from the measurement model to the final path model (Δχ2 (3, N = 1096) = 3.18, p = .36). Table 1 shows the unstandardized estimates (maximum likelihood) for the measurement model and the amount of reliable variance (R2) and internal consistency (α) of each parceled variable. Table 2 contains the intercorrelations among the latent variables for the total sample. In the total sample, 21% of the variance in depressive symptoms was explained by the hypothesized model. Time 1 depressive symptoms explained much of the variance in Time 3 depressive symptoms (β = .45, p < .001). Spirituality had a significant direct effect on personal agency (β = .16, p < .001), a significant indirect effect on direct coping (β = .11, p < .001), but a nonsignificant indirect effect on depressive symptoms (β = −.01, p = .09). Personal agency had a significant direct effect on direct coping (β = .70, p < .001) but a nonsignificant indirect effect on depressive symptoms (β = −.03, p = .07). Direct coping did not have a significant direct effect on depressive symptoms (β = −.05, p = .07).

  • View full-size image.
  • Figure 1 

    Hypothesized structural equation model of the mediated relationship between spirituality and depressive symptoms. Double-headed arrows (psi) represent correlations and single-headed arrows (beta) represent standardized regression paths. Small circles represent residual variance for the dependent variables.

Table 2. Intercorrelations among the latent variables for the total sample in the freely estimated model (N = 1096)
Depressive PersonalDirectDepressive
SymptomsSpiritualityAgencyCopingSymptoms
(Time 1)(Time 1)(Time 2)(Time1)(Time 3)
Depressive Symptoms (Time 1)—
Spirituality (Time 1)−.09∗∗—
Personal Agency (Time 2)−.25∗∗∗.17∗∗∗—
Direct Coping (Time 2)−.20∗∗∗.16∗∗∗.60∗∗∗—
Depressive Symptoms (Time 3).46∗∗∗−.06−.16∗∗∗−.15∗∗∗—

∗∗∗p < .001.

∗∗p < .01.

Moderation of the structural model 

To test for differences in the model shown in Figure 1 by gender, race, and grade level, a series of multigroup models were run. First, the factor loadings of the models were compared to test for measurement equivalence, and second, differences in the structural paths were tested. The cross-group comparison between the three grade levels showed no significant differences. A three-group model assuming measurement equivalence fit the data well (χ2 (321, N = 1096) = 537.21, p < .001, NNFI = .98, CFI = .99, RMSEA = .043); fit statistics were similar to a model in which factor loadings were freely estimated across groups (ΔCFI = .01) and the χ2 difference test comparing the structural paths of the mediation model (spirituality → personal agency → direct coping → depression) across groups was not significant (Δ χ2 (6, N = 1096) = 8.62, p = .20), indicating that the structural model was equivalent across grades. Similarly, there was no moderation by racial group. The two-group model comparing blacks and whites (the only racial groups with sufficient power for latent model comparisons) showed an excellent fit to the data (χ2 (208, n = 888) = 286.69, p < .001, NNFI = .99, CFI = .99, RMSEA = .029); fit was similar to a model with no constraints on factor loadings across groups (ΔCFI = .00) and the χ2 difference test comparing the structural paths of mediation model was not significant (Δ χ2 (3, n = 888) = 3.09, p = .38), indicating that the structural model was equivalent for blacks and whites.

There was evidence of moderation by gender. This two-group model assuming measurement equivalence showed an excellent fit to the data (χ2 (208, N = 1096) = 384.59, p < .001, NNFI = .98, CFI = .98, RMSEA = .039), and fit was similar to a model with no constraints on factor loadings cross groups (ΔCFI = .01). Comparison of the structural paths indicated moderation by gender, however, as indicated by the significant chi-square difference test (Δ χ2 (3, N = 1096) = 10.05, p < .05). Figure 2 shows the structural model with the path estimates separately for girls and boys. For both girls and boys, there were significant direct effects of spirituality on personal agency and of personal agency on direct coping. Notably though, the direct effect of direct coping on depressive symptoms was significant for girls (β = −.17, p < .001), but not for boys (β = .05, p = .41). This path differed significantly between groups (Δ χ2 (1, N = 1096) = 9.97, p < .01). The indirect effect of spirituality on depressive symptoms was significant for girls (β = −.01, p < .05), but not for boys (β = .00, p = .41). The model explained 28% of the variance in depressive symptoms for girls and 16% of the variance in depressive symptoms for boys.

  • View full-size image.
  • Figure 2 

    Final structural equation model with standardized regression coefficients for girls versus boys. Values for girls are placed above values for boys. Dashed line represents a nonsignificant path for both girls and boys. ∗∗∗p < .001, ∗∗p < .01, ∗p < .05.

An additional four-group model was run to test for an interaction between gender and race. This model showed a very good fit (χ2 (452, n = 888) = 735.58, p < .001, NNFI = .97, CFI = .97, RMSEA = .053), and was measurement equivalent (ΔCFI = .00). As in the two-group comparison by gender, the only difference in the latent paths was that the effect of direct coping on depressive symptoms was present for girls, but not for boys (Δ χ2 (1, n = 888) = 6.51, p = .01).

Back to Article Outline

Discussion 

We examined personal agency beliefs and direct coping as mediators of the relationship between spirituality and changes in depressive symptoms among adolescents. Spirituality predicted changes in depressive symptoms via personal agency beliefs and direct coping for girls, but not for boys. This difference is meaningful considering the greater risk for depression among girls compared to boys during adolescence. There is evidence that girls carry more preexisting risk factors for depression that combine with greater biological and social challenges during adolescence [1], [2], [28]. Therefore, it is imperative to understand potential psychosocial factors that may help buffer adolescent girls against depression.

In support of social cognitive theory, personal agency emerged as an important mediator linking spirituality to direct coping, and it was robust across all groups. The main difference in the model was that the link between direct coping and depressive symptoms was significant for girls, but not for boys. Because direct coping was evaluated in the social domain (e.g., solving problems making and keeping good friends), this finding suggests that direct attempts to deal with interpersonal problems more strongly predict lower negative affect for girls. Consistent with this finding, research shows that early adolescent girls may be more interpersonally competent than are early adolescent boys [37]. Future studies might examine whether different types of coping mediate the relationship between spirituality and depressive symptoms among boys.

Contrary to expectations, the model was not moderated by grade level or race. It may be that there was not sufficient variability in age from grades 7 to 9 to find differences in the model by grade level and studies spanning a greater range of adolescence would be required to detect age differences. The lack of racial differences may be due to the mediators selected. Because of the collectivistic nature of black spirituality [38], perhaps collective agency (people's shared belief in their collective power to produce desired results) and religious support (i.e., social support from one's religious community) are more salient mechanisms linking spirituality to well-being among blacks. This is an empirical question that deserves attention in black and other racial minority groups.

In terms of limitations, all measures were self-report. Considering the informant variability with constructs such as depressive symptoms [39], the study would have been strengthened by multiple assessment methods. In addition, this study focused on a non-clinical sample that showed low levels of depressive symptoms with a positively skewed distribution, requiring a square-root transformation. Finally, perceived stress was not measured. Because the spirituality–depression link is mediated by coping, people likely must experience significant stress to activate their coping resources. Indeed, the association between spirituality and depressive symptoms appears to be amplified under stressful conditions [7].

Despite the limitations of this study, there were some noteworthy strengths. The present study demonstrated how spirituality might help maintain low levels of depressive symptoms in a large, school-based sample of adolescents. The majority of studies conducted on the link between spirituality/religiousness and depressive symptoms have focused on older adult populations, often with health problems [6]. In addition, the hypotheses of this study were guided by social cognitive theory and the transactional model of stress and coping. Unfortunately, the majority of empirical studies in religiousness/spirituality and health lack a theoretical framework [40]. Moreover, the longitudinal design of this study permitted an examination of potential mechanisms in the relationship between spirituality and depressive symptoms. Because the effect of Time 1 depressive symptoms on Time 3 depressive symptoms was controlled, the model tested the indirect effects of spirituality on changes in depressive symptoms. Cross-sectional designs, representing the bulk of the spirituality–depression literature, cannot determine the temporal order of effects.

A greater emphasis on spiritual development is needed in the developmental sciences to understand when and how spiritual beliefs and practices emerge as resources for coping. Much work remains to identify the mechanisms that link religion/spirituality to health among adolescents. Compared to spirituality—a multidimensional, distal variable—these mediating variables represent more proximal and more modifiable targets for prevention interventions. If agency beliefs and coping behaviors are mechanisms by which spirituality buffers high-risk adolescents against depressive symptoms, religious as well as non-religious organizations could work together to foster these strengths for mental health and well-being. Theoretically guided mechanism studies with high-risk samples of adolescents are needed to determine whether such pathways are viable targets for depression prevention interventions.

Back to Article Outline

Acknowledgments 

This study was funded, in part, by a National Research Service Award (1 F31 MH12929-01) from the National Institute of Mental Health (to J.P.). The study also was funded by grants from the Smith Richardson Foundation and the Graustein Foundation (to T.L).

Back to Article Outline

References 

  1. Alloy LB, Abramson LY. The adolescent surge in depression and the emergence of gender differences: A biocognitive vulnerability-stress model in developmental context. In:  Romer D,  Walker EF editor. Adolescent Psychopathology and the Developing Brain: Integrating Brain and Prevention Science. New York: Oxford University Press; 2007;p. 284–312
  2. Nolen-Hoeksema S, Girgus JS. The emergence of gender differences in depression during adolescence. Psychol Bull. 1994;115:424–443
  3. Birmaher B, Ryan ND, Williamson DE, et al. Childhood and adolescent depression: A review of the past 10 years. Part I. J Am Acad Child Adolesc Psychiatry. 1996;35:1427–1439
  4. Lewinsohn PM, Essau CA. Depression in adolescents. In:  Gotlib IH,  Hammen CL editor. Handbook of Depression. New York: Guilford; 2002;p. 541–559
  5. Gotlib IH, Lewinsohn PM, Seeley JR. Symptoms versus diagnosis of depression: Differences in psychosocial functioning. J Consult Clin Psychol. 1995;63:90–100
  6. Koenig HG, McCullough ME, Larson DB. Handbook of Religion and Health. New York: Oxford; 2001;
  7. Smith TB, McCullough ME, Poll J. Religiousness and depression: Evidence for a main effect and the moderating influence of stressful life events. Psychol Bull. 2003;129:614–636
  8. Wong YJ, Rew L, Slaikeu KD. A systematic review of recent research on adolescent religiosity/spirituality and mental health. Issues Ment Health Nurs. 2006;27:161–183
  9. Le TN, Tov W, Taylor J. Religiousness and depressive symptoms in five ethnic adolescent groups. Int J Psychol Relig. 2007;17:209–232
  10. Sinha JW, Cnaan RA, Gelles RJ. Adolescent risk behaviors and religion: Findings from a national study. J Adolesc. 2007;30:231–249
  11. Pearce MJ. Little TD, Perez JE . Religiousness and depressive symptoms among adolescents. J Clin Child Adolesc Psychol. 2003;32:267–276
  12. George LK, Ellison CG, Larson DB. Explaining the relationships between religious involvement and health. Psychol Inq. 2002;13:190–200
  13. Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall; 1986;
  14. Saudia TL, Kinny MR, Brown KC, et al. Health locus of control and helpfulness of prayer. Heart Lung. 1991;20:60–65
  15. Strickland BR, Schaffer S. I-E, I-E, & F. J Sci Study Relig 1971;10:366–9.
  16. Bandura A, Pastorelli C, Barbaranelli C, et al. Self-efficacy pathways to childhood depression. J Pers Soc Psychol. 1999;76:258–269
  17. McFarlane AH, Bellissimo A, Norman GR. The role of family and peers in social self-efficacy: Links to depression in adolescence. Am J Orthopsychiatry. 1996;65:402–410
  18. Pargament KI. The Psychology of Religion and Coping: Theory, Research, Practice. New York: Guilford; 1997;
  19. Lazarus RS, Folkman S. Stress, Appraisal, and Coping. New York: Springer; 1984;
  20. Ebata AT, Moos RH. Coping and adjustment in distressed and healthy adolescents. J Appl Dev Psychol. 1991;12:33–54
  21. Herman-Stahl MA, Stemmler M, Peterson AC. Approach and avoidant coping: Implications for adolescent mental health. J Youth Adolesc. 1995;24:649–665
  22. Lopez DF, Little TD. Children's action-control beliefs and emotional regulation in the social domain. Dev Psychol. 1996;32:299–312
  23. Donahue MJ, Benson PL. Religion and well-being of adolescents. J Soc Issues. 1995;51:145–160
  24. Oser F, Gmünder P. Religious Judgment: A Developmental Perspective. Birmingham, AL: Religious Education Press; 1991;
  25. Desrosiers A, Miller L. Relational spirituality and depression in adolescent girls. J Clin Psychol. 2007;63:1021–1037
  26. Taylor RJ, Chatters LM, Jayakody R, et al. Black and white differences in religious participation: A multisample comparison. J Sci Study Relig. 1996;35:403–410
  27. Kass JD, Friedman R, Leserman J, et al. Health outcomes and a new index of spiritual experience. J Sci Study Relig. 1991;30:203–211
  28. Hankin BL, Mermelstein R, Roesch L. Sex differences in adolescent depression: Stress exposure and reactivity models. Child Dev. 2007;78:279–295
  29. VandeCreek L, Ayers S, Bassham M. Using the INSPIRIT to conduct spiritual assessments. J Pastoral Care. 1995;49:83–89
  30. Little TD, Wanner B. The Multi-CAM: A Multidimensional Instrument to Assess Children's Action-Control Motives, Beliefs, and Behaviors (Institute Series No. 55). Berlin, Germany: Max-Planck Institute; 1997;
  31. Little TD, Lopez DF, Wanner B. Children's action-control behaviors (coping): A longitudinal validation of the Behavioral Inventory of Strategic Control. Anxiety Stress Coping. 2001;14:315–336
  32. Kovacs M. Children's Depression Inventory (CDI) manual. North Tonawanda, NY: Multi-Health Systems; 1992;
  33. Graham JW, Hofer SM. Multiple imputation in multivariate research. In:  Little TD,  Schnabel KU,  Baumert J editor. Modeling Longitudinal and Multivariate Data. Mahwah, NJ: Lawrence Erlbaum; 2000;p. 201–218
  34. Jöreskog K, Sörbom D. LISREL 8: User's Reference Guide. Chicago: Scientific Software International; 1996;
  35. Little TD, Cunningham WA, Shahar G, et al. To parcel or not to parcel: Exploring the question, weighing the merits. Struct Equation Model. 2002;9:151–173
  36. Cheung GW, Rensvold RB. Evaluating goodness-of-fit indexes for testing measurement invariance. Struct Equation Model. 2002;9:233–255
  37. Buhrmester D. Need fulfillment, interpersonal competence, and the developmental contexts of early adolescent friendship. In:  Bukowski WM,  Newcomb AF,  Hartup WW editor. The Company They Keep: Friendship in Childhood and Adolescence. New York: Cambridge University Press; 1998;p. 158–185
  38. Taylor RJ, Chatters LM, Levin J. Religion in the Lives of African Americans: Social, Psychological, and Health Perspectives. Thousand Oaks, CA: Sage; 2004;
  39. Kazdin AE. Informant variability in the assessment of childhood depression. In:  Reynolds WM,  Johnson HF editor. Handbook of Depression in Children and Adolescents: Issues in Clinical Child Psychology. New York: Plenum Press; 1994;p. 249–271
  40. Pargament KI. The bitter and the sweet: An evaluation of the costs and benefits of religiousness. Psychol Inq. 2002;13:168–181

PII: S1054-139X(08)00417-5

doi:10.1016/j.jadohealth.2008.08.022

Journal of Adolescent Health
Volume 44, Issue 4 , Pages 380-386, April 2009

Article Tools

* Image Usage & Resolution