The Promises and Pitfalls of Using Panel Data to Understand Individual Belief Change

We investigate whether studies on political belief change can identify change trajectories at the individual level. Using simulations and case studies, we propose a grid-search framework that allows researchers to evaluate the extent to which their target estimates generalize to their study population. We use simulated datasets to estimate plausible values for how many people changed, how much they changed, and who changed, based on observed response trajectories. Our results suggest that researchers should think carefully about the conditions under which they may make claims about belief change at the individual level. We provide an R package to help researchers implement this procedure in their own work.