This study investigates the driving role of full-time equivalent (FTE) R&D personnel in enhancing the regional influence of social science research in China. The study measures this influence by the number of National Social Science Fund (NSSF) projects across 31 provinces from 2003 to 2022. The study draws from 620 province-year observations and multiple national statistical yearbooks, employing a combination of traditional panel regression and four machine learning models—Random Forest, Gradient Boosting, LASSO, and Neural Networks—to assess both linear and nonlinear relationships. The findings of the study demonstrate that research and development (R&D) personnel have a substantial impact on the output of the National Science Foundation (NSSF), particularly when they are supported by internal R&D expenditures and financial contributions. Among the machine learning models, Random Forest and Gradient Boosting demonstrate strong predictive performance, while Neural Networks exhibit instability. Subsequent subgroup analysis reveals pronounced regional heterogeneity: Research and development (R&D) investment has been demonstrated to generate optimal returns in the eastern provinces, while exhibiting moderate and nonlinear effects in the central regions. Conversely, R&D investment in western areas has been observed to yield diminishing returns, and in some cases, negative returns. These findings underscore the necessity of differentiated policy strategies that align R&D investments with local research capacity and structural conditions. The present study makes a methodological contribution through its integration of machine learning into empirical policy analysis, thus offering actionable insights for improving the allocation efficiency of social science funding in China.
Yu et al. (Fri,) studied this question.