Performance evaluation of Natural Science Foundation projects based on hierarchical causal relationships
Document Type
Research-Article
Journal Name
Socio-Economic Planning Sciences
Keywords
AHDT, DEMATEL, Hierarchical causal relationship, Performance evaluation
Abstract
To evaluate the performance of Natural Science Foundation projects more rigorously and transparently and to address the limitations of traditional evaluation methods, which often struggle to capture complex inter-indicator relationships and tend to produce “black box” scores, this study develops a comprehensive evaluation framework that makes explicit the internal logic and value transmission pathways of performance outputs. This study integrates the Decision-Making Trial and Evaluation Laboratory (DEMATEL) approach with the Adversarial Hasse Diagram Technique (AHDT). First, DEMATEL is employed to quantify the strength and direction of causal influence among performance output indicators, thereby identifying core driving indicators and final outcome indicators. Second, based on the DEMATEL results, AHDT is introduced to construct an adversarial hierarchical topological graph of the indicators, thereby transforming implicit network relationships into a clear four-level hierarchical structure. Finally, we combine DEMATEL influence measures with the AHDT-derived hierarchy to derive a weighting scheme that reflects both quantitative influence and structural importance. We further conduct an empirical analysis using a sample of projects from the W Region's Natural Science Foundation. The findings reveal a four-level value transmission chain in the performance outputs of scientific research projects, moving from basic academic output to the application and commercialization of research outcomes and, ultimately, to comprehensive recognition. The analysis indicates significant performance disparities across disciplines and project types. Application-oriented disciplines tend to achieve higher and statistically more robust comprehensive performance scores, whereas some basic sciences may face evaluation challenges due to long research cycles and slower commercialization. The evaluation model constructed in this study helps uncover the internal mechanisms and transmission pathways of performance generation in science foundation funded projects, providing a quantitative basis and decision support for funding agencies to optimize funding allocation, design differentiated support policies, and promote the commercialization of research outcomes. © 2026 Elsevier Ltd.