Introduction

Friendship or personal relationship ties make up many of the networks that we are a part of or can observe. These relationships not only tell us about individual’s closeness to or support of one another, but can also tell us about the sociocultural/historical/economic context at the time, as well as effects of gender, race, and class on interpersonal and group dynamics. Thus it is hardly surprising that one of the first attempts at social network analysis was about a group of young students in Germany.

For my final lab, I chose to work with a classic, well-known social network dataset, “Class of 1880/81: friendship network of a German boys’ school class from 1880/1881”.

Related Visualizations

I entered this exercise with several examples of the ways that others have used this dataset or similar datasets. The examples were clearly done by researchers with more time and greater experience with social network analysis software. Unlike this earlier SI student who had an idea about what kind of design that they wanted for their network, I was inspired more about the specific elements that were pulled out or highlighted in my example networks.

Materials

For this visualization, I used Gephi, an open source network analysis and visualization software. I uploaded one previously downloaded dataset (GEXF file), found on Gephi’s wiki, which included data on a friendship network of a German boys’ school class from 1880/1881. GEXF files are highly compatible with Gephi, so as a new user of the software, I assumed that this file format would be more manageable while learning.

Methods/Process

I downloaded the dataset from Gephi’s list of datasets on Github. While I have previous experience in social network analysis, it has been several years since I have worked with any software or literature, and I did not use Gephi. Therefore, I used this lab as an opportunity to merely explore the software and the dataset.

After opening the dataset in Gephi, I spent most of my time making the visualization more manageable to look at. I first gave the visualization a ForceAtlas2 layout (Fruchterman Reingold was also considered, but the visualization seemed artificially evenly spaced, so I chose ForceAtlas2 instead). I then increased the gravity so that the isolates would not disappear off-screen and would instead just float on the perimeter of the connected network.

To cluster some of the information, I colored and sized the nodes and edges based on degree [of connectedness], and labeled the nodes with each person’s name (which is all of the identifying information that was given).

Result

The resulting network is relatively simplistic, but I do still think that, with some accompanying qualitative information (*Who* were these anonymous students? Why might some of them have been left out? Was it an interpersonal issue between classmates or a result of a greater socioeconomic or political context?), this network could be useful for a researcher studying childhood friendships, classroom relationships, etc.  

Reflections

If I worked with this dataset again, I would like to try to pull out one specific characteristic about the dataset to build a narrative about the classroom dynamic.

While the visualization looks interesting, because I have no context for the dataset, it is not particularly useful for anything other than knowing the relative closeness and connectedness of some students compared to others. A greater quantity and quality of both quantitative and qualitative data would be immensely useful if a researcher wanted to use this dataset to tell a more meaningful story.