What does cell phone usage reveal about the strength of relationships and the make-up of social networks?
That’s one of the important questions that a project on the dynamics of social networks at the Interdisciplinary Center for Network Science and Applications (iCeNSA) investigates.
The project exemplifies the collaborative nature of iCeNSA research by bringing together primary investigators from four departments: Zoltán Toroczkai (physics), Nitesh Chawla (computer science and engineering), David Hachen (sociology), Omar Lizardo (sociology) and Mark Alber (mathematics).
Hachen notes that iCeNSA’s strength is its interdisciplinary, inter-college nature. That’s evident in the vast array of the people involved and studies conducted.
“The social networks project is divided into three groups,” says Hachen, “but we all work together. Toroczkai and his students work on the modeling aspects, Chawla and his group focus on data mining and prediction, and Lizardo and I focus on the sociological issues.”
So, what does cell phone usage reveal about relationships? It turns out a lot.
Hachen and Lizardo study the number and frequency of calls and text messages to explore reciprocity – the way people respond to each other – and how it emerges over time.
“If I called you a lot, and you didn’t call me a lot that’s an imbalanced relationship,” says Hachen. “It turns out that’s extremely important. We now believe what’s really important in networks is not the formation of ties—and that’s important—but the persistence or the dissolution of ties.”
According to Hachen, this social network research is distinctive because it focuses on how strong or weak ties are between people – not just whether ties exist, which is what a lot of previous research examined. Researchers use cell phone data from seven million people – numerical data like the number, frequency and timing of calls and texts, but not the actual content of conversations nor any information, such as phone numbers, that could lead to the identification of individuals – to form a picture of a social network.
That gives them an idea of how many friends a caller has, how many people they interact with and how strong those ties are. It also enables them to predict future behaviors.
To study relationship strength researchers also look at who initiates contact and how often it occurs. One conclusion: ties that are more reciprocal are more likely to remain intact. When there is an imbalance the ties will likely dissolve.
Hachen says that it’s easier for a relationship to become reciprocal if two people have a similar number of ties. “If you have 20 friends, and I have 20 friends we can achieve a balance. But if I have 20 friends and you have two, you’re more likely to try to call me a lot more.”
That exemplifies one key distinction between social and physical networks: in social networks a person tends to connect with others who connect as much as they do, whereas physical networks like air transportation systems must connect major hubs with many regional airports because to connect major hubs only to each other breaks the system.
Lizardo says the type of data affects social network research. A decade ago, he says, most data came from interviews that required people to remember friends and provide subjective information.
Now, the ability to gather information about peoples’ actual behavior rather than memories about their behavior means that researchers can make more accurate observations. Interviews often reveal only strong ties, yet weak ties are very important, too. Behavioral research reveals both.
The next step for the sociology group is to add another layer of data. “In the cell phone project, we have lots of data on the network, but no or very limited information on the people and their behaviors, their tastes, their attitudes,” says Hachen. “Sociologists are constantly interested in the attitudes, beliefs, behaviors from surveys, but they don’t have network information. We have network information, but not that. And now we want to get both.”
So they have begun the legwork for another cell phone project that will provide 250 first-year Notre Dame students with smart phones, in order to address questions like whether ties influence people’s behavior or whether people choose to form ties based on existing similarities. For the project Hachen and Lizardo will collaborate with engineers who are interested in increasing the quality of wireless networks.
For more information, contact David Hachen at dhachen@nd.edu.