The influence of higher education based on machine learning on subjective well-being

Abstract

As higher education becomes increasingly prevalent and accessible in China, a growing number of residents are afforded the option to pursue advanced studies. Can higher education genuinely enhance residents’ subjective well-being? The response to this enquiry necessitates additional investigation. This study selected 5 wave data of Chinese General Social Survey (CGSS), a total of 53,874 samples. Machine learning methodologies, including XGBoost and GBDT, were utilised for the inaugural correlation investigation between higher education and subjective well-being in China. Feature importance sorting elucidated the nonlinear correlations and interaction effects, such as the threshold effect of social fairness cognition on happiness, that typical regression models struggle to capture. (1) The average subjective well-being of the higher education group (4.005309) was significantly higher than that of the non-higher education group (3.835478), and the education level had a significant positive predictive role on subjective well-being (p = 0.000 < 0.05); (2) Machine learning uncovers substantial correlations between higher education and subjective well-being (=0.008, p < 0.01), with social justice cognition (feature weight=0.32) and self-rated health (0.28) identified as primary mediators. (3) The proportion of women in the highest level of well-being ([4.0,5.0)] was slightly higher than that of men. Higher education can markedly enhance the subjective well-being of individuals. Furthermore, it improves residents’ subjective well-being via social justice perception, self-assessed health, social class identity, job satisfaction, and socio-economic status. This establishes a scientific foundation for the government and all societal sectors to augment investment in education, enhance the distribution of educational resources, and foster the comprehensive development of individuals.