Hostile Media Perceptions in the Age of Social Media: Following Politicians, Emotions, and Perceptions of Media Bias

1. Treating W2 emotions as mediating variables suggests emotions should have a more immediate relationship with hostile media perceptions. However, we also tested a model with W1 emotions as mediators. Results are consistent, as following politicians on social media is related to enthusiasm about the supported candidate (b = .13 (.04), p < .001) and anger at the opposed candidate (b = .11 (.04), p = .002). Unlike the original model, following a politician on social media is related to anxiety about the opposed candidate (b = .09 (.03), p < .01), though anxiety is not related to hostile media perceptions. Like the original model, W1 enthusiasm toward the supported candidate is related to hostile media perceptions (b = .20 (.03), p < .001), as is anger at the opposed candidate, though the latter variable falls short of conventional standards of statistical significance if a two-tailed test is used (b = .07 (.04), p = .06).

2. Measuring emotions in both waves allows for models that include a lagged measure of emotion to predict changes in this variable over time. Controlling for W1 emotions, following politicians on social media is not significantly associated with W2 emotions, as most of the variance is explained by emotions in the previous wave. Enthusiasm toward the supported candidate in W2 remains significantly related to hostile media perceptions (b = .34 (.06), p < .001). Anger at the opposed candidate is also related to hostile media perceptions, though the coefficient here does not fall below conventional standards of statistical significance if a two-tailed test is used (b = .10 (.06), p = .065). Although the data do not provide evidence that following politicians on social media changes emotional responses toward candidates over time, there remains strong and consistent evidence that following politicians on social media is associated with enthusiasm toward the supported candidate and anger at the opposed candidate over the course of the election cycle.

3. We also tested the possibility of reverse causal order. Because hostile media perceptions were only measured in W2, we cannot use W1 perceptions to predict W2 social media following behavior or emotions. We therefore examine how hostile media perceptions predict emotional responses and following politicians on social media in W2 only. Individuals who perceive the media as hostile toward their candidate are more angry (b = .18 (.04), p < .001) and anxious (b = .08 (.03), p = .02) at the opposed candidate, and more enthusiastic about the supported candidate (b = .34 (.04), p < .001). Anger directed at the opposed candidate (b = .10 (.04), p = .03) and enthusiasm toward the supported candidate (b = .10 (.04), p = .004) are associated with following politicians on social media cross-sectionally in W2.