American User Party Affiliation Classification from 2020 Twitter Data

Zachary Yang's study analyzes over 50 million tweets from late 2020 to measure partisan polarization in COVID-19 health discussions across the United States and Canada. The dataset, last updated in April 2026, likely contains a polarization index derived from tweets about lockdowns, masks, and vaccines. It explores correlations between online discourse and regional vaccination rates.

Use Cases

• Analyze regional variation in partisan polarization based on the geographic heterogeneity described.
• Model the correlation between online discourse and public health outcomes based on the described relationship with vaccination rates.
• Track temporal spikes in polarization linked to specific political events and policy announcements as mentioned in the description.
• Compare cross-national differences in polarization dynamics between the U.S. and Canada as highlighted in the study.

Strengths

• Analysis is based on over 50 million tweets, providing a large-scale social media sample.
• Explicitly covers a critical period of polarizing discourse in late 2020 during the COVID-19 pandemic.
• Examines three key health interventions: lockdowns, masks, and vaccines.
• Released under a permissive CC-BY-4.0 license for broad reuse.

Limitations

• Row count and column-level documentation are unknown, which limits suitability assessment.
• The 5.5 KB file size suggests the dataset is a summary or index, not the raw tweet data.
• Data may reflect bias inherent to the Twitter (now X) platform and its politically engaged user base.

Provenance

Source

figshare

Collection Method

Computational analysis of Twitter data, as described in the study.

Time Range

Late 2020

Freshness

Last updated 2026-04-20 17:58:10; freshness should be verified.

Geography

United States and Canada, with subnational regional analysis.