Modelling the academic publishing system: a data-driven agent-based approach

Academic publication is used to disseminate and advance knowledge. It also provides a measure of individual research performance for career advancement and personal prestige. Furthermore, it is one measure of university performance and productivity for global rankings and government policy purposes. This research is motivated by the importance of academic publication to these various stakeholders, and aims to address the following research question: How can the academic publishing system be explained by modelling the strategic behaviours of scholars and academic journals? This research uses existing literature to define the strategic publishing behaviours of scholars and the strategic approaches of academic journals associated with publication. These investigations are then used to develop a conceptual model of the academic publishing system. The conceptual model is then operationalised using an agent-based modelling and simulation approach. Finally, real-world data is used to parameterise and validate this model. This research question is addressed through the empirical analysis of bibliometric data, and the implementation of the agent-based model. The analysis aims to fully understand the quantitative status of the academic publishing system, in order to calibrate and validate the agent-based model. Academic journals are characterised using 13 attributes that are captured across bibliographic data sources. The quantitative relationships among these journal characteristics are analysed over four impact-oriented journal quartiles to further characterise journal types. Cluster analysis of data collected for 11,427 scholars, and their 284,128 journal publications, identifies six types of scholars: singleton, small-team low performer, small-team high performer, big-team strategist, free-style follower, and life-time warrior. The empirical analysis identifies new, previously undescribed, growth patterns of academic journals and scholars, and contributes a characterisation of scholar types that provides a robust and holistic conceptualisation of academic publishing behaviours. A methodological approach of coupling the agent simulation with empirical research is introduced. It uses empirical data to set the parameters of the conceptual agent-based model that best reflects the scholars’ strategic publishing-decision processes. The simulation results validate independent historical data at both macro- (growth patterns of scholars and their publications, as well as the performance measures of scholars) and micro-levels (number of scholars and their publications for each calendar year). Thus, the thesis contributes a new typology of academic publishing strategy, and empirical journal characterisation. This informs a new understanding of the individual and collective behaviours that lead to a variability in research outcomes between scholar types. Importantly, it provides a framework that can be used to optimise the distribution of scholar types within teams and institutions, with the aim of maximising research outputs.