Survey Measurement of Complex Latent Constructs: An Example from the National Long Term Care Survey (NLTCS)

• Activity Number: 351
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Elena A. Erosheva*+
• Companies: University of Washington
• Keywords: activities of daily living ; chronic disability ; survey design ; two-phase data collection

Abstract

We examine chronic disability measurement from two linked NLTCS interviews, the screen and detailed, where the detailed interview was typically administered to those cases that were identified as disabled by the screen. We use cumulative logit models for matched pairs to compare activity of daily living (ADL) disability measures from the two interviews, and to examine whether discrepancies between these measures are associated with either survey- or respondent-level characteristics. Our analyses show that the detailed interview produced significantly higher disability than the screen, implying that a number of people who have been identified by the screen as non-disabled on ADLs would have been identified as ADL-disabled by the detailed interview if it was administered. We discuss implications and provide recommendations for survey researchers dealing with complex latent constructs.