One Idea and Many Approaches: A Meta-Analysis of Transparency Literature

This article explores a sample of the literature on transparency for the past thirty-six years (1984-2020) through a systemic meta-analysis of the scholarship. With a sample of 242 works (articles, books, and book chapters) collected from different academic repositories using API queries in the R programming interface, LDA probabilistic topic modelling – an unsupervised machine learning approach – is employed in order to a) narrow down the sample selection process, and b) to classify and construct a typology of topics within the literature. This process covers a two-sided problem; delimiting samples and dynamically pre-classifying them. Taking advantage of this methodological complementarity, three overarching research questions are addressed: a) What analytical approaches are identified in the literature? b) How is transparency conceptualised through such analytical approaches? And, c) where has transparency’s focus been placed in relation to an event-process framework? Following Heald’s (2006) event-process framework of transparency, the results of the literature meta-analysis show unequal methodological approaches, topics and issues investigated. Crucially, the analysis of the works basic bibliographic data also points towards a divide regarding knowledge production in the field, which paired with the LDA model’s results shed light into more problematic trends. These insights and the novel approach utilised outline challenges and opportunities for future research in the field, with shared conceptual and practical elements – potential greater conceptual consistency – and concerns, such as the far-reaching causes and consequences of corruption.