Collaborative Research: A New Design for Identifying Persuasion Effects and Selection in Media Exposure Experiments via Patient Preference Trials

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

General Does media choice cause polarization or does polarization cause media choice? The PIs investigate the extent to which either or both of these two causal processes contribute to polarization of political attitudes. Adopting a procedure from the medical literature, the PIs design and implement an experiment that uses both a forced-choice and a free-choice scenario. Using this procedure allows for estimation of the degree of selective exposure, the degree of polarization by like-minded media, and the degree of persuasion across the partisan aisle by counter-attitudinal media. This method also involves estimation of bounds, or the range of treatment effect sizes that result in different subgroups under certain assumptions about the choice environment. Technical Does media choice cause polarization or does polarization cause media choice? We investigate the extent to which either or both of these two causal processes contribute to attitude polarization. Adopting the patient preference trial procedure from the medical literature, the PIs design and implement an experiment that uses both a forced-choice and a free-choice scenario. Using this procedure allows for estimation of the degree of selective exposure, the degree of polarization by like-minded media, and the degree of persuasion across the partisan aisle by counter-attitudinal media. Methodologically, the PIs also compute nonparametric bounds on the subgroup-specific treatment effects and conduct sensitivity analyses to demonstrate the reliance on certain assumptions of stable choices.