Twitter in the 2019 Indian General Elections: Trends of Use Across States and Parties

Twitter was used widely by various parties, especially political leaders, during the 2019 general election campaign. In this article, a database of 6.9 million tweets is analysed to examine key trends across the parties’ use of social media during election time.

The use of social media as the primary means of public outreach was one of the notable features of the 2019 general elections (Press Trust of India 2019). After Prime Minister Narendra Modi’s successes with the use of Twitter and Facebook in the 2014 elections (Rao 2018), politicians from various states jumped onto the bandwagon, building Twitter profiles for themselves and getting their party politicians to do the same (Pal 2019). The 2019 elections, thus, saw every single major political party create some form of a  digital presence, with several shifting significant parts of their political campaign online, and cutting out the mainstream press as an intermediary in the process (Hindu 2019; Prasad 2019).

The 2014 elections were also important for bringing to light personality-centric politics. That is, Modi was far more central to the social media discourse of the party than its social or economic ideology (Kaur 2015). Modi humanised the campaign and anchored it to his individual professional and personal characteristics. Thus, for him, the story of an earthy, self-made politician working his way up, from delivering tea to helming a political movement, was aided by the affordances of social media that enabled the creation of a “man for all seasons” (Gulati 2016). This was achieved through pictures of him in various avatars, alongside citizens, leaders, celebrities, and family members (Jain and Ganesh 2019; Sen 2016).

The opposition parties in 2014, with the exception of Aam Aadmi Party, were mostly conspicuous in their absence (Goyal 2014). The Indian National Congress (INC) doggedly avoided getting on social media, with most of its key leadership having no organised presence online (Reuters 2013). This was despite the fact that they once boasted of having India’s first political social media celebrity Shashi Tharoor (India Today 2013). For regional parties, the realities of 2014 were very different from those of 2019. The rapid spread of low-cost smartphones and the uptake of high-speed internet connections had not happened yet, and the core population of social media users was still relatively small, at least less than 100 million (India Today 2014). However, the 2017 Uttar Pradesh (UP) elections made it clear that social media played a role, both as an outreach mechanism and as a means of organising on-ground resources for an election (Nanjappa 2017; Datta 2017; Kaushal 2017; News Mobile 2017; Jain 2017).

For this article, we analyse a database of 6.9 million tweets, collated from the accounts of 17,261 politicians across 127 parties in India, to present descriptive patterns of the organisational infrastructures1 related to the 2019 elections. The data was  collected using a machine language classifier. We started with a sample of known politicians2 and featurised their Twitter profile text and tweet content to distinguish them from other Twitter users, using a logistic regression classifier pipeline. We then used their Twitter friendnetwork, listnetwork, hashtagnetworks and the Election Commission of India (EC) database to find out who they were connected to, as a means of discovering new politicians. For this, it was assumed that politicians were likely to be connected to other politicians, and that over multiple iterations of running the classifier, we would have a fairly high coverage of politicians. Once the classifier identified an account as one being associated with a politician, it was verified by a human, and coded for state, party and level of electoral contest. State was coded as the state of the politician’s most recent electoral contest or the state of Rajya Sabha appointment, or noted as national where the account belonged to an individual or party with no regional affiliation, such as any individual who served as national secretary for a party, or the party headquarter Twitter page. 








Politician

Party

Total Retweets recorded in the election season (in million)

Total Likes recorded in the election season (in million)

No of Tweets in the election season 

No of followers at the end of elections (in million)

State / Constituency

Narendra Modi

BJP

430 

98 

1,814

49.9 

Uttar Pradesh

Arun Jaitley

BJP

108 

26 

3,672

14.8 

Gujarat

Amit Shah

BJP

355 

82 

6,140

13.5 

Gujarat

Rajnath Singh

BJP

27 

412

13.0 

Uttar Pradesh

Sushma Swaraj

BJP

21 

174

12.7 

Madhya Pradesh

Kiran Bedi

BJP

8

715

11.9 

Delhi 

BJP Official

BJP

403 

121 

23,927

11.0 

National

Gautam Gambhir

BJP

35 

167

9.8 

Delhi

Akhilesh Yadav

SP

45 

430

9.5 

Uttar Pradesh

Rahul Gandhi

INC

80 

21 

234

9.5 

Uttar Pradesh

Table 1: Most followed political Twitter accounts during the 2019 election season. (Source: Authors’ own data)

Looking at the top 10 politicians, one pattern is clear: the Bharatiya Janata Party (BJP) dominates the list. In fact, 228 of the top 500 most followed politicians across the country are from the BJP.  We also see that politicians with constituencies in northern India dominate the Twitter space, and in fact just four states Delhi, Maharashtra, UP, and Gujarat collectively account for 60% of the top 500 most followed politicians. The BJP also had the most effective use of a party central account. That is, its central account, BJP4India, tweeted over 150 times each day during the election season, and eight of the 10 accounts that tweeted with the highest frequency were all from the BJP, six of which were official party accounts.3 A significant part of this tweeting activity was the retweeting of other leaders or state party accounts at appropriate times, such as immediately following a statement from a major leader’s personal Twitter account. For a comparison, the INC’s official national account was tweeting at the rate of about 20 tweets daily through the election season.

Despite the BJP’s obvious lead, other parties are nonetheless invested in online spaces. During the run-up to the 2019 elections, every major political party had built a Twitter presence. Over 30 politicians had a follower base in excess of 1 million, a list that featured most heads of parties. Several of the professionals who had served one party moved on to working with others, and a cottage industry of specialists who worked on social media for politicians flourished. While the BJP still dominated this space in terms of the sheer numbers of politicians’ followers or the scale of virality of their messages, other parties had followed its leadership on one important metric: top politicians increasingly spoke to the media, and the public in general, only through social media, largely sidestepping professional journalists altogether (Rodrigues 2019).


Figure 1: Weekly (normalised) tweets and their retweet rates by key political leaders during the election cycle. (Source: Authors’ own data)

Figure 1 provides a descriptive overview of the activities of key politicians across parties. We see that Modi is a dominant figure, both in terms of the volume of tweeting as well as the extent to which his tweets, on average, get retweeted. Rahul Gandhi, on the other hand, has a much smaller volume of tweets, but gets a higher average of retweets than Modi. Likewise, Akhilesh Yadav and Mayawati have smaller tweeting volumes than some of the other major politicians, but their tweets tend to get highly retweeted. Arvind Kejriwal and Aaditya Thackeray are among the most prolific tweeters, by number of tweets per period. Most politicians are fairly consistent with their volume of tweeting, though Nitish Kumar and Mayawati all have fairly sizable peaks, and both saw declines in their Twitter engagement. Moreover, they both stopped tweeting right after the election results.

Temporal Trends

Social media use by politicians tends to have temporal patterns of use around the times that topics trend and fall. Such use tends to be driven by antagonistic tweeting around specific topics. For instance, in Figure 2, we see that the Congress and the BJP both had peaks and drops around the same period. In the figure below, we see that one peak around early February is timed around the entry of Priyanka Gandhi Vadra into politics as well as a fairly strong mobilisation behind the ChowkidarChorHain hashtag by the party. The next major period of activity is around early-March 2019, with the BJP first wresting the initiative with momentum from the Balakot incident, and the subsequent MainBhiChowkidar campaign. The period of striking exception, we observe, is the end of the campaign, when BJP has a spurt in tweeting following its victory at the polls. On the other hand, the Congress had a dramatic fall in tweeting.


Figure 2: Weekly (normalised) Twitter activity from all BJP and INC accounts for the election cycle. (Source: Authors’ own data)

These patterns are most evident in locations where there is a two-party race between fairly evenly-matched opponents competing in roughly the same electoral constituencies. A great example of this is seen in the tweeting trends of rivals in Tamil NaduDravida Munnetra Kazhagam (DMK) and the All India Anna Dravida Munnetra Kazhagam (AIADMK). Both parties tend to have spikes and drops in the number of tweets that party leaders put out, around the same times, as seen in Figure 3. Tamil Nadu is a unique case of continuing sparring between two parties that are collectively dominant in the state, but have no practical presence in other states of the country. However, for most other parties, the key rival is the BJP. This can change the nature of online battles since it is not sufficient to counter a state unit of the BJP on Twitter one is essentially up against a national juggernaut, which has a formidable downstream force from the headquarters. 

One challenge of facing down a national party is in countering hashtags. The vast majority of hashtags have a short shelf-life, but are extremely important in terms of influencing the overall discourse. The BJP dominates this discourse. The BJP  features in the highest number of hashtags in 16 states, while the Congress trended the most in nine. Even in states like Tamil Nadu, where the BJP is only a player through a coalition partner, the top trending hashtag for the entire election period was #TNWelcomesModi. This highlights the importance of a national figure in pushing forth a certain agenda. That is, the recognisability of Modi on social media, as opposed to the state party president, became the main driver of Twitter hashtags and discourse in general around the BJP in Tamil Nadu, driven ostensibly by the value of the Modi brand.

While hashtags do not give a good sense of what the overall sentiment is like, given their users tend to be advanced Twitter users, they do provide a window into the exchange between big players (Giachanou and Crestani 2016; Greenhow 2017). In Telangana, for instance, the overall highest number of trending hashtags are of the All India Majlis-e-Ittehadul Muslimeen (AIMIM), followed by the BJP, and finally the Telangana Rashtra Samithi (TRS). This is partly because the AIMIM and the BJP were head-to-head on Twitter,6 while the TRS was relatively subdued online. Though the latter eventually trounced the BJP in the state, they did not have the pitched nationwide battle online that AIMIM did (Jose 2019). This is largely because the AIMIM, and its leader Asaduddin Owaisi, caters primarily to a Muslim constituency, and has, thus, often been pitched against figures on the Hindutva spectrum (Moore 2016).

 

Figure 3: Weekly (normalised) Twitter activity from all AIADMK and DMK accounts for the election cycle. (Source: Authors’ own data).

Personality versus Ideology

There has been much discussion on how Twitter may inherently be personality-driven, and encourages behaviour that focuses on individual traits rather than substantive policy discussions (Cocker and Cronin 2017; Avedissian 2016; Uysal and Schroeder 2019; Ott 2017). A look at the following major parties (Table 1) indicates how the Twitter thrust is aimed at individuals rather than the party of which they are a part. In each of the 23 parties discussed here, we find that a key leader has more followers than the party’s main account. While in several cases the party account is still the account that tweets the most or acts as the official voice of the party, the central leader for many parties come to represent the official face of the party. 






Party

Most Followed Leader

Leader Following

Official Party Account Following

Following Multiplier

JDU

Nitish Kumar

48,42,184

9,093

532.5

JKNC

Omar Abdullah

30,72,587

36,469

84.3

BSP

Mayawati

2,75,314

10,767

25.6

AITC

Mamata Banerjee

33,06,946

1,91,529

17.3

AMMK

TTV Dhinakaran

1,02,113

6,383

16.0

BJD

Naveen Patnaik

25,39,516

1,61,685

15.7

TDP

N Chandrababu Naidu

42,65,776

2,71,602

15.7

JDS

HD Kumaraswamy

97,059

6,545

14.8

AIMIM

Asaduddin Owaisi

5,24,399

43,311

12.1

SS

Aditya Thackeray

18,63,545

1,67,797

11.1

DMK

Udhay Stalin

24,80,817

2,89,634

8.6

SAD

SS Badal

3,31,611

53,101

6.2

SP

Akhilesh Yadav

95,02,948

16,92,198

5.6

TRS

KT Rama Rao

16,93,016

3,03,799

5.6

PDP

Mehbooba Mufti

1,68,632

37,164

4.5

BJP

Narendra Modi

4,96,85,274

1,20,05,992

4.1

NCP

Sharad Pawar

12,27,804

3,06,245

4.0

CPM

Sitaram Yechury

7,71,091

2,41,331

3.2

YSRCP

YS Jagan Reddy

9,19,136

3,55,629

2.6

ADMK

O Panneerselvan

3,39,894

1,39,897

2.4

MNS

Raj Thackeray

1,83,492

84,346

2.2

INC

Rahul Gandhi

94,83,390

51,66,879

1.8

AAP

Kumar Vishwas

54,85,421

48,98,184

1.1

Table 2: Parties and their most followed Twitter accounts during the 2019 election season. (Source: Authors’ own data.)

We find that regional parties are much more likely than national parties to have an individual leader with a significantly higher following than the party. In fact, if we remove the Bahujan Samaj Party (BSP) and the All India Trinamool Congress (AITC), which are classified as national parties per the technical definition of the EC, but for practical purposes are primarily competitive in only one state each, then the relative influence of the central figure of the political party is much greater than that of the party account. The (AAP) was the only party in which the head of the party, Arvind Kejriwal, had less followers online than the official party handle. 

We looked for cases in which a single outsized figure massively dominates the Twitter presence of a party. Of the parties surveyed, we found that Kanhaiya Kumar of the Communist Party of India (CPI), Nitish Kumar of the Janata Dal (Secular) (JD[S]), Omar Abdullah of the Jammu and Kashmir National Conference (JKNC), Naveen Patnaik of the Biju Janata Dal (BJD), and Chandrababu Naidu of the Telugu Desam Party (TDP), significantly dominated the Twitter presence of their respective parties.6 In all but four cases, the most followed politician from a party was the declared or presumptive head of the party; Kumar Vishwas of AAP was the only party politician, not related to the head of the party, who was the most followed individual in the party. 

For JD(S), DMK, and Shiv Sena, the most followed figure is the son of the head of the party.  The pattern of familial domination of party on Twitter is seen in about half the parties studied. For AIMIM, Samajwadi Party (SP), Pattali Makkal Katchi (PMK), Nationalist Congress Party (NCP), TRS, Yuvajana Sramika Rythu Congress Party (YSRCP), JD(S), Shiv Sena, Rashtriya Janata Dal (RJD), DMK, and Shiromani Akali Dal (SAD), two each or more from the same family feature among the most followed Twitter accounts for that party. For instance, for the SP, no individual other than those from the Mulayam Singh Yadav family features among the most followed. Although political theory suggests that the investment of a party into a single individual can have detrimental effects for the longevity or health of the party itself (Schumacher and Giger 2017), dynasty-driven parties have had a long and successful history at the polls in India (Chandra 2016). The two largest communist parties, that is, the CPI and Communist Party of India (Marxist) (CPI[M]) were relatively small players on Twitter, when pegged against their respective influence in state legislatures and their historical importance in the political system. While the CPI(M) had a number of strongly followed accounts in Kerala, as well as active accounts of the general secretary and the main party account, it had a marginal presence in other states. The CPI, on the other hand, had no account with more than 10,000 followers, other than that of Kanhaiya Kumar.

There is a significant difference between the various states in terms of how many important accounts of politicians are situated in respective states. This can be an important factor if those particular politicians are drivers of votes in those respective states7. We plotted the politician accounts with more than 500,000 followers by state and found that there are distinct national trends to the spread of key players. That is, states with a high number of parliamentary seats, such as UP and Maharashtra, are over-represented, while others like West Bengal and Tamil Nadu are under-represented. Delhi has an extremely high number of politicians on this list, in part because of its status as the capital and a metropolis, and also because AAP, with its headquarters in Delhi, has several politicians who are very active on Twitter. 


Figure 4: Twitter accounts with more than 5,00,000 followers, by state. (Source: Authors’ own data.)

Parties have a small number of highly retweeted accounts. For instance, Amit Malviya and Tajinder Bagga from the BJP, and Ashok Tanwar and Akhilesh Singh from the INC are highly retweeted by their respective parties and have strong effects within their networks, despite having a much lower following than some of the major leaders of the respective parties. These leaders share two common characteristics. First, despite their relatively low following outside of the party, when compared to top leaders, these politicians are very highly followed among the ranks and file of their respective parties. Second, leaders either hold key positions that makes them privy to the digital space (Malviya heads IT for BJP, Tanwar heads the Indian Youth Congress [IYC]), or have built reputations as being aggressive trollers of the other side.8

Language Use

Following up on past work that has suggested that the growth and use of Indian languages online has helped fuel a rise in social media adoption in some parts of the country (Seetharaman 2018; Mathur 2019), we examined the role of language and the extent to which they have contributed towards parties adopting Twitter. In Figures 5 and 6, we see the distinctions in the use of English between the states and parties.  We plot the log odds of English being used by politicians of a state, using a mixed effects model. We controlled for the user, their followers count and their party. A similar model was used to evaluate variation by party.


Figure 5:  Log odds of using English on Twitter, by state. (Source: Authors’ own data.)

We find that politicians in states from the Hindi-heartland (with Hindi as their official language) are much less likely to use English on Twitter as compared to their counterparts in the rest of India. Moreover, we found a positive correlation between the use of English and the number of Twitter followers, indicating that, on average, senior, more popular politicians tend to use English more than their local, state-level peers. There may be a range of reasons for this, including that senior politicians tend to appeal to more citizens than just their immediate electoral constituencies, and also serve as spokespersons for the mainstream media, both in India and abroad.

For parties, we find that government handles are most likely to tweet in English compared to political parties, showing that the formal language of business continues to be English.9 Next set are parties from smaller non-Hindi states like the JKNC and Jammu and Kashmir Peoples Democratic Party (JKPDP), SAD (Punjab), and AIMIM (Hyderabad). While the INC has a small and statistically insignificant preference for English, the BJP has a small and significant preference against it. 

Being the national parties, their tweets are made of several languages and vary by state. Moreover, regional parties from large states like TDP, TRS and YSRCP (Telangana/Andhra Pradesh), RJD and JD(U) (Bihar), SP and BSP (UP), NCP and Shiv Sena (Maharashtra) are highly unlikely to use English to Tweet. These parties show a preference for their respective local languages. In particular, Hindi is widely used in the parties that have their constituents primarily in the Hindi-belt states, and has arguably contributed to an expansion of the political constituency of Twitter use. Finally, we find that parties from Tamil Nadu (DMK, AIADMK and Amma Makkal Munnetra Kazhagam [AMMK]) prefer Tamil overwhelmingly, and are least likely to use English (and are incidentally even less likely to use Hindi).

 

Figure 6:  Log odds of using English on Twitter, by party. (Source: Authors’ own data.)

The use of language is important from the perspective of widening the net for those who have access to Twitter outreach online (Seetharaman 2018). The rapidly expanding use of Hindi and Tamil, alongside other regional languages online (Mandavia and Krishnan 2019), enables politicians to engage directly with constituents in local languages, including on hot-button issues where better grassroots mobilisation can be done (India TV News 2019). This goes against assumptions that social media, and particularly Twitter, is a means of appealing primarily to the elite, English-speaking audiences, which was largely true a decade ago (Lloyd 2012). However, Twitter serves an important function of serving as the party’s official outreach vehicle, and the expansion into local languages highlights the outreach to regional-language mainstream media as well.

Conclusions

This article shows some of the large-scale trends in political outreach on Twitter through the elections period. The use of language highlights ways in which Twitter has extended its function well into being a means to mediate the conversation between political actors and the mainstream media, as well as small fractions of the voters who were active on social media in 2014, to being a channel accessible to the electorate outside of the social media elite (Drache et al 2015).10 

While the BJP was, for all functional purposes, the only player in the game in 2014, almost every major party is reasonably well-invested in the fray now. The Twitter presence of several parties offers a window into their branding offline as well, as many of the smaller state parties invest heavily in highlighting the personality and voice of the leaders’ families.11 However, the real story of this work may be to highlight the sheer dominance the BJP has on Twitter. Not only are the vast majority of highly followed leaders from the BJP, the party is both aggressive in terms of the frequency with which its various accounts output messages as well as organised in terms of the  networks of retweeting and favouriting among followers. Despite the challenge posed by smaller parties, the lead that the BJP holds on Twitter will be formidable to counter online in the years to come.

Must Read

Do water policies recognise the differential requirements and usages of water by women and the importance of adequate availability and accessibility?
Personal Laws in India present a situation where abolishing them in the interest of gender justice also inadvertently benefits the reactionary side.   
Concerns have been raised about criminalising triple talaq now that the Muslim Women (Protection of Rights on Marriage) Bill, 2017 has been passed as an ordinance. This reading list is to help...
Back to Top