Abstract:
|
Interest in the analysis of patient-focused measures, including patient-reported outcomes measures (PROMs) and patient-reported experience measures (PREMs) continues to grow as attempts are made to understand and assess latent (i.e., unobserved) traits such as quality of life and mental health. Moreover, large-scale international surveys continue to inform policy makers about latent health-related behaviors and knowledge.
The proliferation of survey instruments to measure PROMs, PREMS, and health behaviors requires valid and reliable measures that are invariant between known population groups and over time. Detection of Differential Item Functioning (DIF) and Response Shift (RS) is relevant for these measures. In addition, instrument length optimization is important to avoid respondent fatigue while still accurately measuring the latent trait of interest.
Borrowing from educational testing literature, discussion will focus on how existing psychometric models (e.g., Item Response Theory), DIF and RS detection techniques (e.g., Rasch Trees, Response Shift Algorithm in IRT), and test construction methods can be used and even improved in health policy research.
|