Skip to main content



Course Social Networks and Persistence in STEM

This paper by Zwolak, Zwolak, and Brewe comes from the physics education research community: a coalition of researchers who study how students understand physics and how students experience physics classes. In this study, the researchers gathered social network data from students in an introductory physics class, and ran logistic regressions to determine whether the social networks and other variables such as grades and gender could predict whether a student would persist in physics and continue taking physics courses. They had students name others who were in both their in-class and out-of-class networks: the students they worked with during class time, and the students they spent time doing homework with outside of class. For these networks, the researchers calculated the indegree, outdegree, closeness, and eigenvector centrality of each student. As discussed in class, a student’s indegree is the number of students who named them as a collaborator, and outdegree is the number of others that that student named. Hence, a student who is well connected in a course would have a high indegree and outdegree. As we learned in class, the distance between two nodes is the length of the shortest path between them, and closeness is a measure of one divided by the sum of the distances from that student to all other students in the network. Since the sum of distances is in the denominator, a low average distance to other nodes leads to a high closeness score. Finally, eigenvector centrality is a more complicated calculation that is based on not only how well connected the student is, but also how connected the student’s connections are. For example, if student A has an outdegree of 10, and each of those 10 nodes also have an outdegree of 10, their eigenvector centrality is higher than student B who has an outdegree of 10, but those 10 students are only connected to two other students each.

I found the results of the paper to be really interesting and to speak to my own educational experiences. The authors found that grade is the most important predictor of persistence for students who receive relatively high grades in the class and relatively low grades in the class. This makes sense; getting a good grade encourages a student to take more similar courses, while getting a bad grade is discouraging. Interestingly, the most important predictors of persistence for students whose grades were in the middle (most students) were closeness and outdegree in the out-of-class network. This also makes sense intuitively: if I have friends that I hang out with outside of class who are taking the next course in the physics sequence, I might as well tough it out with them.  

The fact that in-class networks are unimportant for persistence has interesting implications for instructors. Although our courses often attempt to foster in-class networks (for the physics department at least, we have collaborative iClicker questions, group work in discussion sections, and a corresponding lab course that is completed in groups), it is only the out-of-class network that significantly influences students’ decisions to stay in physics courses. The authors of the paper propose that instructors should administer the network surveys to their students in the middle of the semester, and provide some sort of intervention for students who aren’t part of the giant component (as most students will inevitably belong to a giant component) to help them get connected. However, this presents a huge logistical challenge: how do we nudge students towards building their out-of-class network without overstepping boundaries? I have personally noticed one of my professors taking action to develop an out of class network, though. In INFO 1300 with Professor Harms, we are frequently told in both lecture and “lab” to ask the people around us for their phone numbers so that we can contact them outside of class. I think this is a really clever way to give students a support structure in the course, especially for those who don’t already know other students in the course. Prof. Harms always talks about evidence-based strategies, and developing an out-of-class network is indeed one of them!

 

Source: https://journals.aps.org/prper/abstract/10.1103/PhysRevPhysEducRes.14.010131

Comments

Leave a Reply

Blogging Calendar

September 2022
M T W T F S S
 1234
567891011
12131415161718
19202122232425
2627282930  

Archives