Mokken scale analysis and confirmatory factor analysis of the Health of the Nation Outcome Scales

The Health of the Nation Outcome Scales is/are widely used in many countries and have been suggested for use within the Payment By Results framework in England. There are, however, questions over the best way to score it. Originally it was promoted as twelve independent scales, although more recently several alternative models of its factor structure have been proposed. Research so far has concentrated on the use of parametric methods of analysis that may be inappropriate. In this study, we examined the structure of 80,161 completed HoNOS scores using Mokken scale analysis, which is a nonparametric form of item response theory. Confirmatory factor analysis was also conducted on the proposed scales. Two possible subscales were found, which correspond to a Depression subscale and a Social and Cognitive Problems subscale. Neither scale had strong Mokken scale properties, particularly when compared with other scales. Confirmatory factor analysis suggested that the two subscale model had acceptable fit statistics. It is clear that the items cannot be considered as twelve independent items or indeed as a unidimensional scale. Given the relative psychometric weakness of HoNOS, it may be advisable to develop a new measure or at least to consider alternative measures of outcome.