Network analyses allow researchers to ask questions regarding the relationships between items of interest. The UC Davis DataLab worked with Professor Marc Facciotti and Dr. Kamali Sripathi to visualize and analyze the connections between core teaching concepts in an introductory biological sciences (BIS2) curriculum. This project sought to find possible gaps in student understanding by comparing two custom network datasets. One network was generated and validated by subject matter experts, while the other was generated by student use patterns of the publicly available teaching materials. To allow instructors and teaching researchers to engage more deeply with curriculum design and help identify possible differences between these models of understanding, the DataLab developed a Network Report, a sister service to our Text Report.
The BIS2 series is required for every undergraduate biology student passing through UC Davis, with the BIS2A course alone averaging over 3000 students per year hailing from more than 80 different majors. Because of the foundational nature of this course, which is a prerequisite to almost all other biological sciences courses, all instructional offerings must meet specific core learning objectives and competencies. The specific core components for this curriculum align with national standards from the American Association for the Advancement of Science. But how do those objectives interact, and how do the students navigate through them during the course? In an effort to promote student success and ensure that the materials are taught as effectively as possible, Dr. Marc Facciotti and Dr. Kamali Sripathi are leveraging network analyses to zoom out to map these connections at the course level.
The Network Report
The DataLab Network Report takes in a network dataset, and generates common network metrics, along with accompanying interactive visualizations. Each measure includes an explanation of how it is generated and what it looks like in the context of the specific network used for the report. In the BIS2A network, we quickly found that one learning concept, an exercise of walking through the energy story of a reaction, was central to both student and expert understanding of the course materials. Removing that learning objective, or node, would have a drastic impact on how concepts connect across the curriculum, and how the students navigate through them during their learning journey.
The metrics covered by the report include both network and node level measures. For the network as a whole we include the number of nodes (entities) and edges (connections), as well as the number of disconnected groups (called components), and lone nodes (called isolates). We also calculated the density of the network, a measure of how interconnected it is, and the diameter. For node level metrics, we provide and explain the degree, betweenness, and eigenvector centrality scores, along with the mean geodesic distances. Each of these is a different way to determine the importance or centrality of a specific node in the network. Finally, the report provides a way to compare and contrast known groups in the network, as well as run clustering algorithms to identify groups via unsupervised methods.
Supplemental to the Network Report, we also created an onboarding document for researchers, to help them decide if a network approach will be useful for their project. It details what network analysis can do, when it is useful, and what required data components need to be prepared. If you are interested in learning more about network analyses, DataLab research support services including the network report, and upcoming training opportunities, reach out to the DataLab or stop by our weekly office hours.