Intro to SNA (Essex Summer School)

This course is offered at the Essex Summer School and will provide a practical, but comprehensive introduction to the analysis of social networks.

Why Social Network Analysis?

Social network analysis takes the view that social research should not solely focus on the individual unit of analysis, but rather emphasises that researchers should also incorporate the social relations (networks) that connect these individual units (actors). For example, we might be interested in friendship among schoolchildren, trust among employees, collaboration among NGOs, exchanges of resources among companies, or conflict among nations.

By measuring these social relations, social network analysis tries to get a precise picture of the social processes taking place between these individual units. Hence, social network analysis enables us to investigate how a person’s individual behaviour is influenced by the people surrounding that person (e.g. a person’s smoking behaviour might be influenced by the smoking behaviour of his/her friends), or how and why social relations emerge as a result of social processes (e.g. trust might be more likely to emerge between researchers of the same discipline, or between those who have collaborated before). We might also be interested in why some groups have a different network structure than other groups, or how a group’s network structure affect outcomes such as a group’s efficiency.

Course Overview

The course focuses on the description and visualisation of social network data using UCINET. We will concentrate on uncovering structural properties of the network (e.g. density, homophily, and clustering), as well as on how to identify important persons in a network (e.g. degree centrality, structural holes, …). We will also pay attention to the detection of subgroups and deal with basic hypothesis testing for social network analysis. Throughout the course some classic theories that focus on network processes (e.g. related to homophily, centrality measures, structural holes, Granovetter’s strength of weak ties and small worlds) will be discussed.

Course Details

1. Social Network Analysis: what, how, why?
• What is social network analysis? Why do we need social network analysis?
• How does a social network approach differ from “classic/standard” research?
• What is the difference between egocentric and complete networks?
• What type of social network data are there? How to collect social network data?
• How can we (best) visualize networks? What programs are available?

Exercises with UCINET:
• How to build/import a dataset
• Visualisation of social networks
• Collecting network data and questionnaire design

2. First analysis at the group level and at the individual level
• What programs are available to analyse social networks?
• What is social capital? What is social support?
• How cohesive is my network? What is network density?
• Who is most central in my network? What is degree centrality?
• When is a network centralized, and why is it important? How can we measure it?
• What is a core-periphery structure?

Exercises with UCINET:
• Calculate the density of a network
• Degree centrality
• Freeman’s centralization (and core-periphery)

3. Centrality measures: an overview
• What types of centrality measures are there? What is the difference between degree, closeness and betweenness centrality? What other measures of centrality are there?
• What is the geodesic distance of a network?
• When do we use which central measure (closeness, betweenness, …)? How are they different?
• How can we deal with valued/weighted network relations?
• How can we test effects of individual position?
• What is a permutation test? Why can’t we use classic statistical tests?

Exercises with UCINET:
• Different centrality measures: closeness betweenness, etc.
• Valued networks: centrality, density and centralization
• Permutation tests

4. Homophily, diversity and social contagion
• What is homophily? What does it mean for your network? How can we measure it?
• Why do friends tend to be similar to ourselves (e.g. smoking, music taste)? What is social contagion?
• Why do some people have more diverse ego-networks, and how does it impact their outcomes?
• When can I claim that my ego-network more resourceful?
• What is QAP regression?

Exercises with UCINET:
• Homophily (EI index)
• QAP regression
• Measures of diversity and Blau’s IQV
• Measures of resourcefulness

5. Structural holes, closure and brokerage roles
• What is Granovetter’s “Strength of Weak Ties” argument? Why is it important?
• What is a “small world” network? What is six degrees of separation?
• Is it better to be connected to different groups of others, or have one big group of closely interwoven contacts? What is Ron Burt’s view? And James Coleman’s view? How can we measure this?
• What are Simmelian ties, and why are they important according to David Krackhardt?
• What are Gould and Fernandez’ brokerage types?

Exercises with UCINET:
• Clustering coefficient
• Constraint index and other measures of openness/closure
• Gould & Fernandez brokerage roles

6. Dyad and triad census
• What is reciprocity? What is a dyad census?
• How multiplex is my network?
• What is a random network (distribution)? How can it help to test hypotheses?
• What types of triads are there? How can I interpret different triad configurations in practice?

Exercises with UCINET:
• Dyad census and reciprocity
• Analysis of multiple networks
• Generating random graphs
• Triad census

7. Subgroups and hierarchies
• How can I identify subgroups in my network? What types of subgroups are there? How many components does my network have? What is a clique? What is a k-plex?
• What are the properties of a hierarchical network? To what extent does my network correspond to a hierarchical network?

Exercises with UCINET:
• Subgroup analysis (components, k-cliques, k-clans, …)
• Dimensions of hierarchies

8. Equivalent positions, roles and blockmodeling
• When do two actors have the same (or a similar) position in a network?
• What is regular equivalence? What is structural equivalence? What does it mean to be structural, regular equivalent?
• What is blockmodeling?

Exercises with UCINET:
• Calculate structural and regular equivalence
• Identify roles through blockmodeling

9. Two-mode networks
• What is a two-mode (affiliation/bipartite) network? How is it different from a one-mode network?
• What properties of a two-mode network are interesting?
• How can we identify central persons in a two mode network?

Exercises with UCINET:
• Different ways of dealing with two-mode networks (i.e. transforming them) in order to use available procedures in UCINET.

10. Questions and specific topics of interests
• What more is there to learn? What other (statistical) methods are available?

• Have a short overview of some other social network programs and some more advanced statistical analysis.

Link to course: Essex Summer School.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: