Stability of preferences in longitudinal stated preference surveys: An application to COVID-19 vaccines across 9 countries

Abstract

Researchers commonly discuss possible sources of preference instability in the context of stated choice surveys, primarily linked to learning and fatigue. However, with the reliance on one-off surveys, capturing data at a single point in time, almost all work relies on the assumption that results are stable over time. This paper discusses how, especially in the context of important exogenous changes, analysts should also give credence to the possibility that results might be different if data had been collected at different points in time. We propose a structured approach for testing choice and preference consistency across repeated waves of stated choice surveys, and illustrate this on the basis of a COVID-19 vaccine survey carried out in 3 repeated waves across 9 countries. We show that while there is overall a high degree of consistency, variations in preferences exist at the level of individual explanatory variables, and this differs across countries