Sage Journals: Discover world-class research

Abstract

The influence of partisan identification infiltrates all aspects of a democracy. This study employs an innovative design to explore the presidential debate-viewing experience among young citizens. Data were collected from across the United States for all three 2016 presidential debates between Hillary Clinton and Donald Trump using pretest/posttest surveys and debate viewers’ Twitter posts. Examining Twitter expression as a form of political elaboration, the study employs a social identity theoretical perspective to better understand the process through which viewers form political attitudes. Applying the theory of identity-motivated elaboration (TIME) to presidential debates, the current research illuminates how partisan social identification changes the way viewers think about political issues and, resultantly, evaluate candidates and form political opinions. A strong partisan social identification results in greater identity-consistent elaboration and Twitter expression throughout one’s presidential debate viewing, which results in more biased candidate image evaluations and, subsequently, stronger preference for the in-party candidate.

Keywords

Twitter identity-motivated elaboration partisan social identity presidential debates political uses of social media

The 2016 presidential campaign was compelling not only for its surprise outcome but also for the use of digital technologies that spurred greater political communication. While televised presidential debates have occurred for decades, the use of social media in conjunction with debate viewing is a more recent phenomenon that has emerged over the past few election cycles in the United States and internationally (Chadwick et al., 2017; Freelon & Karpf, 2015; Houston, Hawthorne, et al., 2013; Houston, McKinney, et al., 2013; Wells et al., 2016). As debate viewers witness the back-and-forth dialogue between candidates, they are now also active content creators, sending political commentary into the digital realm that becomes part of the overall debate narrative (Driscoll et al., 2018; Trilling, 2015). In fact, a record-breaking 17.1 million debate-related messages were posted to Twitter during the first debate between Hillary Clinton and Donald Trump in 2016 (Nielson’s Social Content Rating, 2016). As opposed to engaging televised debates as politically neutral observers, many viewers tune in to the debates because they are politically motivated and active partisans (McKinney & Warner, 2013). Through Twitter, citizens are coopting and coproducing meaning, and political actors are actively trying to capitalize on the split screening of political debates (Wells et al., 2016). As such, questions arise as to how live commentators’ political attitudes are shaped not only because of their partisan predispositions but also because of their engagement with social media while debate watching.

In fact, 2016 saw the largest viewership for any general-election debate series in the history of US televised presidential debates (McKinney, 2018). Reasons undoubtedly included the novelty and entertainment value of Donald J. Trump as a candidate and also the presence of Hillary Rodham Clinton, the first woman presidential candidate on the general-election debate stage. Moreover, Twitter in particular as a social media platform allowed debate viewers to discuss candidate debate performances in real time (Hawthorne et al., 2013). Importantly, the hours including the televised presidential debates became the most-tweeted period during the 2016 general campaign season (Jarvey, 2016). It is therefore crucial that we examine how such digital communication phenomena shape the political attitudes of presidential debate viewers.

In this study, we examine the effects of live-tweeting during the 2016 presidential debates. Although other social media are used to second-screen for political news in general (Gil de Zúñiga et al., 2015; Park & Kaye, 2017), second-screening on Twitter during televised debates is a particular phenomenon that has been studied as a cognitive and communicative behavior that informs, persuades, and engages users (Chadwick et al., 2017; Freelon & Karpf, 2015; Houston, Hawthorne, et al., 2013; Houston, McKinney, et al., 2013; McKinney et al., 2014; Wells et al., 2016; Zheng & Shahin, 2020). We begin our examination by reviewing the relevant literature on presidential debates, providing an overview of our theoretical perspective, identity-motivated elaboration, and a review of the extant research at the intersection of elaboration and political attitudes. We next summarize our study design and methodology, followed by presentation of our results. We finish by discussing the implications of our findings for digital technologies on political campaign communication in general and televised presidential debates in particular.

Presidential Debates

Presidential debates represent focal points in the electoral process for informing and persuading voters (McKinney & Carlin, 2004). Debates engage voters psychologically and provide a process for evaluating candidates. This cognitive engagement can occur through on-line or memory processing (Kim & Garrett, 2012), or through elaboration. Importantly, greater cognitive engagement, by more carefully examining candidates’ arguments or paying close attention to how debate messages shape candidate image, can increase the likelihood of persuasion (Warner et al., 2018). Cognitive engagement can be driven through a variety of paths to involvement in a message (H. Cho & Boster, 2005), and in the context of political persuasion, one of these paths can be partisan social identity (Greene, 1999; Jennings, 2019; Mason & Wronski, 2018; Settle, 2018).

Debate viewers evaluate candidates for their issue stances and image characteristics. Presidential debate research, in fact, has long been characterized by a bifurcated focus on the effects of debate viewing on perceptions and knowledge of candidate image and policy issues (Lanoue & Schrott, 1991; McKinney & Carlin, 2004). Meta-analyses have illustrated that debate viewing induces both image and issue effects (Benoit et al., 2003; Benoit & Hansen, 2004). Different studies, however, have produced differing results on the extent to which image or issue has stronger effects (McKinney et al., 2003; Zhu et al., 1994), and other scholars have argued against the existence of an image–issue dichotomy (e.g., Parry-Giles, 2010). Regardless, the information-rich televised debate message provides voters a prime opportunity to assess both candidate image and candidate issue stances.

Live-Tweeting as Cognitive Engagement

Viewers of televised debates have long participated in “social-watching” behavior, which is watching and discussing a debate with others (Fein et al., 2007). Over the past few election cycles, both internationally and in the United States, this social watching has extended to online environments (Chadwick et al., 2017; McKinney et al., 2014; Thorson et al., 2015; Wells et al., 2016). Debate viewers now increasingly engage in second-screening while watching televised debates, a behavior that engages viewers cognitively and communicatively as they process information presented from televised debates (Bramlett et al., 2018; Chadwick et al., 2017; Jennings et al., 2017; Wells et al., 2016).

Voters now engage in second-screening for a range of purposes related to politics such as live-tweeting debates or seeking political news (Barnidge et al., 2017; Gil de Zúñiga et al., 2015). Second-screening can produce political persuasion (Barnidge et al., 2017), online participation (Gil de Zúñiga et al., 2015; McGregor & Mourão, 2017), and more attention toward and engagement with televised political events (Chadwick et al., 2017; Houston, Hawthorne, et al., 2013). Second-screening is both a cognitive and expressive process (Chadwick et al., 2017; J. Cho et al., 2018; Nekmat, 2012; Valkenburg, 2017) and also a process where elaboration occurs (McGregor & Mourão, 2017; Nekmat, 2012). To produce a tweet in the debate-viewing context, for example, users must pay attention to the ongoing debate message, think about information from the debate that one wishes to comment on or respond to, form one’s thought into a tweet, and then complete the process of sharing the tweet in a public arena. Multiple studies have found that viewers of televised debates who live-tweeted the debate did feel more cognitively engaged with the debate content (Chadwick et al., 2017; Houston, Hawthorne, et al., 2013). This notion of greater cognitive engagement is also supported by research that shows live-tweeting debates are associated with higher post-debate information acquisition (Bramlett et al., 2018; Houston, McKinney, et al., 2013; Jennings et al., 2017). However, these findings have been contested, with some research showing that live-tweeting, as a form of media multitasking, distracts from learning (Gottfried et al., 2017). Regardless of the outcome, the behavior of live-tweeting requires cognitive expenditure. The process of live-tweeting could therefore be viewed as a cognitive process and as a proxy for information processing (Bramlett et al., 2018) that can be motivated by partisan identity (J. Cho et al., 2018).

Identity-Motivated Elaboration

Given the practices of cognitive processing and expressive engagement within the act of second-screening, we theorize that partisan social identity plays a significant role. The TIME posits that an individual’s partisan social identity will bias information processing, resulting in valenced elaboration (Jennings, 2019). TIME works as an integration of three important psychological, political, and communicative theoretical concepts: social identity theory, the theory of motivated reasoning, and the elaboration likelihood model. Previous studies (e.g., Jennings, 2019) have investigated the influence of partisan social identities on the processing of information in political news articles; however, this study investigates identity-motivated elaboration in the context of a presidential debate.

Partisan Social Identity

Political parties are a powerful influence on political attitudes and behavior. Downs (1957) utilized a rational choice paradigm and an instrumental perspective to argue that voters make political decisions, including party allegiances, based on party performance, policy support, and other political factors. However, political parties are not simply ideological divides. There is a “psychological attachment” that citizens have to their party (Campbell et al., 1960). As such, we have seen a paradigm shift from citizens as rational voters to citizens as rationalizing voters (Lodge & Taber, 2013). The partisan social identity hypothesis argues that political parties operate as social identities (Greene, 2004; Iyengar et al., 2012; Mason & Wronski, 2018). Like any social identity, party attachment can bias evaluations. Social identity theory explains that people demonstrate ingroup favoritism and outgroup denigration to enhance their group’s identity (Turner, 1991), suggesting that while we treat members of our social group in a favorable manner, we discriminate against individuals in competing social groups.

Applying the motivational and cognitive components of social identity theory to politics, the partisan social identity hypothesis predicts that citizens will attempt to positively differentiate their party from the opposing party. In American politics, the two dominant parties are the Republican and Democrat parties. In short, the partisan social identity hypothesis explains why Democrats increasingly do not like Republicans and Republicans do not like Democrats (Iyengar et al., 2012; Jennings et al., 2016).

Tajfel and Turner (1986) explain that group identity even biases character evaluations. Thus, the partisan social identity hypothesis can be tested by examining image evaluations of in-party and out-party candidates. If the partisan social identity hypothesis is correct, we expect to see biased evaluations of candidates’ image factors. As such, we hypothesize that partisan social identity will lead to Republicans evaluating the Republican candidate more favorably than the Democratic candidate and Democrats evaluating the Democratic candidate more favorably than the Republican candidate. Moreover, although persuasive effects of debate viewing are not completely constrained by partisanship, most often viewers express greater enhancement of support for in-party candidates (Warner et al., 2020). In addition, it could be expected that the strength of partisanship matters (Settle, 2018), and strong partisans have been found to become more politically polarized after viewing a presidential debate (Jennings et al., 2018). Therefore, we anticipate the following:

Hypothesis 1 (H1). The strength of partisan social identification will predict (a) more favorable image evaluations of the ingroup candidate and (b) more negative image evaluations of the outgroup candidate.

Hypothesis 2 (H2). The strength of partisan social identification will predict stronger preference for the ingroup candidate over the outgroup candidate following the debate.

Partisan-Motivated Reasoning and Biased Elaboration

Extending the motivational and cognitive aspects of social identity to politics also has implications for information processing. The theory of motivated reasoning argues that political thinking is biased by prior attitudes and predispositions (Kunda, 1990). In other words, a Republican partisan will be motivated to view a Republican candidate in a positive light. In pursuit of this motivational goal, the partisan may interpret new information in a manner that benefits their perception of the candidate because a more positive view of a representative, or even a prototypical leader of the party (see Turner, 1991), would enhance their perception of the party. Ultimately, a more favorable evaluation of one’s social group results in a more positive self-concept (Tajfel & Turner, 1986). Directional motivational goals (where an individual seeks to support preexisting opinions) are achieved by accepting information that is consistent with one’s attitudes with little or no scrutiny and dismissing or counterarguing that runs counter to preexisting attitudes (Kunda, 1990). This biased processing is used to rationalize predispositions (Lodge & Taber, 2013).

Presidential debates promote partisan-motivated reasoning. Partisan predispositions influence voter cognition (Holbrook, 1996), and voters will prefer the arguments of the candidate they supported before viewing a debate (Jarman, 2016). The result of partisan-motivated reasoning is biased cognitive elaboration. Elaboration refers to the cognitive effort exerted in processing new information and is required for an individual to understand arguments, especially on complex political issues. The elaboration likelihood model explains that message receivers can either use central or peripheral elaborative processing, whereas central route processing is associated with greater cognitive elaboration (Petty & Cacioppo, 1986). Personal involvement with an issue results in greater elaboration (Chaiken, 1980). The social identity theory posits that as individuals become more attached to a group, their self-concept and worth become tied to group success (Tajfel & Turner, 1986). As such, a stronger social identity results in greater personal involvement with issues that influence the success of a group. As the social identity theory, specifically the self-esteem hypothesis (Abrams & Hogg, 1988), posits that one’s self-value is tied to the success of their social group, an individual may actively engage Twitter to advocate for their party’s candidate. As a result, a stronger partisan social identity will produce increased identity-congruent elaboration.

The influence of cognitive elaboration, though, is not limited to the amount of elaboration, but we must also consider the valence of elaboration as elaboration can be either positively or negatively valenced (O’Keefe, 2012). To achieve directional motivational goals, an individual will elaborate more positively on issues and candidates that align with their political beliefs and more negatively on issues and candidates of the opposing party. Therefore, we hypothesize that during a presidential debate, viewers will engage in more positively valenced elaboration on the candidate of their party and in negatively valenced elaboration on the candidate of the opposing party:

Hypothesis 3 (H3). The strength of partisan social identification will (a) positively predict identity-congruent elaboration/tweeting and (b) negatively predict identity-incongruent elaboration/tweeting.

Elaboration and Political Attitudes

Multifactor Approach to Candidate Image Evaluation

Image is classically defined as “a human construct imposed on an array of perceived attributes” (Nimmo & Savage, 1976, p. 8). Candidate image has been measured in various ways, although in recent years a new typology of candidate image has been developed that has significant explanatory strength. Extending the theory of image reasoning articulated by Popkin (1994), Warner and Banwart (2016) proposed six factors (i.e., character, intelligence, leadership, charm, benevolence, and homophily) that form the foundation of a citizen’s image evaluation of political candidates. Subsequent research has refined and adapted the factors and their measurements (e.g., Warner et al., 2018). This six-facet measure of candidate image can influence the likelihood to vote for a candidate (Warner & Banwart, 2016; Warner et al., 2018). Of the six facets, homophily and character may be most important in influencing vote choice (Warner & Banwart, 2016, p. 276).

Candidate image is particularly important in the context of presidential debates. Presidential debates, as televised events, inherently display candidate image through the modality of the medium (Druckman, 2003). Candidate image in presidential debates has been examined since the classic Sidney Kraus edited volumes on the 1960 and 1976 debates (e.g., Dennis et al., 1979; Kraus & Smith, 1962; Simons & Leibowitz, 1979; Tannenbaum et al., 1962). Candidates and campaigns are aware of the power of image in debates, and consideration of candidate image on television looms large in pre-debate negotiations over production details (Jamieson & Birdsell, 1988). Candidates often strategize as to how they may change image perceptions, as exemplified by Barack Obama’s more forceful approach in his second 2012 US presidential debate following a rather restrained performance in the first debate (Denton et al., 2020). Presidential debates do influence evaluations of candidate image (Benoit et al., 2003; Benoit & Hansen, 2004). Strong debate performances influence candidate image evaluations in subsequent debates (Holbrook, 1996), and candidates who exceed pre-debate expectations can see increases in their image evaluations (McKinney et al., 2011). The process of thinking about the candidate, their debate performance, and personal characteristics can influence evaluations of their image. More positive thinking could lead to more positive image assessments, whereas more negative thinking could lead to more negative assessments. Therefore, we hypothesize the following:

Hypothesis 4 (H4). Identity-consistent elaboration will predict (a) improved image evaluations of the ingroup candidate and (b) diminished image evaluations of the outgroup candidate, whereas identity-incongruent elaboration will predict (c) diminished image evaluations of the ingroup candidate and (d) improved image evaluations of the outgroup candidate.

Parry-Giles (2010) argues that evaluation of candidate image interacts with assessments of a candidate’s policy positions and ultimately one’s support for a particular candidate. Importantly, evaluating candidate image may not simply be a low-cost heuristic, as voters may value relevant, observable characteristics such as homophily (i.e., how much a politician is perceived to care about voters), more than an image facet such as charm (Warner & Banwart, 2016, p. 276). In addition, voters with high need-for-cognition can be more likely to weigh image characteristics when considering voting for a candidate (Warner et al., 2018). In summation, thinking about relevant image characteristics could make voters more likely to utilize image evaluations when considering supporting a candidate. Therefore, we hypothesize the following:

Hypothesis 5 (H5). Following the debate, (a) image evaluations of an ingroup candidate will positively predict strength in preference for that candidate over the outgroup candidate, but (b) image evaluations of an outgroup candidate will negatively predict preference for the ingroup candidate.

The previous three hypotheses (H3–H5) construct a hypothetical model, which is presented in Figure 1. The model consists of four levels: (1) partisan social identity, (2) identity-congruent/incongruent elaboration, (3) ingroup/outgroup candidate image evaluations, and (4) change in ingroup candidate preference. In the hypothesized model, there are four indirect pathways through which partisan social identity predicts greater change in preference for the ingroup candidate:

Hypothesis 6 (H6). The strength of partisan social identification will predict a greater change in preference for the ingroup candidate through four double-mediated pathways through identity-congruent/incongruent elaboration/tweeting and image evaluations of the ingroup/outgroup candidate.

Figure 1.

Hypothesized model of the identity-motivated elaborative process of partisan social identity’s indirect influence on candidate preference.

Method

Participants

Participants (N = 657) from 13 universities from across the United States took part in the experimental study. Demographically, more participants identified as female (n = 379; 57.7%) than male (n = 278; 42.3%). A majority identified as White (n = 505; 76.9%), followed by African American (n = 51; 7.8%), Asian (n = 48; 7.3%), Hispanic/LatinX (n = 30; 4.6%), Native American (n = 2; .3%), and a race/ethnicity not specified (n = 21; 3.2%). There were 23 Strong Republicans (3.5%), 112 Republicans (17.0%), 125 Lean Republicans (19.0%), 110 who identified with neither party (16.7%), 117 Lean Democrats (17.8%), 112 Democrats (17.0%), and 58 Strong Democrats (8.8%). The mean age was 20.2 years (SD = 3.28).

Procedure

Participants watched one of the three televised 2016 US presidential debates in real time. Before the debate, participants answered a pretest with demographic and political belief variables, including partisan affiliation and candidate feeling thermometers. They were instructed to tweet their thoughts using their mobile device throughout the debate using a designated hashtag. We chose to use Twitter because of the popularity of live-tweeting debates providing ecological validity (Zheng & Shahin, 2020), the ability to easily collect data, and the precedent of using tweets as a proxy to measure elaboration (e.g., Jennings et al., 2017). Each participant was required to have a Twitter account; if a participant did not have a Twitter account or did not want to use their regular Twitter handle, they were asked to create an account. Immediately following the debate, and before exposure to any post-debate commentary, participants responded to a posttest, which included image evaluations and feeling thermometers for each candidate. Participants who did not tweet during a debate were not included in the analysis.

Measures

Partisan Affiliation

Participants were asked, “To what extent do you consider yourself a Democrat or Republican?” Participants indicating that they did not identify with either party were asked a follow-up question: “I know you don’t have a preference between the two parties, but say you had to vote for a generic candidate from one of the two. In this case, who would you be more likely to vote for?” These two questions were used to classify each participant as either a Democrat or a Republican. There were 347 participants classified as Democrats (52.8%) and 310 participants classified as Republicans (47.2%). Although this was not used as a variable in any hypothesis, partisan affiliation was used in the construction of each subsequent variable, as the ingroup candidate for Democrats (Hillary Clinton) was different than the ingroup candidate for Republicans (Donald Trump).

Partisan Social Identity

A scale measuring identification with social groups (Greene, 1999) was adapted to measure partisan social identity. Importantly, the scale measured identification with the Democratic Party for Democrats and identification with the Republican Party for Republicans. A 7-point Likert-type (1 = strongly disagree; 7 = strongly agree) social identity scale with three items (i.e., “When I talk about Democrats [Republicans], I usually say ‘we’ rather than ‘they’”; “When someone criticizes Democrats [Republicans], it feels like a personal insult”; and “I don’t have much in common with most Democrats [Republicans]” (reverse coded)) was utilized. Scores (M = 3.62; SD = .94) ranged from 1 (extremely weak partisan identification) to 7 (extremely strong partisan identification).

Elaboration

As in previous studies (e.g., Bramlett et al., 2018; Jennings et al., 2017), tweets were used to operationalize elaboration during the debate. Total elaboration was calculated by summing the number of tweets that a viewer tweeted throughout the debate using the designated hashtag (M = 11.65; SD = 11.39). Identity-congruent elaboration (M = 5.12; SD = 5.95) was calculated by summing the number of tweets supporting the ingroup candidate and the number of tweets criticizing the outgroup candidate. For example, identity-congruent elaboration (M = 2.43; SD = 3.31) for a Democrat was calculated by summing positive tweets about Clinton and negative tweets about Trump. Conversely, identity-incongruent elaboration was calculated by summing critical tweets of the ingroup candidate and positive tweets of the outgroup candidate.

Image Evaluations

Using Warner and colleagues’ (2018) scales and terminology, which were adapted from Warner and Banwart (2016), image evaluations were measured before the debate for each candidate. There are six factors of candidate image evaluation, and each factor was measured using three items where the participants were asked, “To what extent do you agree that Hillary Clinton [Donald Trump] is “trustworthy,” “dishonest,” and “believable” (character); “unintelligent,” “knowledgeable,” and “smart” (intelligence); “strong,” “poised,” and “a good leader” (leadership); “capable,” “effective,” and “incompetent” (competence); “understands people like me,” “understands the problems faced by people like me,” and “shares my values” (homophily); and “charismatic,” “likable,” and “unpleasant” (likeability). Overall, evaluations for ingroup candidates (M = 3.60; SD = .68) were higher than evaluations for outgroup candidates (M = 2.79; SD = .69).

Change in Preference for Ingroup Candidate

Before and after viewing the presidential debate, feeling thermometers were used to measure favorability for each candidate. Participants were asked how much they like or dislike Hillary Clinton and Donald Trump (separately) on a 101-point scale from 0 (very much dislike) to 100 (very much like). Preference for the ingroup candidate was determined by subtracting the favorability score for the outgroup candidate (Trump for Democrats; Clinton for Republicans) from the favorability score for the ingroup candidate (Clinton for Democrats; Trump for Republicans). Change in preference for the ingroup candidate was calculated by subtracting the pretest preference score from the posttest preference score. Viewers expressed stronger preference for the ingroup candidate following the debate (M = 5.95; SD = 24.26), as favorability for the ingroup candidate improved by an average of 6.35 (SD = 18.77) compared with only 0.38 (SD = 17.48) for the outgroup candidate.

Content Analysis

Undergraduate research fellows were trained as coders to content-analyze the participants’ tweets. Following coder training and the achievement of satisfactory inter-coder reliability, coders independently coded an assigned portion of the 8,501 tweets generated during the three presidential debates. The unit of analysis was an individual tweet. Each individual sample had an overlap with tweets from other coders’ samples to form the sample for reliability assessment. Three coders each coded roughly a third of the 2,579 total tweets from Debate 1. Two coders each coded roughly half of the 2,379 total tweets from Debate 2. Finally, four coders each coded roughly a quarter of the 3,543 total tweets from Debate 3.

The variables of interest for this study were the valence of tweets toward Hillary Clinton or Donald Trump. This variable was the result of a two-step coding process. First, coders answered the question “Which candidate does this tweet mention or reference?” with options including Hillary Clinton, Donald Trump, Tim Kaine, Mike Pence, Barack Obama, No Candidate, Other Candidate, and Can’t Determine. Then, where candidates were mentioned, coders marked valence as positive (1), negative (2), neutral (3), or not mentioned (0).

Intercoder reliability was assessed using Hayes’ K-alpha (Hayes & Krippendorff, 2007). Reliability above .70 is typically viewed as acceptable when using Krippendorff’s alpha (Krippendorff, 2018), although Krippendorff has used .67 before (Krippendorff, 1980, as cited in Lovejoy et al., 2016). Due to the two-step coding process, candidate valence was in the .67–.70 range in two debates. Intercoder reliability for both candidate focus and candidate valence is presented in Table 1.

Table 1.

Intercoder Reliability as Measured With Krippendorff’s Alpha (α).

N = 7654	Clinton focus	Clinton valence	Trump focus	Trump valence
Debate 1	.80	.69	.72	.67
Debate 2	.81	.70	.80	.70
Debate 3	.89	.77	.84	.71

Results

The first hypothesis predicted that a stronger partisan social identity would predict more biased evaluations, higher ingroup evaluations, and lower outgroup evaluations. For H1a, the results found support for the influence of partisan social identification on image evaluations, as the strength of identification predicted more favorable image evaluations for ingroup candidates, F(1, 655) = 10.30; p < .001. However, the hypothesis (H1b) predicting a negative influence on outgroup evaluations was not supported, F(1, 655) = 1.71; ns. The strength of partisan identification predicted ingroup favoritism, but not outgroup denigration.

The second hypothesis predicted that strong partisans’ evaluations would be more positively affected by the debate performance of the ingroup candidate. Indeed, partisan social identification positively predicted change in the strength of preference, F(1, 655) = 6.72; p < .01. However, it is important to note that this effect was limited to Democrats, F(1, 346) = 9.04; p < .01, as the relationship was not significant for Republicans, F(1, 309) = 3.54; ns.

Hypothesized Model

The remaining hypotheses (H3–H6) construct the hypothesized model presented in Figure 1. It was analyzed using the lavaan package in the R ecosystem (Rosseel, 2012). Structural equation modeling, a multivariate statistical analysis technique, allows for factor analysis to construct latent variable and path analysis with simultaneous estimation of all variable relationships. Chi-square change and probability values are presented in text; unstandardized path estimates, standard error, standardized coefficients, and confidence intervals are displayed in Table 2; and significant pathways are represented in Figure 2. The data fit the model well, χ²(31) = 55.26, p < .001, comparative fit index (CFI) = .945, Tucker–Lewis Index (TLI) = .916, root mean square error of approximation (RMSEA) = .035 [.019–.049], standardized root mean square residual (SRMR) = .034, signifying that the residual variances in the model appear to be behaving randomly with no systematic flaw in the indices or hypothetical model (see Little, 2013).

Table 2.

Direct and Indirect Effects.

	5,000 bootstrap samples
	B(SE)	β	LLCI	ULCI
Hypothesized direct effects
H4a: Partisan Social Identity (PSID) → ID-Congruent Elaboration (ICE)	0.31 (.04)	0.08	0.23	0.37^a
H4b: → PSID ID-Incongruent Elaboration (IIE)	−0.42 (0.06)	−0.06	−0.54	−0.30^a
H5a: ICE → Ingroup Image Evaluation	0.05 (0.01)	0.41	0.04	0.06^a
H5b: ICE → Outgroup Image Evaluation	−0.05 (0.01)	−0.40	−0.06	−0.04^a
H5c: IIE → Ingroup Image Evaluation	−0.10 (0.01)	−0.43	−0.11	−0.08^a
H5d: IIE → Outgroup Image Evaluation	0.09 (0.01)	0.40	0.07	0.11^a
H6a: Ingroup Image Evaluation → Ingroup Preference	2.48 (1.25)	0.10	0.03	4.92^a
H6b: Outgroup Image Evaluation → Ingroup Preference	−3.35 (1.42)	−0.13	–6.23	−0.71^a
Non-hypothesized direct effects
PSID → Ingroup Image Evaluation	0.05 (0.02)	0.06	0.01	0.09^a
PSID → Outgroup Image Evaluation	−0.05 (0.01)	−0.03	−0.06	−0.04^a
Indirect effects
H7a: PSID → ICE → Ingroup Image Evaluation → Ingroup Preference	0.08 (0.04)	–	0.01	0.16^a
H7b: PSID → ICE → Outgroup Image Evaluation → Ingroup Preference	0.40 (0.11)	–	0.22	0.64^a
H7c: PSID → IIE → Ingroup Image Evaluation → Ingroup Preference	0.07 (0.03)	–	0.01	0.13^a
H7d: PSID → IIE → Outgroup Image Evaluation → Ingroup Preference	0.31 (0.07)	–	0.19	0.45^a

B: unstandardized coefficient; SE: standard error; β: standardized coefficient; LLCI: lower limit confidence interval; ULCI: upper limit confidence interval.

Indicates that the 95% confidence interval does not contain zero.

Figure 2.

The Influence of Partisan Social Identity on Change in Ingroup Candidate Preference.

The third hypothesis predicted that elaboration would be more consistent with a stronger identity than a weaker one. H3a was supported, as partisans who identified strongly tweeted more positively about the ingroup candidate and more negatively about the opposing candidate (Δχ² = 4.77; p < .05). Similarly, H3b was supported as partisan social identity predicted fewer favorable tweets about the outgroup candidate or critical tweets of their own candidate (Δχ² = 4.16; p < .05). The third hypothesis, then, was fully supported; partisan social identification biases elaboration.

The fourth hypothesis looks at the effect of both identity-congruent and identity-incongruent elaboration on image evaluations of each candidate. As hypothesized, identity-congruent elaboration (H4a and H4b) predicted higher image evaluations of the ingroup candidate (Δχ² = 132.45; p < .001) and lower evaluations of the outgroup candidate (Δχ² = 124.26; p < .001). Identity-incongruent elaboration (H4c and H4d) predicted lower image evaluations of a partisan’s in-party candidate (Δχ² = 149.12; p < .001) and more favorable image evaluations of the opposing candidate (Δχ² = 115.91; p < .001). H4 is fully supported, as elaboration is associated with candidate image evaluations.

The fifth hypothesis was fully supported, as more favorable image evaluations of the ingroup candidate enhanced the favorability gap between candidates (Δχ² = 8.69; p < .01). Conversely, more favorable image evaluations of the opposing candidate diminished the gap (Δχ² = 6.78; p < .01). Results from H5a and H5b reveal that image evaluations influenced the strength of preference for the ingroup candidate over the outgroup candidate.

Finally, the four indirect effects predicted in the sixth hypothesis (H6a, H6b, H6c, and H6d) were significant. The strength of partisan social identification predicts a greater positive change in preference for the ingroup candidate through four double-mediated pathways. The two mediating levels are (1) identity-congruent and identity-incongruent elaboration, and (2) image evaluations of both candidates.

Discussion

In this study, we examined the effects of live-tweeting televised presidential debates during the 2016 US presidential campaign. Results from our study contribute to the body of literature examining cognitive and behavioral processes and outcomes from presidential debate viewing (e.g., Jarman, 2016; Kim & Garrett, 2012; McKinney & Carlin, 2004; McKinney & Warner, 2013; Mullinix, 2015; Warner et al., 2020). Engagement with and processing of a debate message, measured in the form of live-tweeting while viewing a debate, produced variations in viewers’ candidate image evaluations and candidate preference. These findings provide important contributions to the body of literature examining cognitive and attitudinal outcomes related to second-screening and politics (Barnidge et al., 2017; Chadwick et al., 2017; J. Cho et al., 2018; Gil de Zúñiga et al., 2015; Houston, Hawthorne, et al., 2013; Houston, McKinney, et al., 2013; McGregor & Mourão, 2017; Wells et al., 2016).

The first hypothesis predicted that partisan social identity strength would predict more favorable image evaluations of the ingroup candidate and more negative image evaluations of the outgroup candidate. Results here were mixed, as partisan strength did predict more positive ingroup evaluations but not more negative outgroup evaluations. These findings are in line with other presidential debate research that has also found debate viewing results in a significant increase in ingroup candidate evaluations and a slight decrease in outgroup candidate evaluations (Mullinix, 2015; Warner et al., 2018). Importantly, then, our study replicates past findings related to candidate support and evaluations, particularly candidate image evaluations (Warner et al., 2018). However, the second hypothesis was only partially supported. It had forecast that partisan social identity strength would predict stronger post-debate in-candidate preference over the outgroup candidate. While this was the case for Democrats, it was not the same for Republicans.

These findings may point to the idiosyncratic nature of Donald Trump and Hillary Clinton as candidates in 2016, and particularly their debate performances. Our sample of the millennial generation may also help explain what happened. It is possible that these Democrats, before viewing the debate, perhaps assessed Clinton as something of a “compromise” candidate. However, upon viewing her debate performance, they may have reevaluated their initial assessment based on her strong debate performance, and especially so in light of the barrage of attacks she encountered from Donald Trump. As such, it seems reasonable for Democrats to strengthen their preferences for Clinton. In fact, such a phenomenon bore out as the campaign progressed. According to a Democracy Corps poll in 2016, while millennials initially preferred a “revolutionary” candidate such as Bernie Sanders, they eventually coalesced behind Clinton after comparing her to Donald Trump (Rosenfeld, 2016). At the same time, a similar shift in post-debate reassessment did not occur among Republicans. Among the millennial Republican voters, at least, they came to their debate viewing with a somewhat lower assessment of Trump as a candidate, and his debate performance did not enhance assessment of their party’s nominee—nor did their evaluation of Clinton change much.

The third hypothesis anticipated that the strength of partisan social identity would predict more identity-congruent elaboration and less identity-incongruent elaboration. For both Democratic and Republican viewers, strong partisan social identity predicted more identity-congruent and less identity-incongruent elaboration in the form of valenced tweets. These findings have important implications for our understanding of partisan social identity and televised debate engagement. Politics and partisanship are increasingly about rooting for your own team (Mason & Wronski, 2018), and the televised presidential debate seems to provide the ideal arena for such partisan rallying via second-screening and social media. In the context of live-tweeting a televised candidate debate, we observed that strong partisans were more likely to generate more thoughts (tweets) supporting their candidate, and more thoughts criticizing and attacking the outgroup candidate, than thoughts generated to criticize one’s own candidate or praise the opposing candidate. Such social media expressions are consistent with existing debate effects research that shows viewing presidential debates actually increases viewers’ affective political polarization (McKinney & Warner, 2013). As candidates’ debate argumentation is most commonly framed through partisan appeals and opponent attacks, viewers’ partisan identity is activated, and as our study demonstrates, viewers who are social watching the televised debate exhibit their polarized attitudes with expressions of more positive thoughts (tweets) supporting rather than criticizing their candidate and more thoughts criticizing rather than praising the outgroup candidate. As a result, Twitter, through valenced elaboration and opinion expression, may lead to greater opinion certainty and increased partisan polarization. Political echo-chambers develop on Twitter (Guerrero-Solé, 2018), leading to increased anxiety about the current political state (Auxier & Vitak, 2019).

The fourth hypothesis predicted that identity-congruent elaboration would improve ingroup candidate image evaluations and decrease outgroup candidate image evaluations, and that identity-incongruent elaboration would increase outgroup candidate evaluations while diminishing ingroup candidate image evaluations. Results here were fully supported. The fifth hypothesis predicted that image evaluations would relate to candidate preference. This hypothesis was also supported, affirming prior research showing that image evaluations influence candidate preference (Warner & Banwart, 2016; Warner et al., 2018). Finally, our sixth hypothesis predicted that the strength of partisan social identity would induce the greatest changes in candidate preference through double-mediated pathways. This was also supported, highlighting the intricate ways in which persuasion occurs in debate viewing.

These findings help us better understand the nuanced ways in which viewing a presidential debate might influence voters, and particularly in the current digital media environment in which increasing numbers of viewers engage the televised political message while expressing their thoughts and opinions via social media. The strength of one’s partisan social identity influences assessment of candidates’ image, and as the partisan viewer elaborates upon their evaluation of candidates—both ingroup and outgroup candidates—this increased elaboration affects post-debate candidate preference.

Limitations and Future Directions

First, we acknowledge Twitter is likely overrepresented in political social media research. This is partly due to the accessibility of data. However, this study, which did not collect data from the general public but instead from consenting participants, did not have the constraints that much social media research faces. Because political conversations are common on Twitter, live-tweeting presidential debates is becoming increasingly popular (Nielsen’s Social Content Rating, 2016; Zheng & Shahin, 2020), and as Twitter’s open API (application programming interface) provides for easier public data collection relative to other social media platforms, we deemed Twitter to be the most appropriate social media platform for testing our hypotheses. Future research should examine the influence of debate engagement on various social media platforms, particularly Facebook, on attitude formation as well as knowledge acquisition. In addition, to test the generalizability of the findings, the effects of social media engagement should be examined among different populations, including non-digital natives. Finally, the partisan social identity hypothesis and identity-motivated elaboration need to be investigated among different forms of political communication to ascertain theoretical boundaries and provide more nuanced understanding of psychological mechanisms of political engagement, both online and offline.

The results are limited in generalization because our study utilized a convenience sample. However, in a meta-analysis of presidential debate effects studies, Benoit and colleagues (2003) found there were no significant differences between convenience samples and representative samples of the US population. In this study, results may be influenced by the particular feelings that young people had toward the candidates, particularly then-candidate Donald Trump. We observed less favorability toward Trump from Republican viewers, potentially highlighting a difference between young adults and a fully representative adult sample of voters.

Conclusion

Overall, we find strong evidence that partisan strength predicts candidate preference and that elaboration via live-tweeting a presidential debate spurs change in evaluations of candidates. However, the strength of such change can be contingent on partisan identification, the unique candidate viewers may identify with, and the unique candidate from whom viewers disassociate. Social media dialogue surrounding political events continues to increase, and especially for key political communication moments such as televised US presidential debates (McKinney, 2018). Twitter allows citizens to engage debate information in new ways to play an increasing role in building narratives and constructing meaning (Wells et al., 2016). With President Donald Trump now the nation’s “tweeter-in-chief,” the role of social media expression as part of our political communication landscape has become even more significant. Understanding the effects of such expression by citizens presents a fruitful line of research for political communication scholars.

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

Freddie J. Jennings

Author Biographies

Freddie J. Jennings (PhD, University of Missouri) is a visiting assistant professor of Communication at the University of Arkansas. His research interests include political psychology, partisan social identity, and political communication in the evolving media landscape.

Josh C. Bramlett (PhD, University of Missouri) is an assistant professor of Communication at Eastern New Mexico University. His research interests include political comedy, political campaign communication, and the political uses of social media.

Mitchell S. McKinney (PhD, University of Kansas) is a professor of Communication at the University of Missouri. His research interests include presidential debates, political campaigns, media and politics, and presidential rhetoric. He is the director of the Political Communication Institute.

Molly M. Hardy (PhD, University of Missouri) is an associate instructor of communication at Maryville University. Her research interests include political polarization, political uses of social media, and gender in political communication.

References

Abrams

Hogg

M. A.

(1988). Comments on the motivational status of self-esteem in social identity and intergroup discrimination. European Journal of Social Psychology, 18(4), 317–334. https://doi.org/10.1002/ejsp.2420180403

Auxier

B. E.

Vitak

(2019). Factors motivating customization and echo chamber creation within digital news environments. Social Media + Society, 5(2), 1–13. https://doi.org/10.1177/2056305119847506

Barnidge

Gil

Zúñiga

Diehl

(2017). Seconds screening and political persuasion on social media. Journal of Broadcasting & Electronic Media, 61(2), 309–331. https://doi.org/10.1080/08838151.2017.1309416

Benoit

W. L.

Hansen

G. J.

(2004). Presidential debate watching, issue knowledge, character evaluation and vote choice. Human Communication Research, 30(1), 124–144. https://doi.org/10.1111/j.1468-2958.2004.tb00727.x

Benoit

W. L.

Hansen

G. J.

Verser

R. M.

(2003). A meta-analysis of the effects of viewing U.S. presidential debates. Communication Monographs, 70(4), 335–350. http://doi.org/10.1080/0363775032000179133

Bramlett

J. C.

McKinney

M. S.

Warner

B. R.

(2018). Processing the political: Presidential primary debate “live-tweeting” as information processing. In Warner

B. R.

Bystrom

D. G.

McKinney

M. S.

Banwart

(Eds.), An unprecedented election: Media, communication, and the electorate in the 2016 campaign (pp. 169–188). Praeger.

Campbell

Converse

Miller

Stokes

(1960). The American voter. University of Chicago Press.

Chadwick

O’Loughlin

Vaccari

(2017). Why people dual screen political debates and why it matters for democratic engagement. Journal of Broadcasting & Electronic Media, 61(2), 220–239. https://doi.org/10.1080/08838151.2017.1309415

Chaiken

(1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39(5), 752–766. https://doi.org/10.1037/0022-3514.39.5.752

10.

Cho

Boster

F. J.

(2005). Development and validation of value-, outcome-, and impression-relevant involvement scales. Communication Research, 32(2), 235–264. https://doi.org/10.1177/0093650204273764

11.

Cho

Ahmed

Keum

Choi

Y. J.

Lee

J. J.

(2018). Influencing myself: Self-reinforcement through online political expression. Communication Research, 45(1), 83–111. https://doi.org/10.1177/0093650216644020

12.

Dennis

Chaffee

Choe

(1979). Impact upon partisan: Image and issue voting. In Kraus

(Ed.), The great debates: Carter vs. Ford, 1976 (pp. 314–330). Indiana University Press.

13.

Denton

R. E.

Jr. Trent

J. S.

Friedenberg

R. V.

(2020). Political campaign communication: Principles and practices (9th ed.). Rowman & Littlefield.

14.

Downs

(1957). An economic theory of political action in a democracy. Journal of Political Economy, 65(2), 135–150.

15.

Driscoll

Leavitt

Guth

K. L.

Bar

Mehta

(2018). Beyond Big Bird, binders, and bayonets: Humor and visibility among connected viewers of the 2012 US presidential debates. Social Media + Society, 4, 1–12. https://doi.org/10.1177/2056305118761201

16.

Druckman

J. N.

(2003). The power of television images: The first Kennedy-Nixon debate revisited. The Journal of Politics, 65(2), 559–571.

17.

Fein

Goethals

G. R.

Kugler

M. B.

(2007). Social influence on political judgments: The case of presidential debates. Political Psychology, 28(2), 165–192. https://doi.org/10.1111/j.1467-9221.2007.00561.x

18.

Freelon

Karpf

(2015). Of big birds and bayonets: Hybrid Twitter interactivity in the 2012 presidential debates. Information, Communication & Society, 18(4), 390–406. https://doi.org/10.1080/1369118X.2014.952659

19.

Gil de Zúñiga

Garcia-Perdomo

McGregor

S. C.

(2015). What is second screening? Exploring motivations of second screen use and its effects on online political participation. Journal of Communication, 65(5), 793–815. https://doi.org/10.1111/jcom.12174

20.

Gottfried

Hardy

Holbert

Winneg

Jamieson

(2017). The changing nature of political debate consumption: Social media, multi-tasking, and knowledge acquisition. Political Communication, 34(2), 172–199. https://doi.org/10.1080/10584609.2016.1154120

21.

Greene

(1999). Understanding party identification: A social identity approach. Political Psychology, 20(2), 393–403. https://doi.org/10.1080/1369118X.2014.952659

22.

Greene

(2004). Social identity theory and party identification. Social Science Quarterly, 85(1), 136–153. https://doi.org/10.1111/j.0038-4941.2004.08501010.x

23.

Guerrero-Solé

(2018). Interactive behavior in political discussions on Twitter: Politicians, media, and citizens’ patterns of interaction in the 2015 and 2016 electoral campaigns in Spain. Social Media + Society, 4(4), 1–16. https://doi.org/10.1177/2056305118808776

24.

Hawthorne

Houston

J. B.

McKinney

M. S.

(2013). Live-tweeting a presidential primary debate: Exploring new political conversations. Social Science Computer Review, 31(5), 552–562. https://doi.org/10.1177/0894439313490643

25.

Hayes

A. F.

Krippendorff

(2007). Answering the call for a standard reliability measure for coding data. Communication Methods and Measures, 1(1), 77–89. https://doi.org/10.1080/19312450709336664

26.

Holbrook

T. M.

(1996). Do campaigns matter? SAGE.

27.

Houston

J. B.

Hawthorne

Spialek

M. L.

Greenwood

McKinney

M. S.

(2013). Tweeting during presidential debates: Effect on candidate evaluations and debate attitudes. Argumentation and Advocacy, 49(4), 301–311. https://doi.org/10.1080/00028533.2013.11821804

28.

Houston

J. B.

McKinney

M. S.

Hawthorne

Spialek

M. L.

(2013). Frequency of tweeting during presidential debates: Effect on debate attitudes and knowledge. Communication Studies, 64(5), 548–560. https://doi.org/10.1080/10510974.2013.832693

29.

Iyengar

Sood

Lelkes

(2012). Affect, not ideology: A social identity perspective on polarization. Public Opinion Quarterly, 76(3), 405–431. https://doi.org/10.1093/poq/nfs038

30.

Jamieson

K. H.

Birdsell

D. S.

(1988). Presidential debates: The challenge of creating an informed electorate. Oxford University Press.

31.

Jarman

J. W.

(2016). Motivated reasoning and viewer’s reactions to the first 2012 presidential debate. Speaker & Gavel, 53(1), 83–101.

32.

Jarvey

(2016, October 9). Second Donald Trump-Hillary Clinton showdown unseats first as most tweeted debate ever. Hollywood Reporter. https://www.hollywoodreporter.com/news/second-donald-trump-hillary-clinton-debate-most-tweeted-936761

33.

Jennings

F. J.

(2019). An uninformed electorate: Identity-motivated elaboration, partisan cues, and learning. Journal of Applied Communication Research, 47(5), 527–547. http://doi.org/10.1080/00909882.2019.1679385

34.

Jennings

F. J.

Coker

C. R.

McKinney

M. S.

Warner

B. R.

(2017). Tweeting presidential primary debates: Debate processing through motivated Twitter instruction. American Behavioral Scientist, 61(4), 455–474. https://doi.org/10.1177/0002764217704867

35.

Jennings

F. J.

McKinney

M. S.

Greenwood

M. M.

(2018). Preaching to the choir: Partisan social identity and presidential debate social watching. In Hinck

(Ed.), Presidential debates in a changing media environment (Vol. 2, pp. 131–151). Praeger.

36.

Jennings

F. J.

Molina-Galarza

R. A.

Warner

B. R.

(2016). Political polarization in the 2014 U.S. midterm elections. In Hendricks

J. A.

Schill

(Eds.), Communication and midterm elections: Media, message, and mobilization (pp. 83–97). Palgrave Macmillan.

37.

Kim

Y. M.

Garrett

(2012). Online and memory based: Revisiting the relationship between candidate evaluation processing models. Political Behavior, 34, 345–368. https://doi.org/10.1007/s11109-011-9158-9

38.

Kraus

Smith

(1962). Images and issues. In Kraus

(Ed.), The great debates: Background, perspective, effects (pp. 289-312). Indiana University Press.

39.

Krippendorff

. (1980). Content analysis: An introduction to its methodology. Sage Publishing.

40.

Krippendorff

. (2018). Content analysis: An introduction to its methodology (4th ed.). Sage Publishing.

41.

Kunda

(1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480–498. https://doi.org/10.1037/0033-2909.108.3.480

42.

Lanoue

D. J.

Schrott

P. R.

(1991). The joint press conference: The history, impact, and prospects of American presidential debates. Greenwood.

43.

Little

T. D.

(2013). Longitudinal structural equation modeling. Guilford Press.

44.

Lodge

Taber

C. S.

(2013). The rationalizing voter. Cambridge University Press.

45.

Lovejoy

Watson

B. R.

Lacy

Riffe

(2016). Three decades of reliability in communication content analysis: Reporting of reliability statistics and coefficient levels in three top journals. Journalism and Mass Communication Quarterly, 93(4), 1135–1159. https://doi.org/10.1177/1077699016644558

46.

Mason

Wronski

J. W.

(2018). One tribe to bind them all: How our social group attachment strengthen partisanship. Political Psychology, 39(S1), 257–277. https://doi.org/10.1111/pops.12485

47.

McGregor

S. C.

Mourão

R. R.

(2017). Second screening Donald Trump: Conditional indirect effects on political participation. Journal of Broadcasting & Electronic Media, 61(2), 264–290. https://doi.org/10.1080/08838151.2017.1309418

48.

McKinney

M. S.

(2018). Political campaign debates in the 2016 elections: Advancing campaign debate scholarship. Argumentation and Advocacy, 54(1–2), 72–75. https://doi.org/10.1080/00028533.2018.1446818

49.

McKinney

M. S.

Carlin

D. B.

(2004). Political campaign debates. In Kaid

L. L.

(Ed.), Handbook of political communication research (pp. 203–234). Lawrence Erlbaum.

50.

McKinney

M. S.

Dudash

E. A.

Hodgkinson

(2003). Viewer reactions to the 2000 presidential debates Learning issues and image information. In Kaid

L. L

Tedesco

J. C.

Bystrom

D. G.

McKinney

M. S.

(Eds.), The millennium election: Communication in the 2000 campaign (pp. 43–58). Rowman & Littlefield.

51.

McKinney

M. S.

Houston

J. B.

Hawthorne

(2014). Social watching a 2012 Republican presidential primary debate. American Behavioral Scientist, 58(4), 556–573. https://doi.org/10.1177/0002764213506211

52.

McKinney

M. S.

Rill

L. A.

Watson

R. G.

(2011). Who framed Sarah Palin? Viewer reactions to the 2008 vice presidential debate. American Behavioral Scientist, 55(3), 212–231. https://doi.org/10.1177/0002764210392158

53.

McKinney

M. S.

Warner

B. R.

(2013). Do presidential debates matter? Examining a decade of campaign debate effects. Argumentation & Advocacy, 49(4), 238–258.

54.

Mullinix

K. J.

(2015). Presidential debates, partisan motivations, and political interest. Presidential Studies Quarterly, 45(2), 270–288. https://doi.org/10.1111/psq.12187

55.

Nekmat

(2012). Message expression effects in online social communication. Journal of Broadcasting and Electronic Media, 56(2), 203–224. https://doi.org/10.1080/08838151.2012.678513

56.

Nielson’s Social Content Rating. (2016, September 27). First presidential debate of 2016 draws 84 million viewers. Nielson. https://www.nielsen.com/us/en/insights/article/2016/first-presidential-debate-of-2016-draws-84-million-viewers/

57.

Nimmo

D. D.

Savage

R. L.

(1976). Candidates and their images: Concepts, methods, and findings. Goodyear Publishing Company.

58.

O’Keefe

D. J.

(2012). The elaboration likelihood model. In Dillard

J. P.

Shen

(Eds.), The SAGE handbook of persuasion: Developments in theory and practice (2nd ed.; pp. 137–149). SAGE.

59.

Park

C. S.

Kaye

B. K.

(2017). Twitter and encountering diversity: The moderating role of network diversity and age in the relationship between Twitter use and crosscutting exposure. Social Media + Society, 3(3), 1–12. https://doi.org/10.1177/2056305117717247

60.

Parry-Giles

(2010). Resisting a “treacherous piety”: Issues, images, and public deliberation in presidential campaigns. Rhetoric & Public Affairs, 13(1), 37–64. https://doi.org/10.1353/rap.0.0138

61.

Petty

R. E.

Cacioppo

J. T.

(1986). The elaboration likelihood model of persuasion. In Berkowitz

(Ed.), Advances in experimental social psychology (Vol. 19, pp. 123–205). Academic Press.

62.

Popkin

S. L.

(1994). The reasoning voter: Communication and persuasion in presidential campaigns (2nd ed.). The Chicago University Press.

63.

Rosenfeld

(2016, October 7). Millennials are now moving toward Clinton, but will they turn out and vote? AlterNet. https://www.alternet.org/2016/10/millennials-are-now-moving-toward-clinton-will-they-turn-out-and-vote/

64.

Rosseel

(2012). An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36.

65.

Settle

J. E.

(2018). Frenemies: How social media polarizes America. Cambridge University Press.

66.

Simons

Leibowitz

(1979). Shifts in candidate images. In Kraus

(Ed.), The great debates: Carter vs. Ford, 1976 (pp. 398–304). Indiana University Press.

67.

Tajfel

Turner

J. C.

(1986). The social identity theory of intergroup behavior. In Worchel

W. G.

Austin

W. G.

(Eds.), Psychology of intergroup relations (pp. 7–24). Nelson Hall.

68.

Tannenbaum

Greenberg

Silverman

(1962). Candidate images. In Kraus

(Ed.), The great debates: Background, perspective, effects (pp. 271–288). Indiana University Press.

69.

Thorson

Hawthorne

Swasy

McKinney

M. S.

(2015). Co-viewing, tweeting, and Facebooking the 2012 presidential debates. Electronic News, 9(3), 195–214. https://doi.org/10.1177/1931243115593320

70.

Trilling

(2015). Two different debates? Investigating the relationship between a political debate on TV and simultaneous comments on Twitter. Social Science Computer Review, 33(3), 1–18. http://doi.org/10.1177/0894439314537886

71.

Turner

J. C.

(1991). Social influence. Open University Press.

72.

Valkenburg

P. M.

(2017). Understanding self-effects in social media. Human Communication Research, 43, 477–490. https://doi.org/10.1111/hcre.12113

73.

Warner

B. R.

Banwart

M. C.

(2016). A multifactor approach to candidate image. Communication Studies, 67(3), 259–279. https://doi.org/10.1080/10510974.2016.1156005

74.

Warner

B. R.

Jennings

F. J.

Bramlett

J. C.

Coker

C. R.

Reed

J. L.

Bolton

J. P.

(2018). A multimedia analysis of persuasion in the 2016 presidential election: Comparing the unique and complementary effects of political comedy and political advertising. Mass Communication and Society, 21(6), 720–741. https://doi.org/10.1080/15205436.2018.1472283

75.

Warner

B. R.

McKinney

M. S.

Bramlett

J. C.

Jennings

F. J.

Funk

M. E.

(2020). Reconsidering partisanship as a constraint on the persuasive effects of debates. Communication Monographs, 87(2), 137–157. https://doi.org/10.1080/03637751.2019.1641731

76.

Wells

Van Thomme

Maurer

Hanna

Pevehouse

Shah

D. V.

Bucy

(2016). Coproduction or cooptation? Real-time spin and social media response during the 2012 French and US presidential debates. French Politics, 14(2), 206–233. https://doi.org/10.1057/fp.2016.4

77.

Zheng

Shahin

(2020). Live tweeting live debates: How Twitter reflects and refracts the US political climate in a campaign season. Information, Communication & Society, 23(3), 337–357. https://doi.org/10.1080/1369118X.2018.1503697

78.

Zhu

Milavsky

J. R.

Biswas

(1994). Do televised debates affect image perception more than issue knowledge? A study of the first 1992 presidential debate. Human Communication Research, 20(3), 302–333. https://doi.org/10.1111/j.1468-2958.1994.tb00325.x

Tweeting Along Partisan Lines: Identity-Motivated Elaboration and Presidential Debates

Abstract

Keywords

Presidential Debates

Live-Tweeting as Cognitive Engagement

Identity-Motivated Elaboration

Partisan Social Identity

Partisan-Motivated Reasoning and Biased Elaboration

Elaboration and Political Attitudes

Multifactor Approach to Candidate Image Evaluation

Method

Participants

Procedure

Measures

Partisan Affiliation

Partisan Social Identity

Elaboration

Image Evaluations

Change in Preference for Ingroup Candidate

Content Analysis

Results

Hypothesized Model

Discussion

Limitations and Future Directions

Conclusion

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iD

Author Biographies

References