Contraception is frequently discussed on pregnancy and parenting forums. Before birth, users discuss efforts to become pregnant; after birth, users discuss family planning strategies and postpartum contraception. These conversations include users’ contraception questions, experiences, and second-hand information. Despite the frequency of contraception-related conversations in these settings, contraceptive misinformation has not been broadly studied on these platforms. This project will leverage an existing database of posts and comments retrieved from discussion forums on BabyCenter, a popular pregnancy and parenting app and website, to address three contraception misinformation research questions:

(1) What contraception-related content is discussed on BabyCenter, and how do users engage with and respond to these narratives?

(2) Can human-in-the-loop text clustering methods identify types of contraceptive misinformation and user responses at scale? How do themes identified in such an analysis compare to themes captured by qualitative coding?

(3) Can a qualitative framework developed by human coders be used to enable few-shot classification with a large language model (LLM) to identify and classify contraceptive misinformation at scale?

Our findings will provide a foundational understanding of contraceptive misinformation on pregnancy and parenting forums as well as a methodological foundation for future efforts to automate contraception misinformation surveillance.