Describing patterns in the relationships between people is a fundamental task for network analysts. Network datasets are often difficult to summarize, but general patterns can be found. One common pattern is homophily, the tendency for similar individuals, or members of the same group, to associate with one another.
Along with other members of the Social Dynamics Laboratory at Cornell, I am involved with multiple projects that use large online traces of social networks to better measure and estimate the levels of segregation and integration between different groups in society. Separately, I have worked on a method of measuring homophily along continuous variables within social networks.
Respondent-Driven Sampling (RDS), a network-sampling technique that is used by sociologists and public health workers, leverages the phenomenon of homophily to gather information about hard-to-reach or stigmatized populations. At Cornell I work with Douglas Heckathorn, the developer of RDS, and others to measure and improve the statistical reliability of population estimates from RDS samples.