Abstract:
|
This paper discusses missing data issues for two-mode social network analysis based on influence model. There have been studies about the effects of missing data on the structural properties of social networks and different treatments of missing data. However, there is little literature on how missing data would affect estimated coefficients of influence model in two-mode network analysis. Two mode refers to actors and events here. Influence model is a very useful statistical model in social science. The basic idea is to estimate actors' behavior at time two by actors' exposure to other people's (who they interacted with) behavior or information or norm presented on events they attended from time one to time two, given actors' behavior at time one in the model. The paper will be organized in the following way. In section 1, a formal influence model will be specified. In section 2, different situations of missing data will be described. In section 3, different missing mechanism will be discussed. In section 4, possible imputation methods will be explored. How different imputation methods could affect the estimated coefficients of influence model will be explored as well.
|