Sources of variation in social networks

What explains the large variation in the number of contacts (degree) that different participants of social networks have: age, randomness, or some unobservable fitness measure? To answer this question, I extend the model presented in Jackson and Rogers (2007) to allow individuals to vary in their ability to attract contacts. I estimate the parameters of the extended model, using a social network of citations among high-energy physics papers, and find that the extended Jackson–Rogers model can parsimoniously fit the degree distribution of each age cohort. Moreover, both the length of time spent in the network and the unobservable fitness measure are important in explaining the observed variation in participantsʼ degrees.