Abstract In analyzing data deriving from the administration of a questionnaire to a group of individu-
als, Item Response Theory (IRT) models provide a flexible framework to account for several aspects
involved in the response process, such as the existence of multiple latent traits. In this paper, we focus
on a class of semi-parametric multidimensional IRT models, in which these traits are represented
through one or more discrete latent variables; these models allow us to cluster individuals into homo-
geneous latent classes and, at the same time, to properly study item characteristics. In particular, we
follow a within-item multidimensional formulation similar to that adopted in the two-tier models,
with each item measuring one or two latent traits. The proposed class of models may be estimated
through package MLCIRTwithin, whose functioning is illustrated in this paper relying on examples
based on data about quality-of-life measurement and propensity to commit a crime.


看到一篇方法学的文章,供大家学习参考.