Experiment FindingsPDF Available

A Social Media Network Analysis of Gavin Williamson MP, Secretary of State for Education, and the Characteristics of how he used Twitter during COVID-19 between 24.07.2019 to 18.03.2021.

Authors:

Abstract and Figures

This paper evaluates one of 22 Social Media Network Analysis (SMNA) reports completed between March and May 2021. The research uses a micro-sociological perspective to look closer at the phenomenology of an individual’s use of Twitter. Using NodeXL, I have conducted a social media network analysis using Twitter to investigate the topology network determined from Gavin Williamson MP and his use of Twitter. The aim was to understand the characteristics of his social media activity and what lessons can be learned from an education politician facing a national crisis (during COVID-19). Main research question: 1) What are the characteristics of Gavin Williamson's Twitter activity? Sub-research questions: 2) Why did Gavin Williamson use Twitter less frequently during an educational crisis, and what could be the potential reasons for this? 3) What does Gavin Williamson choose to retweet, and when? 4) What sentiment analysis can be determined from Gavin Williamson’s use of Twitter?
Content may be subject to copyright.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 1 of 49
A Social Media Network Analysis of Gavin Williamson MP,
Secretary of State for Education, and the Characteristics of how he used
Twitter during COVID-19 between 24.07.19 to 18.03.21
by Ross M. McGill
Knowledge, Power and Politics
Faculty of Education,
Wolfson College, Cambridge University
EdD Assignment 8, Year 4
Submission date: November 2021
10,298 words / 34 minutes reading time
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 2 of 49
Abstract
In a collection and analysis of 587 tweets exported and completed on 18th March 2021, this study
aims to identify how Gavin Williamson MP, the former Secretary of State (SoS) for Education in
England, used Twitter to communicate before, during and after the COVID-19 pandemic, during
what can only be described as an educational crisis. At the time of writing, Williamson was a
member of the Conservative political party and was appointed to the cabinet office by Prime
Minister Boris Johnson on 24th July 2019. He was sacked from his position on 15th September 2021
after 840 days in the post (and during this research study). There were several demands for
Williamson to resign; from fellow politicians, teachers and the general public, all observed in the
traditional media and the public sphere (Fuchs, 2014)1across Twitter.
This paper evaluates one of 22 Social Media Network Analysis (SMNA) reports completed between
March and May 2021. The research uses a micro-sociological perspective to look closer at the
phenomenology of an individual’s use of Twitter. “There remains a considerable ignorance of the
nature and procedures of phenomenology, and those who think of it as 'unscientific' or even
'anti-scientific’” Natanson, 1966)2;zooming in on one individual’s activity, to evaluate Twitter data
imported into NodeXL (Network Overview Discovery and Exploration for Excel, version 1.0.1.444),
which provides the ability to collect, analyse and visualise complex social networks from Twitter.
The aim is to find any correlations between any key education stories at the time and the data.
This type of analysis provides the education sector with an opportunity to learn how people are
connected, who they are, what they are saying, how they fit into the online discourse and from
this, what patterns of discourse-data can be understood. This research also allows us to begin to
understand how education politicians communicate online and share policy announcements.
Keywords:
COVID-19, education, Gavin Williamson, network analysis, schools, social media, Twitter, politics,
phenomenology
2Natanson, 1966, The phenomenology of Alfred Schutz
1Fuchs, C.. 2014, Social Media: A Critical Introduction.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 3 of 49
1. Introduction
Between March to May 2021, 22 static Social Media Network Analysis (SMNA) projects were
completed on education topics shared on Twitter’s social media platform. The objective was to
sample several recurring topics over a specific period to understand who, what, and how key
education issues were shared, how the topic discourse evolved, and identify any trends or
patterns in user content at particular periods. Topics and educational issues at this time included:
Keeping schools open or closed, wearing masks in school and whether or not young people
should get vaccinated.
As the initial data emerged, the initial perspective narrowed towards tweets published by Gavin
Williamson MP, Secretary of State for Education. The rationale for selecting one individual was
because often, all of the other trending topics were a by-product of many of his policy
announcements. In essence, the data was much more significant to observe, and the content
appeared to continue to polarise politicians, teachers and families with pupils in school. As such,
because Williamson was the key source to managing the pandemic, it proved to be an obvious
decision to focus on his Twitter activity during the COVID-19 crisis. In summary, to evaluate how he
communicated on Twitter, using a “public sphere of political communication that has
emancipatory political potentials" (Fuchs, 2014)3, particularly throughout the COVID-19 pandemic;
these insights may provide further understanding regarding government policy announcements
and how key decisions are spread across the public sphere. “We call events and occasions
‘public’ when they are open to all, in contrast, to close or exclusive repairs” (Habermas, 1989c, 1).4
Willamson’s data was exported from Twitter on Thursday, 18 March 2021 at 10:48 (GMT) and this
study evaluates his 587 status updates from the point Gavin Williamson was appointed as
Secretary of State for Education on 24th July 2019 up until the date for analysis. For clarity, Twitter
status updates posted during his previous tenure as Defence Secretary of State have been
removed.
4Habermas, Jürgen. 1989. The structural transformation of the public sphere. Cambridge, MA: MIT Press
3Fuchs, C.. 2014, Social Media: A Critical Introduction.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 4 of 49
The paper is organised as follows. In sections 2 to 5, the reader will be presented with the main
research question in section two, a literature review in section three with definitions in section four
to help understand the terminology used in the social network analysis. In section five, the chosen
methodology is shared.
In section six, there is a link to the dataset that has been exported and evaluated. In sections 6.2
to 6.6, there is a focus on five key datasets to analyse the social network and understand the
propagation process to investigate the underlying topology and user activity. These five areas
have been chosen based on the nature of this type of SMNA, particularly related to one vertice
(user) with no analysis of any edges (replies from other accounts). There is scope to evaluate this
information in future projects.
In the final sections 7-12, concluding remarks on the data are presented, the findings and
responses to the main research questions are identified as outlined in section 2 below. This case
study aims to understand why educational topics trend on Twitter, the conceptualisation of
‘influence’ and what factors play out for teachers in England who interact with these networks.
The use of social media now offers “multiple parties inside and outside politics (e.g. teachers) the
opportunity to start bottom-up initiatives and to use their online acquired social capital to exert
real influence on policy processes” (Rehm et al., 2019)5. In future research, evaluating topics on
Twitter shared by the general public and how this may influence education politicians or the
traditional media, would be a potential next step.
5Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 5 of 49
2. Main research question
The critical question for this study is:
1. What are the characteristics of Gavin Williamson's Twitter activity?
Sub-research questions
2. Why did Gavin Williamson use Twitter less frequently during an educational crisis, and what
could be the potential reasons for this?
3. What does Gavin Williamson choose to retweet, and when?
4. What sentiment analysis can be determined from Gavin Williamson’s use of Twitter?
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 6 of 49
3. Literature Review
To understand how information is shared and the exchange of ideas on Twitter, this study explores
the cognitive dimension, “whether participating actors (i.e. teachers interested in education
politics) share a common understanding and terminology”(Rehm et al., 2019) with education
politicians. This case study provides the opportunity to evaluate this research question: What are
the characteristics of Gavin Williamson's Twitter activity (in his role as Secretary of State for
Education)?
A hypothesis is that Twitter is a tool that takes the level of participation (of parents, teachers,
school leaders, academics and the general public) in policy-making from ‘manipulation’ towards
‘consultation’, then onto ‘partnership’ and ‘delegated power’ (Arnstein, 19696). Using NodeXL, a
social media network tool, provides some understanding of these social connections and levels of
participation to evaluate the information from a relational perspective, describing “issues such as
motivations and common values among individuals” (Rehm et al., 2019)7.
In this study, we are provided within a static social graph. The key difference to this approach
compared to a ‘dynamic network', is that the user cannot interact with the dataset in an active
environment (see Power Bi, section 5) to manipulate the data. Therefore, a static dataset does not
offer any relational data with other vertices (users), however, other relational information can be
determined. For example, the time of day, a sentiment analysis of the language and hashtags
used, as outlined in section 6. A second research objective is to determine why Gavin Williamson
MP used Twitter less frequently during an educational crisis. This analysis will help to understand
how an education politician communicates on Twitter, comparing status updates with trending
topics evolving on Twitter, and to map any underlying network.
7Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
6Arnstein, S., A Ladder of Citizen Participation, (1969)
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 7 of 49
Morales et al., (2019)8research, ‘Efficiency of human activity on information spreading on Twitter’
discovered “that most of the information posted on Twitter is hardly propagated through the
network”; 71% of the messages do not travel any farther than the authors timeline (Cheng and
Evans, 2009)9. Note, Twitter technology has advanced significantly since this citation. Given that
there are more users, and more experienced users, it would be interesting to compare how Twitter
activity now performs a decade later. In essence, are an individual’s tweets seen beyond the
individual’s follower network.
In order to understand how people use Twitter, I am keen to learn how an individual can influence
a Twitter trending topic and if this is a common occurrence; learning if this pattern includes any
hallmarks, predetermined by traditional media outlets or any social media algorithms. “How are
people connected on Twitter? Who are the most influential people? What do people talk about?”
are some questions posed by Kwak et al (2010)10 in one of the largest and first quantitative studies
on the entire Twitter platform of its time, evaluating if Twitter is a social network or a news media
channel.
It is also worth considering if trending topics, derived by figures of public interest are accidental,
influenced by the opinions of others, good timing, or as a result of an individual’s public or hidden
network. For example, private direct messages on Twitter or on other platforms such as
coordinated WhatsApp announcements.
A final objective is to attempt to answer some of the sub-research questions posed:
1. Why did Gavin Williamson use Twitter less frequently during an educational crisis, and what
could be the potential reasons for this?
2. What does Gavin Williamson choose to retweet and when?
3. What sentiment analysis can be determined from Gavin Williamson’s use of Twitter?
10 Kwak et al, 2010, Twitter is not a social network but a news media
9Cheng, A., and Evans M., (2009), An in-depth look inside the Twitter world, Available at
https://sysomos.com/inside-twitter/ [Accessed 26.11.2020]
8Morales et al, (2019), Efficiency of human activity on information spreading on Twitter’
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 8 of 49
This data analysis will help to evaluate individual messages and potentially any collective
responses, to learn “the ratio between the emergent spreading process and the activity
employed by the user” (Morales et al., 2019)11.
There is “a lack of empirical research” (Rehm et al, 201912) from Twitter to policy, particularly in
relation to social media discussions and how the network topology influences (or not) education
policy across England. I aim to understand if teachers play their part - in a ladder of citizenship
participation - and this focus will generate future study. For this research, how public figures of
interest use Twitter, how they may shape trending topics and if they influence traditional forms of
media or “whether these types of networks and communicative exchanges are able to exert real
influence on (educational) policy processes.”
Critically, does Gavin Williamson create the trending discourse, with traditional forms of media
gathering around him, for example, press releases to journalists or television interviews, or does the
Twitter platform itself, to keep users engaged with polarised discourse, create the conditions for
topics or users to ‘trend’ that are separate to the traditional communications network. Future
research may also consider if Twitter algorithms determine how the network is influenced. Many
will be familiar with public figures having their Twitter updates quoted in the press or on television.
For example, “Gavin Williamson 'likes' tweet saying he's rubbish at his job”13.In this emerging digital
epoch, it could be argued that Twitter is an established form for communicating news.
Using network analysis methods, the study is to learn how we can “deal with the large amounts of
text data that are being produced within social networking sites” (Rehm et al., 2019)14 and
quantitatively assess keywords, frequency of tweets, plus time, date and key influencers. Emerging
work which is similar in the field and of interest is this research on Ofsted, the English school
14 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
13 The National (The Jouker, 2020)
https://www.thenational.scot/news/18659181.gavin-williamson-likes-tweet-slamming-performance-education-secretary/ [Accessed
19.11.2021]
12 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
11 Morales et al, (2019), Efficiency of human activity on information spreading on Twitter’
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 9 of 49
inspection watchdog: Demonstrating the potential of text mining for analyzing school inspection
reports: a sentiment analysis of 17,000 Ofsted documents (Bokhove and Sims, 2020)15. The authors
“set out to demonstrate the potential of analyzing very large sets of inspection reports using text
mining methods” to investigate any relationship between the language used (sentiment analysis)
of each report, the language used under different education frameworks, and the overall Ofsted
inspection grade awarded. This type of analysis, including how discourse analysis plays out on
Twitter, provides new opportunities for education research at a macro level “presenting the
correspondence between the space… [captured] in a visual and synoptic way” (Bourdieu,
1996)16 to understand the theory and transfer of power and how it mirrors itself from the real world
to online conversation and relationships.
In this study in particular, evaluating large amounts of text data, I have analysed the makeup of
Twitter user, @GavinWillaimson17 (MP) from a micro-sociological perspective. As the evaluation
content consists of one vertice (user), at present there is no discourse (or polarisation) to analyse
between multiple Twitter users. Notwithstanding, it would be interesting to learn in future case
studies how heterogeneous Twitter networks are, in this case, the education politicians in England,
and if this bears any influence on communication with the general public active on Twitter. This
“heterogeneous behavior” (Morales et al, 2019) of the network typically creates some disparity in
how messages are sent and received and consequently, how the information is disseminated
across the network.
To understand the tensions between politicians and in “teachers’ beliefs about children and
young people and their abilities and capabilities” (Bietsa et al, 2015)18, I have observed how social
media has provided teachers with the ability to reach key figures of authority beyond their own
classrooms, seeking solutions from others who may hinder, stifle or block teacher agency. This
emerging use of Twitter, allows individuals to accumulate social capital which systematically
18 Biesta G., Priestley M., Robinson S: The role of beliefs in teacher agency Teachers and Teaching 2015 vol: 21 (6) pp: 624-640
17 Gavin Willamson MP; Twitter account, https://twitter.com/GavinWilliamson
16 Bourdieu, P., (1996), Physical space, social space and habitus, Vinhelm Aubert Memorial Lecture, Report
15 Christian Bokhove and Sam Sims (2021) Demonstrating the potential of text mining for analyzing school inspection reports: a sentiment
analysis of 17,000 Ofsted documents
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 10 of 49
“privileges the forms of cultural and social capital typical of the middle classes” (Skeggs, 200419); it
is these strategies used and any user-motivations which are a focus from an educational context
within England.
We are still learning how to filter literal truth from our social media use. “It is becoming increasingly
difficult for governments to steer the information and design of (educational) policy processes in
traditional ways” (Rehm et al, 2019)20 as more people turn to social media to express their views
and reach a wider audience. As a result, governments have also had to become more active on
social media to communicate information. From a brief perspective, it is interesting to see how
former Secretary of State for Education politicians have used Twitter during their tenure. Morales et
al. (2014) found that “emergent patterns are remarkably influenced by the underlying network
topology. Conversations I have personally had with education politicians were immediate, albeit
brief in nature, during the formative years of Twitter. As we have each learned how to use the
platform, how relationships evolve and how the online network can be used for social profit, or
against us in terms of viral news, outrage or bullying, users have become more savvy with the
technology, its benefits and pitfalls.
Whilst more literature reviews are needed in my evolving research, this study provides a much
needed perspective (as an experienced Twitter user who is) observing Twitter trends across the UK,
particularly within English education. In particular, I hope to understand what teachers are
discussing on Twitter, what topics are important to them, including any individuals’ active on the
Twitter network who have a position of influence on others; what content is shared and how it
spreads throughout “the network [cognitive] dimension” (Rehm et al., 2019)21. To investigate the
dialogue on Twitter allows us to understand which key influencers shape others. “Previous studies
have reported complex properties in [sic] network (Kwak et al., 201022), like degree distribution
with power-law behaviour, the small mean distance between nodes and modular structure”
22Kwak et al, 2010, Twitter is not a social network but a news media
21 As above
20 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
19 Skeggs B (2004) Class, Self, Culture. London: Routledge.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 11 of 49
(Morales et al., 2014)23. For example, does the number of follower connections make any
difference to content influence on others?
This evaluation will seek to determine how social capital is formed using NodeXL using a static
social media network analysis. I am particularly interested in the number of teachers involved in
future discussion as they “have largely been neglected from the analysis of policy processes and
social capital formation within SNS” (Rehm et al., 2019)24.
Using NodeXL as a social network analysis, one can begin to understand the meso-sociological
(sub-group) perspective of users, the structure and culture of users, how they are organised into
heiracrchy “to rationalise life from fundamentally different basic points of view and in very
different directions” (du Gay, 200825) and how they are connected in online relationships. In the
studies I have conducted to date, it can be observed that the Top Influencers (see 6.1) rarely
share content in comparison to other users - the sub-groups found in a wider network analysis. We
can learn that not all “participants must employ the same amount of effort, to accomplish the
same level of retweets”(Morales et al., 2014) which suggests that individual users have to work
harder to get their messages spread by others in their network.
In this case study, no additional network is evaluated. In future projects, there is enormous scope
to learn if teacher-agency (as an underlying network) is a factor in the public discourse, how
content spreads and perhaps, how Gavin Williamson or any other politician changes the way they
use social networks such as Twitter.
25 Du Gay, P., 2008, Max Weber and the Moral Economy of Office
24 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
23 Morales, A., J., Borondo, J., Losada, J., C, and Benito, R., M., (2014), Efficiency of human activity on information spreading on Twitter.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 12 of 49
4. Definitions
For this research to be accessible to those outside of the field of SMNA, I have provided several
definitions for clarification (table 1) based on this static analysis of Gavin Williamson.
Table 1:
Vertices: 1
The people in the network
Unique Edges : 0
Connections between (vertices) people
Edges With Duplicates: 587
The total number of multiple connections between
two (vertices) people
Total Edges: 587
The number of connections
Number of Edge Types: 2
The number of edge types
Tweet: 568
The number of tweets in the network
Retweet: 19
The number of retweets
Self-Loops: 587
An edge that starts and ends at the same vertex (is a
self-loop)
Reciprocated Vertex Pair Ratio: Not Applicable
When two vertices link to each other
Reciprocated Edge Ratio: Not Applicable
When an edge from A to B is joined from another
edge from B to A
Connected Components: 1
A group of vertices (people) that are all connected
Single-Vertex Connected Components: 1
A vertex that has zero connections and is isolated
Maximum Vertices in a Connected Component:
1
A connected component composed of a number of
vertices (people)
Maximum Edges in a Connected Component:
587
A connected component composed of a number of
edges (connections)
Maximum Geodesic Distance (Diameter) : 0
The distance between two vertices
Average Geodesic Distance : 0
The average distance between two vertices
Graph Density: Not Applicable
The number of edges among a group of vertices
over the total possible number. A high graph density
represents many connections to many others.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 13 of 49
Modularity: 0.000851
The measure of fitness of the groups that are created
within the cluster. Low = well-defined. High = low
quality.
To help understand a social network and its topology, in the graphical representation (Figure 1),
according to Hanneman and Riddle (2015)26 the graph is directed, which is a set of nodes
connected by edges and vertices (the number of people or things in the network).
Figure 1: A static description of the network topology
As defined in table 1, unique edges can be defined as a “connection between two vertices”
(NodeXL, 2020)27 and for every Twitter message posted there is an ‘edge’. An edge within Twitter is
defined as a "replies-to" relationship in a tweet, where one user replies to another. An edge for
each "mentions" relationship in a tweet that contains another person’s username anywhere in the
body of the tweet (Twitter, 2020)28.
28 Twitter, (2020), About replies and mentions, Available at https://help.twitter.com/en/using-twitter/mentions-and-replies
27NodeXL, (2020), About Us, Available athttps://www.smrfoundation.org/networks/overall-metrics-defined/ [Accessed 26.11.2020]
26 Hanneman and Riddle (2015), Introduction to Social Network Methods
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 14 of 49
In this study, there is just one vertex (person) with zero unique edges - the number of ‘unique’
connections where multiple connections between A and B are counted only once. This is due to
this particular SMNA research focusing closely on the activity of one Twitter user.
The number of edges with duplicates totalled 587. These are the connections between two
vertices and only count if there are multiple connections between two edges. This data analysis
was accumulated from Gavin Williamson’s 587 tweets - a relatively small number of updates
posted on the social network compared to someone much more active on Twitter.
Aself-loop edge for each tweet that is not a "replies-to" or "mentions" and is observed when a user
posts a tweet on their timeline and “starts and ends on the same vertex (person)” (NodeXL,
2020)29. This is when a user shares a status update and does not tag any other user ID. Sometimes
this can include receiving no replies from other users. In this study, there were 587 self-loops which
is an edge (tweet) that starts and ends in the same vertex (person). In reality, there were
thousands of replies to Gavin Williamson’s tweets, but I have exported only his status updates to
be able to answer the research question.
In total, there were 19 retweets that could offer further analysis. A sub-research question is: What
does Gavin Williamson choose to retweet and when? (See section 2) Replies to Willamson’s
original status updates have not been evaluated, but they would prove to be an interesting future
case study to observe how the social network responds to Twitter accounts that hold public
attention.
The total edges are the total number of connections where multiple connections between A and
B are all counted. If this network analysis factored in multiple accounts, then the number would
exceed 587 reported in this study.
29 NodeXL, (2020), About Us, Available athttps://www.smrfoundation.org/networks/overall-metrics-defined/ [Accessed 26.11.2020]
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 15 of 49
Finally, the total number of edge types in this evaluated network is 2. These two are edge types
(tweets) in this network. Willamson’s personal tweets and the retweets of other Twitter users
Willamson has chosen to share. It is also worth noting that a hashtag is a metatag and is “used to
index keywords or topics'' (Twitter, 2020)30 with a # symbol as the initial character (or Unicode) to
represent text. This makes specific words organised and searchable by all users even if two
vertices (people) are not directly connected (followers).
30 Twitter, (2020), How to use hashtags, Available at https://help.twitter.com/en/using-twitter/how-to-use-hashtags
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 16 of 49
5. Methodology
In this research investigation, the “structural dimension is concerned with the social interactions
between individuals within a particular setting” (Rehm et al., 2019)31. In this study, I have used
NodeXL (version 1.0.1.444) which is a cloud-based application that operates as a social network
analysis tool to investigate Gavin Williamson’s tweets on the microblogging platform Twitter.
The visual graph's vertices is an element of the network (Figure 1), which is the number of people
grouped using the Clauset-Newman-Moore cluster, “a hierarchical agglomeration algorithm for
detecting community structure which is faster than many competing algorithms” (Cornell
University, 2004)32. The graph source was made from a Twitter search and was laid out by NodeXL
using the Harel-Koren Fast Multiscale layout algorithm which provides “a clustered network of
Twitter users using colour to differentiate among the vertices” (Rodrigues et al, 2011)33. This
visualisation provides a method for identifying for example, groups, text data and frequency. This
clustered network approach provides modularity as “a measure of the fitness of a group:
(NodeXL, 2020)34
Once the initial data was collected and clustered using NodeXL, the information gathered,
particularly when evaluating the information of one user, can be further analysed using Power BI
(see figure 2); a data analytics tool provided by Microsoft Corporation. The software provides
interactive visualisations and allows a user to manipulate information to report on specific data
exported from the network. This software uses a range of cloud-based apps and services that
allow users to manage, manipulate and evaluate data from a variety of sources. In essence, the
software pulls together all of the data, tidies it up and presents it in a compelling way. This tool
provides the opportunity to explore the information in greater depth and can be presented
alongside the NodeXL data.
34 NodeXL, (2020), About Us, Available at https://www.smrfoundation.org/networks/overall-metrics-defined/
33 Rodrigues, M., E., Milic-Frayling, N., Smith, M., Shneiderman, B., and Hansen, D.,. (2011), Group-in-a-Box Layout for Multi-faceted
Analysis of Communities.
32 Clauset, A., Newman, E., J., Moore, C., 2004, Finding community structure in very large networks
31 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 17 of 49
In this case study, the data exported from NodeXL was imported into Power BI which provides
further interactive analysis and can be accessed from the two hyperlinks:
1. NodeXL: https://www.nodexlgraphgallery.org/Pages/Graph.aspx?graphID=250844
2. Power BI:
https://app.powerbi.com/links/CnZZ8D0sfr?ctid=29d4c11c-057c-487f-8fda-e86d59939e64&
pbi_source=linkShare
Figure 2: Gavin Williamson MP, Twitter, Power Bi, 18 March 2021, 14:14 GMT
The methodology identified in this study used a grounded theory approach to provide a social
media network analysis of a static network to identify any patterns. NodeXL and PowerBI
supported the document analysis used in this research to synthesise the methodical gathering of
status updates on Twitter. The techniques used for data analysis include coding, counting and
descriptive statistics to summarise the collection of information.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 18 of 49
Ignoring any relational data, the objective was to analyse the structural dimensions to determine
of Williamson’s network; what sentiment content was shared and when to understand what
impact Williamson’s social activity had on influencing current affairs during the peak of the
COVID-19.
Future research can involve how other people responded to Williamson and the relational
dimensions, particularly the cognitive participation and how the general public responded to
messages across the Twitter network. This could include any past or future Secretary of State for
Education and their use of Twitter to learn how policy is communicated, shared and managed,
and compare how sentiment data translates into policy paperwork or traditional news media.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 19 of 49
6. Data Collection
NodeXL Pro software uses Microsoft Excel to enter a network edge list (see definitions, section 4)
into a workbook to see a visual graph and gather a detailed summary of the online environment.
The software enables researchers to customise the network diagram, calculate basic metrics and
filter various vertices and edges. To put this data collection into context, NodeXL analyses the
data and provides a comprehensive breakdown of social connections and conversation. The
challenge is to understand this unstructured data and help researchers take “informed and
organized collective action” (Höchtl et al., 2016)35 to make better choices. NodeXL typically
provides a snapshot of the following information below. I have highlighted in bold the fields of
entry I have been able to evaluate for this case study as a result of analysing one Twitter vertice
(user):
1. Top Influencers
2. Top URLs,
3. Top Domains,
4. Top Hashtags,
5. Top Words,
6. Top Word Pairs,
7. Top Replied To
8. Top Mentioned and,
9. Top Tweeters.
Power BI data looks carefully at the following areas and provides the research with data across
two different SMNA platforms for analysis:
1. Overview
2. Network
3. Time
4. Time Grid
35 Höchtl, J., Parycek, P., and Schöllhammer, R., (2016), Big data in the policy cycle: Policy decision making in the digital era
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 20 of 49
5. Location
6. Influencers
7. Hashtags
8. Top Tweets
9. Words
10. Sentiment
11. Media
12. Web Links
13. Compare 2
14. Compare 4
There is scope to compare the data between NodeXL and Power BI network platforms, I have
analysed this information in greater depth throughout the following section. Those topics
compared include:
10. Top Influencers (Figure 3)
11. Top Hashtags, (Figures 8 and 9)
12. Top Words, (Figures 10 - 12)
13. Top Word Pairs (Sentiment Analysis, Figure 13)
14. Web Links (URLs)
6.1. Top Influencers
The graph (see Figure 3) represents a network of Gavin Williamson MP; a single Twitter user whose
tweets as Secretary of State for Education from 24th July 2019 to the day of export, 18th March
2021, and have been exported and evaluated. Evaluating one vertex explains why the visual
representation is limited in appearance.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 21 of 49
Figure 3: Gavin Williamson MP, Twitter, NodeXL, SNA Map, 18 March 2021, 10:48 GMT
This network was obtained from Twitter on Thursday, 18 March 2021 at 10:48 and can be found
online at: https://nodexlgraphgallery.org/Pages/Graph.aspx?graphID=250844. The tweets in the
network were gathered over 603 days from 24th July 2019 to 18 March 2021 and can be
manipulated further in this online worksheet:
https://docs.google.com/spreadsheets/d/1pJZoWoLn1PEGUsuCJQT6LyrA0CeofBB2/edit?usp=shari
ng&ouid=112138020938601997015&rtpof=true&sd=true
Additional tweets were also collected from Gavin Williamson’s Twitter account, prior to his
appointment as Secretary of State for Education. This Twitter activity has been removed and not
evaluated in this case study as these status updates primarily form part of his role as Defence
Secretary.
The dataset is divided into key groups for analysis, beginning with a summary of the entire graph
by hashtag (Figure 2). In this data collection, because only one Twitter user is evaluated, there is
no evaluation made of the differences between the data collected in the entire network and any
sub-group of vertices.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 22 of 49
It is also worth noting, although there is no evaluation of the vertices surrounding Gavin Williamson
MP, typically listed as top influencers. As defined by NodeXL, “Users [are] ranked by a network
score calculated based on the person's connection to otherwise disconnected groups of people”
are provided. This would provide us with the ability to observe which Twitter users are influential to
the network and instrumental in the social sphere of public discourse surrounding Williamson’s
status updates.
When comparing the NodeXL data (Figure 8) to the first image we gain from Power BI (Figure 3),
with Power Bi one can observe the frequency of hashtags used by the scale of the text. With
NodeXL, hashtag influence is represented by frequency in descending order. A distinction
between both reports is that Power BI presents the original NodeXL graphic and also offers a
one-page summary of the vital information explored in this data collection. For example, the
number of users, number of tweets, connections and top hashtags used are some of the key
information presented. Both platforms offer a helpful representation of the data.
Figure 3: Top Influencers, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 10:48 GMT
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 23 of 49
A SMNA term ‘Betweenness Centrality’ is a ‘Bridge Score’ that measures how much a person is the
only way to connect from one part of the network to another. ‘Betweenness’ is a sociological
proxy for ‘influence’ and “is different from other social media reputation scores in that it is local
and bounded and finds people other people find valuable within a particular topic during a
particular time.” The evaluation of Betweeness Centrality helps to identify “which factors, like the
individual behaviour or the underlying substratum, determine the user’s efficiency to have their
messages spread through the network” (Morales et al, 2014). Note, this measure does not use
follower count, the number of tweets or retweets as a measure of influence.
In the paper, Gavin Williamson MP is the only vertice; therefore this data is not available. However,
future research could evaluate how the social sphere responds to a particular person (vertice) or
tweet.
6.2. Top URLs
The top URL (Uniform Resource Locator) is informally called a website address. It specifies a
particular location on a computer network and is mostly referenced in a web page HTTP.
(Hypertext Transfer Protocol) for transmitting data between web browsers and web servers. In
essence, it is a landing page to collect information and resources.
In this social network study, one particular web page promoting ‘new support for young people’
in care featured 7 times in the entire graph (see Figure 4). When compared to other types of
network analysis, the number is low, yet in this analysis features more than any other URLs for two
reasons. The first is Gavin Williamson MP is the only vertice (person) in this network analysis. Two, it
suggests that www.gov.uk/government/news/vital-new-support-for-young-people-leaving-care
URL Williamson has shared was the most popular, as defined by ‘likes’ and 'retweets’ from those in
the network.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 24 of 49
The URL is a government press release published on the government website (gov.uk) and
discusses ‘support for young people leaving care’ with housing, healthcare and employment
programmes totalling £19M.
Figure 4: Top URL shared Gavin Williamson MP, Twitter, NodeXL, 18 March 2021, 10:48
Future network analysis could evaluate how often this link was shared and evaluate when and
how it was received across the network. For example, what happens in the social sphere when
other edges (connections) are made between other vertices (people), when other vertices
respond, in this case, when Gavin Williamson MP shares a message on Twitter containing a URL
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 25 of 49
address and if and how the network responds and interacts with a single message, another
vertice (person) or the URL, including how far the content spreads online.
I have provided Figure 5 below which is an overview of the top URLs used in this network. The URLs
are a mixture of official government web pages, news articles from the media and retweets and
quoted tweets from other Twitter users. This collection of top URLs has been gathered using
NodeXL. To evaluate this data would require further investigation to determine if Gavin Williamson
MP shared the hyperlinks directly, retweeted or quoted tweets using other vertice’s (people’s)
messages.
Figure 5: Top URL shared Gavin Williamson MP, Twitter, NodeXL, 18 March 2021, 10:48
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 26 of 49
When comparing the above data from NodeXL with the same information on Power BI (Figure 6),
this particular information lacks any detail worthy of analysis. It is potentially my intermediate use
of the software at this stage that explains why no data is represented. It is notable that each
platform offers unique perspectives at various stages, depending on what is being explored.
Figure 6: Top URLs, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 10:48 GMT
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 27 of 49
6.3. Top Domains
Top domains in this particular social network (Figure 7) are representative of the social platform
used to share the URLs and/or Twitter status messages. Focusing on the top domains used in the
entire graph only, the most popular messages shared were communicated directly within Twitter
itself. For example, 160 references to twitter.com.
Future analysis could consider if all of the top
domains shared in Twitter were status updates - a
text message and nothing else - or an update
shared on Twitter with an image, a quote tweet
using someone else’s status update and/or
hyperlink. The second top domain in this network
analysis were 83 URLs signposting the network to
the gov.uk official website as highlighted in
section 6.2., Top URLs.
Figure 7: Top Domains shared Gavin Williamson MP, Twitter, NodeXL, 18 March 2021, 10:48 GMT
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 28 of 49
6.4. Top Hashtags
NodeXL defines Hashtags that are “ranked by frequency of mention reported along with the raw
count of mentions (NodeXL, 2020).” Top Word
Pairs ranks are “pairs of words that appear next
to one another with the greatest frequency”
and will offer some interesting analysis in this
network analysis of Gavin Williamson.
From the entire graph and at the point of data
collection, the top hashtags mentioned (Figure
8), and ranked by frequency of mention
throughout the entire dataset, are as follows:
Figure 8: Top Hashtags shared Gavin Williamson MP, Twitter, NodeXL, 18 March 2021, 10:48 GMT
1. #VoteConservative, is mentioned 34 times in Williamson’s tweets.
2. #GetBrexitDone, 21 times.
3. #BackToSchool, 14 times
4. #Education, 13 times
5. #PlanForJobs, 10 times
6. #StayHomeSaveLives, 10 times
7. #NorthWestDurham, 10 times
8. #LoveOurColleges, 9 times
9. #BackBoris, 9 times
10. #Apprenticeships, 9 times.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 29 of 49
The first observation is that “specific keywords [or hashtags] relate to topics of conversation that
have captured a significant part of the collective attention” (Morales et al., 2014). Several
research studies and questions posed in this study ask if these findings are representative and
shared heterogeneously among the network. For example, was ‘getting Brexit done’ more
popular with the public or those responding to Gavin Williamson’s tweets than returning ‘back to
school’ during a pandemic?
Similar hashtags are cited across the entire data collection because Gavin Williamson MP is the
only user. If multiple accounts were analysed, each group would provide a range of URLs,
hashtags, and word insights unique to each group. Analysing groups enable research in network
analysis to learn how discourse is spread and evolves, particularly how content influences other
users or words used by key influencers shape others and their opinions.
Figure 9: Top Hashtags, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 14:14 GMT
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 30 of 49
The top hashtag used in figure 8 is political rather than related to Williamson’s role in education.
When a hashtag is used, the keyword becomes a metadata tag on microblogging platforms such
as Twitter. Accessing hashtags means that other Twitter users can reach and observe keyword
information outside Gavin Willaimson’s network. In essence, these words become
cross-referenced content—for example, Twitter users who do not follow his Twitter account and
connect with it directly.
In future work, it would be interesting to investigate which hashtags influence traditional forms of
media. For example, television broadcasts and front page tabloids. Secondly, the top hashtag is
the most dominant in all groups, with some lower-ranking groups and users dominating the
sub-group conversations. Morales et al. (2014)36 found that there was “no correlation between the
number of followers and activity employed, which [meant] that the amount of messages posted is
independent of the user position in the followers network.” Twitter users are “motivated to partake
in the discussion by their goal to encourage other Twitter users to participate in the online
discussion” (Rehm et al., 2019)37 and perhaps influence the discussion and the trending topic.
A key observation is how hashtags, “based on human activity taking place around specific topics
of conversation on Twitter” (Morales et al, 2014) are used in conversations to support content
reaching others through the use of a metatag. Christian Fuchs writes that social media is a
reflection of the middle-class bourgeois; that “Twitter constitutes a new public sphere of political
communication that has emancipatory political potentials" (Fuchs, 2014)38. It is worth noting that
this social capital is popular content and is also context-specific within the community itself.
38 Fuchs, C.. 2014, Social Media: A Critical Introduction.
37 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
36 Morales et al, (2019), Efficiency of human activity on information spreading on Twitter.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 31 of 49
6.5. Top Words
Top words in the entire Gavin Williamson network are listed in figure 10. It is reassuring to note that
‘educational’ language is dominant in the network; this is only notable after removing ‘Defence
Secretary’ data prior to his appointment as SoS.
Figure 10: Top Words shared Gavin Williamson MP, Twitter, NodeXL, 18 March 2021, 10:48 GMT
For deeper analysis in this network, future studies could evaluate what keywords were used and
when. Given that this network unpicks the Twitter data of Gavin Williamson, insights could be
obtained with timestamps to correlate what public messaging was used at particular periods to
communicate education policy announcements. For example, the global pandemic, keeping
schools safe and getting pupils back to school safely.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 32 of 49
6.6. Top Word Pairs
Top words are a “ranked list of pairs of words that appear next to one another with the greatest
frequency” (NodeXL, 2020)39. Interestingly, when the metatag is not used, it is interesting to
observe the number of times it is used and perhaps the missed opportunity for these keywords to
reach further afield as a searchable hashtag. An example of this would be two examples (Figure
11) in the Top Words used in the entire dataset. Those include ‘schools’ listed 126 times and
‘mental health’ (x10), both with no (#) metatag, shared 136 times. Given the pandemic crisis and
the impact on young people attending school, Williamson’s tweets may have reached a wider
audience if hashtags were carefully considered in his communication to the general public.
Figure 11: ‘Schools’ ‘Mental Health’, Twitter, NodeXL, Top Words, 20 November 2020 at 17:07 UTC
A future study would be an exciting opportunity to analyse and investigate what keywords are
mentioned and how this may or may not influence other users and sub-group conversations.
Taking a closer look at Power BI, a helpful graphic (Figure 12) is provided with a breakdown of
words offered in the bottom-right hand corner. The words include education (featured x109
times), schools (101), children (95), people (86), work (71). I am keen to learn more about why
some ‘top words’ report a slightly different frequency in both platforms and improve my SNMA
expertise for future analysis.
39 NodeXL, (2020), About Us, Available at https://www.smrfoundation.org/networks/overall-metrics-defined/
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 33 of 49
Figure 12: Top Words, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 14:14 GMT
Power BI’s sentiment analysis (Figure 13) provides the opportunity to evaluate words mentioned by
Gavin Williamson MP in his Twitter status updates. Several adjectives feature in the ‘positive word
by tweet’ analysis. For example, “great (x42), fantastic (33), amazing (15) and brilliant (14).”
Equally, in the “negative word by tweet” breakdown, adjectives which regularly feature are “hard
(x25), difficult (9) and challenging (7).”
Future evaluation would help understand how education policy is worded and communicated in
the public sphere.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 34 of 49
Figure 13: Sentiment, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 14:14 GMT
Other essential data worth noting is the time of day when Williamson decided to post Twitter
updates throughout the COVID-19 period. A breakdown month by month is included below (see
Figure 14):
March 2020 = 34 tweets
April 2020 = 22
May 2020 = 8
June 2020 = 11
July 2020 = 11
August 2020 exams = 7
September 2020 = 5
October 2020 = 29
November 2020 = 23
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 35 of 49
December 2020 = 19
January 2021 = 43
February 2021 = 42
March 2021 = 26.
Figure 14: Tweet Quantity, Gavin Williamson MP, Twitter, Node XL, 18 March 2021, 10:48 GMT
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 36 of 49
Returning to one of the three sub-research questions, Why did Gavin Williamson use Twitter less
frequently during an educational crisis, and what could be the potential reasons for this? In total,
throughout 2020, Williamson posted 215 tweets and 111 tweets in the first, three months in 2021.
What is striking, is that given the academic year commences in late August and early September,
schools, teachers and parents would have been seeking some clarity about returning to school
safely; this was a period in which Williamson rarely posted (Figure 15).
It is also worth observing that the frequency of status updates at the beginning of 2021 doubled
compared to the previous six months.
Figure 15: Frequency, Gavin Williamson MP, Twitter, Node XL, 18 March 2021, 10:48 GMT
The network tools provide further analysis. Using Power BI, it is possible to break down Williamson’s
tweets by the time of day (Figure 16) to help understand when status updates are likely to be
posted. Comparing this data with sentiment analysis would highlight when education policy
information is most likely to be announced. Power Bi offers this data under the ‘Time’ and ‘Time
Grid’ (which is very useful) software features.
A closer analysis of Williamson’s time grid (Figure 17) reveals that most Twitter activity occurs on a
Wednesday. From this data export, 124 tweets in total when compared to 42 tweets on a Sunday.
The most popular hour is at 16:00 hours, with Wednesday at 18:00 being the most popular period.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 37 of 49
Figure 16: Time Grid, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 14:14 GMT
Figure 17: Time Grid, Gavin Williamson MP, Twitter, Power BI, 18 March 2021, 14:14 GMT
If the general public hopes to respond (or receive a reply) to Williamson’s tweets or appear on the
social network supporting or challenging status updates, the time grid data provides an
exceptional overview of where best one should focus their efforts.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 38 of 49
Information propagates using a combination of several factors: 1) Popularity, 2) Frequency, 3)
Novelty, and 4) “the resonance of the message content” (Romero et al., 2011)40. A Twitter user
with few followers may find they post content that resonates with many people, but the other
factors either present or absent at a particular time may or may not influence the social network.
7. Identification of major relationships or patterns.
“The most retransmitted users tend to be the most followed ones as well” (Morales et al, 2014). In
this case study, politicians carry a significant responsibility for the public and generally gather a
large social media audience. In this case, Gavin Wiliamson is just shy of 100,000 Twitter followers.
People in positions of power and influence, whether active on social media or not, frequently
become key discussion items on social media channels. One future project could unpick how the
general public responds to people of public interest and how this unfolds within the social media
network before it is referred to in the news media.
7.1. Structural, cognitive and relational dimensions
When considering the social interactions between individuals, patterns emerge “from human
interactions, and show it to be universal across several Twitter conversations.” Morales et al (2014)41
in their study concluded: “Some influential users efficiently cause remarkable collective reactions
by each message sent, while the majority of users must employ extremely large efforts to reach
similar effects.”
When topics are discussed in the online education community in England, I often receive private
messages with requests for me to amplify other people’s messages. Politicians are no different.
People often use communication strategy to mobilise their followers (or perhaps voting
constituents) “to inform them about specific information” and “understand a wide variety of
phenomena.” Morales et al (2014) suggests that the “novelty of the posted information decays
41 Morales, A., J., Borondo, J., Losada, J., C, and Benito, R., M., (2014), Efficiency of human activity on information spreading on Twitter.
40Romero et al, (2011) Influence and Passivity in Social Media
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 39 of 49
quite rapidly” once it has been published. Prior to reading this research, I had started to realise this
for myself, reducing the number of times I would share others’ content. It appears that as your
audience grows, one can become more reluctant to share information.
In this study, the contextualisation of the social network can “provide important insights into how
people behave” (Rehm et al, 2019)42. To understand what makes content viral, a homogeneous
online “society would allow users to gain an extremely high amount of retweets, only by means of
employing an enormous amount of initial activity as well” which appears to also be supported by
“available connections on the underlying network” (Morales et al, 2014).
In terms of any structural dimensions of a network, the selection to share one another’s content
‘could be’ largely derived by the relationship between each user. There are people who do
influence other Twitter users. The user’s network “represents the way that the collective attention is
organised” (Morales et al, 2014) and is manifested in retweets or quote tweets to support or
disdain, depending on the homogeneous community. By retweeting a message, “users deliver
specific information to their own followers, at the same time that endorse ideas and gain visibility
in the network” (Boyd et al., 2010)43. Social outrage or free speech, depending on the context of
information and one’s worldview are becoming important topical discussion points at the time of
writing. To speak online is to publish, and publish online is to connect with others. With the arrival of
globally accessible publishing [blogs and Twitter], “freedom of speech is now freedom of the
press”, and freedom of the press is freedom of assembly (Shirky, 2011)44.
When we use Twitter accounts alongside any social categories, such as a politician acting as a
social hub (Rehm et al, 2019) or a social authority (for example, a journalist), these accounts are
more likely to influence the social network more often than the general public - who are less likely
to influence the network. Whether this is due to their position in the public eye, the news in general
and how they influence content being shared to the general public, or whether they possess a
44 Shirky, C. 2011. The political power of social media. Foreign Affairs 90 (1): 28–4
43Boyd, D., M. and Marwick, A., E., (2010) Tweet Honestly, I Tweet Passionately: Twitter Users, Context Collapse, and the Imagined
Audience
42 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 40 of 49
degree of expertise others do not have, remains to be evaluated and tested in this type of SMNA.
Twitter is a platform to gather information and “collect human-to-human interactions” (Morales et
al., 2014).
Teachers using Twitter regularly recommend the platform for professional development and space
to source ideas quickly and spread information far beyond their physical workspace. Morales et
al. (2014) concluded that most “people post messages that do not travel at all”, with many
teachers still believing it is a tool worth investing in. Whether they believe they can influence
education politics via Twitter will be considered in future projects. Social organisers (within the
social hub 3) believe “that teachers are insufficiently heard among politicians”. The
phenomenological experiences of using Twitter have attracted “political and financial attention”
(McGill, 2020)45, yet some may believe that Twitter is not a solution to derive influence. This lack of
perception is one possibility why there may be some ignorance or unscientific belief that
micro-celebrities can influence education politics “to increase visibility and have an impact on
how education is shaped and changed” (Rehm et al., 2019)46 by simply posting on Twitter.
8. Identification of strengths and weaknesses.
What is interesting is that “in order for some users to gain attention from the collective, others must
lose it at the same time” (Morales et al., 2014). In this case, having key messages amplified instead
of working hard to have our own ideas seen or heard. Perhaps, this mirrors our relationships at work
and in the real world? In the study, ‘Power to the People?! Twitter Discussions on (Educational)
Policy Processes’ offers a glimpse of hope for the general public who do wish to influence policy:
“Social media offers multiple parties inside and outside politics (e.g. teachers and other
educational professionals) the opportunity to start bottom-up initiatives and acquire social
capital” (Rehm et al, 2019)47.
47 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
46 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
45 McGill, R., M., (2020) Fateful Moments As A Micro-Celebrity, An Autoethnography
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 41 of 49
What Williamson shares on Twitter, in all examples, is most likely to drive more engagement than
the vast majority of other users. The ability to investigate how these social networks are formed
and the vast amount of content being shared rapidly, enables researchers to gain an
understanding of the topology, how it organises itself organically or perhaps how algorithms may
collude to shape our perceptions. More importantly in future research, how individuals may
influence English education policy.
In contradiction, albeit without a focus on education, ‘Efficiency of human activity on information
spreading on Twitter, Morales et al (2014) found “that some influential users acted as information
producers, providing messages that are received by the passive large majority of information
consumers” (Morales et al, 2014). We may think our latest post, petition or video may go viral, but
in reality, much of the information shared disappears on the World Wide Web. However, there is
some scope for influencers with organic content or paid-for advertising to disrupt the trend.
The behavioural patterns of Gavin Willaimson’s status updates on Twitter suggest that in the
context of COVID19 and the challenges, teachers, parents and pupils were facing, policy
decisions are largely “determined in a top-down fashion” (Rehm et al, 2019)48. Discussions
increasingly taking place on Twitter allow teachers to closely follow developments and perhaps
influence the discourse about education policy. Examples include when Williamson threatened to
sue Greenwich council in South London when leaders “advised schools to shut early amid a rise in
coronavirus cases.”49 In terms of social capital, the general public (including teachers) used Twitter
and tried to influence how the government responded to this. Kevin Courtney, joint general
secretary of the National Education Union (NEU), said “We strongly welcome the decision by
Greenwich council to urge all of its schools to close from Monday evening, to all except
vulnerable children and the children of key workers. We urge other councils to take the same
49 Evening Standard
https://www.standard.co.uk/news/education/greenwich-schools-legal-challenge-stay-open-government-b310672.html [accessed 05
November 2021]
48 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 42 of 49
decision.”50 Whether Courtney’s response and its Twitter popularity is a result of his position as a
union leader or as a result of a timely response on Twitter are yet to be evaluated.
“Educational scientists have increasingly acknowledged that the concept of social capital can
contribute to our understanding of how informal networks develop and evolve over time”
(Moolenaar, Sleegers, and Daly, 2012b; Risser, 2013)51. Several years later, we are beginning to
have a better understanding of our social media use. This knowledge may help us to understand
how, in particular, how education policy is curated, published and evaluated in an epoch of
social media where Twitter users can also observe at a distance without publicly interacting.
Rehm et al. (2019)52 conclude that “We have not yet checked on whether the shared information
and content on Twitter might have had an effective influence on the actual policy document.”
With my interests in the field of social networks across English education and Ofsted inspection, the
school watchdog for education standards across England, I am curious to learn if “knowledge is
closely tied to tacit, embodied knowledge of individuals” (Cornelissen, et al. 2015) and if this can
be evaluated on Twitter. Taking a macro-sociological perspective of others should enhance my
understanding of key influencers and how online content shapes English education policy.
52 Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?! Twitter Discussions on (Educational)
Policy Processes.
51 Moolenaar, N., M., Sleegers, P., J., C., and Daly, A., J., 2012b,Teaming Up: Linking Collaboration Networks, Collective Efficacy, and
Student Achievement
50 Kevin Courtney, Twitter 14.12.2020 https://twitter.com/cyclingkev/status/1338396126437904385 [Accessed 22.11.2021]
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 43 of 49
9. Limitations of the study
Prior to using NodeXL, using Hoaxy53, a tool designed by Indiana University “visualises the spread of
articles online” and searches content related to fact-checking from spambot Twitter accounts.
This required me to first export the dataset from Twitter. “Hoaxy visualizes two aspects of the
spread of claims and fact-checking: temporal trends and diffusion networks” (McGill, 2020)54. I
have also attempted to complete my own social media network analysis by using R55, “a free
software environment for statistical computing and graphics” (R, 2020). I decided to discard this
approach until I learn how to use it more efficiently. NodeXL offers a quick export and overview of
the data which is helpful, but takes away some of the required analysis to understand the in-depth
social network behaviours I seek to investigate in this type of case study.
This study could be further developed to scrutinise how Twitter conversations develop teacher
agency in terms of how they voice their opinions online during trending hashtags, and if or when
they have a direct influence or other online users, or traditional forms of media. Understanding
how public figures post updates online and how the general public respond in the social sphere
provide insights for the English education sector which are largely new and unknown.
This era of social interaction provides the opportunity for teachers’ tweets to be cited alongside
figures of public interest, or quoted in educational newspapers and displayed on television news
bulletins across England as they gain traction within the network. Future research should explore
how teachers have a say in these social conversations with education politicians and if they lead
to a degree of virality, influence and change in policy.
55 R Project, https://www.r-project.org/
54 McGill, R., M., 2020, Who is the Mayor?
53 Hoaxy, https://hoaxy.osome.iu.edu//faq.php
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 44 of 49
10. Identification of any conflicting evidence.
It is important to consider a range of Twitter social media network analysis reports to determine the
validity of content and, in particular, if the general public do influence Twitter trends and
government policy and media. This is a relatively new field of social science research with few
studies zooming in on education politicians and how they influence the social sphere to influence
the general public and the communications network
An initial search on Google Scholar using "social media network" analysis "education" "England" as
a search term resulted in only 1,900 sources56 at the time of writing. Refining this search to include
the terms “teachers”, “education policy” and “influence” yields only 55 results57. When the search
entry contains "social media network" analysis "education" "England" "teachers" “policy”
"influence" "Twitter" and "Ofsted" is added, zero results are reported.
This is the area where I hope to contribute new knowledge58.
58 Google Scholar search 3
https://scholar.google.co.uk/scholar?hl=en&as_sdt=0%2C5&q=%22social+media+network%22+analysis+%22education%
22+%22England%22+%22teachers%22+%E2%80%9Cpolicy%E2%80%9D+%22influence%22+%22Twitter%22+-students+%22O
fsted%22+&btnG=
57 Google Scholar search 2
https://scholar.google.co.uk/scholar?hl=en&as_sdt=0%2C5&as_vis=1&q=%22social+media+network%22+analysis+%22ed
ucation%22+%22England%22+%22teachers%22+%22education+policy%22+%22influence%22&btnG= [Accessed
21.11.2021]
56 Google Scholar search 1
https://scholar.google.co.uk/scholar?as_vis=1&q=%22social+media+network%22+analysis+%22education%22+%22Engla
nd%22&hl=en&as_sdt=0,5 [Accessed 22.11.2021]
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 45 of 49
11. Conclusions
In this case study, I have explored relatively new research in the field of social media network
analysis and English education policy; I have conducted a Twitter search to investigate the
topology network determined from Gavin Williamson’s use of Twitter to understand the
characteristics of his social media activity and what lessons can be learned from a politician
facing a national crisis.
The tool NodeXL offered a quantitative analysis of the structure and relationships of Gavin
Williamson MP on Twitter during a period of time when COVID19 implications impacted the closure
(or opening) of all schools across England throughout 2020 and early into 2021. As reported in the
literature review, the activity of the typical Twitter network is fuelled by a small group of users, who
in some cases display very little social activity, but derive a large proportion of the network
conversation in others. In essence, they have social capital with other users gaining “influence in
proportion to the activity they employ” (Morales et al, 2014) which can contribute to a degree of
social transubstantiation, from the real to the online world.
This single static social media analysis still highlights key issues when compared to similar discussions
shown in the other 21 network analysis trials, many that mirror much of the discourse across the
physical landscape in English schools. Observing this data on Twitter highlights how some Twitter
users gain an influential profile with the formation of social capital or from their position of
influence over the general public; there is little or no evidence (at this stage) if any individual with
social capital can impact government decisions at policy level.
One final point worth making.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 46 of 49
At the time of writing (22.11.2021), from the point at which Gavin Williamson MP was removed from
his position as Secretary of State for Education, apart from tweeting his ‘thanks’59 on September
15th 2021 at 13:42 PM (GMT), from this point onwards he has only tweeted four times across 67
days, averaging one Twitter update every 16 days.
This raises several reflections:
1. During the height of the pandemic (March/April 2020), Gavin Williamson tweeted fewer
times when compared to the first three months of 2021. This suggests that he preferred to
keep a quiet profile during a time of crisis.
2. Since being removed from office, Gavin Williamson has been almost silent on Twitter,
tweeting just four times in 67 days. This suggests that he is taking time away from social
media to restock.
3. Being in a position of power not only brings accountability, workload and stress to those in a
specific role, but does social media also generate those additional pressures on an
individual to be connected 24/7 and regularly post updates.
4. More importantly, why did Williamson stop communicating during a time of need.
12. Taking it further
My doctoral research aims are to learn how to use NodeXL, R and Hoaxy to export and analyse
Twitter discourse. In this single case study, how the Secretary of State for Education used Twitter
during a global crisis, evaluating the characteristics and speculating why Twitter activity reduce
during a time of need. Future static social media network analysis in this field will enable the
education community to learn how information is spread on Twitter, who the key influencers are,
what factors influence virality and if this content influences traditional forms of media or
government policy at a later stage. There is then scope to conduct structured interviews with key
influencers and education politicians to learn how perceptions correlate to any sentiment
analysis.
59 Gavin Wiliamson, Twitter https://twitter.com/GavinWilliamson/status/1438120943323320334
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 47 of 49
13. References
1. Arnstein, S., (1969), A Ladder of Citizen Participation,
2. Biesta G., Priestley M., Robinson S: (2015 vol: 21 (6) pp: 624-640), The role of beliefs in teacher
agency Teachers and Teaching
3. Bokhove, C., and Sims, S., (2021), Demonstrating the potential of text mining for analyzing
school inspection reports: a sentiment analysis of 17,000 Ofsted documents, International
Journal of Research and Method in Education
4. Bourdieu, P., (1996), Physical space, social space and habitus, Vinhelm Aubert Memorial
Lecture, Report
5. Bourdieu, P., (1995), The Rules of Art
6. Boyd, D., M. and Marwick, A., E., (2010), Tweet Honestly, I Tweet Passionately: Twitter Users,
Context Collapse, and the Imagined Audience
7. Cheng, A., and Evans M., (2009), An in-depth look inside the Twitter world, Available at
https://sysomos.com/inside-twitter/ [Accessed 26.11.2020]
8. Clauset, A., Newman, E., J., Moore, C., (2004), Finding community structure in very large
networks
9. Cornelissen, F. et al. (2015), ‘Teacher Education’s Challenge of Changing Research
Relationships With Schools
10. Courtney, K., Twitter 14.12.2020 https://twitter.com/cyclingkev/status/1338396126437904385
[Accessed 22.11.2021]
11. Du Gay, P., (2008), Max Weber and the Moral Economy of Office
12. Evening Standard (2020), Greenwich schools face legal challenge from Gavin Williamson to
keep schools open,
https://www.standard.co.uk/news/education/greenwich-schools-legal-challenge-stay-open-g
overnment-b310672.html [Accessed 05.11.2021]
13. Fuchs, C.. (2014), Social Media: A Critical Introduction.
14. Google Scholar search 1
https://scholar.google.co.uk/scholar?as_vis=1&q=%22social+media+network%22+analysis+%22
education%22+%22England%22&hl=en&as_sdt=0,5 [Accessed 22.11.2021]
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 48 of 49
15. Google Scholar search 2
https://scholar.google.co.uk/scholar?hl=en&as_sdt=0%2C5&as_vis=1&q=%22social+media+net
work%22+analysis+%22education%22+%22England%22+%22teachers%22+%22education+polic
y%22+%22influence%22&btnG= [Accessed 21.11.2021]
16. Google Scholar search 3
https://scholar.google.co.uk/scholar?hl=en&as_sdt=0%2C5&q=%22social+media+network%22
+analysis+%22education%22+%22England%22+%22teachers%22+%E2%80%9Cpolicy%E2%80%9
D+%22influence%22+%22Twitter%22+-students+%22Ofsted%22+&btnG=
17. Hanneman, R., and Riddle, M., (2015) Introduction to social methods, Available at
https://faculty.ucr.edu/~hanneman/nettext/index.html [Accessed 23.11.2020]
18. Höchtl, J., Parycek, P., and Schöllhammer, R., (2016), Big data in the policy cycle: Policy
decision making in the digital era
19. Kwak et al, (2010), Twitter is not a social network but a news media
20. McGill, R., M., (2020), Fateful Moments As A Micro-Celebrity, An Autoethnography
21. McGill, R., M., (2020), Who is the Mayor?
22. Moolenaar, N., M., Sleegers, P., J., C., and Daly, A., J., 2012b,Teaming Up: Linking
Collaboration Networks, Collective Efficacy, and Student Achievement
23. Moore, C., (2004), Finding community structure in very large networks, Available at
https://arxiv.org/abs/cond-mat/0408187 [Accessed 26.11.2020]
24. Morales et al, (2019), Efficiency of human activity on information spreading on Twitter.
25. Morales, A., J., Borondo, J., Losada, J., C, and Benito, R., M., (2014), Efficiency of human
activity on information spreading on Twitter.
26. Natanson, M., (1966), The phenomenon of Alfred Schutz
27. NodeXL, (2020), About Us, Available at
https://www.smrfoundation.org/networks/overall-metrics-defined/ [Accessed 26.11.2020]
28. Rehm, M., Cornelissen, F., Notten, A., Daly, A., and Supovitz, J., (2019), Power to the People?!
Twitter Discussions on (Educational) Policy Processes.
29. Rodrigues, M., E., Milic-Frayling, N., Smith, M., Shneiderman, B., and Hansen, D., (2011),
Group-in-a-Box Layout for Multi-faceted Analysis of Communities.
Supervisor: Steve Watson.
Ross M. McGill - Social Network Analysis: Politics, Education and Social Media Page 49 of 49
30. Romero et al, (2011), Influence and Passivity in Social Media
31. Shirky, C. (2011). The political power of social media. Foreign Affairs 90 (1): 28–4
32. Skeggs B (2004) Class, Self, Culture. London: Routledge.
33. Twitter, (2020), About replies and mentions, Available at
https://help.twitter.com/en/using-twitter/mentions-and-replies [Accessed 23.11.2020]
34. Twitter, (2020), How to use hashtags, Available at
https://help.twitter.com/en/using-twitter/how-to-use-hashtags [Accessed 23.11.2020]
35. Williamson, G., Twitter, (2021),
https://twitter.com/GavinWilliamson/status/1438120943323320334 [Accessed 22.11.2021]
Supervisor: Steve Watson.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Social media technologies collapse multiple audiences into single contexts, making it difficult for people to use the same techniques online that they do to handle multiplicity in face-to-face conversation. This paper investigates how content producers navigate 'imagined audiences' on Twitter. We talked to participants with different types of followings to understand their techniques, including targeting multiple audiences, concealing subjects, and maintaining authenticity. For some, the techniques of audience management resemble those of 'micro-celebrity' and personal branding, representing strategic self-commodification. Our model of the networked audience assumes a many-to-many communication through which individuals conceptualize an imagined audience evoked through their tweets.
Article
Many national education systems incorporate a central inspectorate tasked with visiting, evaluating and reporting on the performance of schools. The judgements produced by inspectors often play a part in the way that schools are held to account and also constitute an important source of data in their own right. Hence, they are of great interest to researchers. However, the sheer quantity of inspection reports produced by national school inspectorates creates challenges for analysts. We demonstrate the use of text mining - automated processing and analysis of unstructured textual data - to analyse the complete corpus of school inspection reports released by the English national schools inspectorate since the turn of the century. More specifically, we report the results of a sentiment analysis, comparing the tone of inspection reports across the different grades awarded in each inspection and across periods defined by different Chief Inspectors. In doing so, we hope to demonstrate the potential of text mining for providing representative analysis of very large volumes of inspection reports, making them a useful complement to smaller-scale, manual analyses. Resources and references are provided for researchers looking to use text mining technique
Article
Article Summary Discussion of the political impact of social media has focused on the power of mass protests to topple governments. In fact, social media's real potential lies in supporting civil society and the public sphere — which will produce change over years and decades, not weeks or months. On January 17, 2001, during the impeachment trial of Philippine President Joseph Estrada, loyalists in the Philippine Congress voted to set aside key evidence against him. Less than two hours after the decision was announced, thousands of Filipinos, angry that their corrupt president might be let off the hook, converged on Epifanio de los Santos Avenue, a major crossroads in Manila. The protest was arranged, in part, by forwarded text messages reading, "Go 2 EDSA. Wear blk." The crowd quickly swelled, and in the next few days, over a million people arrived, choking traffic in downtown Manila. The public's ability to coordinate such a massive and rapid response — close to seven million text messages were sent that week — so alarmed the country's legislators that they reversed course and allowed the evidence to be presented. Estrada's fate was sealed; by January 20, he was gone. The event marked the first time that social media had helped force out a national leader. Estrada himself blamed "the text-messaging generation" for his downfall. Since the rise of the Internet in the early 1990s, the world's networked population has grown from the low millions to the low billions. Over the same period, social media have become a fact of life for civil society worldwide, involving many actors — regular citizens, activists, nongovernmental organizations, telecommunications firms, software providers, governments. This raises an obvious question for the U.S.
Book
introduction to social networks, interesting the centrality chapter.
Conference Paper
Twitter, a microblogging service less than three years old, com- mands more than 41 million users as of July 2009 and is growing fast. Twitter users tweet about any topic within the 140-character limit and follow others to receive their tweets. The goal of this paper is to study the topological characteristics of Twitter and its power as a new medium of information sharing. We have crawled the entire Twitter site and obtained 41:7 million user profiles, 1:47 billion social relations, 4; 262 trending topics, and 106 million tweets. In its follower-following topology analysis we have found a non-power-law follower distribution, a short effec- tive diameter, and low reciprocity, which all mark a deviation from known characteristics of human social networks (28). In order to identify influentials on Twitter, we have ranked users by the number of followers and by PageRank and found two rankings to be sim- ilar. Ranking by retweets differs from the previous two rankings, indicating a gap in influence inferred from the number of followers and that from the popularity of one's tweets. We have analyzed the tweets of top trending topics and reported on their temporal behav- ior and user participation. We have classified the trending topics based on the active period and the tweets and show that the ma- jority (over 85%) of topics are headline news or persistent news in nature. A closer look at retweets reveals that any retweeted tweet is to reach an average of 1; 000 users no matter what the number of followers is of the original tweet. Once retweeted, a tweet gets retweeted almost instantly on next hops, signifying fast diffusion of information after the 1st retweet. To the best of our knowledge this work is the first quantitative study on the entire Twittersphere and information diffusion on it.
  • G Biesta
  • M Priestley
  • S Robinson
Biesta G., Priestley M., Robinson S: (2015 vol: 21 (6) pp: 624-640), The role of beliefs in teacher agency Teachers and Teaching
Physical space, social space and habitus
  • P Bourdieu
Bourdieu, P., (1996), Physical space, social space and habitus, Vinhelm Aubert Memorial Lecture, Report
An in-depth look inside the Twitter world
  • A Cheng
  • M Evans
Cheng, A., and Evans M., (2009), An in-depth look inside the Twitter world, Available at https://sysomos.com/inside-twitter/ [Accessed 26.11.2020]