The proposed project considers the role of survey design and question-framing in evaluating subjective response questions commonly used to elicit information about health and wellbeing. It also develops econometric techniques to control for biases that might arise as a result. The project will investigate three widely-used longitudina datasets, the Health and Retirement Study (HRS), the English Longitudinal Study of Ageing (ELSA), and the Survey of Health, Ageing, and Retirement in Europe (SHARE), and will lead to cross-country comparisons of the influences being studied. The proposed project develops econometric techniques to control for biases that might arise from survey design and question phrasing in secondary datasets and will address the following questions: (1) How do different characterizations of the points of a scale or the wording of a question affect the distributions of responses to survey questions about health or well-being? (2) Which individuals are most susceptible to influences of survey design such as question wording and sequencing? 3) To what extent do people evaluate questions in absolute terms, without consideration of their relation to a broader population, versus in relative terms, through consideration of a reference population? (4) If there is predictable variation in the effects of these influences, can that information be used to improve predictions of future health or economic outcomes of interest? (5) What role do perceptions play in health or economic decisions and outcomes? (6) Comparing responses across countries, what adjustments are necessary to account for reporting or attitudinal heterogeneity and improve inference? The extent to which these biases differ across sub-populations is important from a policy perspective since failure to control for attitudinal heterogeneity could lead to flawed assessments of either new or existing policies. On aggregate, incorrect inference could prove costly if behavior does not respond as predicted. The analysis will consider ways to measure reporting and attitudinal heterogeneity and to use this information to improve inference from models of future health and economic outcome variables (e.g., whether an individual has a heart attack, whether inflation outpaces retirees'income).
Often policymakers rely on research findings using survey data to evaluate existing policies and recommend new ones. Yet by their design, many survey questions contain an element of subjectivity that may be influenced by external factors or self-perceptions. The proposed research will develop techniques to identify and measure these influences, so policymakers can better predict the expected response to a proposed policy change.