Main Article Content
Brief Report
Reliability of the Beck Depression Inventory and the Self-Rating Anxiety Scale in a sample of South African adolescents
Abstract
Objective — This study aimed to
assess the reliability of the Beck Depression Inventory (BDI) and the Self-Rating
Anxiety Scale for epidemiological investigations of adolescents' symptoms.
Method — Self-report questionnaires
were administered on two occasions to 104 students in four private high schools
in Cape Town, South Africa. Test-retest reliability was assessed using Cohen's
kappa and observed agreement. Internal consistency of each scale was assessed
using Cronbach's alpha.
Resuls — All items of the BDI and
all but four of the Self-Rating Anxiety Scale had at least fair Kappa scores.
The remaining four items of the Self-Rating Anxiety Scale had reasonable
observed agreement and reflect more transient states that are likely to change
between administrations.
Conclusions — The results give encouraging
indication that these instruments may be reliable in developing contexts.
However, the study is limited by a sample that is small and not sufficiently
representative of the broader South African population. Future studies should
include larger and more representative samples, in order to explore item
bias in different race and gender subgroups. Validity also needs investigation.
Journal of Child and
Adolescent Mental Health 2003, 15(1): 73-75
assess the reliability of the Beck Depression Inventory (BDI) and the Self-Rating
Anxiety Scale for epidemiological investigations of adolescents' symptoms.
Method — Self-report questionnaires
were administered on two occasions to 104 students in four private high schools
in Cape Town, South Africa. Test-retest reliability was assessed using Cohen's
kappa and observed agreement. Internal consistency of each scale was assessed
using Cronbach's alpha.
Resuls — All items of the BDI and
all but four of the Self-Rating Anxiety Scale had at least fair Kappa scores.
The remaining four items of the Self-Rating Anxiety Scale had reasonable
observed agreement and reflect more transient states that are likely to change
between administrations.
Conclusions — The results give encouraging
indication that these instruments may be reliable in developing contexts.
However, the study is limited by a sample that is small and not sufficiently
representative of the broader South African population. Future studies should
include larger and more representative samples, in order to explore item
bias in different race and gender subgroups. Validity also needs investigation.
Journal of Child and
Adolescent Mental Health 2003, 15(1): 73-75
