Homophily in social networks
One of the important observations made by social scientists is a tendency in social groups for similar people to be connected together (after all ‘birds of a feather flock together’). It has a significant impact on the value we get from social media (as often we hear similar voices and interact with like-minded people).
This phenomenon is called Homophily (meaning love of the same) (McPherson et al. 2001).
Homophily can be directly observed in the virtual worlds using analytical techniques, for example Huang et al (2009) showed that in the Massive Online Role-Playing Game Everquest players tended to interact with other players of similar age, experience and who lived near them in the real world. This held across all sorts of interactions, from questing together to trading in the in game auction house. In fact the only way they looked for Homophily and didn’t find it was in gender, something that they put down to the fact that 32% of people play the game with a romantic partner.
Homophily has predictive power in social media, so much so that researchers looking at last.fm could predict real-life friendships by examining on-line interaction, shared interests and location (Bischoff, 2012).
In fact homophily is so powerful a principle that whole communities on Facebook can be modelled by extrapolating from as little as 20% of the population (Mislove et al. 2010). This has real consequences for privacy and anonymity, as merely knowing your place in a network may allow analysis tools to make guesses about your private information with high accuracy.
In Twitter De Choudhury (2011) has shown that different types of homophily hold for different types of users (for example, normal users with roughly the same number of followers and followed have location and sentiment homophily - i.e. they tend to live and work near each other, and show similar reactions and views).
Homophily is a good example of where an existing social theory can now be explored numerically, and be easily verified in a wide variety of different networks, because the data is held digitally.
To what extent are the people that you friend/follow on your own social networks similar to you? Are there exceptions, and do those exceptions contribute something of different value to your personal network compared to people who are similar to you?
A summary of De Choudhury’s article on homophily in Twitter is attached to this article. Optionally take some time to read the summary and understand how the research was undertaken. It’s a good example of how comparing statistics from different types of users can start to reveal how large social networks behave.
Bischoff, K. (2012). We love rock “n” roll: analyzing and predicting friendship links in Last.fm. Presented at the WebSci ‘12: Proceedings of the 3rd Annual ACM Web Science Conference
De Choudhury, M. (2011). Tie Formation on Twitter: Homophily and Structure of Egocentric Networks. SocialCom/PASSAT, 465–470.
Huang, Y., Shen, C., Williams, D., & Contractor, N. (2009). Virtually There: Exploring Proximity and Homophily in a Virtual World. International Conference on Computational Science and Engineering, 4, 354–359.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a Feather: Homophily in Social Networks. Annual review of sociology, 27, 415–444.
Mislove, A., Viswanath, B., Gummadi, K. P., & Druschel, P. (2010). You are who you know. Presented at the third ACM international conference, New York, New York, USA: ACM Press.
© University of Southampton 2017