Exposure to one-sided messaging and belief in voter fraud: the curious case of the US 2020 presidential election

Abstract

In lead-up to the 2020 U.S. presidential election, conservative media amplified Donald Trump’s concerns about the security of US elections by publishing uncritical one-sided coverage that supported the idea that US elections were particularly vulnerable to fraud. The vast majority of existing scholarship would lead us to expect that exposure to such one-sided news stories should increase belief in the vulnerability of US elections. Drawing on a survey experiment conducted shortly before the 2020 election, we show that exposure to one-sided news coverage did not have such an effect. We explain this as a result of treatment heterogeneity among partisan groups, and we use regression and random forest estimation techniques to identify major sources of treatment effect heterogeneity. We conclude with a discussion of the implications for the belief in unsubstantiated falsehoods.