EASP Preconference Workshop: Mapping attitudes as networks: a guided workshop applying the bipartite network approach to opinion-based groups and group-based opinions
30th June 2023, Krakow, Poland
Pre-Conference Workshop, EASP General Meeting 2023
In this workshop we will present a method for mapping survey (and other) data as a bipartite network, where people are linked by the attitudes they jointly hold, and attitudes become socially connected when they are jointly held by people. This structure simultaneously links people into groups and attitudes into clusters. Using a guided workshop format, we will introduce the basics of network analysis. Participants will learn how to analyze and visualize survey data as a bipartite network in R/Python using the recently launched SurveyGraph package (made possible by an ERC Proof of Concept Grant). We show how analysis of this bipartite network structure allows us to quantify multidimensional polarization, and identify group structure and polarization. We will end with a guided brainstorming process where participants can explore the strengths and weaknesses of the method applied to their own research questions and data.
Cost: There are no participation fees for this preconference. (light refreshments and lunch covered by ERC funding; venue kindly provided by the Centre for Social Cognitive Studies, Institute of Psychology, Jagiellonian University)
How to Apply: Send an email to firstname.lastname@example.org with the subject “Application for EASP workshop.” If there are more applicants than spaces, we will aim for diversity in gender, geography, career stage, and research interests, so please include a very brief bio with as many of these details as you are comfortable sharing.
Application Deadline: 19 March 2023. We will continue to process applications ad hoc after this deadline if places are available.
Further opportunities: If you are interested in workshops on this method in the future, email email@example.com and request to be added to our mailing list. If you are interested in this workshop, you might be interested in the opening for a postdoc in our group or know someone who would be. See : https://my.corehr.com/pls/ulrecruit/erq_jobspec_version_4.jobspec?p_id=057968; deadline 23rd March 2023 before 12 noon, Irish Standard Time.
Planned Schedule (subject to improvement):
09:00-09:30 Welcome; tea/coffee
09:45-10:30 Theoretical overview -- Attitudes as Networks: Opinion-Based Groups and Group-Based Opinions as Bipartite Networks
10:30-11:00 Tutorial: Basic network methods and bipartite graphs
11:00-11:15 Comfort Break
11:15-12:15 Tutorial: Using the SurveyGraph R package to convert a secondary dataset to a network
12:15-12:45 Tutorial: Visualizing SurveyGraph output as a network
14:15-15:00 Applying network methods to identify groups and describe them
15:00-15:15 Comfort Break
15.15-16.15 Brainstorming own applications
Background: why examine survey data as an attitude network?
Attitudinal survey data provides the basis for identifying connections between groups of individuals based on shared attitudes. By using individuals as nodes and their attitude similarity scores as edges, we can visualise this data as a network, and detect opinion-based groups. It has been demonstrated that shared opinions can become the basis of group identity (Bliuc et al., 2007; O’Reilly et al., 2022), and opinion-based groups can drive polarization in society.
The network includes a participant projection, which lets us identify clusters of people who share similar attitudes (e.g. see Figure 1).
Figure 1. Network of people connected by vaccine-related attitudes in Bangladesh (Wellcome Global Monitor). From MacCarron et al. 2020.
The method also allows us to produce an attitude projection to identify the clusters of attitudes which are shared by groups of people. Figure 2 illustrates the pattern of attitudes commonly shared by participants in the same survey that generated Figure 1.
Figure 2. Network of attitudes connected by people who share them (MacCarron et al. 2020)
This novel attitude network can also detect patterns of polarization among a sample, based on clusters of attitudes that tend to be held by particular groups. For example, Maher et al. (2020) used this method to detect growing polarization in relation to public health attitudes in the UK during the beginning of the COVID-19 pandemic (see Figure 3).
Figure 3. Participant projection (left) and attitude projection (right) of public health attitudes
(Maher et al. 2020)
At time one, the network method identified two groups connected by similar public health attitudes. As seen in Figure 1, these groups diverged over time due to growing differences in trust in science and health officials between the two groups.
This method can identify polarization and opinion-based groups based on combinations of attitudes, which may not have been evident using conventional mean-based methods.
In this workshop, you will learn how to convert a dataset into a network, how to produce both participant projections and attitude projections, and how to interpret the resulting visualisations. You are encouraged to think about how this method would be useful in your own research.