Supporters of political parties often disagree on fundamental political facts, but it remains unclear whether these differences of opinion are based on genuine convictions or are merely an expression of a particular pattern of behaviour. In a survey experiment involving 888 US participants, Hannon investigated how incentives to answer correctly affect responses to factual political questions. He found that incentives reduced party-political differences in factual reporting, with this effect being driven almost exclusively by Republicans. Democrats and independents showed little change in their answers when offered incentives. When Republicans received incentives, they answered questions favoured by Democrats more accurately, but were less accurate on questions favoured by their own party, without any overall improvement in accuracy. If incentives merely prompted respondents to think more carefully, we would expect a general improvement in accuracy; however, this was not the case. Rather, these patterns are consistent with expressive responses, but are also compatible with strategic responses, in which participants tailor their answers to the perceived expectations of the researchers. Our findings also suggest that a general reduction in polarisation under incentive conditions may mask important asymmetries between party groups and question types, making it difficult to draw conclusions from incentive-based experimental designs about what citizens actually believe.
A video recording of the talk can be found here.