|
Abstract:
|
Social network data consists of social ties, node characteristics and behaviors over time. We all know that people who are close to each other in a social network are more likely to behave in a similar way. one of the reasons they act similarly is due to the peer influence and social contagion that acts along the network ties. A primary interest of social network data is to identify the contagion-based social correlation. To answer this question, based on the generalized linear transformation model we propose a regression model with a time-varying covariate for time to event data to incorporate the network structure and quantify the contagion-based social correlation. Then, consistency and asymptotic normality are proved for the regression parameters under suitable regularity conditions, and an efficient estimation procedure is described. We further apply this framework to a mobile game network of 966 users, using the time of each user beginning to play the game, demographic data, and geographic data. We find that there is significant social contagion in playing game behavior on this data set.
|
