Topic Modelling with Constructivist Grounded Theory: A Way of Big Textual Data Analysis for Theory Building

There is a growing demand for methodological plurality, especially with the emergence of Machine Learning (ML) techniques for analysing big textual data. To address this need, our study introduces a novel methodology that combines a ML technique with a traditional qualitative approach to reconstruct knowledge from existing publications. With its pragmatist and abductive stance, it allows for human-machine interaction. The method employs Topic Modelling (TM), an ML technique, to facilitate Constructivist Grounded Theory (CGT). A four-step coding process (Raw Coding, Expert Coding, Focused Coding, and Theoretical Coding) is implemented to ensure procedural and interpretive rigor. To present this approach, we collected data from an open-source professional project management community website and illustrated the research design, data collection, and data analysis processes leading to theory development. The results revealed the potential of this novel methodology to extract latent meanings and reveal phenomena within published data, thereby offering a new avenue for academics to develop potential theories in various fields.