Rasch Model-Based Evaluation of Toefl Listening Items: Analyzing Difficulty, Discrimination, and Fit

This study analyzed TOEFL Listening Section items using the Rasch. Quantitative analysis of 200 participants revealed significant problems: (1) 7 misfit items (Infit MNSQ 1.5), particularly items of extreme difficulty; (2) 4 items with negative discrimination, indicating non-construct variance contamination; and (3) measurement gaps in the Wright Map (dead zones for low-ability participants and ceiling effects for high-ability groups). The findings confirm the structural weaknesses of the test design, recommending item revision, strategic additions, and redistribution to enhance validity and assessment fairness. This study underscores the need for psychometrically sensitive approaches in high-stakes language assessment.