When coding-and-counting is not enough: using epistemic network analysis (ENA) to analyze verbal data in CSCL research

Research on computer-supported collaborative learning (CSCL) is often concerned with the question of how scaffolds or other characteristics of learning may affect learners' social and cognitive engagement. Such engagement in socio-cognitive activities frequently materializes in discourse. In quantitative analyses of discourse, utterances are typically coded, and differences in the frequency of codes are compared between conditions. However, such traditional coding-and-counting-based strategies neglect the temporal nature of verbal data, and therefore provide limited and potentially misleading information about CSCL activities. Instead, we argue that analyses of the temporal proximity, specifically temporal co-occurrences of codes, provide a more appropriate way to characterize socio-cognitive activities of learning in CSCL settings. We investigate this claim by comparing and contrasting a traditional coding-and-counting analysis with epistemic network analysis (ENA), a discourse analysis technique that models temporal co-occurrences of codes in discourse. We apply both methods to data from a study that compared the effects of individual vs. collaborative problem solving. The results suggest that compared to a traditional coding-and-counting approach, ENA provides more insight into the socio-cognitive learning activities of students.