PRS: Practicing Research Skills

Computational Approaches

in Political and Social

Sciences

Week 3A: Data Analysis / Skills tutorial on network analysis. Mon 17 June

Diliara Valeeva and Eelke Heemskerk

Plan for Week 3

Mon 17 June: Advanced Gephi Tutorial

+ How to write 'Data Analysis'

Wed 19 June: Group meetups

Fri 21 June: Group meetups

=> Submitting 'Data Analysis' assignment

Plan for today

1. Reflections

2. Skills tutorial on network analysis - 2

3. How to write 'Data Analysis'

1. Reflections

2. Skills tutorial on

network analysis - 2

Image source: https://medium.com/the-xplane-collection/anatomy-of-a-social-network-de73f1224ac5

Data on

co-authorship (2015)

Canvas => Modules => Week 3 => 2015_coauthorship.gexf

2 main types of measures

Network-related

Node-related

1. Network Measures

Average Degree

Degree is the number of ties a node has to other nodes
High degree: nodes have a large number of ties with others in a network

Graph Density

Density indicates how densely nodes are connected in a network
= Number of actual ties / Number of potential ties
High density: everyone knows each other in a network

Network Diameter

Diameter shows how far apart are the most distant nodes from each other
High diameter: nodes are far away, it can take time and effort to reach each other

Betweenness Centrality

Betwenness show whether a node obtains "a bridge" positions between others
High betweenness: node has access to diverse information and resources, connects non-connected groups

Closeness Centrality

Closeness shows how close is the node to all others
High closeness: a node can reach others in a network very easily

Modularity

Shows the network communities
Nodes are in one community if they are densely connected with each other
High modularity: dense intra-community, sparse inter-community ties
Higher is better

Connected components

Nodes are in one component if they are connected only with each other
A large number of components: network is highly disconnected

2. Node Measures

Clustering coefficient

It is a measure of how complete the neighborhood of a node is
The friend of my friend is also my friend" effect
High clustering: everyone knows each other

Eigenvector Centrality

Shows how a node is connected with other influential nodes
High eigenvector: the node has a large number of ties and its neighbors also have a large number of ties

Where to find all the results?

Gephi returns reports
Network-related measures are also saved in 'Data Laboratory' tab.
Export results using "Export Table" and explore in Excel

3. Data Analysis

Data Analysis

1. Introduction

2. Description and interpretation of the results

3. Conclusions

* max 2000 words

Deadline: Friday 21 June, 19.00

1. Introduction

Research question(s) and hypothesis(es): A brief reminder

Reformulate if they have changed

Data and methods: Which dataset(s) and method(s) you used?

2. Results

What are your main findings?

How would you explain these findings? What do they mean? Did you obtain what you expect or not?

* Think about the research question

* Return to the literature review if needed

3. Conclusions

A summary of the main findings

Did you answer your research question or not? Did you confirm your hypothesis(es) or not?

What would you recommend for future studies? What can be explored more? What would you like to study more on this topic if you would have more time / better datasets?

Group meetups

Wednesday 19 June

Success offline : 9.00 - 9.20

Success online : 9.20 - 9.40

Past and Present: 9.40 - 10.00

Neighbors: 10.10 - 10.30

Key players : 10.30 - 10.50