Jason Kaufman Nicholas A. Christakis Andreas Wimmer Harvard University
The ?new? science of networks has become one of the most vibrant and promising fields of social scientific inquiry. In particular, the rapid growth of social network sites (SNSs)?online spaces specifically dedicated to articulating users? networks?has spurred a bustle of academic activity. Using one popular SNS, Facebook.com, and collaborating with the administration at an undisclosed American university, the PIs have compiled two waves of network data on a complete college cohort of 1,640 students. The data also include information about the cultural preferences and group activities of our respondents over time. They will study phenomena such as the ethno-racial locus of homophily among students; the relationship between cultural ?tastes? and friendship; and the competing roles of spatial proximity and on-line affiliation in the formation of friendship groups. The PIs will collect data on the students for one additional year (when the majority will graduate from college); continue analyzing these data across multiple research agendas; present and publish this research; and, most importantly, make this dataset available to other researchers. Not only does this dataset contain some of the richest personal and relational data to be extracted from an SNS to date, but it also has a number of important advantages vis-Ã -vis past network datasets. 1) By collecting naturally occurring data, the Facebook instrument avoids a number of prominent measurement issues in social network research, including interviewer effects, memory limitations, and the artificial truncation of respondents? reported network ties; 2) Facebook enables complete network data to be collected on a large, naturally-bounded population. Hence, both ego-centric as well as ?socio-centric? analyses are possible; 3) The dataset is multiplex, containing data on Facebook friends, ?picture friends? (coded using postings in online photo albums), and?in a ?real life? supplement to these online measures?various kinds of housing data provided by the university; 4) Unlike the majority of network datasets, the final dataset will include four waves of longitudinal data; and 5) Facebook enables the collection of extensive, open-ended cultural data in the form of favorite movies, music, and books. As a result of these five features, the dataset can be employed for a number of intellectual projects that have heretofore been difficult or impossible to pursue. Such projects include examining the distinct role of race/ethnicity in network behavior; inductively determining the best way to measure and classify cultural preferences in light of the extant literature on cultural capital; and comparing the respective importance of cultural, demographic, and geographical factors in the formation of entire social networks.
Broader Impact. The final dataset will contain four waves of data corresponding to the four years respondents are in college. It will be available via the Internet to other researchers interested in utilizing this data. The project thus provides an important new resource for the broader community of network scholars, one that can be used for methodological development (refining tools for analyzing large, multiplex, directed, longitudinal, and/or two-mode data) and for examining diverse empirical phenomena (spanning fields from social stratification to cultural sociology). The methodology will be widely emulated and improved upon by other researchers, thereby enhancing our understanding of both general network phenomenon and American college students? social and cultural worlds more specifically.
