Equating test scores with R
Marie Wiberg, Umeå University
Abstract
The aim of test score equating is to adjust the test scores on different test forms so that they can be comparable and used interchangeably. The overall goals of the workshop are to help the attendees understand the principles of equating, to conduct equating, and to interpret the results of equating. Different R packages will be used to illustrate how to perform equating when test scores data are collected using different data collection designs. Traditional equating methods, item response theory (IRT) equating methods and kernel equating methods are included in the workshop.
Presenter
Prof. Marie Wiberg is a Professor at the Professor at Umeå School of Business, Economics and Statistics (USBE). Her research interests include educational measurement and psychometrics in general, especially test equating, parametric and nonparametric item response theory, and international large-scale assessments.
Equating test scores with R
The aim of equating is to adjust test scores on different test forms so that test scores can be used interchangeably. Equating has a central role in standardized testing, and it is an important step in the process of collecting, analyzing, and reporting test scores. Equating is important to be able to provide a fair assessment regardless which background the test takers have or which time the test is given at. This workshop has two explicit goals. The first goal is to introduce the concept of equating both theoretically and in practice through examples and exercises using different data collection designs. The second goal is to give useful tools for the attendees to perform different equating methods in practice using the R software with special focus on the packages; equate, equateIRT and kequate.
Content
The workshop starts by introducing traditional equating methods and different data collection designs. These methods are illustrated with the R package equate. This is followed by an introduction to item response theory (IRT) equating and the attendees will receive practical guidance on how to perform these methods using R by primarily using equateIRT. Then the attendees will be guided through the five steps of kernel equating: 1) presmoothing, 2) estimating score probabilities, 3) continuization, 4) equating and 5) evaluating the equating transformation, using primarily the R package kequate. Included in the workshop is also practical recommendations and examples how to perform equating to conduct a fair assessment regardless of at which time the test is given or if the background of the test takers differs. This workshop intends to give the attendees a broad knowledge of equating as well as how they should be performed within the R environment. A number of examples and exercises are provided. Attendees are expected to bring their own laptop with R and Rstudio installed together with the latest versions of the R packages equate, equateIRT, kequate, and SNSequate. Electronic materials (code and datasets) will be given to the attendees.