Character Deprecation in Fake News: Is it in Supply or Demand?

Jonathon McPhetres; David Rand; Gordon Pennycook

doi:10.1177/1368430220965709

In brief

Character-focused fake news headlines (attacking a politician's character rather than policy) may appear more often in pro-Republican content, but they are not more attractive to Republicans or less reflective people. In controlled experiments, both Democrats and Republicans prefer news with policy implications over character attacks, suggesting character deprecation may be in the supply of fake news but not in actual demand.

Contributions

Identifies character deprecation—false claims targeting a politician's character rather than policy implications—as a distinct type of fake news
Demonstrates in observational Study 1 that Republicans are more likely to share character-deprecation headlines than policy-important ones, compared to Democrats
Shows in controlled experimental Study 2 that when headlines are exactly matched across partisan divides, both groups prefer policy-important news and are less likely to share character-deprecation headlines
Reveals a disjuncture between what fake news creators supply (more character deprecation in pro-Republican content) and what consumers demand (content with policy implications)
Finds cognitive reflection (CRT performance) is unrelated to preference for character-deprecation content, suggesting it appeals broadly rather than to less analytical individuals

Method

Study 1 (Observational). Pre-tested 79 fake news articles for character focus and policy importance. Participants (n=198 Part 1, n=206 Part 2) from Amazon Mechanical Turk and Lucid rated whether articles focused on character, whether they mattered for policy, and partisan lean. Selected 20 articles (10 pro-Democrat, 10 pro-Republican; half high character-focus, half policy-important) matched on content and partisanship strength. Main study participants (n=206) rated sharing likelihood on social media.

Mixed-effects model predicted sharing by character-focus vs. policy importance, political orientation, and their interaction, controlling for age and gender.

Study 2 (Experimental). Created 10 novel fake news headlines with 4 versions each: pro-Democrat/pro-Republican and either generic target ("a Democrat") or specific person ("Nancy Pelosi"). All headlines focused on character with no policy implications. Balanced headline sets such that each participant saw one version of each political article plus neutral headlines. Participants (n=404 Part 1 rating, n=306 Part 2 sharing; from Lucid) rated character focus, policy importance, partisan lean, then indicated sharing likelihood on Facebook. Included CRT (6 items) as a measure of cognitive reflection.

Mixed-effects model with article characteristics and political orientation as predictors.

Results

Study 1: Overall, character-focused headlines (compared to policy-important ones) were more likely to be shared (B = 1.409, p < .001). Critically, there was an interaction: Republicans were significantly more likely to share character-focused (vs. policy-important) fake news (B = .38, p = .01), whereas Democrats were nominally more likely to share policy-important headlines (B = −.20, p = .07). Political orientation × character focus interaction was significant (B = −3.173, p < .001).

Study 2: Strong character-focus score (vs. policy importance) was associated with lower sharing likelihood for both Democrats (B = −.33, p < .001) and Republicans (B = −.63, p < .001). Critically, this reversal contradicted Study 1: when headlines were experimentally controlled and matched exactly across partisan divides, both groups were less likely to share headlines focusing on character and more likely to share those important for policy. Political orientation × character focus interaction was significant (B = −0.087, p = .006), with stronger effect for Republicans. CRT was unrelated to character-focused sharing (B = −.32, p = .008 for overall lower sharing, but no interaction with character focus).

Notes

The paper reveals an important supply-demand mismatch: while character-deprecation headlines may be created disproportionately for pro-Republican audiences (possibly because they are cheaper to produce or more inflammatory to generate), people across the political spectrum do not actually prefer them when given controlled choices. This suggests that mitigating character-deprecation fake news may be less about audience preference and more about changing what fake-news creators choose to publish. The controlled Study 2 is more reliable than the observational Study 1 because real-world fake news headlines differ in content across partisan divides, making it impossible to isolate the effect of character focus alone from confounding variation in topic, quality, or other factors.