Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Persistence of Contrarianism on Twitter: Mapping users' sharing habits for the Ukraine war, COVID-19 vaccination, and the 2022 Midterm Elections

David Axelrod, Sangyeon Kim, John Paolillo

TL;DR

The paper investigates whether users’ stances on three major topics—COVID-19 vaccination, the Ukraine war, and the 2022 US midterm elections—are related on Twitter. It employs a three-stage PCA-based dimensionality reduction workflow to align retweet behaviors across topics into a common space, followed by density-based clustering to identify coherent user groups among highly active cross-topic participants. The results reveal a persistent contrarian stance opposing public health narratives and Biden-era foreign policy, with several cross-topic clusters displaying ideologically coherent but not strictly left–right orientations; no distinct pro-vaccine cluster emerges. The study highlights durable cross-topic ideological structures among motivated Twitter users and offers a framework for examining how misinformation and political content proliferate across multiple salient issues.

Abstract

Empirical studies of online disinformation emphasize matters of public concern such as the COVID-19 pandemic, foreign election interference, and the Russo-Ukraine war, largely in studies that treat the topics separately. Comparatively fewer studies attempt to relate such disparate topics and address the extent to which they share behaviors. In this study, we compare three samples of Twitter data on COVID-19 vaccination, the Ukraine war and the 2022 midterm elections, to ascertain how distinct ideological stances of users across the three samples might be related. Our results indicate the emergence of a broad contrarian stance that is defined by its opposition to public health narratives/policies along with the Biden administration's foreign policy stances. Sharing activity within the contrarian position falls on a spectrum with outright conspiratorial content on one end. We confirm the existence of ideologically coherent cross-subject stances among Twitter users, but in a manner not squarely aligned with right-left political orientations.

The Persistence of Contrarianism on Twitter: Mapping users' sharing habits for the Ukraine war, COVID-19 vaccination, and the 2022 Midterm Elections

TL;DR

The paper investigates whether users’ stances on three major topics—COVID-19 vaccination, the Ukraine war, and the 2022 US midterm elections—are related on Twitter. It employs a three-stage PCA-based dimensionality reduction workflow to align retweet behaviors across topics into a common space, followed by density-based clustering to identify coherent user groups among highly active cross-topic participants. The results reveal a persistent contrarian stance opposing public health narratives and Biden-era foreign policy, with several cross-topic clusters displaying ideologically coherent but not strictly left–right orientations; no distinct pro-vaccine cluster emerges. The study highlights durable cross-topic ideological structures among motivated Twitter users and offers a framework for examining how misinformation and political content proliferate across multiple salient issues.

Abstract

Empirical studies of online disinformation emphasize matters of public concern such as the COVID-19 pandemic, foreign election interference, and the Russo-Ukraine war, largely in studies that treat the topics separately. Comparatively fewer studies attempt to relate such disparate topics and address the extent to which they share behaviors. In this study, we compare three samples of Twitter data on COVID-19 vaccination, the Ukraine war and the 2022 midterm elections, to ascertain how distinct ideological stances of users across the three samples might be related. Our results indicate the emergence of a broad contrarian stance that is defined by its opposition to public health narratives/policies along with the Biden administration's foreign policy stances. Sharing activity within the contrarian position falls on a spectrum with outright conspiratorial content on one end. We confirm the existence of ideologically coherent cross-subject stances among Twitter users, but in a manner not squarely aligned with right-left political orientations.
Paper Structure (8 sections, 1 figure)

This paper contains 8 sections, 1 figure.

Table of Contents

  1. Introduction
  2. Data
  3. Methods
  4. Principal Components Analysis
  5. Clustering
  6. Results
  7. Conclusion
  8. Acknowledgments.

Figures (1)

  • Figure 1: Six largest user groups identified by clustering on principal component scores from the cross-sample common space (see section \ref{['pca']}). Clusters are color encoded equivalently in 1(A) and 1(C). (A) Pair plot of clustered users along the common space principal components. (B) Biplots showing the rotations of the sample-derived principal components into the common space. (C) Co-retweet network for the six main clusters with edge weights proportional to the number of shared retweets and edge colors highlighting the two-chamber structure pulled out by PC1/PC2 and Louvain community detection.