Changes in Adolescent School Bonding, Developmental Assets, and Depression: A Latent Change Score Mediation Analysis

Introduction: School bonding is associated with various health and behavioral factors (e.g. educational attainment, mental health, delinquency) among adolescents (Shochet, Dadds, Ham, & Montague, 2006; Whiteside-Mansell, Weber, Johnson, Ward, & Phillips, 2015). However, most of these relations have been found for cross-sectional data and US-samples. In addition, research on more complex, dynamic relations between school bonding and student mental health is scarce. This study sought to fill these gaps by utilizing longitudinal data from adolescents in Germany to examine the potential role that social and personal developmental assets play in mediating the relation between school bonding and depression. To test for mediation effects, the Latent Change Score (LCS) mediation model from Selig and Preacher (2009) was used which allows for linking both levels and changes in these constructs to each other.

Methods: Self-report data from 722 German adolescents, serving as the control group for the evaluation of a universal school-based life skills program, were collected in four waves (ages 10-14) beginning with fifth-grade (average age= 10.54 years; 56% female; 59% on college-bound school track). Depression was assessed at waves 4 and 5 (8 items; α= .87, .90), school bonding at waves 1 and 3 (9 items; α= .77, .83), and developmental assets at waves 3 and 4 (average across two subscales, i.e., personal (13 items; α= .86, .84) and social (13 items; α= .86, .87)). The LCS mediation model was tested using Mplus version 7 with robust maximum likelihood estimation, full information maximum likelihood, saturated correlates to incorporate auxiliary variables, and type=complex. Bias-corrected bootstrap 95% confidence intervals were used to determine the significance of all indirect effects.

Results: The results of the LCS mediation model, MLRχ2 (0, N = 722) = 0.00, CFI = 1.000, RMSEA = .000 (90% CI: .000, .000), SRMR = .000, indicated that four of the eight indirect effects were significant. Both level and changes in developmental assets significantly mediated the relation of level of school bonding with level of depression. Additionally, both level and changes in developmental assets significantly mediated the relation of changes in school bonding with level of depression.

Conclusion: Our results demonstrate the merits of examining more dynamic mediation hypotheses that link both levels and changes in constructs to each other. Future work seeking to promote stronger ties with one’s school in order to positively influence mental health should take into account the students’ personal and social developmental assets.