Issues in survey measurement of chronic disability: An example of the National Long Term Care Survey

The National Long Term Care Survey, a longitudinal study that consisted of the screen and detailed interviews, provides a unique opportunity to illustrate issues in survey measurement of chronic disability. The original intent of the survey was to use the shorter measure of disability status in the screen interview to maximize the yield of disabled cases and then examine those cases with the longer measure in the detailed interview, identifying possible "false positive" disability cases from the screen. In this paper, we show empirically that the intended relationship between the screen and detailed NLTCS disability measures doesn't hold uniformly - the detailed measure provides significantly higher ADL disability estimates than the screen measure, contrary to the expectation. We examine whether discrepancies between the two disability measures are associated with certain features of the survey design and respondent-level characteristics, discuss implications of our findings for prevalence estimation, and provide recommendations for disability survey design.