Prior research has documented exposure to fake news and online misinformation using large- scale data on individuals’ media use, which has provided important information about the scope and nature of people’s exposure to misinformation online. However, most of this work has made use of data collected during the 2016 US election, and far fewer studies have examined how exposure to misinformation online has changed since 2016. In this paper, we examine exposure to untrustworthy websites in the lead up to the 2020 US election using a dataset of over 7.5 million passively tracked website visits from a nationally representative sample of American adults (N = 1,151). We find that a significantly smaller percentage of Americans were exposed to untrustworthy websites in 2020 compared to in 2016 (as calculated by Guess et al. [2020]). While exposure was concentrated among similar groups of people as it was in 2016, levels of exposure appear to be lower across the board. There were also differences in the role online platforms played in directing people to untrustworthy websites in 2020 compared to 2016. Our findings have implications for future research and practice around online misinformation.
Research using large-scale data on individuals' internet use has provided vital information about the scope and nature of exposure to misinformation online. However, most prior work relies on data collected during the 2016 US election. Here we examine exposure to untrustworthy websites during the 2020 US election, using over 7.5 million website visits from 1,151 American adults. We find that 26.2% (95% confidence interval 22.5% to 29.8%) of Americans were exposed to untrustworthy websites in 2020, down from 44.3% (95% confidence interval 40.8% to 47.7%) in 2016. Older adults and conservatives continued to be the most exposed in 2020 as in 2016, albeit at lower rates. The role of online platforms in exposing people to untrustworthy websites changed, with Facebook playing a smaller role in 2020 than in 2016. Our findings do not minimize misinformation as a key social problem, but instead highlight important changes in its consumption, suggesting directions for future research and practice.
Considering the threat misinformation poses to democratic functioning, little research has studied how misinformation can influence beliefs about elections. We examine the effect of online misinformation exposure on a core democratic belief: the validity of elections. We study this relationship through a two-wave survey of 1,194 American adults and over 21M observations of these individuals’ web-browsing activities before and after the 2020 US Presidential Election. After flexibly adjusting for observed differences and propensity for selective exposure using double machine learning, we find that people exposed to misinformation websites were 4.2% more likely to falsely believe that the election results were fraudulent. We find strong evidence, however, of motivated reasoning: the effect rises to 12.6% among conservatives, but for liberals we observe no effect. There was also a dosage effect. For each exposure to a misinformation website, there was a .03% increase in the likelihood of holding this false belief. We discuss theimplications of these results in relation to broader concerns regarding the effect of misinformation on false beliefs related to democratic backsliding.
In an increasingly polarized political environment, heightened focus has been placed on the potentially devastating effects of misinformation on democracy. Emblematic of this concern is the viral conspiracy theory QAnon, which began during the Trump presidency and evolved to become a frequent distributor of misinformation. In this paper, we investigate the extent to which Americans are exposed to QAnon, in what contexts, and to what effect. To do so, we employ a mixed methods review of 21 million website visits from a nationally representative sample of 1,238 U.S. adults collected during the 2020 presidential election. We find that: (1) exposure to QAnon is limited and stratified by political affiliation and news consumption; (2) exposure tends to occur within right-wing media ecosystems that align with QAnon beliefs; and (3) mixed methods approaches to analyzing digital trace data can provide rich insights that contextualize quantitative techniques.
Social scientists and computer scientists are increasingly using observational digital trace data and analyzing these data post hoc to understand the content people are exposed to online. However, these content collection efforts may be systematically biased when the entirety of the data cannot be captured retroactively. We call this unstated assumption the problematic assumption of persistence. To examine the extent to which this assumption may exist, we examine over 21 million URL-logged web browser visits from 1,515 participants over four months and record the degree to which hard news and misinformation URLs individuals visited were persistent, inaccessible, or ephemeral. While we find that the URLs collected are largely persistent, we find there are systematic biases in which URLs are ephemeral and inaccessible. For example, conservative misinformation URLs are more likely to be ephemeral than other types of misinformation. To standardize the reporting and understanding of the problematic assumption of persistence, we offer a set of metrics, PersistenceRate, InaccessibilityRate, EphemeralityRate (PIE metrics), that future research should report when using digital trace and web scraping data.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.