Psychometric properties of the revised Diabetes Knowledge Test using Rasch analysis

Abstract

OBJECTIVE: The revised Diabetes Knowledge Test (DKT2) is the instrument that is the most widely used for measuring diabetes knowledge; nevertheless, its structural validity has never been evaluated. This study aimed to determine the psychometric properties of the DKT2. METHODS: This study performed a secondary data analysis of people with diabetes recruited at university hospitals. Rasch analysis was used. RESULTS: One item in the 14-item DKT2 measuring general diabetes knowledge exhibited a poor-fit, and so it was eliminated. The person-item map showed that items of greater difficulty need to be added to the instrument. The principal-components analysis of residuals revealed a unidimensional structure. The person reliability was 0.50, with a person separation index of 1.01. Measurement invariance was not satisfied for items 11 and 2 according to gender and age. CONCLUSIONS: The unidimensional structure of the 13-item DKT2 demonstrated poor person reliability and a low person separation index. Females and elderly persons found it more difficult to respond to items 11 and 2, respectively. PRACTICE IMPLICATIONS: Careful consideration is necessary when using the 13-item DKT2 in practice and research. It is recommended for future studies to add items of greater difficulty to the instrument.