Advanced Social Network Analysis

Recent decades have witnessed substantial progress in the methods of analysis and statistical modeling of social networks. This includes models for the static network structure as well as models for network and behavioral change. This course will introduce participants to these modern, advanced methods in a hands-on manner. The methods will be discussed using real research cases and data. While (sociological/psychological/economic) theories will be touched upon, the focus will be on modeling assumptions, proper interpretation, and other practical aspects of conducting statistical analysis.
The course will start with types and methods for testing statistical hypotheses about network structure (e.g., permutation tests, CUG tests) and then move on to Exponential-family Random Graph Models (ERGM). The ERGMs for static networks will be discussed in the context of sociocentric data and in the context of egocentrically-sampled data. Moving on to modeling dynamic networks, the course will discuss Temporal ERGMs (TERGM) and Stochastic Actor-Oriented Models (SAOM). The course will almost exclusively rely on R.
Every meeting will consist of two parts. The first part will include a presentation, demonstration, and discussions of concepts, models, and methods. The second part will be devoted to hands-on training in using R to apply the presented methods to real data. While the instructors will provide datasets for these exercises, participants are encouraged to bring their own data.

Course leader

Michal Bojanowski, Kozminski University, Poland.

Target group

You will find the course useful if:

- You are a social science researcher (PhD student, post-doc or senior) conducting or preparing to conduct network-analytic studies.
- You are a researcher intending to extend your arsenal of empirical research skills to statistical network analysis and related techniques.

Course aim

By the end of the course you will be able to:

- test statistical hypotheses about network structure
- conduct Quadratic Assignment Procedure
- fit Exponential-family Random Graph Models (ERGM) and their temporal variants (TERGM)
- fit ERGMs to egocentrically-sampled data
- fit Stochastic Actor-Oriented Models (SAOM).