D01-02AM: Discourse Network Analysis

Discourse Network Analysis is a methodological toolbox for measuring and analysing policy debates and their development over time. The software Discourse Network Analyzer (DNA) allows researchers to manually code actors’ opinions about policies in text data. Similar to other qualitative content analysis tools, the user annotates statements of political actors about their preferred or rejected concepts, policy instruments, frames, or beliefs. Text sources can include newspaper articles, parliamentary testimony, press agencies, or social media. DNA then exports various kinds of network data based on the user’s coding.

These networks capture relationships between political actors based on their congruence or conflict around concepts. Aggregated networks help identify discourse coalitions, advocacy coalitions, brokers, opinion leaders, central actors and concepts, dimensionality of discourse, frames composed of different concepts, and the evolution of debates over time. Researchers can apply ideological scaling techniques to measure actors’ ideal points or model contributions to the debate using statistical methods.

The course explores connections between discourse network analysis and policy process theories, covering best practices for coding statements in text data. It introduces statistical and exploratory methods for network analysis and dedicates about half the time to software implementations, including DNA, R, and visone, and a practical example.

Contents:

Discourse Network Analysis is a methodological toolbox for measuring and analysing policy debates and their development over time. The software Discourse Network Analyzer (DNA) allows researchers to manually code actors’ opinions about policies in text data. In a way similar to other qualitative content analysis tools, the user annotates statements of political actors about their preferred or rejected concepts, policy instruments, frames, or beliefs. Useful text sources can be newspaper articles, parliamentary testimony, press agencies, or social media. DNA then allows the researcher to export various kinds of network data based on what the user coded.

The network data capture the relationships between political actors based on their congruence or conflict around concepts. As these relationships are aggregated into a network, the user can identify discourse coalitions or advocacy coalitions in these networks, identify brokers, opinion leaders, and central actors and concepts, examine the dimensionality of the discourse, find frames composed of different concepts through co-agreement by multiple actors, track the evolution of the policy debate over time (for example before policy change occurs), apply ideological scaling techniques to measure actors’ ideological ideal points relative to each other, or model the contributions by actors to the debate using statistical techniques.

The course explores the connections between discourse network analysis and a number of policy process theories. We will cover best practices for coding statements in text data. The course will introduce a variety of statistical and exploratory methods to analyse the resulting network data. In addition to the methodological and theoretical foundations, we will dedicate about half of the time to software implementations, including DNA, R, and visone.

Learning outcomes:

After attending the course, participants will be able to confidently code their own policy debates using the DNA software, possibly in teams, and will be able to analyze the resulting network data competently in visone and, if a participant has prior R skills, also using the rDNA package. Most importantly, participants will learn how to operationalize important aspects of policy process theories using discourse network analysis and apply the methodology to their own research questions.

Admission requirements:

Participants should be interested in thinking about social phenomena and policy making as complex systems. Basic experience with the statistical programming environment R would be a plus, but only a small part of the course will actually use R. If you are willing to listen without using R during these parts, R skills are not strictly required. Participants should ideally bring their own laptops.

Daily schedule:

Monday:

• Introduction: Networks and policy debates
• DNA methodology: How do we create those networks?
• Examples: German pension politics and EU software patents
• Software lab: Coding in DNA 3.0
• Homework: development of a codebook

Tuesday:

• Software lab: Analysis of discourse networks in visone
• Coding and validity
• Coalitions: Clusters, communities, and subgroups
• Homework discussion and new homework: Coding

Wednesday:

• Measuring polarisation in discourse networks
• Ideological scaling of discourse networks using item response theory
• Phase detection: Measuring change in discourse networks
• Examples: U.S. climate change and UK health policy debates
• Backbone identification: Reducing the set of concepts
• Software lab: data management using rDNA
• Homework discussion and new homework: coding revision and analysis in visone

Thursday:

• Homework discussion and group work on empirical analysis in visone
• Presentations on group work and/or students’ own projects (on a voluntary basis)
• Software lab: Analysis using rDNA
• Agent-based simulation of discourse networks

Friday:

• Statistical modelling of discourse networks, including relational event models
• Outlook on future developments

This description is subject to change at the discretion of the Instructor

2 Credits

For completion of all work before and during the course, as outlined by the Instructor, and 90% participation and attendance of the course.

2 Additional Credits

Course specific extra assignments. These can include submitting assignments before the course, daily assignments, and/or a final assignment to be completed after the course as decided by the Instructor.  This description is subject to change at the discretion of the Instructor.