2007 HSR&D National Meeting Abstract
2012 — Applications of Item Response Theory Methodologies in Monitoring Healthcare through the Continuum of Care
Velozo CA (Rehabilitation Outcomes Research Center)
Item response theory (IRT) statistical methodologies are gaining considerable attention in health services research due to their usefulness in evaluating instruments, generating crosswalks between healthcare measures, and developing item banks for computer adaptive testing (CAT). Using IRT, instruments can be evaluated at the item-level, allowing the identification of ceiling/floor effects and other gaps where item difficulties do not match person abilities. Since the traits measured with IRT are "test free", this methodology provides a statistically-sound means to create crosswalks between instruments that measure the same trait, thereby allowing the monitoring of healthcare outcomes across settings that use different instruments. CATs allow assessments to be tailored for the individual, whereby the respondent receives only those items most relevant to his/her ability level. Through the combination of IRT methodologies and computer technology, CATs dramatically reduce respondent burden with little loss in measurement precision. IRT measurement methodologies offer a number of advantages that have not been previously available. These methods provide a means for improving veteran healthcare by increasing measurement sensitivity, permitting continuous healthcare monitoring across the continuum of care and decreasing the participation burden of veterans who volunteer for our healthcare studies. These advantages are critical as we attempt to monitor and improve the healthcare of returning Operation Enduring Freedom/Operation Iraqi Freedom veterans. The purpose of this workshop is to introduce participants to IRT models, provide examples of applications in health services research, and show participants how to run basic Rasch analyses.
IRT statistical formulas will be presented, showing how the mathematical formulas lead to measures that are test free. Illustrative examples will be used to show participants how to identify ceiling/floor effects and produce output reports that connect the measures obtained on instruments to the qualitative content of the instruments. Participants will be shown the basics of using a downloadable version of the Winsteps Rasch software program to analyze data commonly collected by health services researchers.
Research Health Scientists
Assumed Audience Familiarity with Topic:
Basic to intermediate