Sage Journals: Discover world-class research

Abstract

The 2016 US Presidential Election provided an opportunity to examine how political candidates’ use of social media can affect voting intentions. This study considers how political candidates can use social media to increase potential supporters’ perceptions that they will win the election, providing them extra motivation to go out and vote. Results from a two-wave survey provide evidence that following the in-group candidate (Trump or Clinton) relates to voting intentions through the increased belief that the candidate would win. However, this mediation effect occurs for only supporters of Trump or Clinton, but not for partisans of the opposing party.

Keywords

elections candidates social media voting partisan motivated reasoning

Introduction

The 2016 US Presidential Election cycle was historic. A woman won the Democratic Party’s nomination and a businessperson with no experience in public office won the Presidency. Mainstream news media reported on the campaign and candidates in relatively predicable ways: who was ahead in the polls, campaign events, and to a lesser extent, about specific policies. Despite constant news attention and a high level of citizen interest in the election, many citizens were unenthusiastic about the prospects of voting for either candidate, as Trump and Clinton were among the least popular presidential candidates in recent history (Saad, 2016). Consequently, an atypically large number of voters, including some partisans, were undecided about whom they would vote for (Silver, 2016). In this context, the candidates were faced with the task of rallying potential voters to support their campaign. Trump and Clinton utilized the opportunities afforded by social media platforms to communicate about their campaign, critique the opposing candidate, and provide an optimistic outlook about their chances of winning the election. This study considers how political candidates can use social media to increase potential supporters’ perceptions that they will win the election, providing them extra motivation to go out and vote.

Perception of a candidate’s viability can have an important effect on citizens’ voting behavior (Abramowitz, 1989; McAllister & Studlar, 1991; Utych & Kam, 2013). In a tumultuous election, apathetic partisans may look for confirmation that their candidate is likely to win. Partisans who followed Trump and/or Clinton on social media were likely to find an overly optimistic narrative. Trump’s social media messages often described himself as an outsider, looking to “drain the swamp,” “Make America Great Again” at the same time suggesting he would beat “Crooked Hillary” in spite of “Fake News” and “Rigged Elections.” Comparatively, many of Clinton’s messages maintained that Trump was unfit to be President and subtly assured followers that her presidency was inevitable.

Political candidates commonly use social media platforms to craft alternative election narratives (Howard, 2006; Kreiss, 2012). Different political narratives can affect citizens’ political perceptions and orientations, which, ultimately, provide the framework through which political action ensues (Edelman, 1988; Entman, 1993). However, the extent to which a citizen is likely to buy into a candidate’s portrayal of election outcomes should depend upon whether they share the same party affiliation. When partisans follow an in-group candidate on social media, receiving information directly from the “horse’s mouth,” they should readily accept a narrative that supports their partisan goals, leading to an increased perception that the candidate is poised to win the election. Conversely, members of the opposing party may dismiss the information or think about how to counter-argue it (see Slothuus & De Vreese, 2010). Such perceptions may be particularly important in elections where enthusiasm is low, as a citizen’s perceptions about the likelihood a candidate will win the election may help determine whether that citizen is motivated to vote for that candidate (Abramowitz, 1989; McAllister & Studlar, 1991; Utych & Kam, 2013).

This study considers the relationship between attention to social-mediated candidate communication, perceptions of candidates’ likelihood of winning an election, and intention to vote for candidates. The 2016 US Presidential Election provides a unique case study to examine these relationships. This study employs a two-wave survey with a national sample to test our expectations that following the in-group candidate (Trump and Clinton) on social media relates to intentions to vote for them through the perception that a candidate was likely to win the election. Our results show the importance of using social media to facilitate favorable attitudes among candidates’ supporters.

Literature Review

Controlling the Message: Candidate Social Media Communication

Today, citizens can choose to get their political information from a wide variety of options. Many citizens do, of course, still turn to the news media to retrieve political information (Massanari & Howard, 2011; Sydnor & Psimas, 2017). During election periods, news media cover candidates in different ways. They may report on who is ahead in polls, debates, and speeches, and sometimes to a lesser extent, policy issues. Exposure to news frames can affect citizens’ political attitudes and orientations (De Vreese, 2004; Entman, 2003) and influence how they think and feel about political issues and events (Lecheler, Keer, Schuck, & Hänggli, 2015; Pfau, Cho, & Chong, 2001; Scheufele, Kim, & Brossard, 2007; Schuck et al., 2013).

Although political candidates have limited control over how mainstream news media frame the election, candidates are increasingly using social media platforms to craft alternative election narratives (Howard, 2006; Kreiss, 2012). Candidates have adopted digital communication for strategic purposes, allowing supporters to engage with the campaign in ways beneficial to the candidate (Stromer-Galley, 2014). Truly interactive communication between candidates and constituents via social media is limited. Kreiss (2016) showed that during the 2012 Presidential Election, campaign staffers understood the “virality” of a campaign moment and disseminated social media information that would influence journalists and fulfill expectations of social media followers.

During the 2016 US Presidential Election, Donald Trump and Hillary Clinton used social media to befit their campaigns in divergent ways (Enli, 2017). Enli suggests Trump used an “amateurish” approach and turned away from the traditional norms of organization, strategy, and content, whereas Clinton’s more “professional” posts resembled established norms. Trump’s continual posts about “fake news” were a rallying point for his supporters, but he also posted contestable and sometimes categorically inaccurate content. By doing so, Trump was able to deflect the impact of criticisms of his own inaccuracies by saying news media were doing the same thing (A. S. Ross & Rivers, 2018). Importantly, candidates may come across as most honest on social media, compared with talk shows, speeches, and news interviews, primarily because they are perceived as created by the candidate, users can provide feedback, and individuals may trust candidates posting on a relatively “new” medium (Enli & Rosenberg, 2018).

Altogether, social media affords politicians the ability to frame political moments in ways that align with their campaign strategy. This may include providing an overly optimistic portrayal of their chances of winning an election, regardless of the realities of the campaign. Posting about positive poll numbers and endorsements or attacking the shortcomings of another candidate or media coverage may foster a positive feeling for those supporting the candidate. In the case of Trump, posts about “Crooked Hillary,” “Fake News,” or “Making America Great Again” likely spurred confidence and perhaps anger for his supporters. Similarly, Clinton posting about endorsements, mainstream news coverage showing her leading polls, and attacks on Trump’s temperament likely encouraged optimism about her likelihood of winning. The degree to which citizens buy into and internalize the candidate’s portrayal will depend, in larger part, on whether they are members of the same or opposing party as the candidate.

Social Media Following and Partisan Motivated Reasoning

According to partisan motivated reasoning, partisans often interpret and process information in a way that is consistent with their party identification (Druckman, Peterson, & Slothuus, 2013; Leeper & Slothuus, 2014; Taber & Lodge, 2006). Partisans tend to dismiss or counter-argue political arguments if they are inconsistent with their partisan goals. Information that supports their party, on the other hand, is often sought out, eagerly accepted, and perceived to be more convincing (Druckman et al., 2013; Kim, Taber, & Lodge, 2010; Slothuus & De Vreese, 2010). This biased processing can lead to attitude polarization, as partisans accept and internalize congruent information, while rejecting incongruent information (Taber & Lodge, 2006).

Partisan motivated reasoning can often be activated by source cues. When partisans receive information sponsored by the political in-group, they will be motivated to support the opinions, attitudes, or policy positions expressed by their party (Petersen, Skov, Serritzlew, & Ramsøy, 2013; Slothuus & De Vreese, 2010). Conversely, if the political out-group presents the same information, they would be more likely to oppose and counter-argue the information (Bolsen, Druckman, & Cook, 2014; Druckman et al., 2013). Slothuus and De Vreese (2010) show that partisans are more likely to accept and favorably assess a political frame disseminated by members of their party. Furthermore, this tendency to accept a party sponsored frame, while rejecting frames promoted by the opposition, is heightened during partisan conflict (Slothuus & de Vreese, 2010).

Social media provides ample opportunities to accept or reject the information based on partisan identification. Bode and Vraga (2015) show exposure to social media news feed posts about an issue reflecting their opinion resulted in rating it favorably. Similarly, users exposed to posts aligning with their political views are more likely to rate the post as accurate, compared with users seeing the post that did not align with their views. Posts with misinformation are also rated differently based on personal views (Kahne & Bowyer, 2017).

In the context of an election, when partisans encounter political frames sponsored by their party’s political candidate (i.e., candidate’s social media), they should be more likely to accept and internalize the arguments. When they encounter arguments sponsored by the opposing candidate, they should be more likely to either dismiss the message outright, or engage in effortful counter-arguing. Thus, when partisans are regularly exposed to messages sponsored by the in-group candidate that indicate they are likely to win the election, partisans should assimilate this perspective and grow more confident in their candidate’s prospects. Alternatively, if they see such an optimistic portrait from the opposing candidate, they are likely to reject the message as misguided, perhaps becoming even more confident that the perspective is wrong (i.e., attitude-polarization; see Taber & Lodge, 2006). Based on this rationale, we predict:

H1: Partisans will be more likely to believe the in-group candidate will win the election when following that candidate on social media.

H2: Partisans will be less likely to believe the out-group candidate will win the election when following that candidate on social media.

Perceived Likelihood of Winning Election

Believing a candidate will win an election can have an important effect on citizens’ voting behavior during elections (Abramowitz, 1989; Miller, Wang, Kulkarni, Poor, & Osherson, 2012). When candidates are perceived to be viable, individuals are more likely to seek out information about that candidate, to favorably evaluate that candidate, and to vote for that candidate (Utych & Kam, 2013). It has long been argued that citizens are more likely to join the “bandwagon” and vote for candidates that they believe are doing well in the polls (McAllister & Studlar, 1991). Polls can encourage individuals to follow the majority opinion, while discouraging those with minority opinions (Rothschild & Malhotra, 2014). Crucially, citizens’ perceptions about candidates’ electoral prospects are more influential on their likelihood of voting than the actual reality of the electoral competition (McDonald & Tolbert, 2012).

Therefore, the degree to which a candidate can affect a citizen’s perceptions about the likelihood that they will win the election may help determine whether the citizen is motivated to vote for that candidate. These perceptions should be highly influenced by the political information a citizen has been exposed to. Politicians, news media, and partisans can discuss polling numbers, political momentum, public opinion, and recent events in very different ways. Thus, an individual’s exposure to a candidate’s social media communication can have important consequences for their perception of what will happen in an upcoming election. Candidates’ social media posts and users’ attitudes developed after seeing them likely carries over to participating politically, including voting (Towner & Dulio, 2011; Williams & Gulati, 2008). We therefore predict:

H3: Perceived likelihood that a candidate will win the election will be related to higher intention to vote for that candidate.

Finally, based on the predicted chain of events hypothesized above, we also predict the following indirect effects:

H4: Partisans who follow the in-group candidate on social media will be more likely to vote for that candidate through increased perceptions that the candidate will win the election.

H5: Partisans who follow the out-group candidate on social media will be less likely to vote for that candidate through decreased perceptions that the candidate will win the election.

Method

Data

To test our hypothesized model, we employed a two-wave national panel. Participants were recruited by Survey Sampling International (SSI). SSI recruits potential participants from its panels, online communities, social media, and websites and compensates them via cash, prizes, charitable donations, and various other incentives. 1201 US adults were initially recruited over a 1-week period in early October 2016. Participants who completed the survey were invited to participate in a follow-up survey during the first week of November 2016. A total of 498 participants completed the second survey. The sample was 45.6% Democrat, 25.5% Republican, and 27.1% independent. Our analysis focused on Democrats and Republicans, giving us a final sample of 354 participants (64.7% Democrat and 35.3% Republican).

Measures

Candidate Social Media Following

The degree to which participants followed each presidential candidate on social media was measured by asking respondents: “How often do you use the following sources to get news and information about the 2016 presidential election: Facebook page, Twitter account, Snapchat account, Instagram account, and personal website” (1 = never; 7 = very frequently). Participants’ scores for each candidate’s social media sites were merged to create a social media following variable. Participants’ scores for following Hillary Clinton (Cronbach’s α = .98; M = 2.51; SD = 2.05) or Donald Trump (Cronbach’s α = .97; M = 2.20; SD = 1.81) were then matched to partisan affiliation to create in-group and out-group social media following variables. That is, the in-group social media following variable (M = 3.01; SD = 2.22) included Democrats’ scores for Clinton following and Republicans’ scores for Trump following; the out-group social media following variable (M = 2.22; SD = 1.89) included Democrats’ scores for Trump following and Republican’ scores for Clinton following. To note, the large standard deviations for these variables suggest that while some respondents likely don’t get news and information from candidate’s social media accounts, others do at least occasionally and still others somewhat frequently.

Perceiving Likelihood of Candidate Winning Election

To measure respondents’ perception that the in-group or out-group candidate would likely win the 2016 US Presidential Election, participants were asked, “Just your best guess—how likely do you think it is that the following candidates will win the presidential election?”(1 = very unlikely; 7 = very likely). Respondents reported believing Clinton had a better chance to win the election (Wave 1: M = 5.10; SD = 1.78; Wave 2: M = 5.05; SD = 1.91) than did Trump (Wave 1: M = 3.68; SD = 1.99; Wave 2: M = 3.76; SD = 2.03). Perceiving Likelihood of the In-group Candidate Winning (Wave 1: M = 5.60; SD = 1.55; Wave 2: M = 5.60; SD = 1.64) was constructed by including Democrats’ perception that Clinton would win the election and Republicans’ perception that Trump would win. Perceiving Likelihood of the Out-group Candidate Winning (Wave 1: M = 3.32; SD = 2.00; Wave 2: M = 3.38; SD = 1.99) was constructed by including Democrats’ perception that Trump would win the election and Republicans’ perception that Clinton would win.

Likelihood of Voting for Candidate

Our dependent variable measured the likelihood that a participant would vote for either the in-group or out-group candidate. Participants were asked, “How likely are you to vote for the following presidential candidates?” (1 = very unlikely; 7 = very likely). Overall, participants expressed a greater likelihood of voting for Hillary Clinton (Wave 1: M = 4.45; SD = 2.55; Wave 2: M = 4.41; SD = 2.61) than Donald Trump (Wave 1: M = 3.40; SD = 2.50; Wave 2: M = 3.48; SD = 2.54). Likelihood of Voting for the In-group Candidate (Wave 1: M = 5.97; SD = 1.65; Wave 2: M = 5.85; SD = 1.86) was constructed by combining Democrats’ expressed likelihood of voting for Clinton and Republicans’ expressed likelihood of voting for Trump. Likelihood of Voting for the Out-group Candidate (Wave 1: M = 2.26; SD = 2.00; Wave 2: M = 2.34; SD = 1.99) was constructed by combining Democrats’ expressed likelihood of voting for Trump and Republicans’ expressed likelihood of voting for Clinton.

Controls

Candidate Favorability

To account for the influence of pre-existing support for a candidate, we controlled for how favorably participants felt toward the candidates at Wave 1. Participants’ favorability toward the in-group and out-group candidates was measured with a 100-item slider scale where 0 represented unfavorable evaluations and 100 represented favorable evaluations. Participants were asked how favorably they felt about Clinton (M = 48.61, SD = 36.76) and Trump (M = 35.75, SD = 35.69). In-group Candidate Favorability (M = 69.81, SD = 30.02) was constructed by combining Democrats’ favorability toward Clinton with Republicans favorability toward Trump. Out-group Candidate Favorability (M = 20.74, SD = 27.84) was constructed by combining Democrats’ favorability toward Trump with Republicans favorability toward Clinton.

Demographics

Demographic control variables included age (M = 40.22; SD = 14.66), gender (41.6% female), race (86.7% Caucasian), education level (some high school = 1.6%, high school/GED equivalent = 17.6%, some college = 21.9%, college graduate = 39.1%, and postgraduate degree = 19.8%), and income level (less than US$29,999 = 18.3, US$30,000 to US$59,999 = 26.4%, US$60,000 to US$99,999 = 31.8%, US$100,000 to US$149,999 = 14.5%, and US$150,000 or more = 9.1%).

News Media Exposure

Participants were asked whether they read or watch news from a wide range of sources (1 = never; 7 = very frequently). Indices were created for mainstream news, conservative news, and liberal news. Mainstream news included national newspapers, local newspapers, national television news broadcast, local television news broadcast, online national TV news websites, online local news websites, online national newspaper websites, and online news magazine websites (Cronbach’s α = .93; M = 4.31; SD = 1.70). Conservative news included cable news programs on Fox News and conservative news websites (Spearman–Brown = .62, M = 3.73; SD = 2.06). Liberal news included cable news programs on MSNBC, cable news programs on CNN, and liberal news websites (Cronbach’s α = .89; M = 3.84; SD = 2.02)

Political Orientation

We also controlled for participants’ general political orientation. These variables included ideology (1 = very liberal; 7 = very conservative; M = 4.08; SD = 1.82), and political interest (“How interested are you in what’s going on in government and politics?” [1 = not at all interested; 7 = extremely interested; M = 5.61; SD = 1.44]).

Analytic Technique

Analysis was performed using the SPSS Macro PROCESS Model 4 (Hayes, 2012), which examines the relationship between an X variable (social media following), a Y variable (intention to vote), and an M (mediating) variable (perceived likelihood candidate will win election). We used an autoregressive model, where the Wave 1 score on social media following was used to predict the Wave 2 scores for intention to vote and perceived likelihood a candidate would win the election, with the Wave 2 measures (T2) for voting intentions and perceived likelihood of winning regressed on the Wave 1 measures (T1) to to control for the influence of the prior levels of the variables (Maxwell, Cole, & Mitchell, 2011). The model also included candidate favorability at Wave 1, along with several demographic and political orientation control variables. Figure 1 displays our analytic model.

Figure 1.

Conceptual model for following political candidate on social media.

Prior to analysis, all variables were standardized so that the resulting regression coefficient can be interpreted as a fully standardized score. Indirect effects were examined using a bootstrap analysis with 5,000 bootstrap samples and a 95% confidence interval (CI). Using this method, when an indirect path’s CI does not overlap zero (e.g., .3 to .5) it is considered significant at p < .05, and if it does overlap zero (e.g., −.3 to .5) it is considered nonsignificant.

Results

Likelihood of Voting for In-Group Candidate

Our first hypothesis predicted that following the in-group candidate on social media would lead to a higher perception that they were likely to win the election. As shown in Table 1, following the in-group candidate on social media is significantly related to perceiving that he or she will win the presidential election (β = .25; p = .002). Thus, H1 was supported.

Table 1.

Following In-Group Candidate on Social Media Model.

	Perceived likelihood of candidate being elected^T2	Likelihood of voting for candidate^T2
	β(SE)	β(SE)
Age	.02 (.06)	−.05 (.03)
Gender	.03 (.05)	.00 (.03)
Race	−.05 (0.5)	.03 (.03)
Education	−.02 (.05)	−.01 (.03)
Income	.06 (.05)	−.02 (.03)
Ideology	−.04 (.05)	.01 (.02)
Political interest	.00 (.06)	−.03 (.03)
Mainstream news	.07 (.10)	−.05 (.03)
Liberal news	−.12 (.10)	.03 (.05)
Conservative news	−.02 (.08)	.08 (.05)
Candidate favorability	.20 (.07) **	.13 (.04) *
Perceived likelihood candidate elected^T1	.32 (.06) ***	−.12 (.03) **
Likelihood of voting for candidate^T1	.12 (.06) #	.55 (.04) ***
Candidate social media following^T1	.25 (.07) **	−.14 (.05) *
Perceived likelihood candidate elected^T2	–	.40 (.03) ***
Following^T2 → likelihood elected^T2	–	.10 (.03) *
Total effect following^T2	–	−.04 (.07)

< .10.

p < .05

p < .01.

***

p < .001.

Next, we hypothesized that citizens who believed a candidate was going to win the election would be more likely to express intentions to vote for that candidate. As expected, we found that those who believed the in-group candidate would be elected were significantly more likely to vote for the in-group candidate (β = .40; p < .001). We therefore find support for H3.

Finally, as expected, we find a significant indirect effect from following the in-group candidate on social media on likelihood of voting for the in-group candidate through perceived likelihood that candidate would be elected b = 0.10, SE = 0.03 (CI: 0.04, 0.17), p < .05. Thus, H4 was supported.

Likelihood of Voting for the Out-Group Candidate

Our second hypothesis predicted that following the out-group candidate on social media would lead to a decreased perception that the candidate would win the election. As seen in Table 2, following the out-group candidate on social media was not significantly related to perceiving that he or she will win the presidential election (β = .03; p = .656). H2 was not, therefore, supported.

Table 2.

Following Out-Group Candidate on Social Media Model.

	Perceived likelihood of candidate being elected^T2	Likelihood of voting for candidate^T2
	β(SE)	β(SE)
Age	−.06 (.06)	.01 (.04)
Gender	.02 (.05)	−.03 (.04)
Race	.06 (.05)	.04 (.04)
Education	.02 (.05)	.02 (.04)
Income	−.01 (.06)	−.05 (.04)
Ideology	.07 (.05)	.01 (.03)
Political interest	.01 (.06)	−.06 (.04)
Mainstream news	.04 (.10)	.19 (.07) **
Liberal news	−.04 (.11)	−.11 (.07)
Conservative news	−.02 (.09)	−.02 (.06)
Candidate favorability	−.01 (.08)	.09 (.05)
Perceived likelihood candidate elected^T1	.42 (.06) ***	−.06 (.04)
Likelihood of voting for candidate^T1	.29 (.08) ***	.64 (.05) ***
Candidate social media following^T1	.03 (.07)	.06 (.05)
Perceived likelihood candidate elected^T2	–	.14 (.04) ***
Following^T2 → likelihood elected^T2	–	.00 (.01)
Total effect following^T2	–	.06 (.05)

< .10.

p < .05.

p < .01.

***

p < .001.

Next, we see that those who believed the out-group candidate would be elected (β = .14; p < .001) were significantly more likely to vote for the out-group candidate. This provides additional support for H3.

Finally, there was not a significant indirect effect from following the out-group candidate on social media on likelihood of voting for the out-group candidate through perceived likelihood that the candidate would be elected b = 0.00, SE = 0.01 (CI: − 0.02, 0.04), ns. We therefore fail to find support for H5.

Discussion

Candidates’ social media use for campaign communication continues to evolve, and its role in influencing electoral politics is undeniable. Candidates can control the message they communicate to supporters via these media, painting an overwhelmingly positive and optimistic portrayal of their candidacy. Doing so is important to contrast other media coverage that may provide a more negative portrayal of their political campaign. Indeed, both mainstream news sources as well as partisan sources influence how voters perceive the candidates and their campaigns. Our study considers how during a period shortly before the 2016 US Presidential Election individuals’ attitudes and voting intentions were influenced by Hillary Clinton’s and Donald Trump’s social media posts. The results suggest that following these candidates relates to intending to vote for the candidate through the increased belief that the candidate would win. However, this mediation effect occurs for only supporters of Trump or Clinton, but not for partisans of the opposing party.

Results of our study show that when partisans follow the in-group candidate, they become more likely to perceive the candidate will win the election, which increases their intentions to vote for the candidate. When partisans follow the opposing candidate, on the other hand, they do not appear to buy into the narratives being promoted by that candidate—as following is not related to perception that the candidate will win the election. These results are consistent with partisan motivated reasoning, which suggests that partisans are willing to quickly accept and internalize the political frames promoted by the in-group party, while resisting the messages promoted by the out-group party (Petersen et al., 2013; Slothuus & De Vreese, 2010). We believe this is important, contributing to the literature signifying the power social-mediated campaign communication plays in influencing attitudes in a complex news media environment (Bimber, 2014; K. Ross & Bürger, 2014).

The way that various sources explain electoral politics can have important implications for citizen attitudes. News media play an important role in explaining politics, oftentimes reporting on current events in ways that contribute to individuals’ attitudes about those issues. This explanation tends to be repeated in news media over time, leading individuals to understand complex issues in relatively simple ways. When specific coverage puts candidates in a negative light (e.g., they are likely to lose or a controversy is likely to hurt them), candidates must attempt to refocus on positive characteristics about their candidacy, highlighting specific themes, topics, and interpretations that favor them. Although candidates have limited control of how mainstream news media report about them, social-mediated communication provides an opportunity for candidate to reshape the election narrative (Kreiss, 2012), potentially providing some much needed optimism and enthusiasm for partisans who may be feeling apathetic about their party’s candidate.

This study demonstrates partisans who followed Donald Trump or Hillary Clinton on social media had different perspectives of the 2016 Presidential Election. The election included Trump, who wanted to “drain the swamp” of political elites in federal government. Despite Trump’ populist appeal, mainstream news media consistently reported on public opinion polling showing him losing to Clinton. Via social media, especially on Twitter, Trump lambasted “fake news” reports on these polls and assured his followers he would win the election. He also repeatedly tweeted about how “crooked Hillary” was unfit to be President and that he would “make American great again.” These messages, we argue, created a distinct narrative for his supporters who followed him on social media—one where Trump’s chances of being elected were far greater than reported in the news media. Supporters who followed Trump saw consistent information: mainstream news media couldn’t be trusted and were attempting to ruin his candidacy, he was a better choice for president compared with Hillary Clinton, and he would win the election.

Clinton also used social media to control messages about her campaign and critique Trump’s candidacy. To be sure, these messages likely enhanced positive feelings for followers who supported her. Clinton used social media to promote herself as a conventional candidate for president and critique Trump as an unimaginable candidate. She also used social media to supplement mainstream news media’s reporting on her ahead in the polls—helping to reinforce beliefs that her election was inevitable. Clinton’s supporters saw consistent and similar messages via her social-mediated campaign communication and mainstream news media. From this perspective, not only were the polls correct that Clinton would win the election, but the alternative (Trump winning) was likely interpreted as out of the realm of possibility.

The mediation role of perceiving the candidate would win the election mattered for those from the same party of either Trump or Clinton, not those opposing party’s candidate. We did expect that partisans who followed the opposing candidate on social media would engage in effortful counter-arguing (e.g., finding reasons to believe that candidate was unelectable), which would actually make them more confident that the opposing candidate would lose the election. While we did not find a significant negative relationship between following the out-group candidate and perceived likelihood of winning, the lack of a significant relationship still implies that, at the very least, partisans from the opposing party are not likely to accept the narratives promoted by the out-group candidate.

Thus, our results provide evidence suggesting partisans interpret and process political candidates’ social media messages in ways consistent with their party identification. Clinton’s traditional, professional use of social media such as posting endorsements, popular public opinion polls showing her in the lead, and other mainstream news media reports likely did not alter attitudes for those reporting affiliation with the Republican Party, even if they were skeptical of Trump as President. These followers may have perceived politics as usual or a perpetuation of the mainstream Trump lambasted. Likewise, those following Trump but not supporting him may have done so for the spectacle of their perceived amateurization of social media use as well as to confirm their beliefs he was unfit for President. These results support research showing people tend to accept messages from their in-group and reject messages from the out-group, especially during partisan conflict (Slothuus & De Vreese, 2010), including on social media (Bode & Vraga, 2015; Kahne & Bowyer, 2017).

Importantly, we employ a two-wave panel approach using an autoregressive model in an attempt to draw causal conclusions like other research examining similar relationships (Dimitrova, Shehata, Strömbäck, & Nord, 2014; Kruikemeier & Shehata, 2017; Maxwell et al., 2011). We show that during the final month of the campaign, the candidates’ social media messages affected their followers’ attitudes and subsequently, their voting intentions. As mainstream news media continued to report about the election in similar ways, the candidates’ social media messages simultaneously continued to influence their supporters’ attitudes. Trump’s social media posts against the mainstream and Clinton’s posts with the mainstream mattered. The two-wave panel approach allows us to be more confident that there is a causal effect of the candidates’ message dissemination on individuals’ political behavior than could be inferred from cross-sectional date. In addition, by controlling for candidate favorability, we are able to account for the high correlation between following a candidate and perceiving they will win the election and/or being likely to vote for them.

At the same time, our study does have some notable limitations. One main limitation revolves around variable operationalization. The study does not measure actual voting behavior, rather only intentions to vote. Although we anticipate voting intentions to be similar voting, especially because we asked days before the election, we cannot rule out respondents changing their vote or not voting at all. Another potential limitation is that we measured social media following as a whole, limiting the understanding of the extent to which sites like Facebook and Twitter and content may play different roles in campaign communication. The nuanced nature of social media suggests that there is value in examining platform-specific and content-specific effects on political attitudes and behavior. These distinctions may be particularly important given that candidates can use social media platforms in different ways. Readers should also note some respondents rarely get news and information from the candidates’ social media accounts, but many others do so at least occasionally.

Furthermore, political campaigns are increasingly using sophisticated data analytics to direct targeted messages to potential voters (Bimber, 2014; Kreiss, 2016; Stromer-Galley, 2014). Indeed, algorithms often determine which messages a citizen will encounter on a specific social media platform (Yeo, Cacciatore, & Scheufele, 2015). This is all to say that we cannot assume that by “following candidates on social media” partisans were encountering the same messages. The very high reliability scores for our social media following variables, however, do provide some degree of confidence that citizens who do follow a candidate on social media are likely to follow them on multiple platforms, leading to exposure to an overarching narrative. Furthermore, we should expect that these messages generally contain an optimistic and overly positive portrayal of the candidate’s campaign, regardless of message personalization. However, the study did not measure the actual message frames; interpretation of the results should be in the realm of patterns of information gathering. Finally, the low attrition rate for our sample is a limitation, as is our over-representation of Democrats. However, because our expectations were based on partisan motivated reasoning, Democrats and Republicans should have the same underlying motivations for how they process and interpret messages from the in-group and out-group candidates.

Despite these limitations, our findings provide insight into an important mechanism by which candidate social media communication influences election outcomes. Information from individuals’ social network is vital for their understanding and attitudes about politics, and these attitudes influence participating in democracy (Ancu & Cozma, 2009; Blumler & Kavanagh, 1999; Serazio, 2014; Xenos & Foot, 2005). Because candidates need to convince citizens to vote for them and citizens get information from a variety of sources, social media provides candidates an opportunity to persuade followers about their campaign and elections as they see fit (Gerodimos & Justinussen, 2015). Doing so is important when other information sources, including mainstream news media, discuss a candidate’s campaign is ways not consistent with the candidate’s message. A candidate’s social media campaign communication can combat this information, leading to a narrative that can help being elected.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Nathian Shae Rodriguez

Author Biographies

Timothy Macafee (PhD University of Wisconsin, Madison) is an assistant professor of Communication at Concordia University Wisconsin. His research interests include uses of social media for political information gathering and expression and how these information behaviors relate to political attitudes and participation.

Bryan McLaughlin (PhD University of Wisconsin, Madison) is an assistant professor of Advertising at Texas Tech University. His research interests include how social identity categories are constructed and maintained through communication, how these social identity categories are processed cognitively, and the effects of this processing on sociopolitical attitudes and behaviors.

Nathian Shae Rodriguez (PhD Texas Tech University) is an assistant professor of Digital Media at San Diego State University. His research interests include minority representation in media, specifically LGBTQ and Latinx portrayals and identity negotiation.

References

Abramowitz

A. I.

(1989). Viability, electability, and candidate choice in a presidential primary election: A test of competing models. The Journal of Politics, 51, 977–992.

Ancu

Cozma

(2009). MySpace politics: Uses and gratifications of befriending candidates. Journal of Broadcasting & Electronic Media, 53, 567–583.

Bimber

(2014). Digital media in the Obama campaigns of 2008 and 2012: Adaptation to the personalized political communication environment. Journal of Information Technology & Politics, 11, 130–150.

Blumler

J. G.

Kavanagh

(1999). The third age of political communication: Influences and features. Political Communication, 16, 209–230.

Bode

Vraga

E. K.

(2015). In related news, that was wrong: The correction of misinformation through related stories functionality in social media. Journal of Communication, 65, 619–638.

Bolsen

Druckman

J. N.

Cook

F. L.

(2014). The influence of partisan motivated reasoning on public opinion. Political Behavior, 36, 235–262.

De Vreese

C. H

. (2004). The effects of strategic news on political cynicism, issue evaluations, and policy support: A two-wave experiment. Mass Communication & Society, 7, 191–214. doi:10.1207/s15327825mcs0702_4

Dimitrova

D. V.

Shehata

Strömbäck

Nord

L. W.

(2014). The effects of digital media on political knowledge and participation in election campaigns: Evidence from panel data. Communication Research, 41, 95–118.

Druckman

J. N.

Peterson

Slothuus

(2013). How elite partisan polarization affects public opinion formation. American Political Science Review, 107, 57–79.

10.

Edelman

(1988). Constructing the political spectacle. Chicago, IL: The University of Chicago Press.

11.

Enli

(2017). Twitter as arena for the authentic outsider: Exploring the social media campaigns of Trump and Clinton in the 2016 US presidential election. European Journal of Communication, 32, 50–61.

12.

Enli

Rosenberg

L. T.

(2018). Trust in the age of social media: Populist politicians seem more authentic. Social Media + Society, 4(1), 1-11. doi:10.1177/2056305118764430

13.

Entman

R. M.

(1993). Framing: Towards clarification of a fractured paradigm. Journal of Communication, 43, 51–58. doi:10.1111/j.1460-2466.1993.tb01304.x

14.

Entman

R. M.

(2003). Cascading activation: Contesting the White House’s frame after 9/11. Political Communication, 20, 415–432. doi:10.1080/10584600390244176

15.

Gerodimos

Justinussen

(2015). Obama’s 2012 Facebook campaign: Political communication in the age of the like button. Journal of Information Technology & Politics, 12, 113–132.

16.

Hayes

A. F.

(2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. http://www.afhayes.com/public/process2012.pdf

17.

Howard

P. N.

(2006). New media campaigns and the managed citizen. New York, NY: Cambridge University Press.

18.

Kahne

Bowyer

(2017). Educating for democracy in a partisan age: Confronting the challenges of motivated reasoning and misinformation. American Educational Research Journal, 54(1), 3–34.

19.

Kim

S. Y.

Taber

C. S.

Lodge

(2010). A computational model of the citizen as motivated reasoner: Modeling the dynamics of the 2000 presidential election. Political Behavior, 32(1), 1–28.

20.

Kreiss

(2012). Acting in the public sphere: The 2008 Obama campaign’s strategic use of new media to shape narratives of the presidential race. Research in Social Movements, Conflicts and Change, 33, 195–223.

21.

Kreiss

(2016). Seizing the moment: The presidential campaigns’ use of Twitter during the 2012 electoral cycle. New Media & Society, 18, 1473–1490. doi:10.1177/1461444814562445

22.

Kruikemeier

Shehata

(2017). News media use and political engagement among adolescents: An analysis of virtuous circles using panel data. Political Communication, 34, 221–242.

23.

Lecheler

Keer

Schuck

A. R.

Hänggli

(2015). The effects of repetitive news framing on political opinions over time. Communication Monographs, 82, 339–358. doi:10.1080/03637751.2014.994646

24.

Leeper

T. J.

Slothuus

(2014). Political parties, motivated reasoning, and public opinion formation. Political Psychology, 35, 129–156.

25.

Massanari

A. L.

Howard

P. N.

(2011). Information technologies and omnivorous news diets over three US presidential elections. Journal of Information Technology & Politics, 8, 177–198.

26.

Maxwell

S. E.

Cole

D. A.

Mitchell

M. A.

(2011). Bias in cross-sectional analyses of longitudinal mediation: Partial and complete mediation under an autoregressive model. Multivariate Behavioral Research, 46, 816–841.

27.

McAllister

Studlar

D. T.

(1991). Bandwagon, underdog, or projection? Opinion polls and electoral choice in Britain. The Journal of Politics, 53, 720–741.

28.

McDonald

M. P.

Tolbert

C. J.

(2012). Perceptions vs. actual exposure to electoral competition and effects on political participation. Public Opinion Quarterly, 76, 538–554.

29.

Miller

M. K.

Wang

Kulkarni

S. R.

Poor

H. V.

Osherson

D. N.

(2012). Citizen forecasts of the 2008 US presidential election. Politics & Policy, 40, 1019–1052.

30.

Petersen

M. B.

Skov

Serritzlew

Ramsøy

(2013). Motivated reasoning and political parties: Evidence for increased processing in the face of party cues. Political Behavior, 35, 831–854.

31.

Pfau

Cho

Chong

(2001). Communication forms in U.S. presidential campaigns: Influences on candidate perceptions and the democratic process. The International Journal of Press/Politics, 6, 88–105.

32.

Ross

A. S.

Rivers

D. J.

(2018). Discursive deflection: Accusation of “fake news” and the spread of mis- and disinformation in the tweets of president Trump. Social Media + Society, 4(2). doi:10.1177/2056305118776010

33.

Ross

Bürger

(2014). Face to face (book) social media, political campaigning and the unbearable lightness of being there. Political Science, 66, 46–62.

34.

Rothschild

Malhotra

(2014). Are public opinion polls self-fulfilling prophecies? Research & Politics, 1(2) 1-12. doi:10.1177/2053168014547667.

35.

Saad

(2016). Trump and Clinton finish with historically poor images. Retrieved from https://news.gallup.com/poll/197231/trump-clinton-finish-historically-poor-images.aspx

36.

Scheufele

D. A.

Kim

Brossard

(2007). How split-screen debate coverage influences evaluation of presidential debates. Communication Research, 34(1), 3–24.

37.

Schuck

A. R.

Vliegenthart

Boomgaarden

H. G.

Elenbaas

Azrout

van Spanje

De Vreese

C. H.

(2013). Explaining campaign news coverage: How medium, time, and context explain variation in the media framing of the 2009 European parliamentary elections. Journal of Political Marketing, 12(1), 8–28.

38.

Serazio

(2014). The new media designs of political consultants: Campaign production in a fragmented era. Journal of Communication, 64, 743–763. doi:10.1111/jcom.12078

39.

Silver

(2016). Election update: Where are the undecided voters? Retrieved from https://fivethirtyeight.com/features/election-update-where-are-the-undecided-voters/

40.

Slothuus

De Vreese

C. H.

(2010). Political parties, motivated reasoning, and issue framing effects. The Journal of Politics, 72, 630–645.

41.

Stromer-Galley

(2014). Presidential campaigning in the Internet age. Oxford, UK: Oxford University Press.

42.

Sydnor

Psimas

(2017). Easing political digestion: The effects of news curation on citizens’ behavior. Journal of Information Technology & Politics, 14, 189–213.

43.

Taber

C. S.

Lodge

(2006). Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science, 50, 755–769.

44.

Towner

T. L.

Dulio

D. A.

(2011). An experiment of campaign effects during the YouTube election. New Media & Society, 13, 626–644.

45.

Utych

S. M.

Kam

C. D.

(2013). Viability, information seeking, and vote choice. The Journal of Politics, 76, 152–166.

46.

Williams

C. B.

Gulati

(2008). The political impact of Facebook: Evidence from the 2006 midterm elections and 2008 nomination contest. Politics and Technology Review, 1(1), 11–24.

47.

Xenos

M. A.

Foot

K. A.

(2005). Politics as usual, or politics unusual? Position taking and dialogue on campaign websites in the 2002 US elections. Journal of Communication, 55, 169–185.

48.

Yeo

S. K.

Cacciatore

M. A.

Scheufele

D. A.

(2015). News selectivity and beyond: Motivated reasoning in a changing media environment. In Jandura

Petersen

Mothes

Schielicke

(Eds.), Publizistik und Gesellschaftliche Verantwortung (pp. 83–104). Wiesbaden, Germany: Springer VS.

Winning on Social Media: Candidate Social-Mediated Communication and Voting During the 2016 US Presidential Election

Abstract

Keywords

Introduction

Literature Review

Controlling the Message: Candidate Social Media Communication

Social Media Following and Partisan Motivated Reasoning

Perceived Likelihood of Winning Election

Method

Data

Measures

Candidate Social Media Following

Perceiving Likelihood of Candidate Winning Election

Likelihood of Voting for Candidate

Controls

Candidate Favorability

Demographics

News Media Exposure

Political Orientation

Analytic Technique

Results

Likelihood of Voting for In-Group Candidate

Likelihood of Voting for the Out-Group Candidate

Discussion

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iD

Author Biographies

References