Kronecker-Structured Factor Analysis of Covariance Structure of Cross-Classified Health Care Quality Measures

• Activity Number: 376
• Type: Invited
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Alan M. Zaslavsky*+ and Laura Hatfield
• Companies: Harvard and Harvard Medical School
• Keywords: factor analysis ; health care quality ; hierarchical models ; repeated measures ; CAHPS survey

Abstract

In repeated surveys of complexly-structured populations, questions of interest often concern variables with a cross-classified structure. Examples include multiple measures assessed at multiple time periods or for multiple subgroups of interest, or coefficients of several covariates in models for several correlated outcomes. Conventional exploratory factor analysis summarizes patterns of covariance among an a priori unstructured collection of variables, but does not accommodate these multiway structures. We present a factor-analytic approach based on Kronecker products for level-2 parameters of hierarchical models. Our motivating applications arise from analyses of data from the CAHPS surveys of quality in the Medicare system.