This package is intended to be a "plug-and-play" tool for Bayesian Item Response Modeling. Currently, specific unidimensional IRMs are implemented, but future versions will aim on adding more flexibility to intermediary and advanced users.
birm depends on LaplacesDemon, an open-source statistical package that is intended to provide a complete environment for Bayesian inference.
This package should be considered experimental in this point of development. The following models are implemented:
- Rasch model (
rasch
function) - Partial and Generalized Partial Credit models (
pcm
function) - Rating and Generalized Rating Scale models (
rsm
function) - 1, 2, 3, 4, and 5-Parameter Logistic models (
plm
function) - Strict Rasch model (
sirm
function) - Bayesian Optimal Scoring (
optscr
function) - Generalized Graded Unfolding model (
ggum
function) - Fishburn models (
fishburn
function) - [Zero-Inflated] (Generalized) Rasch Poisson Counts Model (
rpcm
function)
All of the following estimation methods are available to use:
- Laplace Approximation (
method = "LA"
argument) - Variational Bayes (
method = "VB"
argument) - Hit-And-Run Metropolis (
method = "MCMC"
argument) - Population Monte Carlo (
method = "PMC"
argument) - Iterative Quadrature (
method = "IQ"
argument) - Maximum a Posteriori estimation with Simulated Annealing, Genetic Algorithm, Steepest 2-Group Gradient Descent, or Steepest Adaptive Moment Estimation (
method = "MAP"
andalgo="SANN"
,algo="GA"
,algo="SD"
, oralgo="ADAM"
; these methods were not originally implemented in LaplacesDemon)
Using the 'remotes' package:
install.packages("remotes")
remotes::install_github("vthorrf/birm")