Methodological Rigour in Algorithm Application: An Illustration of Topic Modelling Algorithm

Abstract

The rise of advanced computational algorithms has opened new avenues for computationally intensive research approaches to theory development. However, the opacity of these algorithms and lack of transparency and rigour in their application pose methodological challenges, potentially undermining trust in research. The discourse on methodological rigour in this new genre of research is still emerging. Against this backdrop, I attempt to offer guidance on methodological rigour, particularly in the context of topic modelling algorithms. By illustrating the application of the structural topic modelling algorithm and presenting a set of guidelines, I discuss how to ensure rigour in topic modelling studies. Although the guidelines are for the application of topic modelling algorithms, they can be applied to other algorithms with context-specific adjustments. The guidelines are helpful, especially for novice researchers applying topic modelling, and editors and reviewers handling topic modelling manuscripts. I contribute to the literature on topic modelling and join the emerging dialogue on methodological rigour in computationally intensive theory construction research.