From Textual Data to Theoretical Insights: Introducing and Applying the Word-Text-Topic Extraction Approach

Abstract

Text analysis, particularly custom dictionaries and topic modeling, has helped advance management and organization theory. Custom dictionaries involve creating word lists to quantify patterns and infer constructs, while topic modeling extracts themes from textual documents to help understand a theoretical domain. Building on these two approaches, we propose another text analysis approach called word-text-topic extraction (WTT), which enhances the efficiency and relevance of text analysis for the sake of theoretical advancement. Specifically, we first identify relevant words for a researcher's theoretical area of interest using word-embedding algorithms. That step is followed by extracting text segments from the textual corpus using a collocation process. Finally, topic modeling is applied to capture themes relevant to the specific theoretical area of interest. To illustrate the WTT approach, we explored one research area needing further theory development—innovation. Using 841 CEOs’ letters to shareholders, we found that our WTT approach provides nuanced features of innovation that differ across industry contexts. We guide researchers on decisions and considerations related to the WTT approach in order to facilitate its use in future studies.