An Introduction to Item Response Theory (IRT) for public health professionals

The purpose of this session is to introduce participants to the interpretive theory and application of item response theory (IRT) models and provide experience interpreting the outcomes of analyses and studies employing these models. Health care decisions, population estimates, treatment efficacy and effectiveness studies, and prevention efforts frequently rest on fallible individual self-report measures. To adequately understand differences in disease, health status, and outcome; research must establish the psychometric adequacy of measures used to assess these statuses. This essential task is faced by all public health scientists. IRT, a modern measurement modeling technique, offers a powerful method to empirically establish the measurement properties of instruments used in research. Whether conducting or consuming public health research, individuals increasingly face studies employing IRT models, yet few individuals receive even basic training in these models. The purpose of this session is to address this problem by providing an interpretive introduction to modern psychometrics and IRT. The course will briefly cover basic psychometric concepts, describing the importance of establishing reliable and valid measurement in public health research, and then focus on modern IRT measurement models. Material will cover the Rasch , 2PL, and 3PL models and their associated measurement parameters. Through the Graded response model, Polytomous models will see similar addressment. The course will provide examples of the models in real data and describe their interpretation. The program will address the settings and manner in which IRT models may be employed to increase the validity and reliability of public health research. Finally, it will describe the use of IRT models in a variety of settings, e.g. computer adaptive testing, measurement bias studies, and test equating studies.