Title: Psychometrics in R: A Roundtrip through Latent and Not-so-Latent Variables

Author: Patrick Mair

Affiliation: Harvard University

Abstract:
From a methodological point of view it is pretty exciting to work in the field 
of psychology at the moment. Over the last few years, the advent of new
technologies such as eye-tracking and hand motion devices, and especially fMRI
(functional magnetic resonance imaging) scanners, have completely changed the
way statistical data analysis is done in psychology. The field seems to have
shifted from classical latent variable models towards a much broader
psychometric methods portfolio. This includes machine learning approaches in
fMRI, Bayesian cognitive modeling, all sorts of mixed-effects models, special 
functional methodology for analyzing eye-tracking and hand motion experiments,
as well as old-school latent variable models. How well does R accommodate the
needs of modern psychological research? The first part of the talk takes the
audience back to the early/mid 2000s when the first psychometric packages were
implemented and some initiatives were started in order to boost psychometric
developments in R (JSS special issue, Psychometrics Task View on CRAN). Since
these early developments there has been a tremendous increase in psychometric
package submissions developed by a vital psychometric computing community. Thus,
the second part provides an overview of the most important package developments,
discusses the status quo of psychometric computing and, more generally, the
current role of R in psychology. The third part sets the stage for future
developments, since there are still several gaps in the psychometric R package
ecosystem that need to be filled.