Friendship Networks and Mobile Data
Link to article:
http://www.pnas.org/content/106/36/15274.full.pdf
The article “Inferring Friendship Network Structure by Using Mobile Phone Data” by Eagle, Pentland, and Lazer critiques the study of social networks through an analysis of the Reality Mining study. Through the analysis of this specific study, the authors argue that behavioral data as well as self-report data give rise to a deeper understanding of the behavior of social networks as well as how to decipher a strong versus weak tie.
The Reality Mining Study was conducted in 2004 with 94 participants. During the study, the participants used cell phones that had software already installed on them to track call logs and bluetooth data about the participant’s proximity to cell phone towers. This data was then sent directly to the researcher. Unlike traditional, social network studies this study used self-report data combined with behavior patterns to determine whether or not each individual was connected by a strong or weak tie. After evaluating data, the researchers found that self-report data gave a better picture of how the population’s network was connected. This was due to the social norms that constrained the definition of friendship, thus making people evaluate their relationship as either positive or negative. However, the data did not provide enough information on how some relationships fulfilled basic qualifications of friendship but by artificial standards. For example, if the questionnaire asked how many hours you spent with a person and if the hours exceeded five daily, they would be considered a friend, then most co-workers would fit this definition. However, through the behavioral analysis of interactions on the participant’s cell phones researchers saw that the time of day and the location of interactions impacted how individuals defined their friendships. With this behavioral data, it led researchers to see that a positive tie isn’t necessarily dependent on time, rather it is dependent on if an individual makes an effort to spend time with someone who is not automatically in their environment.
The research conducted in the Reality Mining Study brings light to the discrepancies that exist between self-report data and behavioral data. Through an analysis of these differences, researchers were able to see the complexities of defining a strong or weak tie. Rather than only seeing a picture of the population of participants, the behavioral data allowed researchers to understand how an individual’s definition of a strong tie can starkly contrast another persons. Thus, revealing how the population’s behavior in a network does not tell the full extent of the relationship’s strength.