Abstract:
Recent studies explain the rising trend of cohabitation in China using the Second Demographic Transition theory. A lack of survey data collecting ideas and values on cohabitation, however, created a lacuna in understanding if self-realization is the main driver of the behavioral shift. This study collected text data from two online discussion questions on a popular Q & A website, Zhihu (zhihu.com). 1,083 posts were obtained discussing whether premarital cohabitation is necessary. 8,149 posts were obtained on discussions of whether users are concerned about their partners’ prior cohabitation experiences. A Large Language Model, GPT-4, was used to automatically infer user attitudes from the texts, followed by supervised Topic Modeling to analyze the text data by gender and question groups. Results show that cohabitation is not perceived as a value-neutral choice by the Zhihu users. Further, female users were found to exhibit more liberal attitudes than male users on cohabitation.
June Yang is a research scientist at the Center for Studies in Demography and Ecology and the eScience Institute.
As a Computational Demographer, June focuses on applying Natural Language Processing methods to the study of population family formation processes, gender disparity, and demographic inference. She is expanding her skillset by using Large Language Models in text data annotation and measurement development. A second strain of her current research focuses on complex survey analysis, particularly using network-based samples to study vulnerable, hard-to-reach populations. June also has extensive experience working with administrative data sources.
June received her PhD in Sociology from UW with concentrations in Demographic Methods and Social Statistics. Before starting a PhD at the UW, she worked as a research analyst at the Development Data Group of the World Bank.
