ABSTRACT
Item response theory (IRT) is an approach used to estimate how much of a latent trait an individual possesses. The theory aims to link individuals’ observed performances to a location on an underlying continuum of the unobservable trait. Because the trait is unobservable, IRT is also referred to as latent trait theory-the literal meaning of “latent” is hidden. An example of a latent trait is a test taker’s English reading ability. The construct of reading ability is not observable, so measuring this ability requires observable performances designed to assess it. For instance, test takers could read short passages and then answer multiple-choice questions designed to measure their comprehension of the passages. IRT can then be used to relate these observable performances, scores on the multiple-choice reading test items, to test takers’ underlying reading abilities. IRT can be used to link observable performances to various types of underlying traits. For instance, it can be used to connect individuals’ observed or self-reported anxiety about an assessment to their underlying levels of test anxiety. The following definitions are helpful for understanding IRT concepts as discussed in this
chapter. Latent variables are unobservable traits like second language listening ability, which influence observable behaviors, such as individuals’ responses to items on a second language listening assessment. The term latent variable is used synonymously with the terms construct and underlying trait throughout the chapter. A model is a mathematical equation in which independent variables are combined to optimally predict dependent variables (Embretson and Reise, 2000). Various IRT models, some of which are discussed in this chapter, are used to estimate individuals’ underlying traits on language ability constructs. Parameter is used in IRT to indicate a characteristic about a test’s stimuli. For instance, a 1-parameter (1PL) model mathematically defines only one characteristic of a test item (Osterlind, 2010).
