Title: The optimal number of response alternatives for a rating scale
Authors: Maria Iannario - University of Naples Federico II (Italy) [presenting]
Anna Clara Monti - University of Sannio (Italy)
Pietro Scalera - University of Naples Federico II (Italy)
Abstract: The most widespread models for the analysis of ordinal responses are the cumulative models with the proportional assumption, which implies that the covariates have the same effect on the cumulative odds regardless of the category of the response. This assumption allows for generating ordinal responses with a different number of categories from the same latent variable. By exploiting this latter property, the analysis investigates the asymptotic efficiency of the estimators and the power of the tests when the number of response alternatives varies. The impact of collapsing categories is also considered with special focus on dichotomization. A discussion in case of Additive Location-Shift Models is dealt with in its initial stage.