Evaluating and Predicting Journal Performance for Serials Assessment: A Multidimensional Bibliometric and Machine Learning Analysis of First Quartile Strategy and Management Journals

This is the authors' abstract. We don't add key points for this paper.

Journal performance evaluation is central to serials management, collection development, and research assessment, yet reliance on single bibliometric indicators captures only partial dimensions of journal quality. This study develops a multidimensional and reproducible framework for evaluating and anticipating journal performance using longitudinal Scopus data for Q1 Strategy and Management journals from 2020 to 2024, comprising 915 journal–year observations. Descriptive trend analysis, regression modeling, artificial neural networks (ANN), ARIMA forecasting, and publisher–journal network analysis were integrated. A Composite Impact Score (CIS), based on the geometric mean of CiteScore, SNIP, and SJR, was used as an analytical heuristic for multidimensional assessment. Average CiteScore increased by 66.4%, while citation efficiency, percentage of cited documents, and field-normalized impact positively predicted performance; excessive publication volume showed a significant negative association. ANN models outperformed regression models (R= 0.91 vs. 0.66). The framework supports responsible, evidence-based, and forward-looking serials evaluation and collection management.