Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Nov 21;13(1):7144.
doi: 10.1038/s41467-022-34769-6.

Measuring exposure to misinformation from political elites on Twitter

Mohsen Mosleh  1   2   3 , David G Rand  4   5   6
Affiliations

Measuring exposure to misinformation from political elites on Twitter

Mohsen Mosleh et al. Nat Commun. .

Erratum in

Abstract

Misinformation can come directly from public figures and organizations (referred to here as "elites"). Here, we develop a tool for measuring Twitter users' exposure to misinformation from elites based on the public figures and organizations they choose to follow. Using a database of professional fact-checks by PolitiFact, we calculate falsity scores for 816 elites based on the veracity of their statements. We then assign users an elite misinformation-exposure score based on the falsity scores of the elites they follow on Twitter. Users' misinformation-exposure scores are negatively correlated with the quality of news they share themselves, and positively correlated with estimated conservative ideology. Additionally, we analyze the co-follower, co-share, and co-retweet networks of 5000 Twitter users and find an ideological asymmetry: estimated ideological extremity is associated with more misinformation exposure for users estimated to be conservative but not for users estimated to be liberal. Finally, we create an open-source R library and an Application Programming Interface (API) making our elite misinformation-exposure estimation tool openly available to the community.

PubMed Disclaimer

Conflict of interest statement

M.M. and D.R. have received research funding from Google, and D.R. has received research funding from Meta.

Figures

Fig. 1
Fig. 1. Descriptives of the fact-checking dataset that forms the basis of our elite misinformation-exposure measure.
a Distribution of number of fact-checks per elite provided by PolitiFact. b Number of fact-checks per each PolitiFact category (T True, MT Mostly True, HT Half True, MF Mostly False, F False, POF Pants on Fire). c Distribution of falsity scores associated with each elite. d Distribution of number of Twitter followers of elites. e Distribution of number of elites followed by each Twitter user in our sample. Source data are provided as a Source Data file.
Fig. 2
Fig. 2. Exposure to elite misinformation is associated with sharing news from lower-quality outlets and with conservative estimated ideology.
Shown is the relationship between users’ misinformation-exposure scores and (a) the quality of the news outlets they shared content from, as rated by professional fact-checkers, (b) the quality of the news outlets they shared content from, as rated by layperson crowds, and (c) estimated political ideology, based on the ideology of the accounts they follow. Small dots in the background show individual observations; large dots show the average value across bins of size 0.1, with size of dots proportional to the number of observations in each bin. Source data are provided as a Source Data file.
Fig. 3
Fig. 3. Exposure to elite misinformation is associated with the use of toxic language and moral outrage.
Shown is the relationship between users’ misinformation-exposure scores and (a) the toxicity of the language used in their tweets, measured using the Google Jigsaw Perspective API, and (b) the extent to which their tweets involved expressions of moral outrage, measured using the algorithm from ref. . Extreme values are winsorized by 95% quantile for visualization purposes. Small dots in the background show individual observations; large dots show the average value across bins of size 0.1, with size of dots proportional to the number of observations in each bin. Source data are provided as a Source Data file.
Fig. 4
Fig. 4. In the co-share network, a cluster of websites shared more by conservatives is also shared more by users with higher misinformation exposure scores.
Nodes represent website domains shared by at least 20 users in our dataset and edges are weighted based on common users who shared them. a Separate colors represent different clusters of websites determined using community-detection algorithms. b The intensity of the color of each node shows the average misinformation-exposure score of users who shared the website domain (darker = higher PolitiFact score). c Nodes’ color represents the average estimated ideology of the users who shared the website domain (red: conservative, blue: liberal). d The intensity of the color of each node shows the average use of language toxicity by users who shared the website domain (darker = higher use of toxic language). e The intensity of the color of each node shows the average expression of moral outrage by users who shared the website domain (darker = higher expression of moral outrage). Nodes are positioned using directed-force layout on the weighted network.
Fig. 5
Fig. 5. Estimated ideological extremity is associated with higher elite misinformation-exposure scores for estimated conservatives more so than estimated liberals.
a Political ideology is estimated using accounts followed. b Political ideology is estimated using domains shared (Red: conservative, blue: liberal). Source data are provided as a Source Data file.
Fig. 6
Fig. 6. Estimated ideological extremity is associated with higher language toxicity and moral outrage scores for estimated conservatives more so than estimated liberals.
The relationship between estimated political ideology and (a) language toxicity and (b) expressions of moral outrage. Extreme values are winsorized by 95% quantile for visualization purposes. Source data are provided as a Source Data file.
See this image and copyright information in PMC

References

    1. Lazer DM, et al. The science of fake news. Science. 2018;359:1094–1096. doi: 10.1126/science.aao2998. - DOI - PubMed
    1. Pennycook G, Rand DG. The psychology of fake news. Trends Cogn. Sci. 2021;25:388–402. doi: 10.1016/j.tics.2021.02.007. - DOI - PubMed
    1. Pennycook, G. & Rand, D. G. Examining false beliefs about voter fraud in the wake of the 2020 Presidential Election. The Harvard Kennedy School Misinformation Review (2021).
    1. Slothuus, R. & Bisgaard, M. How political parties shape public opinion in the real world. Am. J. Polit. Sci.65.4, 896–911 (2021).
    1. Slothuus R, Levendusky MS. Party over Pocketbook? How Party Cues Influence Opinion When Citizens Have a Stake in Policy. Am. Polit. Sci. Rev. 2010;115:1090–1096. doi: 10.1017/S0003055421000332. - DOI

Publication types

LinkOut - more resources