Work

Conspiracist attributes differentiate pro- and anti-vaccine online discourses about data

Year: 2026
Type: article
Abstract: The public frequently and increasingly participates in discussions about vaccines and data on social media. However, such public engagement often includes claims and rumors that misuse data to support anti-vaccine perspectives and lead to negative public health impacts. Often, people expressing anti-vaccine views use the same rhetorical strategies as vaccine advocates to engage with data and evidence. Prior work finds that machine learning systems can distinguish pro- and anti-vaccine content as well as other types of misinformation and conspiracy theories. However, prior classification approaches prioritize predictive performance over human-interpretable features that explain why content differs. We propose a theory-driven approach to distinguish anti-vaccine social media discussions of data by identifying textual characteristics of conspiracy discourses, including mentions of authority figures, expressions of certainty, and causal claims. We conduct statistical analyses to evaluate this approach using a dataset of public Twitter posts containing pro- and anti-vaccine discourse on data about COVID-19 vaccines during the first six months after the beginning of Phase 3 clinical trials. We identify pro- or anti-vaccine Twitter posts using hashtags and keywords. Compared to pro-vaccine data discourses on Twitter, we find that anti-vaccine data discourses are less likely to mention authority figures but more likely to express certainty and make causal claims. The findings can inform ongoing vaccination and public health interventions.
Language: English
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7160932004
FWCI:
Cites: 49
Cited by:
Related to: 10