potentially clinically depressed adolescents (Kovacs 1992). The
internal consistency for the NSCAW sample is 0.81 for 7–12-
year-olds and 0.87 for 13–15-year-olds (NSCAW 2009).
Data analysis
Adolescents whose caregivers reported on the presence or
absence of chronic illness were compared on their self-reports
of internalizing and externalizing symptoms and delinquency,
and their caregivers’ reports of internalizing and externalizing
symptoms. Additionally, adolescents with and without chronic
illness were compared on YSR and CBCL scale scores. Cases
were excluded from analyses for missing values of predictor or
outcome variables. These comparisons were completed using
analysis of variance (anova; a = 0.05, one-way).
To test whether adolescent depressive symptoms functioned
as a mediator between caregiver ratings of the adolescent’s
overall health and externalizing outcome variables, we used
bootstrapping methodology, a non-parametric resampling procedure
(Preacher & Hayes 2008). This process constructs 95%
confidence intervals (CI) for the size of the indirect effect of the
mediator using 2000 bootstrap resamples. A significant mediation
effect is indicated if the CI does not include zero; we
examined the bias-corrected and accelerated intervals as
a recommended improvement on traditional bootstrapping
methods (Efron 1987). To ensure accurate results, we used the
continuous measure of adolescent health quality in which caregivers
rated the child’s health on a five-point scale as the exogenous
variable.