Exploring the Effects of Collapsing Rating Scale Categories in Polytomous Item Response Theory Analyses

• In Event: I-Presentation Gallery

Abstract

The purpose of this study is to examine how various approaches to collapsing categories for ordinal rating scales impact rating scale functioning. We used empirical data to demonstrate several category-collapsing schemes and their effect on rating scale functioning. Next, we used a simulation study to explore the effects of various collapsing schemes on rating scale category ordering and precision. Overall, our results suggested that when collapsing the original rating scale to three categories, disordered thresholds and narrow-ranged categories were greatly reduced. Our simulation findings further suggested that aligning the collapsing scheme with the location of the rating scale malfunctioning was generally effective in improving scale performance. We discussed implications for research and practice and directions for further investigation.

Authors

• Chia-Lin Tsai, University of Northern Colorado

• Stefanie A. Wind, University of Alabama

• Samantha Estrada, University of Texas - Tyler