Reliability of self report questionnaires for epidemiological investigations of adolescent mental health in Cape Town, South Africa.

Objective: This study aimed to assess the reliability of a number of self report questionnaires for epidemiological investigations of adolescents' mental health in Cape Town, South Africa. The scales used were: the Short Mood and Feelings Questionnaire (SMFQ), Zung Self-rating Anxiety Scale (SAS), Self-esteem Questionnaire (SEQ), Harvard Trauma Questionnaire (HTQ) and Multi-Dimensional Scale of Perceived Social Support (MSPSS).

Method: The self-report questionnaire (available in Afrikaans, English and isiXhosa) was administered to 237 grade 8 students (14-15 years) on 2 occasions in metropolitan Cape Town high schools. The mean interval between first and second administration of the questionnaire was 8.3 days. Test-retest reliability was assessed using Cohen's kappa and observed agreement. Pearson's correlation coefficients were used to assess consistency across total scores between occasion 1 and occasion 2. Cronbach's alpha was used to assess the internal consistency of each scale.

Results: All items had at least fair test-retest reliability (kappa > 0.20) apart from two items on the Self-rating Anxiety Scale and one item on the Harvard Trauma questionnaire. Test-retest reliability was strong for the HTQ (Pearson's correlation coefficient >0.75), moderate for the SAS, SEQ and MSPSS (0.50-0.74) and weak for the SMFQ (0.25-0.49). Cronbach's alpha for all scales was acceptable (>0.60). Analysis by the different language versions (Afrikaans/English and isi-Xhosa/ English) of the questionnaire indicated good internal consistency for most measures for all three languages.

Conclusions: The results indicate that many of these instruments may be used reliably in South Africa to assess adolescent mental health and that the different language versions of the instruments used in the questionnaire are generally reliable for use in South African schools. However, some caution is required with the use of the SAS and SEQ in different language groups.