Abstract:
|
Mappings of spatially-varying Item Response Theory (IRT) parameters are proposed, allowing for investigation of potential Differential Item Functioning (DIF) based upon geographic location without need for pre-specified groupings. This localized approach to IRT modeling provides a flexible framework, with current emphasis being on 1PL/Rasch and 2PL models. Applications to both simulated examination data and empirical survey data are presented to demonstrate the method, illustrate its benefits, and advocate for the use of local IRT modeling in a variety of contexts such as in the analysis of International Large-Scale Assessment data for education or in the analysis of Patient Reported Outcome Measure data for use in Mobile Health Applications. There is not only practical value with this method but also visual appeal when initial attempts to consider measurement invariance are made across national, state, or other political boundaries. The approach and the visualization it affords have great potential for affecting policy; regional disparities and latent spatial trends in item functionality can be identified and used in a beneficial manner regardless of geographic location.
|