Limits on Learning? Selective Incorporation of Political Information

• In Event: Elections, Accountability, and Representation

Abstract

Does motivated reasoning limit the degree to which voters learn in response to political (mis)information, and the efficacy of interventions that aim to help them make informed decisions? A challenge in answering these questions is that motivated beliefs can be hard if not impossible to distinguish from sincere beliefs. In this paper, we develop a diagnostic tool to detect motivated reasoning which does not focus on changes in beliefs but on whether new information is accepted or remembered in the first place. We develop a formal model which shows how certain kinds of motivated reasoning make distinct predictions about what kinds of information are incorporated and remembered, setting aside how this information affects underlying beliefs. We then use this tool to analyze data from five field experiments – involving over 10,000 voters across Benin, Brazil, Mexico, and Uganda –, which find mixed effects of information treatments on voting behavior. We find that subjects who receive informational treatments are more likely to answer recall questions correctly when the information aligns with their partisan and policy preferences, which would appear to be a sign of motivated memory. However, we find similar patterns among the control group who received no information, which indicates the differences are likely driven by prior beliefs or guessing rather than selective memory. In sum, we find little evidence of motivated reasoning in this set of studies and provide guidance for how to test for this phenomenon in other settings.

Authors

• Melina Raquel Platas Izama, NYU Abu Dhabi

• Andrew Little, University of California-Berkeley

• Pia Raffler, Harvard University