CLASSICAL AND BAYESIAN INSTRUMENT DEVELOPMENT
Issue Date
2016-05-31Author
Garrard, Lili
Publisher
University of Kansas
Format
142 pages
Type
Dissertation
Degree Level
Ph.D.
Discipline
Biostatistics
Rights
Copyright held by the author.
Metadata
Show full item recordAbstract
Both patient-reported outcome measures (PROMs) and clinician-reported outcome (ClinRO) measures are recognized as essential tools for advocating patient-centered care, an important driving force behind the current U.S. health care system. Close collaborations among the research community and regulatory bodies have been initiated to form standardized guidelines for the development and evaluation of PROMs and many ClinRO measures that often are designed as psychometric instruments with ordinal response scales. Classical (i.e., frequentist) instrument development often is time-consuming and challenged by small samples (e.g., cases of rare diseases). An innovative Ordinal Bayesian Instrument Development (OBID) approach within a Bayesian Item Response Theory (IRT) framework is introduced to overcome both small sample size and ordinal data modeling challenges, through efficient integration of content validity and construct validity analyses. The performance of OBID is evaluated under a simulation setting with three different types of expert bias (i.e., unbiased, moderately biased, and highly biased), and further evaluated with an exact Bayesian leave-one-out cross-validation (LOO-CV) approach using real data applications. Results successfully demonstrated the OBID approach as a promising tool in future PROMs and ClinRO measures development for small populations or rare diseases. Alternatively classical psychometric methodologies are efficient and reliable with relatively large sample sizes. This study also presents the classical psychometric evaluation of the National Database of Nursing Quality Indicators® (NDNQI®) falls with injury measure, an essential ClinRO measure that supports health care quality improvement efforts and continuous injurious falls research.
Collections
- Dissertations [4701]
- KU Med Center Dissertations and Theses [464]
Items in KU ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
We want to hear from you! Please share your stories about how Open Access to this item benefits YOU.