aiDIF: Differential Item Functioning for AI-Scored Assessments

Detects and quantifies differential item functioning (DIF) in AI-scored educational and psychological assessments. Provides a fully self-contained robust DIF engine (M-estimation via iteratively re-weighted least squares with the bi-square loss) alongside the Differential AI Scoring Bias (DASB) test, which detects item-level scoring shifts that differ across subgroups when comparing human and AI scoring conditions. Supports independent and paired scoring designs, robust linking of the cross-condition contrast, multiplicity control, conversion of fitted 'mirt' models to package inputs, simulation utilities, anchor weight diagnostics, and an AI-effect classification framework. Methods follow Halpin (2024) <doi:10.1007/s11336-024-09957-6>.

Version: 0.2.0
Depends: R (≥ 3.5.0)
Imports: Matrix, stats, graphics, utils
Suggests: mirt, testthat (≥ 3.1.5), knitr, rmarkdown
Published: 2026-09-05
DOI: 10.32614/CRAN.package.aiDIF
Author: Subir Hait ORCID iD [aut, cre]
Maintainer: Subir Hait <haitsubi at msu.edu>
BugReports: https://github.com/causalfragility-lab/aiDIF/issues
License: GPL (≥ 3)
URL: https://github.com/causalfragility-lab/aiDIF
NeedsCompilation: no
Citation: aiDIF citation info
Materials: README, NEWS
CRAN checks: aiDIF results

Documentation:

Reference manual: aiDIF.html , aiDIF.pdf
Vignettes: Introduction to aiDIF (source, R code)

Downloads:

Package source: aiDIF_0.2.0.tar.gz
Windows binaries: r-devel: aiDIF_0.1.0.zip, r-release: aiDIF_0.1.0.zip, r-oldrel: aiDIF_0.1.0.zip
macOS binaries: r-release (arm64): aiDIF_0.1.0.tgz, r-oldrel (arm64): aiDIF_0.1.0.tgz, r-release (x86_64): aiDIF_0.1.0.tgz, r-oldrel (x86_64): aiDIF_0.1.0.tgz
Old sources: aiDIF archive

Reverse dependencies:

Reverse suggests: aiEvalR

Linking:

Please use the canonical form https://CRAN.R-project.org/package=aiDIF to link to this page.