Trump supported it?! A Bayesian source credibility model applied to appeals to specific American presidential candidates' opinions

The credibility of politicians is crucial to their persuasiveness as election candidates. The paper applies a parameter-free Baysian source credibility model (integrating trustworthiness and epistemic authority) in a real-life test predicting participants’ posterior belief in the goodness of an unnamed policy after a named candidate has publically supported or attacked it. Two studies test model predictions against policy support and attack of five presidential candidates from the USA. Model predictions were measured against observed posterior belief in the goodness of the policy. The results strongly suggest the model captures essential traits of how participants update beliefs in policies given appeals to a candidates’ support of attack. Further, individual differences suggest that people consider other factors than the ones elicited for the study. More studies into appeals to specific candidates are warranted to construct more accurate models of the influence of source credibility on political reasoning.