Homophily and Social Tie Indicators
The principle of homophily states that “similarity breeds connection”. Numerous studies on homophily have provided strong evidence that similarity in various aspects of our daily life such as ethnicity, gender, age, religion, education, occupation, behavior and attitude breeds the formation of social ties (for a review, see McPherson et al., 2001).
One good example is the study done by Crandall and colleagues (2010), in which they examined the relationship between spatial-temporal co-occurrences and the existenceof real social ties among users on Flickr.com, an online image sharing community. The authors used geo-tags and time stamps (either user generated or automatically recorded bythe photo taking devices) of photos uploaded by users to infer whether two users were at the same location (as defined by a certain latitude-longitude range) within a certain period of time (e.g., a day, a month or a year). Their results suggested that as the number of spatial-temporal co-occurrences increased, the probability of two users actually having a social tie as indicated in Flickr’s public social network increased drastically. It is quite striking to learn that, for example, if two users had were at the same location (within 1 latitude-longitude degree) on a given day three times, they were 300 times more likely to know each other than chance level. As we all probably notice, we are co-located with quite a large number of strangers at various locations (such as bus stops, super markets) everyday. I am quite amazed by the fact that the co-location information is so much more informative of our social relations than I originally thought, especially given the fact that the co-location information provided on Flickr is by no means “complete”, and in fact, very sparse.
However, I am also wondering if there are factors other than homophily that would also be good indicators of social ties. Sometimes two individuals might also be strategically avoiding each other for other purposes, such as optimizing allocation of limited resources. For example, two individuals might strategically choose to attend different sessions during a multi-session conference to maximize coverage of the conference content. This “avoidance” behavioral pattern then indicates a strong social tie since it foreshadows a higher probability of communication between these two individuals due to heightened need of information exchange. I am wondering how other social tie indicators can be identified and used in social network analysis.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27 , 415-444.
Crandall, D. J., Backstrom, L., Cosley, D., Suri, S., Huttenlocher, D., & Kleinberg, J. (2010). Inferring social ties from geographic coincidences. Proceedings of the National Academy of Sciences, 107 (52), 22436-22441.