Scholarlink

<p>The evaluation of scientific research plays a critical role in academic decision-making, influencing giving tenure, funding allocation, and award recognition. Although traditional bibliometric indicators such as the h-index and its variants are widely used, their effectiveness in accurately identifying high-impact researchers remains debated. Moreover, many existing studies rely on hypothetical datasets, limiting empirical validation across real-world award-based benchmarks. In this study, we propose a data-driven composite bibliometric index specifically designed for researcher evaluation in the neuroscience domain. The analysis is conducted on a balanced dataset of 1,060 researchers, comprising 530 awardees and 530 non-awardees collected from major international neuroscience societies. We systematically evaluate 64 bibliometric parameters and rank them according to their ability to identify award-winning researchers within the top 100 positions. The top-performing indices are further analyzed through 45 pairwise combinations using seven statistical aggregation models, including arithmetic mean, harmonic mean, geometric mean, quadratic mean, cubic mean, logarithmic mean, and contra-harmonic mean. Empirical results demonstrate that the h2-upper index and A-index exhibit the strongest individual performance. Among the aggregation techniques, the contra-harmonic mean achieves the highest effectiveness in distinguishing awardees. Based on these findings, we introduce a weighted composite metric that integrates the h2-upper and A-index using a contra-harmonic formulation. The proposed index improves awardee identification compared to the best-performing individual metric, providing a more balanced and discriminative framework for researcher evaluation in neuroscience. A supplementary bootstrap resampling analysis further confirms that this discriminative advantage is preserved under realistic class-imbalance ratios of up to approximately 1:22, beyond the balanced 1:1 design used in the primary evaluation. This study offers a systematic, empirically validated approach for bibliometric index integration and contributes toward more accurate and domain-sensitive research performance assessment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A data-driven composite index for evaluating research impact: evidence from neuroscience researchers

  • 期刊论文

  • 发表日期：

    2026年8月1日

查看全文
  • Hafiza Zarafshan,
  • Ghulam Mustafa,
  • Muhammad Tanvir Afzal

摘要

The evaluation of scientific research plays a critical role in academic decision-making, influencing giving tenure, funding allocation, and award recognition. Although traditional bibliometric indicators such as the h-index and its variants are widely used, their effectiveness in accurately identifying high-impact researchers remains debated. Moreover, many existing studies rely on hypothetical datasets, limiting empirical validation across real-world award-based benchmarks. In this study, we propose a data-driven composite bibliometric index specifically designed for researcher evaluation in the neuroscience domain. The analysis is conducted on a balanced dataset of 1,060 researchers, comprising 530 awardees and 530 non-awardees collected from major international neuroscience societies. We systematically evaluate 64 bibliometric parameters and rank them according to their ability to identify award-winning researchers within the top 100 positions. The top-performing indices are further analyzed through 45 pairwise combinations using seven statistical aggregation models, including arithmetic mean, harmonic mean, geometric mean, quadratic mean, cubic mean, logarithmic mean, and contra-harmonic mean. Empirical results demonstrate that the h2-upper index and A-index exhibit the strongest individual performance. Among the aggregation techniques, the contra-harmonic mean achieves the highest effectiveness in distinguishing awardees. Based on these findings, we introduce a weighted composite metric that integrates the h2-upper and A-index using a contra-harmonic formulation. The proposed index improves awardee identification compared to the best-performing individual metric, providing a more balanced and discriminative framework for researcher evaluation in neuroscience. A supplementary bootstrap resampling analysis further confirms that this discriminative advantage is preserved under realistic class-imbalance ratios of up to approximately 1:22, beyond the balanced 1:1 design used in the primary evaluation. This study offers a systematic, empirically validated approach for bibliometric index integration and contributes toward more accurate and domain-sensitive research performance assessment.