Using IRT variable maps to enrich understanding of rehabilitation data.

Abstract

One of the benefits of item response theory (IRT) applications in scale development is the greater transparency of resulting scores. This feature allows translation of a total score on a particular scale into a profile of probable item responses with relative ease, for example by using the variable map that often is part of IRT analysis output. Although there have been a few examples in the literature using variable maps to enrich clinical interpretation of individual functional assessment scores, this feature of IRT output has received very limited application in rehabilitation research. The present paper illustrates the application of variable maps to support more in-depth interpretation of functional assessment scores in research and clinical contexts. Two examples are presented from an outcome prediction study conducted during the standardization of a new functional assessment for elementary school students with disabilities, the School Function Assessment. Two different applications are described: creating a dichotomous outcome variable using scores from a continuous scale, and interpreting the meaning of a classification cut-off score identified through Classification and Regression Tree (CART) analysis.