Sustainable Political Social Media Marketing: Effects of Structural Features in Plain Text Messages

The success of Barack Obama’s 2008 U.S. presidential campaign led politicians and voters all over the world to pay attention to social media. Including Donald Trump for his upcoming 2020 re-election, many politicians around the world have used social media for their political campaigns. While some social media can deliver information in various forms (i.e., video, audio, and interactive content), some popular ones, such as Twitter, are still focused mostly on plain text messaging. With political marketing using simple text messages via social media, there is a need to examine ways of creating messages that ultimately help shape voters’ perception of politicians and eventually win the election. Based on communication science, this study attempts to test the limited capacity model of motivated mediated message processing by examining whether this model can be applied to the simplest form of mediated message, which is plain text. In order to do so, structural features of text messages exchanged on social media engaged in political campaigns, namely linguistic formality and network-mediated human interactivity, are manipulated in an experiment. Findings suggest that linguistic formality and human interaction in plain text messages influence perceived friendliness, truthfulness, and dependability of the message source (politicians), as well as the receivers’ (constituents’) behavioral intent to vote for the message source in an upcoming election. This implies that politicians should pay more attention on sustainable political marketing through appropriate manipulation of structural features in social media messages.


Introduction
Advances in information and communication technologies (ICT) have brought sweeping changes to contemporary life including communication and decision making [1]. The most recent change that ICT brought to the media landscape is the advent of social media, such as Facebook and Twitter. Social media now serves millions of users with a significant portion of young adults as a tool for everyday communications. Beyond communication between friends and acquaintances, social media is now serving as a channel for the public to communicate with corporations and politicians [2,3]. One of the reasons is the cost-effectiveness of social media as a communication tool to enrich existing relationships and forge new ones [4].
There is a long-held belief among the public and the academics that money spent on a political campaign has direct relationship with voting behavior [5][6][7]. However, there are counter arguments. For example, Levitt [8] reported that when everything else is held constant, doubling campaign finance will buy only one extra percent of the popular vote, arguing that campaign financing is almost Hypothesis 2b (H2b). Politicians using informal language in their text messages on social media will be perceived to be as more dependable than sources using formal language.

Network-Mediated Human Interactivity
Rafaeli [39] pointed out that "interactivity is quintessentially a communication concept" and is worthy of scholars' attention. To date, research on interactive media has built a large body of literature with a wide range of topics varying from interactivity as a concept to human-computer interaction. Stromer-Galley [40,41] argued that there are fundamentally two types of interactivity: one is network-mediated human interactivity, and the other is media interaction. Though what constitutes interactivity is heavily debated by communication scholars, reciprocal communication or feedback is widely accepted as an essential phenomenon for defining interactivity [40,42], which applies to both media and network-mediated human interactivity. The effect of media interaction has been extensively studied by scholars including a number of studies on websites used for political campaigns [43]. However, media interaction mostly applies to some social media that relies heavily on delivering information in various forms including video and audio.
Van Dijk [44] and Rogers [45] suggested that network-mediated human interaction has a higher level of interactivity than media interaction. Network-mediated human interactivity is the universal core for all social media, including plain text-based ones such as Twitter. The key feature of human Sustainability 2020, 12, 5997 4 of 10 interaction, mediated or not, is responding through communication. Communication has a profound effect on social behavior. A meta-analysis by Sally [35], comparing over 100 studies, reveals a robust influence of communication in increasing cooperation rates. Even non-verbal communication, such as eye contact, is known to decrease social distance [46]. Though it is relatively slower compared to face-to-face communications, network-mediated communications do foster human relationships and related values including trust [47]. Hence, a politician, normally perceived as an authoritative figure, seen interacting with lay people in a mediated environment, is more likely to be perceived as less socially distance from constituents. The perceived social closeness of the politician should positively influence perceived friendliness and trustworthiness. Hence, the following is predicted: Hypothesis 3 (H3). Sources of responsive political text messages on social media will be perceived to be friendlier than individuals who only make statements.
Hypothesis 4a (H4a). Sources of responsive political text messages on social media will be perceived to be more truthful than individuals who only make statements.
Hypothesis 4b (H4b). Sources of responsive political text messages on social media will be perceived to be more dependable than individuals who only make statements.
Though there is no study in the literature that specifically reports interaction of linguistic formality and human interactivity on message receiver's impression, it is possible that the interaction might intensify or attenuate it. Hence, we propose the following research question (instead of formal hypothesis): Research Question 1: How will the interaction between formality and human interactivity in plain text messages on social media engaged in political campaigns affect perceived friendliness, truthfulness, and dependability of politicians?
Based on the U.S. national election study data, Jacobson [5] argued that though voters' familiarity with candidates is not automatically advantageous, voters' evaluations of candidates do have a strong influence on how they vote. Since this study hypothesizes that elements of human interactivity and formality in plain text messages on social media engaged in political campaigns will influence one's evaluation of a politician, it is natural to consider whether the influences extend to voting intent [48]. Therefore, we propose our second research question (instead of formal hypothesis): Research Question 2: How will formality and human interactivity in plain text messages on social media engaged in political campaigns affect readers' intent to vote for politicians?
In this study, political campaign through social media or the Internet, means fair, clean, and transparent communication-no lies or rumor-spreading, allowing everyone to have equal opportunity to speak up, and being open (or at least giving readers a sense that message sources are open) for communication.

Research Design
The experiment was a mixed Formality (2) × Human Interactivity (2) × Repeat: Politician (3) × Repeat: Message (6) factorial design. All factors are within-subject factor, while Politician and Message are both for repetition. Stimuli were administered in a fully randomized order.
The stimuli were designed based on the screen interface of a mobile phone application for Twitter. Twitter was chosen for its popularity, and its messages are short text (up to 140 characters until 2017, up to 280 characters since 2018), which provides an ideal yet realistic environment for testing the effectiveness of structural features in plain text. The increased chance of its users being exposed to messages from non-friends (including politicians) forwarded ("retweet" function) by friends was another reason for choosing Twitter over other social media. The content of all stimuli was created by researchers, including the profile of message source. Message source's gender, age, and position were controlled. The three fictional politicians were middleaged city councilman from different parties purportedly serving small cities remote from where the experiment took place. This was to avoid participants, who were from a large city, noticing that the messages were from imaginary politicians. The content of the messages consisted of a positive description of the politician's work/political activities, presumably written by the politician himself. Each message was shown with a short source bio (also made up by researchers) including a photograph.

Independent Variables
Formality. Labov [49] noted that "the most immediate problem to be solved in the attack on sociolinguistic structure is the quantification of the dimension of style." Though it may not be easy to exactly quantify the level of formality in language, the formality level may be manipulated by including/excluding certain structural features (e.g., words, linguistic formats), as in this study. For example, nouns (e.g., dude, ya), abbreviations (e.g., c'mon), or cultural signs acronyms (e.g., -p, lol) can be a device for indicating how (in)formal a message is. Based on the literature [50], all Korean verbs used in the stimuli either took a very formal form (hapsyo-che) or a quite casual form (heyo-che) to manipulate linguistic formality level of the message.
Interactivity. As discussed above, most digital media provide devices to indicate network-mediated human interaction in messages. Twitter provides a "mention function" that allows users to respond to one another by including her/his username, and to quote the message by that user. This function allows users to interact with each other, which Stromer-Galley [40] referred to as network-mediated human interactivity. In this study, stimuli (Twitter messages) with human interactivity used the mention function, quoting a user's question or comment and the politician's response to it. Specifically, politicians with human interactivity had half of their messages responding to someone else and half making statements, while politicians without human interactivity had statements only.

Dependent Variables
Perceived Truthfulness and Perceived Dependability. As explained above, trust, being a construct of a multidimensional nature, consists of affective and cognitive dimensions [34]. Perceived truthfulness (affective dimension) and perceived dependability (cognitive dimension) were measured by asking 'how truthful/dependable this politician seems to be' on a 7-point Likert scale.
Perceived Friendliness and Voting Intent. A 7-point Likert scale asking "how friendly this politician seems to be" was used to measure perceived friendliness. For voting intent, participants were asked to rate "how much I am willing to vote for this politician if he was to run in my district during upcoming election" on a 7-point Likert scale because politicians featured in the stimuli were supposedly from remote cities.

Procedure
Politicians, not only in Korea but also in the U.S., have been trying to engage young voters via social media [13,14,17,18]. In order to test the proposed hypotheses on this particular group, a total of 100 college students who use Twitter from two large Korean universities participated in the experiment for monetary compensation. The mean age was 23.70 (SD = 3.88). Total of 42 females and 58 males participated. The study was conducted in a laboratory equipped with desktop computers with monitors, where participants were instructed to complete an online survey.
Participants were administered total of 72 stimuli in a fully randomized order and asked to evaluate each of them in terms of how friendly, truthful, and dependable. After the stimuli presentation was over, participants completed a distractor task (read commercial messages from corporates' social media accounts and evaluate them for their effectiveness) to prevent the interaction of short-term memory (or working memory) and long-term memory during the final task. Finally, questions asking about their willingness vote for each politician in an upcoming election were asked. Demographic Sustainability 2020, 12, 5997 6 of 10 information was collected at the end of the session. Upon completion of the session, participants were thanked, received KRW 30,000 (about USD 25), debriefed, and dismissed.

Results
To test hypotheses 1 to 4, a series of repeated measure ANOVA (analysis of variance) tests using a model of Formality (2) × Human Interactivity (2) × Repeat: Politician (3) × Repeat: Message (6) on corresponding dependent variables was performed.
Political sources using informal language in their text messages on social media engaged in political campaigns were expected to be perceived as friendlier (H1), more truthful (H2a), and more dependable (H2b). The ANOVA test using the model mentioned above revealed that the main effect of formality was significant on perceived friendliness (F (1,99)  Research Question 1 asked if there would be any interaction between formality and interactivity in messages on the dependent variables. There was a significant interaction effect on perceived dependability (F (1,99) = 6.78, p = 0.011, partial η 2 = 0.064), but none on perceived friendliness or truthfulness (both F < 1). As shown in Figure 1, perceived dependability for messages with no human interactivity is low, but informal language brings this up. On the contrary, an opposite pattern was observed for messages with human interactivity.
To be more specific, this study demonstrates that structural features in social media messages clearly have effects on voters' perception of politicians and their intent to vote. Between the two dimensions of trust, formality influenced the affective dimension (perceived truthfulness), while interactivity affected the cognitive dimension (perceived dependability). Formality also has a significant effect on perceived friendliness, which is closer to the affective domain in nature. This suggests that linguistic formality, which reflects social distance, generally has an effect on evaluative measures of affective nature. Interactivity had a negative impact on perceived friendliness.
Interestingly, formality and interactivity significantly affect perceptions without interactively influencing voting intent. In this current study, all interactive messages in this study were related to the politician's work, which might have led subjects to see the politician as a serious workaholic, Sustainability 2020, 12, 5997 7 of 10 making him seem more dependable. If the content of interactive messages was not all about work, the outcome may have differed. It is possible that these two might interact if the messages include non-work-related subjects, leading readers to feel closer in social distance. partial η 2 = 0.048), but in the opposite direction (MInteraction = 4.40, SD = 0.060 vs. MNo_Interaction = 4.46, SD = 0.067). Main effect of interactivity was also significant on perceived dependability in the predicted direction (F (1,99) = 9.59, p = 0.003, partial η 2 = 0.088; MInteraction = 4.62, SD = 0.065 vs. MNo_Interaction = 4.51, SD = 0.071) but not on perceived truthfulness (F < 1), though the results for perceived truthfulness were in the direction predicted by H4a (MInteraction = 4.40, MNo_Interaction = 4.38).
Research Question 1 asked if there would be any interaction between formality and interactivity in messages on the dependent variables. There was a significant interaction effect on perceived dependability (F (1,99) = 6.78, p = 0.011, partial η 2 = 0.064), but none on perceived friendliness or truthfulness (both F < 1). As shown in Figure 1, perceived dependability for messages with no human interactivity is low, but informal language brings this up. On the contrary, an opposite pattern was observed for messages with human interactivity.

Conclusion and Discussion
Vitak et al. [51] reported that, though political activity on social media is a significant predictor of other forms of political participation, most of the young people surveyed inclined to engage in the least intrusive, least time-consuming activities. Simply reading messages posted on social media, like in this study, would qualify as such "least effortful" activity. While social media proved its effectiveness as a tool of political communication, it is rare to find research regarding the effectiveness of plain text messages based on structural features. The objective of this study is to explain how structural features of plain text based social media messages, such as linguistic formality and network-mediated human interactivity, affect constituents' intent to vote as well as perceived friendliness, truthfulness, and dependability of the politicians. This study demonstrated that simple manipulation of structural features in plain text messages on social media clearly has effects on voters' perceptions of politicians and their intent to vote.
Findings from this study provide support to findings that interactivity on social media (Twitter) in political communication had no significant effects on electoral results, based on the analysis of U.K. election data reported by Bright et al. [48]. That is, while interactivity alone may seem to have limited effect on (potential) voters, the interaction between interactivity and (in)formality shows that trying to be friendly (by using informal language) cancels out the effect of interactivity while portraying authoritative image (by use of formal language) brings interactivity back to life (Figure 1). This particular finding suggests that interactivity still has potential as an important component in the context of political communication.
Beyond sustainable political campaigning, the findings also have implications for sustainable personal marketing and public relations using social media. Politicians are trying to use social media to engage young voters, including the recent U.S. presidential campaign, but only a small amount of research has been undertaken on the impact of this media on them [21]. In light of this situation, findings from this study may have important theoretical and managerial implications, though findings Sustainability 2020, 12, 5997 8 of 10 are limited by using one specific group of young voters, college students, as subjects. Using fictional politicians for creating stimuli is another limitation of this study.
Nevertheless, findings from this study which used Twitter, a micro-blogging service, may be applicable to blogging services in general, and to a broader range of Internet services which allow for the posting of messages and responses to them openly. Future research may apply these findings to broader contexts such as corporate PR and personal marketing, and also explore other aspects of social media and diverse structural features in mediated messages [52]. In addition, linguistic formality is operationalized differently from language to language, and this interacts with how hierarchical/egalitarian the society is.
In addition, individual's interest in politics can influence how political messages are processed, and also influence the formation of attitude toward political candidates-which is a subject that definitely may be broaden beyond politics. It would be worthwhile for future studies to explore this venue. Likewise, we leave the following questions for future research: (1) what kind of people prefer friendlier sources (i.e., politicians), (2) who is more likely to get engaged and influenced by social media in the context of political communication and beyond, (3) will users more familiar with creating and disseminating messages through social media (or the Internet) lead to a new form of authoritarianism in the cyber space? All of these are important questions that are worth exploring, not only in the context of political communication, but mediated communication in general.
Lastly, this study was conducted in Korea, which is a fairly hierarchical society, and linguistic formality matters to people. It would be interesting to compare different cultures and languages on this subject.