Isolating a primary dimension within the Cook-Medley hostility scale: a Rasch analysis

The empirically derived Cook-Medley Hostility scale (Ho) has been shown to be related to a number of poor health habits and general mortality. Researchers attempting to attach psychological meaning to Ho scores have had difficulty reaching consensus in defining a single construct to organize Ho items' content. While previous psychometric analyses of the 50 Ho items have revealed a strong primary dimension underlying responses, a number of small and unstable factors also influence Ho scores and thus cloud interpretations. The goal of this study was to better understand the relationship between individual Ho items and the primary latent dimension underlying responses to the Ho scale. We believe that this is a first step to clarifying interpretation of Ho scores as a psychological risk factor for poor health outcomes. Using a Rasch logistic item response model, we identified a subset of items that formed a reliable and unidimensional index that maintained strong relationships with the primary construct assessed by the 50-item Ho scale. The subset of 17 Ho items was not differentially influenced by gender in a large psychiatric sample, and the psychometric properties of the 17 Ho items were confirmed in a large medical sample.