Parental Maltreatment, Bullying, and Adolescent Depression: Evidence for the Mediating Role of Perceived Social Support

Abstract

The support deterioration model of depression states that stress deteriorates the perceived availability and/or effectiveness of social support, which then leads to depression. The present study examined this model in adolescent depression following parent-perpetrated maltreatment and peer-perpetrated bullying, as assessed by a rigorous contextual interview and rating system. In 101 depressed and nondepressed community adolescents between the ages of 13 and 18 (M = 15.51, SD = 1.27), peer bullying and father-perpetrated maltreatment were associated with lower perceptions of tangible support and of belonging in a social network. These forms of support mediated the association of bullying and father-perpetrated maltreatment with greater depression severity. In contrast, mother-perpetrated maltreatment was associated with higher perceptions of tangible support.

This research was supported by a New Investigator Award from the Hospital for Sick Children Foundation (to Kate L. Harkness). We are very grateful to our adolescent participants for being forthcoming about their sometimes painful childhood experiences. We are grateful to the Child and Youth Wellness Centre of Leeds and Grenville, Hotel Dieu Hospital, Kingston Military Family Resource Centre, and the Limestone District School Board for their collaboration in referring study participants. We also thank Angela Digout-Erhardt, Laura Johnson, Jennifer Laforce, and Margaret Lumley for helping with the clinical interviews; Krista Bromley, Alanna Truss, Angela Collins, Jorden Cummings, Nina Rytwinski, and Bernadette Zackher for rating the childhood adversity interviews; and Michael Barnett-Cowan, Stephanie Cerutti, Pilsu-qua Lloyd, and Jonathan McGregor for scheduling participants and managing the database.

Notes

¹The Interpersonal Support Evaluation List (ISEL) has a fourth subscale that assesses respondents' perceptions of social support that enhance self-esteem (e.g., “Most people who know me think highly of me”). This subscale was not used in the current study because the construct of self-esteem as assessed by the ISEL overlaps too strongly with the construct of depression as assessed by the Beck Depression Inventory–Second Edition (r = −.74, p < .001).

²Among the participants with a depressive disorder, the average age at first onset was 13.06 (SD = 2.77) and the average number of previous episodes was 1.76 (SD = 1.68). Thirty-three (59%) of these adolescents were suffering from a comorbid Axis I disorder: dysthymia (n = 6), panic disorder (n = 4), generalized anxiety disorder (n = 8), posttraumatic stress disorder (n = 4), specific phobia (n = 5), social phobia (n = 8), bulimia nervosa (n = 2), oppositional-defiant disorder (n = 3), conduct disorder (n = 1), attention-deficit/hyperactivity disorder (n = 1), substance abuse (n = 7), and enuresis (n = 1). These numbers do not add up to 33 because some adolescents had more than one comorbid diagnosis.

³Recent work in the area of childhood maltreatment has made an important distinction between the constructs of physical discipline and child maltreatment (see Gershoff, 2002). However, it should be noted that the “physical abuse” variable of the CECA includes physical abuse that is not disciplinary in nature.

Note: BDI–II = Beck Depression Inventory–Second Edition; ISEL = Interpersonal Support Evaluation List.

^a n = 45.

^b n = 56.

Note: N = 101. BDI–II = Beck Depression Inventory–Second Edition; ISEL = Interpersonal Support Evaluation List.

^a n = 12.

^b n = 89.

⁺ p < .10. *p < .05.

⁴Duration of maltreatment could not be included as a covariate in our path analyses because this variable was only relevant for those in our sample who had experienced maltreatment or bullying.

*p < .05. **p < .005.

⁵The relation of age to the ISEL subscales raises the potential that age might also be associated with higher depressive symptomatology and/or duration of depression. We note that all of our models contain age as a covariate to account for its potential confounding effects. We also note that age was not significantly correlated with depression severity (i.e., BDI–II scores), either in the total sample, r(100) = .17, p = .10, or in the subset with a depressive disorder, r(55) = .15, p = .27. Further, among the depressed adolescents, age was not significantly correlated with total number of episodes, r(54) = .12, p = .39.

⁶In contrast to earlier investigations, the three ISEL subscales in the current investigation were highly correlated. For this reason, we were unable to include all three ISEL subscales in the same model. The inclusion of all social support variables in the same model would result in unexpected suppression effects due to the partialing that occurs within path models. More important, however, a model including all perceived social support variables would be much more complex than those estimated, rendering the size of the sample in the current article insufficient to the task. We therefore retained four separate models to optimize the robustness and clarity of the parameter estimates we report. It is notable, however, that we do find a similar pattern of results when a model including all three subscales is run. The results of this model are available from the authors by request.