Searching for misinformation

In the days leading up to the 2020 U.S. Presidential Election, many voiced concerns that misinformation and conspiracy theories would circulate online, spread to the general public, and undermine trust in the voting process and results. There was a substantial fear that this misinformation, combined with a contested outcome and prolonged uncertainty, could spark protests and political violence. To determine the public’s engagement with pieces of misinformation and their information seeking for inflammatory claims and conspiratorial content, I collected and analyzed Google Search data from the U.S. during the week of the election.

It is well known that social media content is not representative of the public’s attitudes. A proportionally small amount of content from Twitter, Facebook, and online discussion boards spreads beyond these platforms to mainstream news. Although misinformation certainly spreads online and attracts engagement, it is historically contained within small networks of individuals. To prompt a public response, misinformation typically must cross an extremely high threshold of online activity or be promoted by elites and eventually spread through mainstream channels. However, 2020 has seen widespread misinformation regarding COVID-19 and mainstream discussion of conspiracies like QAnon. These have resulted in unrest and threats of violence, and the tense and polarized political climate may be fertile grounds for widespread misinformation about the election.

Some questions about the integrity of U.S. elections have already moved from the realm of online conspiracy into the public sphere, in no small part because of persistent elite messaging. President Trump has been a key promoter of misinformation about electoral fraud and political opponents “cheating” the election. In January 2017 he claimed that millions had voted illegally and appointed the Presidential Advisory Commission on Election Integrity. After pushback from states and a court order to share working documents with Democratic members, the commission eventually dissolved. Even though no evidence of wide-spread fraud was ever released, Trump continued to revisit the idea that substantial voter fraud had occurred in the 2016 election. As the spread of COVID-19 raised health concerns about in-person voting on Election Day, President Trump, again without evidence, regularly promoted the idea that early and mail-in voting would be fraught with fraud.

Although President Trump’s repeated claims about election fraud may have damaged trust in institutional processes for some, it is unknown if the average citizen is now more susceptible to explicit and inflammatory conspiratorial claims.

One way to determine if a broad audience is interested in learning more about a topic is to track the Google Search frequency of keywords using Google Trends. By plotting the search frequency of different keywords, the prevalence of searches for topics can be assessed and compared.

Table 1. Google Trends keywords and maximum normalized hourly search volume

Using the gtrendsR package and an R script written to scrape and combine Google Trends data across the search terms presented in Table 1, I collected hourly observations of 41 keywords from the morning of Nov. 1 to the morning of Nov. 7 (all times are presented in Eastern Standard Time). I then plotted the search frequency of all the terms (Figure 1A) and then the frequency of only conspiratorial and misinformation terms (Figure 1B). This descriptive data is a first look at salient topics and is not intended to be exhaustive or representative of all the potential topics around the election. That work is ongoing.

Figure 1A. Google Search hits
across time for all keywords
Figure 1B. Google Search hits across time for conspiratorial and misinformation terms

The data show an increase in searches for specific pieces of misinformation and disinformation over the days after the election (Figure 1B). However, the overall frequency, even for topics that trended on social media, is very small compared to the rest of the terms. People do search for general ideas of voter, electoral fraud, and official activities like lawsuits, but even these topics attract a paltry amount of attention compared to information seeking for results and candidates (Figure 1A). These findings suggest most conspiracies are not widespread and the most incendiary claims go ignored. When a President and campaign message misinformation over the course of four years, the public wants to find out more, but the conspiracy of the minute is most likely to quickly vanish. So far, the American public seems accepting of the results and political violence has been avoided. Whether or not a former president and his allies can still drive attention to misinformation remains to be seen.