Item Response Theory: A Conceptual Introduction
Faculty - Karon F. Cook, PhD
This module would introduce item response theory (IRT) to researchers wishing to gain a general understanding. It is conceptual in approach and intentionally avoids the use of mathematical equations (the few that are presented are described element by element). This approach was selected for several reasons. First, substantial training would be required to become a practitioner of IRT. The software that develops IRT models and calculates scores based on them is notoriously technical and "un-friendly". In addition, IR has a number of assumptions that must be investigated using other technical and "un-friendly" software (e.g., using MPLUS to evaluate local dependency based on residual correlations). Training in these methods and full comprehension of the "whys" of each is outside the limits of a one hour presentation.
What can be accomplished in one hour is an intuitive introduction to IRT. With such an introduction, for example, clinical researchers would be prepared to: 1) consult with a statistician or psychometrician regarding what IRT could (and could not) bring to their project, 2) have a working understanding of the differences between IRT and classical methods, and 3) be able to read a study that used IRT and have a conceptual understanding of the methods used.
The content for this module will be divided into 6 subsections, each lasting less than 20 minutes. The sections are: 1) Logic of IRT Scoring, 2) IRT is a Probability Model, 3) Plotting Along: How Item Responses are Related to Person and Item Characteristics, 4) Understanding Item Parameters: Difficulty and Discrimination, 5) Information: A Better Measure of Measurement Precision, and 6) Applications of IRT.