UCLA

In this paper, we focus on the relationship of public figures using statistical network analysis methods. Specifically, we analyze the network of Instagram following relationship of NBA All-Stars in the most recent 5 seasons. Based on Latent Space Position Modeling, we find that there might be an International “gang” and a veteran "gang" within this network although it is not so clear. Triad Census Model implies that this network is more likely to be transitive. With the help of Exponential-family Random Graph Models (ERGM), we analyze this network from 4 aspects, including Demographical Characteristics, Social Media Characteristics, Business Factor and Basketball Factor. We find strong Nationality and Team homophily effects. As for Business Factors, it does show some Brand homophily effect. Also, players who have more NBA experience, more honors and are more active on social media are more likely to form a tie with others.