How philchalmers/mirt works
mirt is an R package for fitting item response theory (IRT) and related latent-variable models to discrete response data. The principal workflow is to provide response data and a model, estimate item and latent-trait parameters, then inspect results or derive scores and diagnostics. Separate capabilities include multi-group and mixed-effects analyses, latent-class and mixture models, and probabilistic unfolding. The README documents these capabilities; sampled C++ excerpts support the item-trace and E-step roles, but do not establish all R-level call wiring.
Main entry points: mirt.R Fit API · EMstep.group.R EM estimation · 01-itemclass.R Item models · multipleGroup.R Multiple groups
Components 21
Model setup
- Fit API
R/mirt.R - Model specification
R/mirt.model.R - Prepare data
R/04-PrepData.R - Item models
R/01-itemclass.R
Model estimation
- EM estimation
R/EMstep.group.R - MHRM estimation
R/MHRM.group.R - Likelihood
R/07-calcLogLik.R
Model variants
- Multiple groups
R/multipleGroup.R - Mixed effects
R/mixedmirt.R - Latent classes
R/mdirt.R - Mixture models
R/Mixture-methods.R - Unfolding models
R/Discrete-methods.R
Scoring and analysis
- Trait scores
R/fscores.R - Item analysis
R/itemfit.R - Fit diagnostics
R/M2.R - DIF analysis
R/DIF.R
Numerical core
- Item probabilities
src/traceLinePts.cpp - Expectation step
src/Estep.cpp - Parameter derivatives
src/dpars.cpp
Other components
- Analyst
- Response data
Connections 25
- Analyst → Fit API: fits model
- Response data → Fit API: supplies responses
- Fit API → Prepare data: prepares data
- Fit API → Model specification: uses specification
- Fit API → Item models: builds item models
- Fit API → EM estimation: estimates by EM
- Fit API → MHRM estimation: estimates by MHRM
- Fit API → Likelihood: evaluates likelihood
- Multiple groups → EM estimation: supports group fitting
- Mixed effects → MHRM estimation: supports mixed fitting
- Latent classes → EM estimation: fits class models
- Mixture models → EM estimation: fits mixture models
- Unfolding models → Item probabilities: computes probabilities
- Item models → Item probabilities: computes item traces
- EM estimation → Expectation step: computes expectations
- EM estimation → Parameter derivatives: computes derivatives
- Fit API → Trait scores: derives scores
- Fit API → Item analysis: supports item analysis
- Fit API → Fit diagnostics: checks model fit
- Fit API → DIF analysis: analyzes DIF
- Analyst → Multiple groups: requests analysis
- Analyst → Mixed effects: requests analysis
- Analyst → Latent classes: requests analysis
- Analyst → Mixture models: requests analysis
- Analyst → Unfolding models: requests analysis