ENA refined an ethnography of socioemotional development in a learning community

Hod, Katz, and Eagan examine how a computational model can strengthen, question, and extend a qualitative ethnography rather than replace it. Their 2020 Computers & Education article studies socioemotional development in a Humanistic Knowledge Building Community and uses Epistemic Network Analysis to revisit a previously developed stage-based interpretation.

The setting was an intensive 13-week graduate course at the University of Haifa with 18 students. The course was designed as a learning community in which participants investigated learning communities while also building one themselves. The researchers analyzed 1,170 notes posted to Knowledge Forum, producing 1,884 coded lines. Six socioemotional categories were used: desire, dynamics, feelings, life outside, empathy, and likeness. These codes represented patterns such as motivation, interpersonal dynamics, emotional expression, connections beyond the course, empathic attention, and perceived similarity.

The authors first drew on the existing qualitative ethnography, which described four stages of group development. They then used ENA Web Tool version 1.6.0 to model how socioemotional codes were connected within each stage. A note was the unit of analysis, indexed by its day, stage, author, codes, and readership. The moving window included each note and the three previous notes, allowing nearby contributions to contribute to one network. Networks were aggregated and compared visually, statistically, and by returning to the underlying discourse.

The ENA model robustly distinguished the four stages. It supported important parts of the original ethnography, including changing relationships among group dynamics, emotional expression, connectedness, and prior experience. It also refined the account by revealing connections that were less visible in the stage narrative alone. The method gave the researchers a shared geometric space in which every note-level network could be compared while still permitting close reading of the original messages.

The value of the paper is methodological as much as substantive. ENA did not independently discover a universal sequence of group development, nor did the network positions prove that one socioemotional pattern caused a group to advance. The model depended on a specific community, coding scheme, stanza choice, and moving window. The authors used the quantitative representation to test the coherence of an ethnographic interpretation and locate further qualitative questions. The study therefore provides a strong example of mixed-method reasoning: theory and close reading define meaningful codes; ENA compares their relational structure; and interpretation returns to the data and setting that gave those relationships meaning.