irt-psychometrics-edu

You can estimate item parameters using 2PL or 3PL item response theory by fitting logistic IRT probability functions to a binary response matrix. This Skill calculates discrimination, difficulty, and guessing parameters to generate item characteristic curves for ability testing.

Mantel-Haenszel DIF detection identifies biased test items by computing stratified odds-ratio statistics on response and group arrays. This Skill classifies differential item functioning using category thresholds and p-values to screen for group comparability issues.

To compute test information functions and standard error of measurement in IRT, this Skill aggregates item information functions across the ability scale. It derives measurement precision metrics by applying information and SEM formulas to the calibrated IRT parameters.

Yes, you can use this Skill for test equating across different test forms. It supports the Stocking-Lord concepts needed to convert scores between different test forms after you calibrate the 2PL or 3PL item parameters from your response data.

For IRT model calibration and DIF screening, you need a binary response matrix of test scores alongside demographic group arrays. The Skill processes these response and group arrays to estimate parameters and perform Mantel-Haenszel DIF classification.

You should use 3PL over 2PL models for educational measurement when guessing behavior is expected, as the 3PL model estimates an additional guessing parameter. This Skill implements both logistic IRT probability functions to fit your item characteristic curves accordingly.