Social Network Analysis

Optional for SRMS
Lecturer: Mark Tranmer, Mark Elliot and Nick Crossley (Sociology)

Timetable: Tuesday 2-4 Venue Williamson 3.59

Aims:
1. To introduce the concepts of social networks and the various kinds of relation that can occur between members of the network.

2. To explain how do describe social networks, including visualisation.

3. To show how statistical models can be used for social network analysis.

4. To demonstrate the use of software for describing and modelling social networks.

Objectives
On completion of this unit successful students will be able to:
• Understand the concept of a social network, and the various kind of relations that can
occur with it.
• Know how to describe and visualise the network using appropriate software and
summary measures.
• Know how to model a social network using appropriate software, and understand the
substantive reasons for doing so.
• To critically assess the use of social network analysis in the social sciences.
• Use Pajek and Pnet and organise the network data for use with each of these software
packages.
• Participate in a discussion about the strengths and weaknesses of a given piece of
research that involves social network analysis.
• Understand the main argument of methodological journal articles on social network
analysis.

Content

Part I: Concepts, description, visualisation. Social networks occur in many situations in the social sciences. In this unit, we begin with some illustrative examples, and consider the various relations that can occur in a social network such as directed relationships, undirected relationships, reciprocation, valued relations. We then consider ways to visualise a network, making use of the free software pajek. To complement the visualisations we consider summary statistics for networks such as density and degree. We then move on to other important ideas such as the centrality and between-ness of network members. Substantively these are extremely important concepts: e.g. to find out who are the key people in the network that facilitate information flow in an organisation. We mainly focus on one-mode networks, but we also consider affiliation networks. Finally we briefly touch on other topics, including collection of network data, longitudinal network analysis.

Part II: Statistical models for social networks. In some situations, the researcher might wish to model the network, to see if a particular configuration, such as triangulation, occurs more often than would be expected at random given the number of people in the network and the total number of relations observed. When this is the case, exponential random graph (p*) models, can be used to characterise the network in terms of configurations. We cover this on the unit, making use of the software pnet. Whilst the mathematics is slightly more complicated, p* models are strongly related to logistic regression and hence we advise participants to have covered logistic regression on a previous course unit.

Teaching Methods:
The course is taught through a series of eleven weekly lectures and practical classes (2 hours combined). Lectures introduce the concepts and methods with practical classes providing an opportunity for immediate hands on learning though computer based exercises.


Preliminary Reading:
Scott, J (2000) Social Network Analysis: A handbook. Sage
Knoke D and Yang S (2008) Social Network Analysis (2nd edition). Sage
De Nooy W, Mvrar A and Batagelj V (2005) Exploratory Social Network Analysis with Pajek.
Cambridge Uni. Press.
Robins, G., Pattison, P., Kalish, Y., & Lusher, D. (2007). An Introduction to Exponential
Random Graph (p*) Models for Social Networks.