Seeing the Forest through the Trees Using Network Analysis: Exploring Student Responses to Conceptual Physics Questions

This study explored how scientific ideas interacted in college students’ scientific explanations using text analysis techniques and network analysis. A task was given to first-year college students asking them to answer a series of physics questions associated with a computer simulation. Students’ written responses were classified into three models—non-normative, mixed, and normative—and then subjected to text analysis using IBM SPSS Modeler with text analytics software. Using the text analysis result, a decision tree model was applied to find what ideas were significantly attributed to the different models of responses. Finally, network analysis was used to reveal and illustrate how the important attributes were used and connected with each other across the three models. The findings support the importance of teaching a small number of core ideas as the premise of the Framework for K-12 Science Education, and indicates that integrated knowledge should emphasize organizing knowledge around the core ideas.