Mapping Fertility Narratives at Scale Using Large Language Models

• In Event: Advances in Computational Methods (Co-Sponsored by Mathematical Sociology Section)

Abstract

Fertility planning is increasingly discussed online, yet we know comparatively little about how fertility intentions, emotional tone, and perceived constraints are articulated at scale in these settings. Using 32,985 fertility-related posts from four Reddit communities between November 2020 and July 2023, this study provides a descriptive map of fertility narratives using a large language model (LLM)-based workflow. We infer three post-level indicators: expressed fertility intention (0-10), sentiment (0-10), and a dominant constraint frame (medical, financial, legal, partner-related, age-related, or unknown). We examine how these indicators vary over time, across communities, and by inferred demographic attributes, and we complement structured inference with abstract topic labels that summarize underlying concerns.

Aggregate fertility intention remains comparatively stable over the study window, with no sustained discontinuities around major policy moments. Fertility constraints are substantially reweighted across social contexts: medical constraints consistently anchor fertility discourse, while financial, relational, age-related, legal, and indeterminate (“unknown”) framing varies systematically by subreddit, age, and income. Topic analyses further show that indeterminate constraints are not residual: they are dominated by themes of uncertainty, emotional distress, mental health, and support-seeking, indicating difficulty anchoring concerns to a single explanatory frame rather than absence of engagement.

These findings describe patterns in online narratives rather than population-level estimates. They characterize fertility planning as sensemaking under uncertainty. Methodologically, we demonstrate how LLM-assisted, schema-constrained anno- tation can be used to construct high-dimensional, interpretable labels for online social discourse at scale.

Authors

• Yan Zhang, Texas State University

• Mingfei Zhang