lsasim: an R package for simulating large-scale assessment data

Abstract

This article provides an overview of the R package lsasim, designed to facilitate the generation of data that mimics a large scale assessment context. The package features functions for simulating achievement data according to a number of common IRT models with known parameters. A clear advantage of lsasim over other simulation software is that the achievement data, in the form of item responses, can arise from multiple-matrix sampled test designs. Furthermore, lsasim offers the possibility of simulating data that adhere to general properties found in the background questionnaire (mostly ordinal, correlated variables that are also related to varying degrees with some latent trait). Although the background questionnaire data can be linked to the test responses, all aspects of lsasim can function independently, affording researchers a high degree of flexibility in terms of possible research questions and the part of an assessment that is of most interest.