Skip to content Search
Search our website:

Twitter General Election 2019

Posted: Wednesday, 18 December 2019 15:00

Twitter General Election 2019

Tweets and retweets from the main political parties leading up to the 2019 General Election -
An extended abstract

Martyn Harris and Mark Levene

Department of Computer Science and Information Systems
Birkbeck University of London

Social media platforms generate a vast amount of data on a daily basis, on a variety of topics and consequently represent a key source of information for anyone interested in a current snapshot of online presence in society. In recent years, there has been an increased amount of research into social media data across a wide variety of disciplines, including sociology, computer science, marketing, and political science.

Twitter is the most popular platform for academic research, as it provides limited access to data via a number of freely available Application Programming Interfaces (APIs). Studies in social media can be framed by drawing on a wide-variety of theories, constructs and conceptual frameworks from a wide-variety of disciplines. Twitter is one of the main social media platforms used to promote views and opinions of people and communities on a wide variety of topics. One particular strand on which there is a large amount of traffic within Twitter is politics, where views are expressed on news events that take place. Often these days, it is quite common that the news starts on Twitter rather than outside it. For these reasons Twitter data has become a good source of data for studying and evaluting attitudes towards political parties and for parties to express their views directly through this medium. Many studies based on Twitter data look to evaluate attitudes towards certain political parties through the mining of Twitter data produced by individuals.

We conducted a study using Twitter data collected from Party Political Candidates (PPC) in the lead up to the 12th December 2019 Elections. Our study focuses on how politicians use Twitter to promote their policies and opinions on particular topics. The data was collected through the Twitter API, which is a small sample selection of tweets from the total population of all tweets. This research concentrates on a quantative analysis of political party tweets and the retweets through a time series analysis during the month leading up to the General Election. This post summarises our preliminary findings.

We collected the Twitter account names for each of the Prospective Parlimentary Candidates (PPC) of the major parties, together with the official party Twitter account names (i.e. @Conservative, @LabourParty, and @LibDems), over the period from the 1st November to 6th December 2019. The proportion of PPC(s) with a Twitter account are summarised in Table 1, where we observed that there are more PPCs for the Liberal Democrats than any other party.

Table 1: The number of PPC Twitter accounts collected for each of the main parties, and the percentage of PPC who have a Twitter account * Source: mpsontwitter.co.uk
Party #Accounts % PPC on Twitter *
Conservative 266 71%
Labour 242 86%
Liberal Democrats 499 79%

Analysing tweets using time series analysis

We summarised the number of tweets sent out by the PPC(s) and the number of times each tweet was retweeted to produce a raw time series on an hourly basis for each party, and a moving average with a window set to a 7 day period. The raw time series shows the distribution of tweets and retweets over time, whilst the moving average provides a method for identifying the weekly trend for each party.

How are the parties tweeting in the run-up to the election?

Table 2: The average number of tweets and retweets sent per hour from PPC and Party twitter accounts for the main parties.
Conservative Labour Liberal Democrats
Time Tweets
(mean)
Retweets
(mean)
Tweets
(mean)
Retweets
(mean)
Tweets
(mean)
Retweets
(mean)
01:00 4.42 433.31 7.43 1423.16 21.97 701.17
02:00 2.77 395.20 3.24 2745.94 8.74 1061.98
03:00 3.00 437.56 2.28 3934.80 6.03 379.82
04:00 3.33 1327.11 2.22 2423.29 7.13 935.15
05:00 3.79 245.34 5.16 837.99 13.53 808.84
06:00 13.50 462.13 19.89 840.49 42.92 752.22
07:00 46.56 499.78 59.08 1089.60 120.31 645.51
08:00 73.19 348.78 104.92 947.44 188.67 515.71
09:00 72.31 356.38 102.50 825.42 186.36 389.88
10:00 61.17 268.06 92.11 969.80 161.31 430.72
11:00 57.06 270.41 88.31 873.35 157.53 291.74
12:00 67.64 295.86 94.00 754.39 160.25 340.19
13:00 65.58 258.31 89.56 680.41 155.86 342.81
14:00 58.44 292.77 76.86 776.29 141.83 409.58
15:00 58.50 306.97 83.54 809.78 137.42 300.14
16:00 67.44 272.17 103.80 583.26 153.97 380.21
17:00 69.25 278.99 97.94 606.99 172.54 375.16
18:00 68.33 247.46 89.60 630.33 173.94 342.60
19:00 67.54 270.96 103.29 612.67 167.03 424.70
20:00 92.74 263.78 128.37 783.86 194.74 450.36
21:00 66.89 352.81 116.40 881.29 201.49 438.13
22:00 57.89 320.47 115.83 876.38 197.71 506.86
23:00 37.06 401.17 75.94 942.49 139.86 625.56
00:00 13.28 571.64 26.53 1217.44 63.03 856.67

We first analysed the raw time series and the weekly trends for the number of tweets sent out by each party during the period. The average number of tweets and retweets generated by hour of the day for each party are presented in Table 2. The plots in Figure 1, show that the Conservatives tweet less compared to the other parties. The Liberal Democrats are shown to be tweeting in much higher volumes than any of the other parties, which is partially due to their larger number of PPCs, although the volume of tweets per day are fairly consistent across all parties. The raw time series also shows a noticeable spike across the parties, occurring on the 19th November. This was due to the large number of tweets and retweets on the topic of the Leader Debate between Boris Johnson and Jeremy Corbyn, which was televised on ITV. A further spike appears in the Conservative time series on the 6th December, which was generated by Twitter activity on the topic of the head-to-head debate held in Maidstone on the BBC's prime ministerial debate. The weekly trends are fairly flat which indicates a relatively steady number of tweets throughout the month, the indication is that the trend is increasing towards the end of the election campaign.

Figure 1: The raw time series for the volume of tweets generated by each of the main parties (top), and the weekly trend for tweets (bottom).

The retweets reveal that the Conservative party tweets are being less retweeted compared to either Labour or the Liberal Democrats, with the Liberal Democrats receiving slightly more retweets compared to the other parties (see also Table 2 above). The weekly trends reveal interesting patterns. The number of retweets for the Conservative party seem fairly steady over the month, while the Labour and Liberal Democrat retweets are noticeably increasing; in particular, the Labour retweets are increasing at a much higher rate than those of the Liberal Democrats. This may indicate that there are many more active Labour supporters on Twitter; this warrants further investigation.

Figure 2: The raw time series for the volume of retweets received by each of the main parties (top), and the weekly trend for retweets (bottom).

What are the parties discussing on Twitter?

In order to understand the parties' topics of discussion on Twitter we extracted the hashtags from each tweet. (A hashtag is a short keyword preceeded by a '#', which allows the social media network to index the tweet so that it can be discoverable by users interested in similar topics who search for the hashtag, even if they’re not following the account.) We also looked at the most popular hashtags, and selected the top-5 most popular hashtags for tweets and retweets. The hashtags for the tweets indicate some of the main topics the parties have been promoting, and the retweets amplify some of these topics, and generate others.

The plot in Figure 4 and Figure 5 shows the volume of tweets, and the number of retweets for each of the top-5 hashtags per party, respectively. In Figure 4, we observe the main party campaign hashtags of "#GetBrexitDone" for the Conservatives, "#RealChange" from Labour, and the Liberal Democrats' "#StopBrexit". These are then followed by other party-specific hashtags, such as a call to action by the Conservatives in the form of "#VoteConservative", whereas both Labour and Liberal Democrats have opted for the more generic hashtag of "#GE2019" referring to the much broader topic of the General Election 2019, which appears further down in the Conservatives top hashtags. The picture is slightly different with respect to the most popular hashtags according to the number of retweets, as presented in Figure 5. Both the Conservative party and Liberal Democrats have a relatively low number of retweets over their official hashtags compared to the Labour Party. Aside from official party hashtags, we also see that across the parties the most retweeted hashtags are for recent televised debates, including "#LeadersDebate", "#ITVDebate", "#Marr", "#MarrShow", and "#bbcqt". In addition, we strangely find the hashtag "#RT" occurring amongst the most popular hashtags for the Liberal Democrats, which may be the result of creating new tweets from tweets prefixed with "RT" standing for a retweeted message. Twitter extracts characters "RT", from the original text and classifies it erroneously as a hashtag, this will be investigated further as the research develops.

Party Hashtags (tweets)

Figure 4: The top-5 hashtags included in tweets published by each party in descending order.

Party Hashtags (retweets)

Figure 5: The top-5 hashtags according to the number of retweets received by each party in descending order.

Monthly view of party hashtags (tweets)

We summarised the main hashtags contained in the tweets sent each day of the campaign by the various parties, and present them in Table 3. Among the entries we see the emergence of the most common hashtags (see Figure 4) over the course of the period. These represent the topics that the parties would like to draw attention to and we see that some are reintroduced over the course of the campaign, almost as a reminder of the key issues the parties would like us to keep at the forefront of our minds.

Monthly view of party hashtags (retweets)

In addition, we analysed the tweets according to how many retweets each tweet received given the hashtag, and report on the most retweeted hashtag for each day of the campaign by party in Table 4. The results show that some of the official party hashtags promoted by the various parties have been picked up by other Twitter accounts, with some hashtags related to the recent televised leadership debates and BBC Question Time. Interestingly, the top retweeted hashtags for Labour and Liberal Democrats vary and focus on some of the main topics that have been discussed such as the Labour manifesto, and the Liberal Democrat hashtag #DebateHer, which campaigns for Boris Johnson and Jeremy Corbyn to debate the political issues with the Liberal Democrat candidate Jo Swinson.

On going work

Here we have described a work in progress, and are currently working on an in-depth analysis of the data, which may reveal more about the use of social media by politicians during the 2019 UK election campaign.

Table 3: The top hashtags generated by each party per day according to the number of tweets sent.

Date

Conservatives

#Tweets

Labour

#Tweets

Liberal Democrats

#Tweets

2019-11-01

GetBrexitDone

22

GE2019

20

StopBrexit

95

2019-11-02

GE2019

9

RealChange

60

StopBrexit

85

2019-11-03

Brexit

8

Marr

53

StopBrexit

55

2019-11-04

Eleanor4Speaker

32

RealChange

24

DebateHer

101

2019-11-05

GetBrexitDone

40

BonfireNight

19

StopBrexit

109

2019-11-06

GetBrexitDone

124

RealChange

25

BrighterFuture

140

2019-11-07

VoteConservative

87

GE2019

23

StopBrexit

70

2019-11-08

GetBrexitDone

43

RealChange

40

StopBrexit

45

2019-11-09

VoteConservative

23

RealChange

64

StopBrexit

67

2019-11-10

RemembranceSunday

63

RemembranceSunday

117

RemembranceSunday

59

2019-11-11

GetBrexitDone

19

RealChange

25

StopBrexit

25

2019-11-12

GetBrexitDone

19

VoteLabour2019

16

StopBrexit

42

2019-11-13

GE2019

11

GE2019

23

StopBrexit

33

2019-11-14

GetBrexitDone

20

RealChange

33

GE2019

60

2019-11-15

GetBrexitDone

13

RealChange

36

StopBrexit

131

2019-11-16

VoteConservative

18

labourdoorstep

29

StopBrexit

60

2019-11-17

GetBrexitDone

29

labourdoorstep

26

StopBrexit

37

2019-11-18

GetBrexitDone

21

GE2019

25

StopBrexit

38

2019-11-19

LeadersDebate

271

ITVDebate

240

ITVDebate

269

2019-11-20

GE2019

17

ITVDebate

40

BrighterFuture

107

2019-11-21

CostOfCorbyn

164

RealChange

146

GE2019

93

2019-11-22

BBCQT

74

RealChange

81

BBCQT

83

2019-11-23

GetBrexitDone

34

GE2019

41

bbcqt

58

2019-11-24

ConservativeManifesto

309

WASPI

61

StopBrexit

60

2019-11-25

ConservativeManifesto

65

GE2019

42

GE2019

88

2019-11-26

VoteConservative

43

RegisterToVote

61

GE2019

121

2019-11-27

GE2019

43

NotForSale

62

GE2019

62

2019-11-28

VoteConservative

45

ClimateDebate

73

ClimateDebate

41

2019-11-29

BBCDebate

18

GE2019

18

GE2019

35

2019-11-30

GE2019

16

labourdoorstep

15

BrighterFuture

48

2019-12-01

ITVDebate

53

ITVDebate

56

ITVDebate

76

2019-12-02

VoteConservative

28

OnYourSide

30

GE2019

53

2019-12-03

VoteConservative

45

GE2019

29

StopBrexit

85

2019-12-04

VoteConservative

37

GE2019

25

StopBrexit

48

2019-12-05

VoteConservative

44

GE2019

55

GE2019

70

2019-12-06

VoteConservative

32

GE2019

15

GE2019

17

Table 4: The top hashtags generated by each party per day according to the number of retweets for each tweet.

Date

Conservative

#Retweets

Labour

#Retweets

Liberal Democrats

#Retweets

2019-11-01

GetBrexitDone

15949

GP

52146

GP

52146

2019-11-02

VoteConservative

2812

GP

104288

Johnson

30639

2019-11-03

GE2019

2934

Marr

72504

EqualityTownHall

21585

2019-11-04

GetBrexitDone

17230

publicduty

6365

DebateHer

90781

2019-11-05

GetBrexitDone

32776

Grenfell

6400

DebateHer

186703

2019-11-06

GetBrexitDone

41430

Tories

21755

DebateHer

129274

2019-11-07

VoteConservative

66857

GE2019

27962

UniteToRemain

29229

2019-11-08

LestWeForget

117985

bbcqt

10941

GE2019

21494

2019-11-09

LestWeForget

101130

RealChange

13040

LestWeForget

16862

2019-11-10

RemembranceSunday

30987

RemembranceSunday

83964

LestWeForget

39465

2019-11-11

OurNHS

9270

RealChange

9869

investigation

8980

2019-11-12

GE2019

973

CantTrustTheTories

7215

ImWithHer

19843

2019-11-13

NHS

3117

GeneralElection19

6577

RemainAlliance

6511

2019-11-14

VoteConservative

3837

r4today

12652

r4today

70143

2019-11-15

ChildrenInNeed2019

4449

NHScrisis

21161

StopBrexit

92033

2019-11-16

GetBrexitDone

2181

ge2019

12936

StopBrexit

11786

2019-11-17

GetBrexitDone

2539

Marr

12104

BrighterFuture

4630

2019-11-18

GetBrexitDone

6427

r4today

11705

BishopAucklandFarmer

10695

2019-11-19

LeadersDebate

105759

ITVDebate

340857

ITVDebate

93150

2019-11-20

LeadersDebate

10643

ITVDebate

92348

BrighterFuture

21186

2019-11-21

CostOfCorbyn

35810

RealChange

457180

GE2019

36952

2019-11-22

BBCQT

45975

RealChange

248885

bbcqt

25227

2019-11-23

GetBrexitDone

30008

bbcqt

64473

bbcqt

85696

2019-11-24

ConservativeManifesto

91033

Ridge

99300

StopBrexit

8125

2019-11-25

ConservativeManifesto

15701

Gogglebox

45252

RegisterToVote

35949

2019-11-26

andrewneil

13271

RegisterToVote

58824

RegisterToVote

20300

2019-11-27

NHS

18827

SaveOurNHS

302570

GE2019

20396

2019-11-28

NHS

12234

ClimateDebate

94728

climatedebate

7133

2019-11-29

LondonBridge

5772

BBCQT

31701

StopBoris

6471

2019-11-30

FakeLaw

7643

RealChange

14773

Bigotgate

19599

2019-12-01

Marr

15864

Marr

153068

MarrShow

124267

2019-12-02

GetBrexitDone

5480

OnYourSide

36835

MarrShow

37279

2019-12-03

VoteConservative

8788

GE2019

26357

StopBrexit

10454

2019-12-04

VoteConservative

13150

SaveOurNHS

34563

Corbyn

11180

2019-12-05

GetBrexitDone

11066

GE2019

22028

Corbyn

5160

2019-12-06

VoteConservative

5342

GE2019

4654

GE2019

5730