Abstract

The application of item response theory (IRT) models and analysis in health-related and other noneducational outcomes is progressing rapidly, and with these applications new problems are emerging. As applications to new health domains and diverse populations increase, we anticipate that new challenges will continue to be identified. The large-scale application of IRT models across a variety of health constructs as part of the National Institutes of Health–sponsored Patient Reported Outcomes and Measurement Information System (PROMIS®) project has identified a number of such challenges (Cella et al., 2007; Cella et al., 2010). For example, as we raised in Chapter 1, problems with multidimensionality, narrowly defined constructs, unipolar constructs, and non-normal data are becoming well recognized as real challenges for IRT analyses in the patient-reported outcomes (PRO) domain.