Visualizing the Networks of Bicycling Groups for, of, and by Women in Philadelphia and New York City

Background

As an avid cyclist and a woman, I have taken part in two groups dedicated to supporting and raising awareness about women who chose to ride bicycles in urban areas: WE Bike, based in New York City; and Women Bike PHL, based in Philadelphia. Both of these groups have closed facebook groups to facilitate skill sharing and community building: The Female Bike Forum by WE Bike; and Women Bike PHL. Although I live in New York City, I personally have many more friends and am more attached to the Women Bike PHL group. This is mainly because my friend, Katie Monroe, founded the group, and it has become wildly successful! She was inspired to create the group after speaking with someone from WE Bike, so Women Bike PHL is a spin-off of sorts.

Method

My odd loyalties caused me to be interested in seeing how the groups differ in terms of the density of the networks in terms of who is friends with whom and how the group members interact with each other. To achieve this goal, I found I could use an application called netvizz through Facebook. The program describes itself as “a tool that extracts data from different sections of the Facebook platform – in particular groups and pages – for research purposes.” I simply had to find the number assigned to the groups, and netvizz extracted information about friendships and interactions in the group, such as who responded to someone’s post. However, it does not give users’ names so I do not know exactly who is friends with whom.

Research Questions

I was interested to see if the friendship structure and the interaction structures would be similar or did people interact with a wider range of people. Which group had more members? Was one more active than the other? I also wanted to see if visualizing the Women Bike PHL group would prove to show it to be more of an ego-centric network since it was the brain child of an individual.

The Visualizations

These are single node as each node represents a person. They show both propinquity and homophily because they are people living in the same place loving the same thing (cycling). These are force directed diagrams showing structure level networks, the entire system (facebook group). The visualizations produced allow us to analyze these social networks. They are not socio-centric because the relationships cannot be contained in a box; they are not in a closed system in which everyone has a relationship to each other.

women_bike_phl_friendships The Women Bike PHL Friendships visualization shows a few well defined subgroups or cliques. I chose to have the colors and weight show this modularity for all of the visualizations. The largest node I believe to be the founder of the group. I’m guessing that the orange or purple subgroups are either friends of Katie’s who went to our alma mater, Haverford College or women she knows through the Bicycle Coalition of Greater Philadelphia. There are many nodes that have no edges, showing that these are people who have no friendship connections through facebook, but are a member of the Women Bike PHL facebook group. I chose to have darker colors for the Women Bike PHL visualizations and brighter colors for the WE Bike to help viewers have a visual cue to distinguish the two groups.

486536454743309_women_bike_phl_interactions2

 

This shows a central node interacting with many other nodes.
298653100250671_we_bike_group_interactions2This visualization shows that several nodes contribute regularly to the interactions and slightly dominate the group.

298653100250671_we_bike_group_friendships2In this visualization we can see four distinct colors: pink, aqua, yellow, and sea foam green. This suggests four different distinct in which nodes are connected, are interested in, or are brought to the group. If only it would tell us how and why!

Interpreting the Statistics

From here on out I will refer to the groups by their city’s abbreviations: PHL for Women Bike PHL and NYC for We Bike.

At the time of this research there were:

  • PHL friendships: 1321 nodes (people) and 3494 edges (connections)
  • PHL interactions: 510 nodes and 1953 edges
  • NYC friendships: 585 nodes and 1276 edges
  • NYC interactions: 207 nodes and 801 edges

This shows us that PHL has many more group members than NYC, over double the amount. My hypothesis is that since Philadelphia with 1.5 million residents has a much smaller population than New York City with 8.4 million, there is much greater potential for visibility and circulation of the group. Whereas in NYC people have many more competing and potential outlets for their interests. The node to edge ratio for friendships in PHL is 0.378 and NYC is 0.458. The node to edge ratio for interactions is exactly the same at 0.26.

It was interesting to see that the modularity for the friendships within the groups was very similar: PHL: 0.56 and NYC: 0.554. This means that they have almost exactly the same density of connections in the communities. The modularity for interactions is nearly the same as well: PHL: 0.261 and NYC: 0.246. This makes sense that in a more public forum the density would be lower as people who do not know each other through a facebook friendship would interact less with strangers.

The clustering coefficients for the friendships show that statistically with 0.394 NYC is slightly more of a small world/clique-ish than PHL with 0.266. However, in terms of interactions, PHL is slightly higher with 0.556 and NYC at 0.46.

The six degrees of separation theory is very popular, but is it true in this case? The friendship average path length is almost the same in both groups: NYC with 3.4 and PHL with 3.8. This means that the women are likely be connected to each other with three degrees of separation. The interactions path length for both is 2.6. This makes sense since theoretically people who are more closely related would interact with each other more often.

-J.E. Molly Seegers