The Structure of Social Situations: Insights from the Large-Scale Automated Coding of Text

Abstract: Social situations are highly complex, with intricate structures that standard survey-based tools cannot fully capture. For this reason, psychology lacks an integrative framework for representing situations and relating them to the myriad psychological variables at play in daily life. We address this problem by analyzing more than 20,000 detailed textual descriptions of dyadic social interactions obtained from participant-generated stories, published fiction, blogs, and autobiographical narratives. Our main methodological contribution is to use generative artificial intelligence (AI) to code the textual descriptions along eighty-eight psychological dimensions, including variables proposed by existing situational taxonomies as well as closely related variables like goals, relationships, and activities. We begin by showing that AI can code situational variables at a human-level of accuracy, validating our approach. We then explore the distributional statistics of our data, and document systematic correlations between groups of variables. Notably, we find that situational valence and actor goals are the most predictive variables, highlighting their central role in situation representation. We also show that situational descriptions cluster into core templates involving themes such as social conflict, family life, and achievement. Our final analysis finds that these situational structures are largely consistent across different datasets, with some minor differences between published media and survey-based participant responses. Overall, our paper shows how large-scale text data, combined with recent advances in AI, can bring analytical precision to classic research problems in social and personality psychology, offering a novel and integrative approach to understanding the complexity of social situations.

Keywords: social situations; artificial intelligence; situation representation; text analysis