Maximum information approach to scale description for affective measures based on the Rasch model

Using the Rasch model for ordered categories, this paper provides a method for qualitative interpretations of data from an affective measure such as an attitude scale or survey instrument. The Bock procedure is first used to partition the total item information to each response category. The location of each response category is determined by maximizing the information associated with this category. The response categories that cluster around a given point on the latent trait are then used to provide a qualitative description of this point. The description is explicit in terms of the behaviors or activities that are likely to be displayed by a respondent at this point. An illustration is provided using an alumni survey used at a large university in the southeast.