CTT-based quiz difficulty rankings are validated by IRT analysis with near-perfect rank correlation¶

Claim · All claims
Evidence · 1 study · associational r2 · q3
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person · recheck by 2028: tool and model accuracy changes quickly

Subclaims¶

q3 i? A 1-parameter logistic (Rasch) IRT model produces quiz difficulty rankings that correlate nearly perfectly with classical test theory rankings for Immunology. → Parker 2025

Evidence¶

Parker 2025¶

Parker, M. J., & Zavala-Cerna, M. G. (2025). What Don't You Understand? Using Large Language Models to Identify and Characterize Student Misconceptions About Challenging Topics. https://arxiv.org/abs/2605.00294 doi:10.48550/arxiv.2605.00294

q3 · i? · associational · r2

IRT validation using the py-irt package on results from 765 Immunology students across three cohorts. "The Spearman correlation was .99, p < 10-5" between IRT-based and CTT-based quiz difficulty rankings, supporting the simpler CTT approach.

"the Spearman correlation against the CTT ranking of those same quizzes was taken. The Spearman correlation was .99, p < 10-5"

Discussion¶

Related Claims¶

• CTT and MFRM yield a positive, perfect correlation between item difficulty indices for open-ended items under a crossed design — related
• CTT and MFRM yield similar item difficulty results for open-ended questions regardless of crossed or nested design — related
• Models with similar rank correlations differ substantially in estimating difficulty magnitude (RMSE and MAE) — related
• Quiz-level first-attempt performance analysis across cohorts identifies persistently challenging topics in online biomedical science courses — related
• Items ranked from easiest to most difficult are identical under CTT and MFRM — related
• In operational reading-test data, item parameter estimates from a standard 2PL/GPCM that ignores local dependence closely match those from the testlet model (average correlations 0.9893 and 0.9978) — related