Replication data and code for: Auditing the Impact of Social Media's Policy Shift on Anti-Vaccine Discourse

Replication data and code for the PLOS ONE manuscript "Auditing the Impact of Social Media's Policy Shift on Anti-Vaccine Discourse: A Large Language Model-Driven Empirical Study." This package contains: (1) tweet IDs and derived annotation labels (vaccine stance, anti-vaccine theme, and stance consistency) for 15,788 tweets from November 16–30, 2022; (2) the three GPT-4o classification prompts used for annotation; and (3) a Python script reproducing the logistic regression analyses. No tweet text or user information is included, in compliance with Twitter/X's Terms of Service. Researchers may rehydrate tweet text using the Twitter/X API. The original tweet IDs were drawn from the COVID-19 Twitter dataset by Banda et al. (2021).