Research Design

The relevant literature calls for more analysis of the causal micro-mechanisms that go beyond correlations in surveys and macro-level dataFootnote ⁶ as well as for uncovering the context-dependent determinants of turnout.Footnote ⁷ Focusing on Turkey, this study attempts to narrow the gap in the literature on negative campaigning.Footnote ⁸

We analyze Twitter dataFootnote ⁹ for the following consecutive period when the voters went to the polls in Turkey: 16 April 2017 Referendum on Constitutional Change, 24 June 2018 General Elections and Presidential Elections, 31 March 2019 Local Elections, and 23 June 2019 Re-run Local Elections in Istanbul. We analyze tweets starting two months prior to election day.

We used search terms (see Appendix) indicating that a voter first planned to abstain or boycott but later decided to vote. We visited each tweet link in order to get a proper sense of the tweets and the references made in the tweets. We removed tweets that had the proper terms but were either re-tweets, tweets that indicated votes for AKP or its ally, the ultranationalist MHP. We also removed tweets that included the search words but were either irrelevant or did not provide any discernable motive for the switch in voting intention. This left us with a significant database of tweets. Based on the relevant literature, we categorized the tweets according to the drivers of the change in turnout decision. Some tweets would fit into more than one category. In that case, we count them in all the relevant categories. We looked at the four election periods both separately and in conjunction. For each period, we focused further on the most dominant dynamics that motivated voters to eventually go to the polls.

Our research design gives us an unobstructed view of the dynamics of change in turnout decision. Most of the literature on turnout and the effect of negative campaigns relies on experiments or surveys. In our case, people speak to and reflect in detail what made them switch their voting decisions both spontaneously and without prompting. In addition, in contrast to studies that try to correlate individual or aggregate data on turnout with the content of election campaigns, our study traces the direct connection between exposure and reception. The fact that people in our case self-identify negative campaigning as the reason for their voting behavior, without having been prompted by researchers to talk about this issue, is significant.

Furthermore, looking at multiple elections within the same country enables us to control for institutional factors that may influence turnout, such as compulsory voting laws or the electoral system.Footnote ¹⁰ These factors are constant in our case. Hence, they would not explain why people would first publicly declare that they would not vote and then reverse their decisions. Similarly, macro-level explanations based on political culture can also be controlled for in our research design.Footnote ¹¹ In addition, since these are people who initially publicly declared that they would abstain, institutional settings, or general desirability biases would not be sufficient to explain their situation.

We believe that the switch from abstention to turnout in our case can best be explained by the emotional responses that grew as a response to the negative campaigning of the incumbent, the AKP, during the electoral process. Negative campaigning, coupled with the unlevel playing field characteristic of competitive authoritarian regimes,Footnote ¹² angered the dissident electorate and triggered their sense of injustice and thereby mobilized them to vote. Their decision to vote was mostly expressive and intrinsically so, as the discussion below will demonstrate. This study will show how essential it is to utilize the literature on emotions and expressive voting in understanding the dynamics that determine a switch from an intended abstention to a turnout decision.

Negative Campaigning and Turnout

The literature on the relationship between negative campaigns and turnout is inconclusive at the cumulative level.Footnote ¹³ Some studies, most famously those of Ansolabehere and colleagues,Footnote ¹⁴ maintain that negative campaigning affects turnout negatively by causing cynicism among the electorate, making them feel less efficacious and demobilizing them. The authors first exposed participants to negative campaign ads in experimental settings and then tested their findings with aggregate election data by looking at how campaign tone affected voter turnout.

Others argued either that negative campaigns and turnout are not strongly relatedFootnote ¹⁵ or that the relationship is positive.Footnote ¹⁶ Finkel and Geer, combining content analysis of presidential campaigns with aggregate data on turnout and survey data, find no clear association.Footnote ¹⁷ They argue that the causal process suggested by Ansolabehere and othersFootnote ¹⁸ may be offset by “alternative processes whereby negative adverting spurs turnout by increasing political knowledge and concern about the election's outcome.” They find it more plausible that negative and positive advertising have different effects on the vote, which may result in weak net effect of advertising tone.Footnote ¹⁹ Kahn and Kenney distinguish between “useful negative information presented in an appropriate manner and irrelevant and harsh mudslinging” and argue that voters do, too.Footnote ²⁰ While the former tends to increase turnout, the latter tends to decrease it. Lau and Pomper focus more on the degree of negativity and argue that, while at most levels negativity increases turnout, at the extreme levels it decreases turnout.Footnote ²¹ Phillips finds that the sponsorship of negative campaigns matters for turnout purposes: while independently sourced negative campaigns increase turnout in favor of the candidate, candidate-associated campaigns do not have such an effect.Footnote ²²

Clear mechanisms linking negative campaigns to turnout behavior are not laid out specifically in most of the literature.Footnote ²³ As Martin argues: “neither side has offered a compelling theory of the causal mechanisms that connect negative campaigns and voter turnout.”Footnote ²⁴ Most of the literature that sees negative campaigns leading to an increase in turnout is based on the finding that negative campaigns stimulate attention and awareness of the campaign, yet the literature does not lay out specifically why this should increase turnout.Footnote ²⁵ Martin identifies three complementary mechanisms, which he calls “stimulation of republican duty, anxiety toward the candidates, and perceptions of increased closeness of the race.”Footnote ²⁶ The first maintains that exposure to negative ads can heighten perceptions of public problems and thereby stimulate republican duty since the election will appear of greater importance. According to the second mechanism, negative ads could raise anxiety and thereby “foster interest in campaigns, and plausibly participation.” The final mechanism suggests that negative ads can make it more likely for people to vote by signaling the relative closeness of an upcoming race.Footnote ²⁷

None of these mechanisms capture the dynamics we find at work in our case, in line with Martin's predictionFootnote ²⁸ that negative ads could stimulate other mechanisms as well. Based on our data, we find that negative campaigns facilitate voting decisions to switch from abstention to turnout as follows: Negative campaigning by the political incumbent and by his or her allies in politics, conventional media, and social media triggers a sense of anger and injustice among dissidents, which urges them to participate. In order to delve into how this process works, we need to understand what role emotions play in the expressive characteristics of voting behavior.

Expressive Voting

In addressing the paradox that so many people vote despite the marginal influence of their one single vote on election outcomes,Footnote ²⁹ scholars have considered the intrinsic benefits of fulfilling one's voting duty. As Schuessler pointed out,Footnote ³⁰ voting is, in part, an act of self-expression, and as such it can capture motivations related less to the outcome of the elections (i.e., instrumental motivations) than to the act of voting itself.Footnote ³¹ Brennan and Buchanan,Footnote ³² emphasizing the symbolic value of elections, liken voting to watching a sporting event: “neither the act of voting nor the direction of a vote cast can be explained as a means to achieving a particular outcome, any more than spectators attend a game as a means of securing the victory of their team.”

Expressive voting is focused on the benefits of the voting act itself such as those derived from “expressing an opinion, confirming an identity, associating with a particular cause or candidate, acting as some moral code requires.”Footnote ³³ As Hamlin and Jennings argue, identification can mean social identification, as is the case with Schuessler's account of expressive voting. SchuesslerFootnote ³⁴ claims that who and how many people participate in voting determines its expressive utility. That way, voters confirm their identities through association with a group of people who vote for “x” candidate.Footnote ³⁵

Identification can also mean identification with parties or candidates.Footnote ³⁶ Brennan and Hamlin,Footnote ³⁷ for example, argue that voters go to the polls if candidates fall within their expressive domains. Voters can identify with “the candidate's moral character, good looks or ethnic origin or with the candidate's or party's general ideology.”

When it is understood as moral choice, Hamlin and JenningsFootnote ³⁸ argue that expressive voting corresponds to things such as voting for moral or desirable characteristics of particular politicians and policies. We can include all moral considerations that enter the turnout and voting decisions under the ambit of expressive voting, as we discuss further with illustrations from our cases.

The scholarly literature has provided a useful distinction between intrinsically and extrinsically expressive motivations for voting. Those that are linked to expressing a certain image and to fulfill internal satisfaction (such as fulfillment of civic duty and responsibility, following a moral imperative) are seen as intrinsically expressive.Footnote ³⁹ Extrinsically expressive factors include pressure from family, friends, and colleagues to vote.Footnote ⁴⁰ In our findings external factors do not play as significant a role as emotions triggered by negative campaigning, as we discuss in detail below.

Not many studies delineate the link between negative campaigning and expressive voting. Hamlin and JenningsFootnote ⁴¹ argue that Schuessler's expressive approach to voting can be used in understanding the link between negative campaigning and turnout in the following way: negative campaigning is seen as a useful tool for each party to discredit the opponent and make it less attractive. Yet when all parties employ this approach, all become less attractive, which implies lower turnout, especially among the least committed. In our case, negative campaigning did not lead to the withdrawal of the voters. Instead, potential abstainers reversed their decisions as a response to negative campaigns. In order to understand why, we need to turn to the role of emotions.

Emotions and Political Participation

Based on their literature review, Crigler and HevronFootnote ⁴² find that given the crucial role of emotion in explaining issues such as political communication, preference formation, and political participation, more work is needed on its short- and long-term impacts. Demertzis also points out that it is only recently that scholars have realized the importance of studying emotions as “microfoundations of political action and macropolitical institutionalization.”Footnote ⁴³

So far, the literature on emotions and political behavior has found that anger is more likely to lead to participatory political action.Footnote ⁴⁴ Based on experimental research, Valentino et al.Footnote ⁴⁵ find that anger, rather than anxiety or enthusiasm, motivates people's participation in campaigns, though they focus on campaign participation, not voting.Footnote ⁴⁶ Only few studies so far have examined the effect of emotion on turnoutFootnote ⁴⁷.

Weber finds that there has been a “scarcity of scholarly work examining whether specific emotions aroused in political campaigns influence political behavior” and that much of the work on negative campaign ads “fail to consider the effect of emotions.”Footnote ⁴⁸ Marcus and Mackuen show that different emotions trigger different cognitive processes during election campaigns; while anxiety “stimulates attention toward the campaign and political learning and discourages reliance on habitual cues for voting,” enthusiasm “powerfully influences candidate preferences and stimulates interest and involvement in the campaign.”Footnote ⁴⁹ Similarly focusing on the effect of emotions on the cognitive process, BraderFootnote ⁵⁰ finds that “when added to a negative message, fear-eliciting images and music stimulate ‘bottom-up’ reasoning on the basis of contemporary evaluations. Enthusiasm-eliciting images and music, when added to a positive message, encourage fidelity to prior beliefs.” While these studies help us establish links between the emotions we feel and the political decisions we make, they do not tell us about how these emotions influence turnout or political mobilization in general.

Based on experiments showing different campaign ads evoking different emotions, ad-tracking, and follow-up survey questions, WeberFootnote ⁵¹ finds that those who show anger express more willingness to participate in political processes whereas fear and sadness lessen civic engagement. Experimental, survey, and aggregate data sets showing correlation between ad exposure and turnout give us the general picture of the emotion-turnout nexus but in these settings participants do not get to elaborate how particular emotions lead them to act in certain ways. That way, the mechanism we capture in our research may go unnoticed. Based on our analyses, we find a different route through which negative campaigns increase turnout and switch the decision from abstention to turnout: negative campaigning by the incumbent party and its allies has triggered a set of emotions (such as injustice, outrage, anger, or bad conscience over not voting), which led to decisional changes.

Recent work on emotions and turnout, further notes the special place of voting as a distinct form of political participation, which makes it important to trace closely the relationship between emotions and turnout,Footnote ⁵² as we will try to do below. In that direction, JungFootnote ⁵³ tracks how moral rhetoric used by political parties increases positive emotions such as pride, hopefulness, and excitement. Groenendyk,Footnote ⁵⁴ based on experimental and survey data, argues that candidates who can induce “good gut feeling” among the voters are more likely to bring voters to the polls, while negative emotions toward candidates did not have a similar effect. Panagopoulos,Footnote ⁵⁵ also based on experimental data, does not find any connection between inducement of hope and voter turnout.

Aytaç, Rau, and StokesFootnote ⁵⁶ find that when unemployment is high, the unemployed are led by the challengers to blame the incumbent and demonstrate anger, which increases turnout rates. This study gives us important insights on how blame for unemployment, as a form of negative campaigning, affects electoral turnout. We extend these findings toward capturing the unintended aspect of negative campaigns. The literature thus far assumes that “politicians anticipate and manipulate” responses of people to emotions such as anger, fear, and shame.Footnote ⁵⁷ Based on this, Aytaç, Rau, and Stokes,Footnote ⁵⁸ for example, call for further research that sorts out “which emotions come into play and how elites anticipate and encourage.”Footnote ⁵⁹ Similarly, BraderFootnote ⁶⁰ also assumes campaigns elicit certain emotions among the electorate for certain mobilizational purposes. While it is true that both politicians and campaign producers are aware of the role of emotions and try to use them accordingly, it is also vital to understand which unintended emotions can be triggered during negative campaigns and how these emotions affect the electoral process. As we will show, AKP was hoping to instill emotions among the electorate that would distance them from the opposition yet inadvertently caused anger, moral outrage, spite, and similar feelings and thereby managed to mobilize an important segment of the opposition against them.

Determinants of Turnout for Those Who Intended to Abstain

In our analysis, we first looked at whether tweets expressed instrumental motives for turnout decision by mentioning the likelihood that their vote can actually change the election outcome. We then looked at whether voters changed their decisions to avoid regretting a tie or loss by one vote caused by their abstention.Footnote ⁶¹ As exhibited in Table 1, we found that only a minority of tweets indicated instrumental motivation. In those cases, the voters became convinced that their votes would have an effect on the election outcome. It was actually only three of the voters who stated that they had changed their decision not to vote based on their belief that their vote would have an impact on the election outcome. Similarly, those who decided to vote because they did not want to be responsible for a tie or loss by one vote were also very rare. This does not necessarily mean that instrumental and minimax rationale do not play a role in turnout decision. It could very well be that those who believe in these motives did not decide to abstain to begin with. Nevertheless, the findings show us that these two factors did not play a significant role in the voters’ change of turnout decision.

Table 1. Distribution of Tweets

Note: Notice that the total number of tweets differs from the sum of tweets in different categories as in some cases one tweet fits into more than one category.

As seen in the table, the remaining explanations can all be placed in the realm of expressive voting. As far as externally expressive factors, such as pressure to vote, we observe limited effect. Only thirteen people counted such pressure as a reason for the switch in their turnout decision (1, 4, 5, 3, respectively). The pressure was reported to have come from a variety of sources, including Kurdish friends, fathers and mothers, professors, friends, Twitter friends, and unspecified. Except in one instance, in which the tweeter stated that his father yelled at him, in most of the cases it seems to be that persuasion and pathic statements such as “for my/our sake,” played a role. For the rest of the tweets, intrinsically expressive factors seem to play the most important role.

Let us start with the factors that are less significant in this category. Civic duty plays some, though very limited role. In the second period, it was actually a university professor who reminded his/her student that voting was a civic duty. In the third period, tweeters expressed civic duty in the following terms: “doing his/her share,” “duty as a Turkish woman, who was granted the right to vote by Atatürk,” (x2), “fulfilling one's citizenship duty,” “for my country (x2),” “duty toward humanity,” “for our children,” “for our grandchildren,” “for Atatürk, for the republic, for laicism.” In the fourth period, one tweeter stated that s/he voted for the future of their children.

Expressive voting is not just related to whom you vote for but also whom you vote against. So expressive voting signifies not just that a voter stands for x but also that they stand against y. This point has played an important role when we think broadly about this category. We divided this category into subcategories in order to further disentangle the switch from abstention to turnout. In one category, there are tweeters who state that they decided to vote in reaction to Erdoğan. They do not further specify how their opposition to Erdoğan led to this decision. There are only eight tweets in this category, five of which were sent in the third period. Most of these tweets were sparked by a speech or declaration made by Erdoğan. Another twenty-one tweets (sixteen in the third and five in the last period) demonstrated that people decided to vote as a reaction to a perceived opposing political camp. In these tweets, it is evident that the pushing effect of the opposing camp overcame the alienation or resentment people were feeling against their parties. Cases in point are these quotes below:

Another category accounting only for a small portion of the tweets are reactions against recent policy decisions related to, e.g., retirement age, military service, soccer teams, gender policies, etc. Such tweets are observed as the following numbers in the respective periods: 1, 1, 6, and 2.

Turning toward factors that show very high prominence, we observe that identification with a party or candidate plays a very important role, second only to negative campaigning. CHP voters became weary of the party leader, Kılıçdaroğlu, and his consistent losses, yet some of the party's other candidates were appealing enough for voters to give it another chance. Especially Muharrem Ince, who ran for president against Erdoğan, Ekrem Imamoğlu, who became the mayor of Istanbul, and Mansur Yavaş who became the mayor of Ankara after having to endure a very unfair election process, were important names in that regard. Demirtaş, an HDP candidate who opened the party to non-Kurdish voters, was another such candidate. To a lesser degree so were the leaders of the other two main opposition parties, namely Meral Akşener and Temel Karamollaoğlu. A significant portion of the electorate was also angry with the CHP yet when push came to shove, they could not help but vote for it, which they strongly identified with Atatürk, the Republic, and laicism. In fact, as can be seen in Table 1, during the presidential campaign the candidate and party effect were leading factors in turnout decisions. What attracted voters to candidates were personality traits, such as honesty, trustworthiness, work ethic, courage, equanimity, etc., or emotions, such as hope, enthusiasm, or conscience. Just principles, lifestyle, worldviews, and attitude, especially in the face of polarizing behavior, injustice, and negative campaigns also appealed to the voters.

Even if not running in elections themselves, journalists and columnists such as Uğur Dündar and Emre Kongar and politicians such as Eren Erdem and Muharrem Ince, played an important role in inspiring people to vote. Among the 39 tweets that fall under this category, Selahattin Demirtaş is the key figure. Demirtaş sent a message from prison via the Yeni Yaşam newspaper on 23 March 2019, about a week before the mayoral elections urging his supporters to go to the polls and vote against fascism, despite their reservations, “for his sake.”Footnote ⁶² On 29 March, he sent a set of tweets that repeated this call. HDP had no candidates running in the mayoral elections in the biggest metropolitan cities of Turkey, and this call was meant to convince HDP supporters to vote for the main candidates of the opposition: Imamoğlu, Yavaş, and Soyer. In one tweet, a Twitter user stated, “I was not thinking of voting. I had many reasons for this. However, I will go to the polls for Demirtaş, the apple of my eye, who despite being imprisoned still continues to instill hope in us and never despairs” (“1112001119742509056”, 30 March 2019). Another wrote: “I was going to boycott the election before chairman Selo's [short for Selahattin] message. But now, it's going to be necessarily CHP” (“1109757093899313152”, 24 March 2019).

Before discussing the remaining categories, which are linked to the negative campaign strategy followed by the AKP, let us first turn to the role of negative campaigns. This strategy became especially prominent and received responses accordingly in the third period, which were local, mayoral elections, as the table also indicates. Multiple reasons can account for why the AKP increased its aggressive tone in this period. First of all, this was the first election in its history when the AKP realized that it could lose even its strongholds. In fact, top AKP members frequently told their voters they understood their grievances but urged them not to use this election to punish them.Footnote ⁶³ What is more, due to the fragmented nature of the opposition, for the first time very unlikely partners came together and made alliances in order to beat the AKP. These included secularists, nationalists, Islamists, and even, indirectly, some pro-Kurdish groups. In this context, it looks like the AKP, particularly Erdoğan, decided to execute a negative campaign and try to chip away as many voters as possible from the opposition bloc. Finally, as Aytaç and ElçiFootnote ⁶⁴ show, the longer AKP was in power and captured the state bureaucracy the harder it became for the party to blame the establishment to increase its populist appeal. Thus, the party turned to the opposition in its friend-enemy distinction in order to further represent itself as the true embodiment of the will of the people.

The plan backfired spectacularly. An electorate that had previously lost its motivation to vote was suddenly energized to participate after all. A Twitter user summarized this dynamic well: “I was not going to vote in the local election but the ugly election campaign of the AKP, its provocations, and plots became a turning point for me. The ugliness they showed in order to gain votes is on the contrary making them lose votes” (13 March 2019, “1105863832042377217”). In addition to negative campaigning, the unlevel playing field also motivated people to participate in the elections in which they did not previously see any stake. This polarizing attitude further pushed potential abstainers to vote. As we will see, across all these factors emotions played a pivotal role.

Starting with negative campaigns, this category accounts for most of the decisions to depart from abstention. It is most visible in the third period (270 tweets), followed by the fourth (32 tweets), second (8 tweets), and first (1 tweet). President Erdoğan himself, AKP's Istanbul candidate Binali Yıldırım, AKP's Ankara candidate Özhaseki, former AKP mayor of Ankara Melih Gökçek, Interior Minister Süleyman Soylu, members of AKP's ally MHP as well as pro-AKP media outlets all joined this process of negative campaigning. Allegations included calling the HDP terrorist, calling CHP an ally of terrorists, calling CHP a coup supporter, and so on. Tweeters described these practices by the incumbent party and its allies with phrases such as “dirty politics,” “hitting below the belly,” and “ugly language.”

The most infamous tweet in this regard was tweeted by Erdoğan himself on 27 February 2019 [Figure 1]. In this tweet, the country was divided into two blocs, namely the Cumhur İttifakı (People's Alliance) and Zillet İttifakı (Alliance of Abominations). During that election period, there were two major political alliances in Turkey. The main reason to form alliances was to cooperate in several districts in order to gain the majority of votes. Erdoğan and his supporters labeled their opponent, Millet İttifakı (Nation Alliance), as Zillet İttifakı. The first one, Cumhur İttifakı, is formed by the AKP and MHP. According to Erdoğan, only this bloc is at the service of the country, while Zillet İttifakı, accusing the Millet İttifakı of supporting terrorist organizations such as PKK and FETO. This tweet clearly demonstrates that the country was divided into two opposite blocs by the head of state himself.

Figure 1: Erdogan's Tweet on The Two Camps in Turkey

Here are some of the responses to this tweet that illustrate how negative campaigning affected the potential abstainers:

Another speech given by Erdoğan featuring his comment “Our security forces have detected that the PKK has put all its hope on the success of the Zillet Alliance on March 31st,” prompted many responses similar to this one:

Such accusations, equating the opposition to terror organizations or foreign powers, led to an outburst of citizens declaring, on social media, that they have changed their minds and would vote as a reaction.

Continuing with the negative propaganda line, Interior Minister Soylu likened the mayoral opposition candidates of Izmir, Ankara, and Istanbul to Trojan horses and claimed that they were working to weaken Turkey and bring chaos. One response to this statement on 11 March 2019 reveals how negative campaign even worked to overcome the anger people felt against the CHP leader: “I was not planning to vote as I was angry at Kılıçdaroğlu but after reading this, I decided to vote [angry face emoji]” (11 March 2019, “1105094514962845697”).

Similarly, the AKP candidate of Ankara maintained that “If the CHP candidate gets elected, PKK and DHKP-C will say ‘we have supported you, it is your turn now.’ Imagine, those who will bring your water bill will be militants.”Footnote ⁶⁵ In response to this, he received two similar tweets:

In addition to associating the opposition with terror groups and alien forces, another negative campaign strategy of the AKP was to associate the CHP with archaic, ineffective rule. Since the CHP is the oldest party in the country, it is also open to being selectively attacked for some of the wrongdoings and inefficiencies of the previous decades in the country. To tap into this repertoire former AKP MP and member of Erdoğan's Council for Law Policies, Burhan Kuzu (6 March 2019), sent the following tweet:

What does CHP stand for? It stands for:

• • Mud

• • Hollow

• • Garbage

• • Poverty

Long line-ups for

• • Gas

• • Fuel

• • Gas tube

• • Oil

• • Sugar

• • Bread

• • Milk

• • Meat

• • Hospital

• • Jobs.

Again, responding to this tweet, one Twitter user wrote: “Keep going like that. I was not going to vote in this election but from now on, my vote goes to the CHP” (7 March 2019, “1103557467915644929”).

Not just people who would consider themselves in the opposition camp, but also those close to Erdoğan on the political spectrum were affected by this negative campaign and polarizing attitude as seen in this tweet: “Dear Erdoğan, as someone with MHP roots, I was not going to vote just to avoid voting for a left-wing party. However, due to your speeches in the recent days, I have changed my mind. I believe now that it is just to change an incumbent that is dividing and looking down on its people to such a degree” (2 March 2019, “1101729313320329217”).

We can multiply these examples. Striking proof of how triggering negative campaigning became in this process is a tweet of Armağan Çağlayan, a well-known TV personality and lawyer, which got 17,000 likes and was retweeted nearly 2,000 times by other Twitter users: “I was not thinking of voting. But they managed to get me to the polls with the propaganda method they used. I will vote” (14 March 2019, “1106128316392706048”). The amount of interaction with his tweet showed there were thousands of people who either shared the same view or empathized with it at least.

In more granular detail, one can see that the negative campaigns even convinced many former CHP supporters to leave their grudges behind and vote for the party regardless. One Twitter user directly addressed Kılıçdaroğlu in his tweet and told him that he needed to thank Erdoğan since were it not for Erdoğan and for Soylu, he would not vote for the CHP even if they chose his/her father as the candidate (31 March 2019, “1112481830165508096”). Another user declared that s/he would not have voted if the AKP had not used such a harsh tone by saying things like “zillet” and that s/he broke his/her vow to not vote due to Kılıçdaroğlu as a result of this (31 March 2019, “1112458857584967681”).

An uneven playing field, and especially unfair treatment by the media and state institutions, also played a significant role in mobilizing people for turnout. A total of 193 tweets were found under this category, 183 of which were in the third period. There is substantial overlap between reactions against negative campaigning and reactions against the unlevel playing field for the opposition in the government-aligned media. If there is only perceived injustice, such as the allocation of no or less time to an opposition candidate, the adoption of a mocking tone, or cutting off coverage of an opposition leader, we coded it as being a reaction against the unlevel playing field. But, when the media outlet itself also went into negative campaigning, we coded it as negative campaigning. There are many tweets showing outrage against the unfair treatment of the opposition by the media. For example, in responding to a broadcast on the state television channel TRT TV that hosted president Erdoğan and the mayoral candidates of his party, a person tweeted: “I was not going to vote but upon seeing this injustice, I said to myself: ‘It is my duty to cast that vote” (30 March 2019, “1111911164966043650”).

The most climactic episode in this regard was a show on Beyaz TV, one of many pro-AKP channels, owned by one of Gökçek's sons, a former AKP mayor of Ankara. This program, hosted by Turgay Güler, triggered intense reactions on Twitter due to Güler's attitudes toward İmamoğlu, the Istanbul mayoral candidate of the opposition alliance. In that program Turgay Güler approached İmamoğlu in a very aggressive manner. He asked opinionated questions to ambush him. For example, he wanted to give the impression that İmamoğlu was in a tacit alliance with the PKK and other terrorist groups. Hence, he was both being very unfair to İmamoğlu and at the same time amplifying the negative election campaign conducted by the AKP. This program prompted the tweet below, which exemplifies the emotional outcry against the negative campaigning of the government camp:

Another tweet, also illustrates similar sentiments:

Numerous such tweets on Twitter point at how Güler's program triggered change in their intention to vote.

Rank injustice, including lawsuits against the main opposition's Ankara mayoral candidate Yavaş as his popularity was reaching its peak, cancellation of mayoral election results in Istanbul, and threats against the opposition, was also an important element of the AKP's negative campaign strategy. To give a few examples, when Erdoğan declared the government would appoint trustees if certain opposition candidates got elected, alleging that these candidates were legally unfit to govern, one Twitter user responded, “I was not going to vote but in this mock election, I am definitely going to vote.” (“1100008976626958336”, 25 Feb 2019.) Similarly, when Minister of the Interior Soylu, declared that they were going to depose Yavaş, someone tweeted “I was not going to vote. But my vote goes to Mansur Bey. (27 March 2019, “1110958626305396737.”) The legal prosecutions against Mansur Yavaş led to the hashtag #MansurYavaşYalnizDegildir (MansurYavaşisnotAlone), which went viral on Twitter.

The polarizing attitude of the AKP also contributed to the switch from abstention to turnout. Ten tweets fall under this category where the tweeters explicitly stated that they got sick of the polarizing attitude of the government. Two examples are as follows:

When we look at emotions, we see that they play a leading role in changing voting decisions. In 152 tweets we see emotions being actively expressed. In the case of negative campaigns, we see outrage. One tweet, for example, reads, “I was not going to vote in this election, but the other side drove me crazy by keeping saying ‘terrorist’” (10 March 2019, “1104793255953289217”). Another states, “I am serious, as an HDP supporter I was not going to vote because I was angry at a CHP candidate. However, I was going to have a bad conscience for the rest of my life if I did not cast an opposing vote after Erdoğan's disgusting election campaign” (30 March 2019, “1111905580690694145”). Similarly, one Twitter user wrote: “I was not going to vote but the propaganda based on lies and defamation made me nauseous. I am going to vote” (14 March 2019, “1106143192444006401”).

Aytaç and StokesFootnote ⁶⁶ find that during the Gezi protests in Turkey, the fact that protests surged as the police attacked demonstrators can be explained by the fact that anger and moral outrage, which made abstention psychologically costly for bystanders turned those bystanders into participants. We see a similar logic operating here. In fact, a tweeter put it in striking terms. Though outside the time span considered by this study – one day after the elections – it is still worth mentioning: “I joined the Gezi protests upon seeing a video showing how they attacked a man opposing violence with pressured water from close distance. Thanks to this guy [Turgay Güler], I gave up on the idea of boycott and made sure those around me do so as well” (01 April 2019, “1112651860576518145”).

Emotions also play a key role for the party or candidate effect. Hope, moral imperative, love, solidarity, and happiness are examples of the emotions triggered by candidates who made people decide to vote despite intending otherwise. Our data revealed that emotions can also lead to turnout by themselves, without any underlying prompting factor mentioned. This can be encountered in tweets that express the turnout decision with phrases such as “raise to the bait” (gaza gelmek), hope, because their conscience dictated them to do so. This point confirms the finding of Aytaç and Stokes that “people turn out to vote because doing so can help them avoid frustration and psychic discomfort”Footnote ⁶⁷ and shows that these emotional urges go even as far as turning an abstainer into a voter.