Development of a reliable and valid nutritional knowledge questionnaire for urban South African adolescents

We wanted to develop a valid and reliable nutritional knowledge test for urban South African adolescents who were participating in the Birth-to-Twenty cohort study. The questionnaire was intended for use every second year, from ages 13 to 14 y until age 20 y.

The initial steps involved the development of a conceptual framework and identification of nutritional concepts in collaboration with nutritional experts, and this defined the construct of the questionnaire. The South African national teaching curriculum objectives for nutritional education and other relevant issues were selected as the desired concepts, and most items were phrased in accordance with the recently developed South African Food-Based Dietary Guidelines. Thereafter, 140 items (questions) were developed and in turn assessed by an expert panel, and the result was that only 88 items remained. This was done to ensure content and face validity of the items. The 88 items were constructed into a questionnaire and piloted for appropriateness and understanding by adolescents, ages 13 to 14 y, as a means of assessing face validity by non-experts. The edited preliminary questionnaire (still with 88 items) was administered to a nutrition expert group (n = 71) versus a non-expert group (n = 82), referred to as sample 1, for the purposes of performing item analysis and assessing construct validity of the questionnaire. The result of the analysis, a 63-item questionnaire, was administered to adolescents at three school grade levels, 8 (n = 128), 10 (n = 143), and 12 (n = 98), referred to as sample 2, which was representative of the grades in which the Birth-to-Twenty group will be in when the questionnaire is administered. The questionnaire was administered to the sample to assess its content validity and internal consistency reliability. The final questionnaire had 60 items, and its construct, content, and internal consistency reliability were reassessed.

The final 60-item questionnaire displayed a significant difference (P < 0.0001) in the mean scores of the expert and non-expert groups tested. It had internal consistencies (Cronbach's alpha) of 0.71, 0.79, and 0.82 for grades 8, 10, and 12 respectively, and an overall value of 0.77 for all groups combined. However, it was less than 0.7 for most grade 8 pupils and for all grades at a historically disadvantaged school.

A nutritional knowledge questionnaire with construct, face, and content validities and internal consistency was developed for use in South African adolescents to evaluate their nutritional knowledge. Internal consistency was low in children at a disadvantaged school and those in grade 8 compared with multiracial groups at a multiracial school. It is recommended that pupils at disadvantaged schools be assisted by trained interviewers when taking the test.