Jason Petroski,1 David Ferraro,2 and Adam Chu2
1Centers for Medicare & Medicaid Services—Office of Information Products and Data Analytics
2Westat, Inc.
Objective: The Medicare Current Beneficiary Survey’s (MCBS) Access to Care (ATC) file is designed to provide timely access to information on the Medicare population, yet because of the survey’s complex sampling design and expedited processing it is difficult to use the file to make both “always-enrolled” and “ever-enrolled” estimates on the Medicare population. In this study, we describe the ATC file and sample design, and we evaluate and review various alternatives for producing “ever-enrolled” estimates.
Methods: We created “ever enrolled” estimates for key variables in the MCBS using three separate approaches. We tested differences between the alternative approaches for statistical significance and show the relative magnitude of difference between approaches.
Results: Even when estimates derived from the different approaches were statistically different, the magnitude of the difference was often sufficiently small so as to result in little practical difference among the alternate approaches. However, when considering more than just the estimation method, there are advantages to using certain approaches over others.
Conclusion: There are several plausible approaches to achieving “ever-enrolled” estimates in the MCBS ATC file; however, the most straightforward approach appears to be implementation and usage of a new set of “ever-enrolled” weights for this file.
Keywords: Medicare, survey research and questionnaire design, access to care, complex sample design, sampling weights, biostatistical methods
doi: http://dx.doi.org/10.5600/mmrr.004.02.a05
Full text HTML for this article is not yet available. Please click the PDF download link to access this article.