Grade of Membership Analysis: Newest Development with Application to National Long-Term Care Survey Data

Mikhail Kovtun, Duke University
Igor Akushevich, Duke University
Kenneth G. Manton, Duke University
H. Dennis Tolley, Brigham Young University

Newest development in the theoretical foundation of the Grade of Membership (GoM) analysis performed by authors brings new insights into the value of GoM analysis, providing new ways to interpret estimates as well as new numerical methods to obtain estimates. In this presentation we expose most important results recently obtained by the authors and apply the new methodology to the National Long Term Care Survey (NLTCS) data. We analyse the results obtained by providing interpretation for the estimates and by investigating the errors arising from the limited size of the sample. This analysis shows that GoM model performs well in the case under consideration.