The Multinomial Logistic Regression Model.

ObjectiveTo explore how fertility-related sentiments expressed by urban Chinese women varied over time and were framed in online discourse between 2011 and 2024.MethodsThis study employed natural language processing (NLP), TF-IDF-based semantic analysis, and multinomial logistic regression to analyze user-generated content from Douyin, Xiaohongshu, and Kuaishou. Sentiment categories (positive, neutral, negative) were classified, and predictive variables such as education, marital status, and urban tier were modeled to identify structural correlates of fertility attitudes.ResultsNegative fertility intentions accounted for 71.0% of all discourse, rising sharply from 56.8% in 2021 to 79.8% in 2024. Neutral and positive expressions comprised 28.5% and 0.5%, respectively. Economic constraints—including housing costs (TF-IDF = 2.34)—and environmental concerns (TF-IDF = 2.00) predominated in negative sentiment, while child-rearing costs (TF-IDF = 4.50) were central to neutral positions. Regression analysis revealed marriage was associated with lower odds of negative intention (OR=0.229, 95% CI: 0.194–0.271), while postgraduate education (OR=2.819) and residence in first-tier cities (OR=4.05) were linked to higher odds. Socio-cultural pressures were the most influential predictors of negative sentiment (OR=11.11, 95% CI: 9.07–12.12).ConclusionFertility intentions among urban Chinese women increasingly represent conscious adaptations to complex structural realities rather than simple expressions of personal reluctance. This shift reflects a broader societal transition in which autonomy, well-being, and ethical responsibility gradually supersede traditional reproductive imperatives. Rather than indicating demographic crisis, these changing intentions mark a natural phase of social development, underscoring the need for institutional reforms and a cultural ethos that affirms reproductive freedom and supports parenthood as an informed, empowered choice.