Understanding narratives of uncertainty in fertility intention: a neural topic modeling approach

Xiao Xu , Netherlands Interdisciplinary Demographic Institute (NIDI)
Anne H. Gauthier , Netherlands Interdisciplinary Demographic Institute (NIDI)
Gert Stulp , University of Groningen
Antal van den Bosch, Meertens Institute

Uncertain responses in fertility intentions are common in many demographic surveys. These responses are often considered noise to be filtered out, rather than as a topic of investigation. Here, we delve deeper into underlying reasons for uncertainty, by examining responses to open-ended questions (OEQs) in a large population-based survey. Open-ended questions (OEQs) provide opportunities for respondents to "expand on" their ideas about uncertainty without pre-defined categories. While such answers can provide a wealth of information, the subsequent quantitative analysis of text data is often challenging, especially in large-scale surveys, and is not yet widely applied in demographic surveys. We aim to integrate unsupervised machine learning techniques into the interpretation of open-ended responses in demographic surveys. In this paper, using a topic modeling approach, we identify topics and logic behind the uncertainty of responses in OEQs in an online survey in the Netherlands. Thus, we explore how the sets of topics are discussed in groups with different uncertainty levels regarding their self-reported fertility intention. We will conduct our analysis based on SCHOLAR, a neural topic models for documents with metadata, and assess the simultaneously identified patterns with further qualitative interpretations.

See extended abstract

 Presented in Session 48. Data and Methods