Topic models as a novel approach to identify themes in content analysis

Abstract

In this paper, we demonstrate the usage of topic modeling as a computer aided content analytic tool in the larger context of methods for analyzing text data. We present some key features of topic modeling based on Latent Dirichlet Allocation (LDA), and demonstrated its application by analyzing the articles published in Organization Research Methods (ORM) since its inception. Our analysis, based on 421 ORM articles reveals 15 topics, which are quite similar to other, more human intensive review exercises. We also identified quantitative measures of relative importance of different topics, which could be used as a variable for further analysis in quantitative studies. To further demonstrate the usage of topic modeling, we identified how the emerged topics varied depending on the disciplinary background of the authors. We conclude by providing some examples of the usage of topic modeling in management research