Increasing Diagnosticity of Low-Stakes Tests With Adaptive Testing

• In Event: Modern Psychometric Methodologies and Applications

Abstract

The purpose of this study was to use variable length computerized adaptive testing (VCAT) of reading comprehension to reduce testing time and test length while maintaining or improving measurement precision, compared to a traditional fixed-item version of the reading comprehension assessment. Then, some of the saved testing time was allocated to administering additional items to gather diagnostic information for struggling readers. This diagnostic information was used for classification purposes, which would be useful to teachers in designing additional support for those struggling readers. While an earlier simulation study by Davison et al. (2022) supported the hypothesis that VCAT can reduce test length while maintaining measurement precision and diagnostic accuracy, we examined this hypothesis using live testing data.

Authors

• Yun Leng Wong, University of Minnesota

• Joseph N. DeWeese, University of Minnesota

• Jiayi Deng, HumRRO

• Mark L. Davison, University of Minnesota