Rigorous Inference for Social Network Models based on Egocentric Data, and its Implications for Epidemiology of HIV
10 April 2017
What to do when you want to understand disease spread over a social network… but you can’t actually observe the network?
NIASRA member Dr. Krivitsky has recently published a paper in Annals of Applied Statistics describing an approach for fitting whole-network models to egocentrically-sampled data: data where the researchers can only observe limited information about a small sample of individuals and non-identifying information about their partners. This facilitates simulation of what the whole network might look like, and how it might mediate the spread of disease.
Reference: Pavel N. Krivitsky and Martina Morris (2017). Inference for social network models from egocentrically sampled data, with application to understanding persistent racial disparities in HIV prevalence in the US. Annals of Applied Statistics, Volume 11, no 11, p427 - 455