PolitiFact COVID-19 Fact-Checked Statements with Ratings

Aggregating statements fact-checked by PolitiFact regarding COVID-19, with original and adjusted rating labels. It was gathered using an automated web scraper targeting the PolitiFact COVID fact checker. The author is justinqbui, and it was last updated in December 2021.

Use Cases

• Train a text classifier to predict the adjusted rating (true, misleading, false) from the statement text.
• Analyze the distribution of original PolitiFact ratings (e.g., half-true, pants-fire, mostly-true) across COVID-19 claims.
• Build a model to map the original rating to the adjusted rating based on the defined transformation rules.

Strengths

• Data is sourced from a well-known fact-checking organization, PolitiFact.
• Includes an adjusted rating column that simplifies the original multi-class labels into three categories.
• Dataset was last updated in December 2021, providing a snapshot of COVID-19 misinformation from that period.

Limitations

• The dataset size and number of rows are unknown, limiting assessment of statistical power.
• Data is temporally stale, covering information only up to late 2021, missing recent COVID-19 developments.
• Potential bias exists as it only includes statements reviewed by PolitiFact, not a comprehensive set of all claims.

Provenance

Source

PolitiFact COVID fact checker.

Collection Method

Gathered using an automated web scraper.

Freshness

Last updated 2021-12-13.

Geography

Regionus tag suggests a US focus, but not explicitly confirmed.