a b s t r a c tScholars have long recognized the potential of Internet-based communication technologies for improving network research-potential that, to date, remains largely underexploited. In the first half of this paper, we introduce a new public dataset based on manipulations and embellishments of a popular social network site, Facebook.com. We emphasize five distinctive features of this dataset and highlight its advantages and limitations vis-à-vis other kinds of network data. In the second half of this paper, we present descriptive findings from our first wave of data. Subgroups defined by gender, race/ethnicity, and socioeconomic status are characterized by distinct network behaviors, and students sharing social relationships as well as demographic traits tend to share a significant number of cultural preferences. These findings exemplify the scientific and pedagogical potential of this new network resource and provide a starting point for future analyses.

The rapid growth of contemporary social network sites (SNSs) has coincided with an increasing concern over personal privacy. College students and adolescents routinely provide personal information on profiles that can be viewed by large numbers of unknown people and potentially used in harmful ways. SNSs like Facebook and MySpace allow users to control the privacy level of their profile, thus limiting access to this information. In this paper, we take the preference for privacy itself as our unit of analysis, and analyze the factors that are predictive of a student having a private versus public profile. Drawing upon a new social network dataset based on Facebook, we argue that privacy behavior is an upshot of both social influences and personal incentives. Students are more likely to have a private profile if their friends and roommates have them; women are more likely to have private profiles than are men; and having a private profile is associated with a higher level of online activity. Finally, students who have private versus public profiles are characterized by a unique set of cultural preferences-of which the ''taste for privacy'' may be only a small but integral part.

Disentangling the effects of selection and influence is one of social science's greatest unsolved puzzles: Do people befriend others who are similar to them, or do they become more similar to their friends over time? Recent advances in stochastic actor-based modeling, combined with self-reported data on a popular online social network site, allow us to address this question with a greater degree of precision than has heretofore been possible. Using data on the Facebook activity of a cohort of college students over 4 years, we find that students who share certain tastes in music and in movies, but not in books, are significantly likely to befriend one another. Meanwhile, we find little evidence for the diffusion of tastes among Facebook friends-except for tastes in classical/jazz music. These findings shed light on the mechanisms responsible for observed network homogeneity; provide a statistically rigorous assessment of the coevolution of cultural tastes and social relationships; and suggest important qualifications to our understanding of both homophily and contagion as generic social processes.T he homogeneity of social networks is one of the most striking regularities of group life (1-4). Across countless social settings-from high school to college, the workplace to the Internet (5-8)-and with respect to a wide variety of personal attributesfrom drug use to religious beliefs, political orientation to tastes in music (1, 6, 9, 10)-friends tend to be much more similar than chance alone would predict. Two mechanisms are most commonly cited as explanations. First, friends may be similar due to social selection or homophily: the tendency for like to attract like, or similar people to befriend one another (11,12). Second, friends may be similar due to peer influence or diffusion: the tendency for characteristics and behaviors to spread through social ties such that friends increasingly resemble one another over time (13,14). Though many prior studies have attempted to disentangle these two mechanisms, their respective importance is still poorly understood. On one hand, analytically distinguishing social selection and peer influence requires detailed longitudinal data on social relationships and individual attributes. These data must also be collected for a complete population of respondents, because it is impossible to determine why some people become friends (or change their behaviors)* unless we also know something about the people who do not. On the other hand, modeling the joint evolution of networks and behaviors is methodologically much more complex than nearly all past work has recognized. Not only should such a model simulate the ongoing, bidirectional causality that is present in the real world; it must also control for a number of confounding mechanisms (e.g., triadic closure, homophily based on other attributes, and alternative causes of behavioral change) to prevent misdiagnosis of selection or influence when another social process is in fact at work (15).Using a unique social network dataset (5) and ad...

No abstract