A Shared Parameter Location Scale Item Response Theory (IRT) Model for Repeated Ordinal Questionnaire Data
*Donald Hedeker, University of Chicago
Keywords: location-scale, ordinal data, variance modeling
Questionnaires are commonly used in studies of health to measure severity of illness, for example, using ordinal items. For such questionnaires, item response theory (IRT) models provide a useful approach to obtaining summary scores for subjects and characteristics of the items. These item parameters characterize the mean model and indicate the level of item endorsement (difficulty) and the degree to which the item separates subjects of varying ability levels (discrimination). In this presentation, we also model the within-subjects variance and allow for a random subject scale effect. Furthermore, we model the ordinal responses from a questionnaire administered at two timepoints that share the random effects (both location and scale). We illustrate application of this shared parameter location scale IRT model using data from the Nicotine Dependence Syndrome Scale (NDSS) assessed in an adolescent study at two timepoints. We show that both the location and scale parameters from the first timepoint are related to the location effects at the second timepoint.