To develop a psychometrically valid questionnaire for testing knowledge on micronutrients and to assess the relationship between knowledge and biomarkers of micronutrient status among adolescents.

Cross-sectional, institution-based, validity and reliability study.

Seven higher secondary schools were covered in the limits of Greater Hyderabad Municipal Corporation, Hyderabad, India.

Students aged 15–19 years, n 92 for the pre-test, n 108 for test–retest and n 109 for studying the relationship between knowledge and biomarkers of Fe, retinol, ascorbic acid, α-tocopherol, folic acid and vitamin B12 status.

From an item pool of 106, thirty-one items were selected based on content validity. Statistical tools to obtain a valid and reliable questionnaire among adolescent boys and girls resulted in eighteen items with a difficulty index of 0·11–0·86, discrimination index of 0·20–0·72 and validity index (point bi-serial correlation) of 0·10–0·62. Reliability as measured by Cronbach's α was 0·71 and the intra-class correlation coefficient was 0·80. A Bland–Altman plot showed good agreement between test and retest scores. The mean response score to the eighteen-item questionnaire was 5·2 (sd 2·68). The mean values of serum retinol were significantly different (P = 0·022) between groups below (24·8 (sd 6·64) μg/dl) and above (28·0 (sd 7·67) μg/dl) the 50th percentile of knowledge score. The relationship persisted after controlling for economic status as a covariate using analysis of covariance (P = 0·018). Other micronutrients did not show any significant relationship.

A valid and reliable eighteen-item knowledge questionnaire was constructed and found to have a significant positive relationship with plasma retinol status alone.

The objectives of the present study were to (i) develop and validate a norm-referenced performance-rating scale to interpret a nutrition knowledge test developed for urban adolescents and (ii) develop a prototype for other researchers to follow when developing nutrition knowledge tests.

For norm development the nutrition knowledge test (questionnaire) was administered to a sample representative of the questionnaire target group, referred to as the norm group. These included 512 adolescents in grades 8 (n 158), 10 (n 149) and 12 (n 205) at three randomly selected schools in Soweto and Johannesburg. The performance scores (in percentages) obtained by the norm group were transformed to Z-scores which were categorised into stanines using established Z-score cut-off points. For validation purposes the questionnaire was completed by 148 volunteers: sixty university dietetics students, nineteen non-nutrition university students and sixty-nine primary-school teachers.

As required of an ideal norm group, the Z-scores formed a normal distribution (a bell-shaped curve). To facilitate interpretation of the results, the Z-score cut-off points for these categories were transformed back to performance scores (percentages) so that the performance of a testee could be interpreted directly from his/her performance in percentage. As is recommended, the nine stanine categories were reduced to five: very poor, fair/below average, good/average, very good/above average and excellent. The discriminatory validity of the norms was substantiated by showing that groups with known nutrition knowledge levels were rated appropriately and that the performance ratings of these groups differed significantly, with university dietetics students scoring 98·3 %, primary-school teachers 20·3 % and non-nutrition university students 31·6 %.

The norm-referenced performance-rating scale can be used with confidence to interpret the performance score achieved by a testee on the nutrition knowledge test developed for urban adolescents in South Africa. The methodology used in the study serves as a prototype for other researchers who are developing knowledge tests.