How knowledge diversity in research proposals affects the academic performance of funded social science scholars

This study examines the impact of external and internal knowledge diversity in research grant proposals on academic performance in social sciences. External knowledge diversity is measured by calculating the semantic similarity between 10,482 social science research grant proposals from the National Science Foundation (NSF) between 1970 and 2019, and 74,056 proposals from 22 other disciplines, using SciBERT and cosine similarity. Internal knowledge diversity is determined by the variety of research categories derived from clustering within social science proposals. Additionally, we analyze 635,910 scientific publications of social science principal investigators (PIs). We investigate how the intellectual proximity between PIs' doctoral and postdoctoral publication abstracts and their proposals affects academic performance. Our findings reveal that increasing external or internal knowledge diversity in grant proposals is not significantly related to social science scholars' productivity or research impact. This lack of a statistically significant relationship suggests that while social science PIs may possess sufficient adaptive capacity to manage diverse knowledge portfolios without performance degradation, the anticipated ‘diversity premium’ did not materialize. In contrast, greater intellectual proximity between research proposals and PIs' early-career research during doctoral and postdoctoral stages significantly improves research impact. The impact of postdoctoral intellectual proximity scales with knowledge diversity, providing the essential expertise required for successful interdisciplinary integration and high-impact research.