Abstract
BACKGROUND: The attitudes of healthcare professionals towards HIV positive patients and high risk groups are central to the quality of care and therefore to the management of HIV/AIDS related stigma in health settings. Extant HIV/AIDS stigma scales that measure stigmatising attitudes towards people living with HIV/AIDS have been developed using scaling techniques such as principal component analysis. This approach has resulted in instruments that are often long. Mokken scale analysis is a nonparametric hierarchical scaling technique that can be used to develop unidimensional cumulative scales. This technique is advantageous over the other approaches; as the scales are usually shorter, while retaining acceptable psychometric properties. Moreover, Mokken scales also make no distributional assumptions about the underlying data, other than that the data are capable of being ordered by item and by person. In this study we aimed at developing a precise and concise measure of HIV/AIDS related stigma among health care professionals, using Mokken scale analysis. METHODS: We carried out a cross sectional survey of healthcare students at the Monash University campuses in Malaysia and Australia. The survey consisted of demographic questions and an initial item pool of twenty five potential questions for inclusion in an HIV stigma scale. RESULTS: We analysed the data using the mokken package in the R statistical environment providing a 9-item scale with high reliability, validity and acceptable psychometric properties, measuring and ranking the HIV/AIDS related stigmatising attitudes. CONCLUSION: Mokken scaling procedure not only produced a comprehensive hierarchical scale that could accurately order a person along HIV/AIDS stigmatising attitude, but also demonstrated a unidimensional and reliable measurement tool which could be used in future studies. The principal component analysis confirmed the accuracy of the Mokken scale analysis in correctly detecting the unidimensionality of this scale. We recommend future works to study the generalisability of this scale in a new population.