Network Methodology

Network science is an area of active methodological development across many disciplines, from sociology to statistics to public health to physics. Our lab identifies methodological problems and attempts to develop solutions. We have published methodological advances related to the longitudinal analysis of large social networks; the estimation of peer effects (we introduced the "directionality test" for identification in 2007); and the understanding of spreading phenomena in partially observed networks. We have also pioneered the use of new tools for doing experiments with networks online. Our lab's interest in the genetic bases of social interactions has also led to advances in this area (e.g., the use of genes as instrumental variables in the estimation of peer effects).
Ongoing work involves improving methods for matching in the conduct of randomized network experiments and improving methods for understanding fundamental network dynamics. Our lab continues to develop and place into the public domain powerful software useful in network science, including Breadboard for conducting online experiments and performing diagnostic assessments of groups, and Trellis for mapping networks in the field and online.
Finally, we have also given thought to the ethical use of network technology and to the ethics of human subjects research related to the use of online and other network data.