This paper aims to analyze the tradeoffs inherent in using text analysis, an increasingly popular methodology for analyzing teams, using McGrath’s (1981) ABC framework as a guide.
Noting team research examples, the authors review the ABC framework, which outlines that all research choices have implications for the generalizability over actors (A), precision of measuring behavior (B) and realism of context (C). The authors then use the framework to situate team researchers’ choices around text data collection and text analysis techniques.
The paper suggests that there are systematic tradeoffs in how team interaction text is collected and analyzed to investigate team processes. With respect to collection, text data from naturalistic settings offers high contextual realism but lacks the behavioral precision of text obtained in experimental studies. With respect to text analysis techniques, human annotation can capture team behaviors with nuanced contextual details but may lack generalizability. Word count methods offer broad generalizability but sacrifice contextual realism. Meanwhile topic modeling, natural language processing and large language models can uncover team dynamics in large text data sets but often with lower behavioral precision.
This paper underscores the importance of aligning text analysis methods with research objectives based on the ABC framework. The authors build on the analysis of the team text analysis tradeoff to provide key recommendations for researchers studying teams and team processes.
