Toward Responsible Scientometrics: Normative Data Practices for Research Evaluation

Conducting scientometric research in a rigorously normative manner requires careful attention to responsible metrics and data practices throughout the research life cycle. This study offers practical suggestions from a data perspective, addressing the key stages from source selection to data sharing while emphasizing transparency and policy implications for evaluation systems. First, the selection of data sources necessitates the alignment of data quality, inclusion criteria, and research objectives. Researchers must ensure that dataset characteristics align with their analytical goals and accurately document sources, including aspects such as coverage and granularity. Data retrieval and acquisition require a clear understanding of the database's scope and update mechanisms. A deliberate approach—utilizing tailored search terms, following platform-specific syntax, and transparently reporting retrieval strategies—ensures reproducibility. Data pre-processing involves a clear comprehension of the field and specific tasks (such as deduplication and normalization). Standardized protocols should connect raw data to analytical requirements while maintaining transparency in methodological choices. For analysis, selecting appropriate tools and differentiating standardization methods (e.g., citation versus collaboration network metrics) is critical to support responsible metrics in evaluation systems. Visualization should enhance interpretability without distorting evidence. Interpretation must strike a balance between robustness and restraint, avoiding over-claiming patterns or neglecting contextual nuances. Engaging domain experts helps mitigate disciplinary biases. Data storage requires the systematic documentation of fields, derived variables, and backup protocols to ensure auditability—a key aspect of transparent and accountable research practices. Finally, obligations for sharing and citations include disseminating data through repositories and formally crediting open-access datasets to maintain scholarly standards. By integrating these guidelines, researchers can enhance scientometrics' methodological rigor, replicability, and ethical accountability, contributing to more responsible and policy-relevant evaluation systems. This framework highlights how normative advancements depend on meticulous data stewardship at each procedural stage, with broader policy implications for research assessment.