Magical Thinking and the Psychological Roots of the Hostile Media Effect: Evidence from the 2016 Presidential Election
Abstract
The hostile media effect is a robust, well-known media effect in the study of political communication. Typically, hostile media effect studies show participants content that is balanced and find that partisans on both sides perceive bias. This paper uses data collected from a survey experiment (N = 1,093) conducted in the week leading up to the 2016 election to test for two variations on traditional HME research. First, we test for relative hostile media effects in balanced election poll reports by presenting participants articles manipulate who is numerically ahead in each condition (Clinton ahead, Trump ahead, dead heat), keeping everything else constant. Beyond examining the hostile media perception with imbalanced, but unbiased information, we also introduce a measure to help unpack the psychological roots of the hostile media effect: magical thinking - the idea that there are imperceptible forces (essences) that drive, carry, or provide the mechanism for causal effects. Magical thinkers not only assume that hidden, invisible powers are behind much of what happens in the world, but that this explanation is more correct than an empirical one. At base, magical thinking involves a correspondence or conflation between the subjective, internal world and external reality. We develop a measure of magical thinking within individuals and suggest that those who experience the strongest hostile media effect are strong partisans who are magical thinkers in unbalanced conditions in which the imbalance favors a candidate the participant opposes. However, we hypothesize that the most extreme magical thinkers are the most likely to experience a hostile media effect, even when the imbalance is in their favor. The results demonstrate both the presence of strong relative hostile media effect and that magical thinking items predicted increased perceptions of bias against Trump and Clinton even after controlling for party identification, personality traits, perception of journalists/media, and the content of the poll report they read, but the results were asymmetrical between candidates. The implications of these findings as well as suggested refinements of our measures are offered.