IRT原理介绍
Extended Rasch Modeling: The R Package eRm
Patrick Mair
Wirtschaftsuniversität Wien
Reinhold Hatzinger
Wirtschaftsuniversität Wien
Marco J. Maier
Wirtschaftsuniversität Wien
Abstract
This package vignette is an update and extension of the papers published in the Journal
of Statistical Software (special issue on Psychometrics, volume 20) and Psychology
Science (Mair and Hatzinger 2007b,a). Since the publication of these papers, various
extensions and additional features have been incorporated into the package.
We start with a methodological introduction to extended Rasch models followed by
a general program description and application topics. The package allows for the computation
of simple Rasch models, rating scale models, partial credit models and linear
extensions thereof. Incorporation of such linear structures allows for modeling the effects
of covariates and enables the analysis of repeated categorical measurements. Item
parameter estimation is performed using cml, for the person parameters we use joint
ml. These estimation routines work for incomplete data matrices as well. Based on
these estimators, item-wise and global (parametric and non-parametric) goodness-of-fit
statistics are described and various plots are presented.
Keywords: eRm Package, Rasch Model (rm), lltm, rsm, lrsm, pcm, lpcm, llra, cml estimation.