Decoding Item Fit Statistics in Generalized Partial Credit and Graded Response Models

Objective In this instructional article, we aim to provide an accessible guide for counseling researchers and practitioners on interpreting item fit statistics within two common polytomous IRT models: the Generalized Partial Credit Model (GPCM) and the Graded Response Model (GRM).Method Using a simulated dataset (N = 839) from the 20-item Center for Epidemiologic Studies Depression Scale (CES-D), we demonstrate how to interpret item parameters, category response curves, item information, key item fit statistics (e.g. infit, outfit, χ2, G2, S-χ2, RMSEA), and graphical analysis methods.Results Results demonstrate how different item fit statistics may yield complementary or contrasting information, depending on the model structure and sample characteristics, emphasizing the need for contextual interpretation.Conclusion Evaluating item fit provides essential validity evidence based on internal structure and offers practical strategies for improving measurement precision and conceptual clarity in counseling research and assessment.