Measuring Individual-level Issue Importance with Repeated Binary-choice Questions

Measuring individual-level issue importance is crucial for the study of voting behavior and public opinion, yet existing methods struggle to balance estimation accuracy with respondent burden. Standard approaches often sacrifice either ordinal precision or cardinal intensity to reduce cognitive load. To overcome this trade-off, I propose a pairwise comparison framework where respondents repeatedly choose the more important of two issues. Using a hierarchical Bayesian model, this method recovers latent continuous importance weights from simple binary choices, capturing both rank order and preference intensity. Simulation studies demonstrate that the method accurately recovers preference structures even with limited data. In an empirical application, the method exhibits concurrent validity with traditional ``Most Important Issue” (MII) questions while revealing granular individual differences. Finally, I discuss an application to a weighted spatial voting model, where estimated issue weights are used to weight policy dimensions in modeling vote choice.