At the conclusion of the webinar, participants would have an expanded knowledge of the basic principles of CAT as well as a deeper conceptual understanding of why it works and what the costs/benefits are. The webinar will be structured around the three words in the phrase “computerized adaptive testing” and a deeper exploration of what each implies.
Computerized refers to the fact that CAT is most often done in some kind of computer-based environment.Computerized adaptive testing (CAT) refers to a collection of systems and statistical models that assemble an assessment based on the observed item responses of each individual. This happens in real time, such that a response to the first item will help determine which item should be administered next.
Measuring patients, whether it is using their native devices within a clinical trial setting or a tablet while waiting in a clinic’s waiting room, is an increasingly common practice. Doing assessment in a computerized environment adds unique capabilities such as tracking response times and greatly increasing the scope of what constitutes an “item.” In addition, respondents could select their preferred language or even increase font size in real time. Adaptive assessment occurs when the items being presented are chosen, at least in part, by the responses to previous items. There is a long history of adaptive design in the clinical trial realm and this is a good analogy for the core statistical ideas underlying and supporting CAT. Items are chosen to be administered from a pool of items (called an item bank) based on a statistical assessment of which item would be most informative for a given individual. Testing reminds us that, despite the first two words, we still need to conduct sound measurement. CAT enables us to make valid inferences based on reliable scores while administering as few items as possible.
After the webinar, the learner will be able to:
- To articulate in a non-statistical way how adaptive testing works
- To articulate costs and the benefits of an adaptive testing strategy within research and clinical PRO frameworks
- Dr. Michael C. Edwards, PhD, Managing Partner/Associate Professor, Vector Psychometric Group, LLC/Arizona State University
- Dr. Bryce Reeve, PhD, Director, Duke University, Center for Health Measurement
RJ Wirth, RN, Vector Psychometric Group, LLC