Research impact evaluation based on effective authorship contribution sensitivity: h-leadership index

Traditional bibliometric measures like the h-index have long been used to evaluate research performance. However, they often overlook the nuances of collaborative authorship, particularly the varying contributions of co-authors based on their position in the author list. This has led to inflated metrics and academic integrity concerns. This project introduces the h-leadership index, a novel metric that assigns weighted citations based on authorship position using a modified complementary unit Gaussian curve. It offers a fairer and more nuanced evaluation of academic contributions, especially in multi-authored publications. We apply this metric to analyze the top 50 researchers across Australia’s Group of Eight (Go8) universities and provide open-source tools for further exploration.

Table of Contents

• Overview
• Methodology
• Results
• Usage
• Citation

Overview

The h-leadership index addresses key limitations of traditional metrics:

• accounts for authorship position beyond just first and last authors.
• uses a modified complementary unit Gaussian distribution to assign weights.
• ensures middle authors receive appropriate recognition.
• avoids penalising large collaborative studies.

This metric is especially relevant in the context of Stanford’s Top 2% Scientists list, which emphasises first and last authorship but overlooks middle contributors. Our approach provides a more balanced and equitable assessment of research impact.

Methodology

The project follows a four-step pipeline:

1. Review of Existing Metrics

   • We analyze h-index, hm-index, g-index, AR-index, and composite indicators like the c-score.

2. Design of the h-Leadership Index

   • Weights are assigned using a modified complementary unit Gaussian curve.
   • First and last authors receive the highest weights (up to 1.0).
   • Middle authors receive gradually decreasing weights, with a minimum of 0.3 beyond the 50th position.

3. Data Retrieval from Scopus

   • Top 50 researchers from each Go8 university were identified via Google Scholar.
   • Publication metadata was retrieved using Scopus APIs.
   • Data includes titles, citations, authorship, venues, and affiliations.

4. Metric Computation

   • Weighted citations are calculated per author.
   • The h-leadership index is computed as the maximum number of papers with weighted citations ≥ h.

For reproducibility, the pipeline is modular and configurable via affiliations.json.

Results

We analyzed 400 researchers across Go8 universities, covering 168,563 publications:

• Journal Articles: 91.56%
• Conference Proceedings: 5.05%
• Books/Book Series: 3.25%

Key files

1. results/{inst}/metrics.csv

   • Contains author-wise data on number of publications, citations, h-index, h-frac-index, hm-index, h-leadership-index, author position and coauthor count.

2. results/{inst}/h-leadership-index.csv

   • Aggregate stats on institutional h-leadership.

3. results/{inst}/correlation_matrix.csv

   • Pairwise Pearson correlation between bibliometrics under evaluation.

4. results/{inst}/correlation_authorship_position.csv

   • Pearson correlation between bibliometrics and authorship position.

Usage

TO use our code for Scopus dataset preparation or analysis, follow below steps:

1. Install Python 3.9 or higher on your system
2. Clone this project

git clone https://github.com/nepython/metrics.git
cd metrics

3. Install dependencies

pip install -r requirements.txt

4. Preparing bibliometric dataset

• scopus.py
  • Contains the code for fetching data from Scopus.
• metrics.py
  • Contains the various bibliometrics evaluated in this study.
• calculate.py
  • Contains the various utility methods for calculating the metrics for an author.

5. Performing bibliometric analysis

• data_analysis.ipynb
  • Contains aggregate plots for bibliometrics vs affiliation.
• h_leadership design iteration.ipynb
  • Contains a comparison of the different ways h-leadership index could have been implemented

Citation

If you use this dataset or code in your research, please cite our paper:

@article{jain2025research,
  title={Research impact evaluation based on effective authorship contribution sensitivity: h-leadership index},
  author={Jain, Hardik A and Chandra, Rohitash},
  journal={arXiv preprint arXiv:2503.18236},
  year={2025}
}