This blogpost was written by the moderators of the guest lecture by Tina Ternes and Anastasia Glawion on Network Analysis in the Digital Social Reading course.
Authors: Selma Tajic and Alice Lemma
Network analysis is a powerful method that goes beyond traditional statistical analyses by providing a contextual understanding of relationships within a structure. Consider two individuals exchanging letters – while content analysis is one aspect, placing these letters in the context of a broader network reveals shared relations and the significance of intermediary connections. In the basic framework, a graph has nodes (vertices) and links (edges) representing relationships. Abstractions simplify real situations but need caution against oversimplification. Knowing edge types (undirected, directed, reciprocal, weighted) is crucial. Different graph types offer insights into network strength.
Moving to our main discussion, we explore different methods by three researchers – Daniel Allington, Anastasia Glawion, and Tina Ternes. A network can provide insights at studies with both a micro and macro scope. Allington focused on representing similarities between Amazon customer and professional reviews on the novel "The Inheritance of Loss." Ternes examined the similarities between reviews in the genres of horror/thriller, science fiction, mystery, and romance. Glawion concentrated on representing similarities between the practices of literature production in fanfiction across a variety of genres, stories, and their protagonists. These research methods, each undertaken by Allington, Ternes, and Glawion, offer diverse approaches to network analysis, enriching our understanding of this field by showing the many sides of network analysis and its applications.
Guiding us is the question: "How does using a social network graph help in analysing data?" This question guides us through understanding why and how social network graphs are useful in data analysis.
For helpful insights and a more in-depth understanding, consider watching Martin Grandjean’s informative videos that explain network analysis.
It is crucial to compare the factors that influenced each researcher’s study to understand how a network is constituted, and what makes it useful to visualise and analyse the collected data. As a network is an aid to visualise the relationships between concepts and/or entities, researchers are required to tailor the network based on the study’s objectives by (a) giving meaning to nodes and edges based on the specificities of the data, and (b) by using additional tools.
(a) Each researcher provided the meanings assigned to the nodes and edges of their network to provide insight into the relations within the data, and as Wasserman and Faust (1994, p. 20) argue:
‘A social network consists of a finite set or sets of actors and the relation or relations defined on them. The presence of relational information is a critical and defining feature of a social network.’
Table 1 provides each researcher’s given meaning to nodes, edges, and weight density. From what can be observed, we can conclude that nodes represent the data to be analysed and the edges represent the variables on which the analysis is based.
*"Thematic analysis is a type of content analysis that emphasises latent content, i.e. aspects of data requiring a notable degree of interpretation, over manifest content, i.e. aspects of data that can be identified more mechanically (for example, the word frequencies analysed in Pihlaja’s, 2016 a study of online responses to pornographic videos)." (Allington, 2016, p.260)
(b) The larger the scope of the study, the more data is collected, the more a network must be personalised. For example, Ternes used different sizes for the nodes to show the degree of centrality and different colours by genre for nodes, while Glawion used directed edges to show the influence of characters on one another and different colours by story for clusters. Even if there is an attempt to encompass most data in a network, not all variables can be represented in a single network, hence using additional charts for the analysis of specific clusters may be beneficial. Ternes used column charts to represent reviews’ absorption based on the categories of the Story World Absorption Scale, and Glawion used column charts to represent the rate of genres in fanfiction stories based on each cluster or number of users per cluster, and pie charts to represent the percentage of stories within each cluster. Interestingly, Allington used multiple networks with only white and grey nodes to find the similarities between two variables at a time, such as in Figure 2 that focuses on customer and professional reviews and Figure 3 that focuses on geographically located reviews in the UK and US (pp.270-271).
Overall, social network graphs serve as a powerful tool for researchers, providing a visual and analytical means to unravel complex relationships and patterns in their data. Here are key points to keep in mind when creating your network graph:
Clear Visualization of Relationships
Identification of Patterns and Trends
Node and Edge Attributes
Comparative Analysis
Cluster Analysis
Customization for Specific Objectives
Integration with Additional Tools
Allington, D. (2016). ‘Power to the reader’ or ‘degradation of literary taste’? Professional critics and Amazon customers as reviewers of The Inheritance of Loss. Language and Literature, 25(3), 254-278. https://doi.org/10.1177/0963947016652789
The Historical Network Research Community (Regisseur). (2021, Juli 12). Introduction to Social Network Analysis [1/5]: Main Concepts. https://www.youtube.com/watch?v=lnLW6ITFY3M
Wasserman S, Faust K. (1994). Social Network Analysis: Methods and Applications. Cambridge University Press. https://doi.org/10.1017/CBO9780511815478