Pro- and Anti-Vaccine Twitter Discourse on COVID Data

Dehydrated tweet IDs for pro-vaccine and anti-vaccine online discourses about COVID data, collected for a published academic study. The processed analysis files include variables for certainty, causal claims, and authority figures. The data requires rehydration via the Twitter API for full use.

Use Cases

• Analyze linguistic certainty in pro- and anti-vaccine tweets using constructed certainty variables from the processed CSV files.
• Compare the prevalence of causal claims between discourse groups using the causal_claims_analysis.csv dataset.
• Examine references to authority figures in vaccine-related Twitter data using the authority_analysis.csv file.
• Rehydrate tweet IDs from anti_vax_ids.json and pro_vax_ids.json to study the original text of vaccine discourse.

Strengths

• Data supports replication of a peer-reviewed academic paper's findings.
• Provides three distinct processed analysis datasets (certainty, causal claims, authority figures) ready for statistical modeling.
• Includes the original collection code and a full analysis pipeline across five Python and one R notebook.

Limitations

• Core tweet data is provided only as dehydrated IDs, requiring API access and rehydration effort.
• The sample size and demographic/geographic coverage of the original tweets are unknown.
• Use of the LIWC2015 dictionary is required for variable construction, which is a proprietary tool.

Provenance

Source

Harvard Dataverse repository for the published paper.

Collection Method

Collected from Twitter/X using code provided in the repository, resulting in separate pro- and anti-vaccine discourse datasets.

Freshness

Last updated in February 2026.