Identifying Response Styles Using Person Fit Analysis and Response-Styles Models

• In Event: Rasch Models: Considerations and Use

Abstract

Response styles are systematic patterns of responses in which respondents tend to select certain response categories regardless of their locations on the construct. Researchers have proposed approaches such as the Partial Credit Model Response Styles (PCMRS) to model and control for response styles. Other researchers have observed that person fit statistics can identify response styles. We used simulated and real data to explore the correspondence between the PCMRS and person fit statistics from the Partial Credit Model (PCM). Preliminary results suggest strong linear relationships between PCMRS response style estimates and person fit statistics from the PCM. These results suggest that researchers can identify response styles as a type of construct-irrelevant variance using simple models such as the PCM.

Authors

• Stefanie A. Wind, University of Alabama

• Yuan Ge, University of Alabama