README.md

In nando11235813/raschreg: Item Response Theory models including explanatory variables of the ability parameter

raschreg

   

Overview

The library raschreg was created and mantained by a group of statisticians from Universidad de la República, Uruguay. It's primary pourpose is to provide R users with a framework for Item Response Theory (IRT) models, allowing for the possibility to include regression effects for the person parameters.

The models

Having a set of p items (Y1, Y2, ..., Yp) the probabilistic structure for the models is: logit(P(Yj=1|θi)) = αj(θi - δj) θi ~ N(0, 1) Where θi, δj and αj represent person, difficulty and discrimination parameters respectively.

raschreg provides a set of functions to fit, analyse and plot IRT models intended to analyse binary items. Currently supported models are: Rasch model (α1=α2=...=αp=1) Rasch model with free discrimination parameter (α1=α2=...=αp) * Two parameter logistic model

All previous models allow the posibility to incorporate explanatory variables (qualitative or quantitative) for person parameters. In this case: θi ~ N(X'β, 1)

Features

1. Models are estimated by maximum marginal likelihood, integrating person parameters via Laplace approximation. Estimated models are stored in rasch objects similarly to lm or glm objects.
2. Usual R methods for linear models (lm, glm) are implemented, such as:
   • coef
   • summary
   • vcov
   • anova
   • update
   • among others
3. Stored rasch objects con be visualized using:
   • Item characteristic curves and information curves
   • Person item maps
   • Forest plots

Installation

You can install the development version from GitHub using library devtools:

devtools::install_github("nando11235813/raschreg")

nando11235813/raschreg documentation built on Oct. 2, 2021, 3:11 p.m.