Personalization of Politics on Facebook: Examining Content & Effects (PDF)

INFORMATION, COMMUNICATION & SOCIETY 2020, VOL. 23, NO. 10, 1481–1498 https://doi.org/10.1080/1369118X.2019.1581244 Personalization of politics on Facebook: examining the content and eﬀects of professional, emotional and private self- personalization Manon Metza, Sanne Kruikemeierb and Sophie Lecheler c a Mannheim Center of European Social Research, Mannheim, Germany; bAmsterdam School of Communication Research, University of Amsterdam, Amsterdam, the Netherlands; cDepartment of Communication, University of Vienna, Wien, Austria ABSTRACT ARTICLE HISTORY With the integration of social media in political communication Received 8 January 2018 repertoires, politicians now permanently campaign for support Accepted 7 February 2019 online. By promoting their personal agenda, politicians KEYWORDS increasingly proﬁle themselves independent from their associated Social media; communication parties on the web (i.e., self-personalization). By focusing on self- style; Facebook; personalization as a multi-layered concept (i.e., professional, personalization; user emotional, private self-personalization), this study investigates engagement; content both the use and consequences of self-personalization on analysis Facebook. A manual content analysis of politicians’ Facebook posts (N = 435) reveals that self-personalization is indeed often used as a communication style on Facebook and is most often present in visual communication. Moreover, the study shows that the use of a more emotional and private style provides a beneﬁcial tool for politicians’ impression management. Publishing emotional and private content yields positive eﬀects on audience engagement, suggesting audiences’ demand for more intimate and emotional impressions of public ﬁgures on the web. With the rise of social media, political communication has undergone a fundamental change. Politicians permanently campaign for support from the electorate, often indepen- dent from the electoral cycle. Social media provide the infrastructure for such permanent campaigning by giving politicians the opportunity to directly create, choose, and send con- tent to citizens (Larsson, 2015, 2016), thereby circumventing traditional mass media (Den- nis, Chadwick, & Smith, 2016, p. 11). This ultimately creates a more personalized environment, in which individual politicians’ proﬁle themselves independent from their political party. Consequently, scholars have documented a trend towards more individua- lized campaign patterns in politics and media today (Vergeer, Hermans, & Sams, 2013). Studies have started to examine the extent to which self-personalization occurs online, and the consequences of self-personalized styles as a political tactic (McGregor, 2018). First indications hint towards considerable amounts of self-personalization online (e.g., Gerodimos & Justinussen, 2015; Small, 2010) with certain types of self-personalization CONTACT Manon Metz [email protected] Mannheim Center of European Social Research, 68159 Mannheim, Germany © 2019 Informa UK Limited, trading as Taylor & Francis Group 1482 M. METZ ET AL. inducing citizen engagement both online (e.g., Bene, 2017; Borah, 2016) and oﬄine (e.g., Kruikemeier, van Noort, Vliegenthart, & de Vreese, 2013). However, while both the extent and consequences of online self-personalization have been recurring subjects of interest, previous work has not frequently examined diﬀerent types of self-personalization present on social media. This study adopts the theoretical conceptualization of Van Santen and Van Zoonen (2010) and Van Aelst, Sheafer, and Stanyer (2012) who show that personalization is a multilayered concept.1 This under- standing diverges from the traditional news media-oriented perspective seeing personali- zation as the visibility of politicians at the expense of political parties (e.g., Holtz-Bacha, Langer, & Merkle, 2014). So, to bridge the gap, we ﬁrst conduct a manual content analysis of politicians’ Facebook posts. Thereby, we take into account visual communication on social media (Abidin, 2017; Bene, 2017) by examining self-personalization in both texts and visuals. This is important as it has been suggested that visuals have the capacity to easily transmit personalized communication (Parry, 2015). Secondly, we test the eﬀects of each self-personalization type (i.e., professional, emotional, private) on audience engagement (i.e., sentiments, shares, and comments), which puts the results into a broader perspective. Thus far, research has focused extensively on who uses what type of social media with what kind of motivation and eﬀort (e.g., Baxter & Marcella, 2012; Lappas et al., 2016; Larsson & Skogerbø, 2018; Lilleker, Tenscher, & Štětka, 2015; Magin, Podschu- weit, Haßler, & Russmann, 2017; Oelsner & Heimrich, 2015; Quinlan, Gummer, Roßmann, & Wolf, 2017), providing us with a clear understanding of how political actors deploy social media. Yet, we add to this, by taking into account the perspective of the audi- ence, as this is examined far less often (expect for: Bene, 2017; Borah, 2016; Gerodimos & Justinussen, 2015). Taken together, this study asks (a) to what extent diﬀerent types of self- personalization are present in political communication on Facebook, and (b) how self-per- sonalization aﬀects audience engagement online. Personalization of politics on social media: examining the use and eﬀects Personalization online The theoretical concept of political personalization addresses the focus on individual poli- ticians at the expense of political parties and institutions (Rahat & Sheafer, 2007), and is often discussed in relation to media coverage (Van Aelst et al., 2012). From this perspec- tive, personalization is a process in which news stories increasingly feature single political ﬁgures to make a story more newsworthy (Strömbäck, 2008). Recently, however, political personalization has been introduced as a frequently used communication strategy on the web. The individual politician as the central communicator promotes a more personalized agenda, sharing a more intimate, private, and less party-centered perspective (Vergeer et al., 2013, pp. 4–6). Individual politicians position themselves as ‘ordinary’ humans behind the oﬃcial oﬃce (Bene, 2017; Gerodimos & Justinussen, 2015; Manning, Pen- fold-Mounce, Loader, Vromen, & Xenos, 2017; Ross & Bürger, 2014), emotional charac- ters that share personal feelings and emotional narratives (Bene, 2017; Borah, 2016; Bronstein, 2013; Douglas, Maruyama, Semaan, & Robertson, 2014), and as professionals who hold individual qualities and individually exert political tasks and activities (Hermans & Vergeer, 2013; Jung, Tay, Hong, Ho, & Goh, 2017; Kruikemeier, 2014). INFORMATION, COMMUNICATION & SOCIETY 1483 A more personalized communication style online closely ﬁts the personalization typol- ogies of Van Aelst et al. (2012) and Van Santen and Van Zoonen (2010), conﬁrming others who also note that personalization is a multi-layered concept (for an overview, see Langer & Sagarzazu, 2018). While Van Aelst et al. (2012) have deﬁned personalization as a two- folded concept (i.e., individualization vs. privatization), the authors themselves accept and state that behavioral aspects of personalization, such as emotions, are excluded from their deﬁnition (Van Aelst et al., 2012, pp. 214–215). Moreover, their deﬁnition is less applicable in online environments, since social media are, by nature, personalized tools that provide individual politicians with their own communication channel (Ekman & Widholm, 2015; Kaplan & Haenlein, 2010, pp. 63–64; Stieglitz & Dang-Xuan, 2013). Following the deﬁnition of Van Aelst et al. (2012), communication stemming from individual politicians would always be identiﬁed as personalized communication. In contrast, the personaliza- tion typology by Van Santen and Van Zoonen (2010) distinguishes between three content dimensions: professional, emotional, and private communication. While professional per- sonalization targets qualities and individual activities related to the oﬃcial oﬃce, emotional personalization puts the personal feelings of the politician to the forefront, and private personalization can be identiﬁed as intimate information about the private persona. This latter conceptualization is chosen in the present study, because it under- stands personalization as individualized communication, but acknowledges the multidi- mensionality of the theoretical concept. We acknowledge that Van Santen and Van Zoonen (2010) examined television portraits, that is biographies produced by journalists, which of course diﬀers from politicians who create original content themselves on their social media platforms. However, we believe that the conceptualization is inclusive and encompasses the most important aspects of online personal narratives (which is also shown in previous work, see Kruikemeier, 2014). Van Santen and Van Zoonen argue that it is ‘useful to distinguish between diﬀerent foci of personal narratives, because they have diﬀerent relevance for the political content … ’ (p. 64). Fewer studies focus on the actual communication styles of politicians and, by adopting this conceptualization of Van Santen and Van Zoonen (2010), we examine diﬀerent aspects of these communi- cation styles and investigate their relevance to the political performance of politicians. Communication that is focused at politicians’ competence, such as their professional activities, are relevant for their political performances. Politicians’ emotions say something about how a politician feels about social and political issues. Politicians’ private life says something about the person behind the political oﬃce (McGregor, 2018). We therefore apply their personalization conceptualization but extend their work by examining perso- nalization as a narrative stemming from the politician themselves on Facebook (following McGregor, 2018) in contrast to media narratives stemming from journalists. However, we also adapt McGregor’s work in so far that we use diﬀerent levels of such self- personalization. All in all, since voters may be signiﬁcantly aﬀected by personal communication styles as they increasingly rely on candidate-speciﬁc information to form political evaluations (McGregor, 2018), and because diﬀerent aspects of self-personalized communication have diﬀerent relevance for the political performance of politicians, we believe that the deﬁnition by Van Santen and Van Zoonen (2010) is most appropriate to conceptualize self-personalization on social media. Therefore, we diﬀerentiate between professional, emotional, and private self-personalization in this study. 1484 M. METZ ET AL. Presence of self-personalization on social media Scholars have argued that social media are more personally oriented networks, providing politicians with the opportunity to bypass traditional mass media by creating and choosing content themselves (Dennis et al., 2016, p. 11; Van Santen & Van Zoonen, 2010, p. 65). Hence, individualized messages, detached from their political parties (Hermans & Verg- eer, 2013), have the potential to be featured more prominently on social media. This ulti- mately creates a more personalized environment in which individual politicians’ proﬁle themselves (Enli & Skogerbø, 2013; McGregor, Lawrence, & Cardona, 2017); a process termed self-personalization (see, McGregor, 2018; McGregor et al., 2017). Overall, studies analyzing the amount of self-personalization on politicians’ social media proﬁles conﬁrm this idea: Analyzing the Twitter accounts of Australian politicians in 2009, Small (2010) shows that 63% of tweets from party leaders contained personal con- tent. Looking at the Facebook accounts of Obama and Romney, Bronstein (2013) ﬁnds 15% of Obama’s, and 25% of Romney’s posts during the course of the 2012 campaign to provide private information about the candidates and their political activities. More- over, analyzing Obama’s 2012 Facebook account, Gerodimos and Justinussen (2015) ﬁnd that more than half of all posted photos featured Obama himself. In contrast, in the 2011 campaign for the New Zealand Parliament, Facebook posts mixing personal and political issues and posts containing personal anecdotes were overshadowed by much more prominent campaign-oriented posts, issuing events, policies, and political par- ties (Ross, Fountaine, & Comrie, 2015). In a nutshell, the few existing studies suggest that social media stimulate the use of personalized communication styles. However, the deﬁnition of personalization varies in the aforementioned studies, which make it diﬃcult to compare outcomes. To date, only two studies have performed an analysis based on the deﬁnition proposed by Van Santen and Van Zoonen (2010) or similar conceptualizations. Hereby, both Krui- kemeier (2014) studying Dutch politicians’ Twitter accounts, and Hermans and Vergeer (2013) analyzing candidate website strategies across 17 countries in the 2009 European Parliament election ﬁnd high levels of professional personalization online, that is, a focus on individual professional impressions of politicians. In contrast, emotional and pri- vate forms of personalization have been used less. On the basis of these latter studies, we expect that professional self-personalization (i.e., focus on individual professional qualities and activities of a politician) is most often used by politicians on Facebook, followed by emotional (i.e., politician expresses emotions in their communication) and private self- personalization styles (i.e., focus on private life). H1: Self-personalization is present in politicians’ Facebook posts, in which a professional self- personalization style is most often used, followed by a private and emotional self-personali- zation style. The role of visuals in self-personalization on social media Neither Kruikemeier (2014) nor Hermans and Vergeer (2013) take into account visuals, when discussing personalization online. So far, this task has mainly been addressed by qualitative studies, with some recent exceptions (e.g., Filimonov, Russmann, & Svensson, 2016; Gerodimos & Justinussen, 2015; Samuel-Azran, Yarchi, & Wolfsfeld, 2016). INFORMATION, COMMUNICATION & SOCIETY 1485 However, image impression management has become more important with the rise of digital networks (Marland, 2012) and visuals have the capacity to transmit personalized communication easily (Parry, 2015). Both McAllister (2007) and Zamora (2010) suggest that visuals render presenting concrete political personas much easier than abstract politi- cal ideas; Loader, Vromen, and Xenos (2016) show that viewing politicians’ visuals on social media can sketch a more human and humorous image. This claim is supported by Filimonov et al. (2016), who ﬁnd that personalization on Instagram accounts of Swed- ish parties contained particularly high levels of personalization in visuals. Based on this initial evidence, we can assume that self-personalization ﬁnds stronger portrayal in visual communication. Moreover, as images have been argued to be especially helpful in trans- mitting the full range of emotions and intimacy in a quick and easy manner (Samuel- Azran et al., 2016; Schill, 2012), it seems plausible that particular forms of self-personali- zation (i.e., emotional, private) are more represented in visuals than other forms of self- personalization (i.e., professional). Due to the relative paucity of investigations, we pose the following research question: RQ1: Is self-personalization more present in posts including visuals than in posts without visuals? Eﬀects of self-personalization on audience engagement Due to the interactive nature of social media, people are able to provide feedback on the content politicians post on their accounts. On Facebook, this direct feedback is expressed via sentiments (i.e., likes and emojis), shares, and comments. These features serve as an indicator of how engaged people are with the content (|Gerlitz & Helmond, 2013). Some more recent studies have investigated how self-personalization styles impact engage- ment levels within Facebook. A common ﬁnding is that particularly private self-persona- lization positively impacts user engagement. Analyzing Obama’s 2012 Facebook campaign, Gerodimos and Justinussen (2015) ﬁnd that photos including private elements (e.g., the presence of Michelle Obama) generated more likes than photos that showed no private elements. Bene (2017) shows that in the Hungarian election campaign of 2014, candidates’ Facebook posts induced signiﬁcantly more likes and comments when content incorporated information on the candidate’s family or disclosures of the person behind the oﬃce. Furthermore, experimental evidence suggests beneﬁcial outcomes of private perso- nalization. Meeks (2017) shows that personalized Tweets containing aspects of private life and self-disclosure induce more positive evaluations of the respective politicians (e.g., friendliness, honesty); Kruikemeier, van Noort, Vliegenthart, and de Vreese (2016) pro- pose an indirect positive eﬀect of privatized Twitter posts on political involvement. Moreover, studies examining emotions on social media agree that emotional content can trigger audience engagement. Borah (2016) suggests that emotionality played an important role in inducing audience engagement in the US presidential election cam- paigns 2008 and 2012. Humor, enthusiasm and fear exerted positive eﬀects on both likes and shares on Facebook. Also, in a US-context, Gerodimos and Justinussen (2015) show that posts with emotional appeals induced more likes, shares, and comments than posts without emotional appeal, albeit Bene (2017) ﬁnds only posts with negative emotions leading to more reactions. 1486 M. METZ ET AL. On the contrary, professional self-personalization, focusing on individual political qual- ities and activities, resembles more traditional presentations of politics that has been referred to as ‘politics as usual’ (Larsson, 2016). This political focus may collide with the social nature of social media. To sum up, we assume that private and emotional self-personalization receive more audience engagement than professional self- personalization. H2: Private and emotional self-personalization receive more audience engagement than pro- fessional self-personalization. Method The study draws on a quantitative content analysis of German parliament members’ Face- book posts. These posts were collected over the course of the last two weeks in November 2016, which were initially identiﬁed as regular weeks with no major political events happening. While most research on personalized communication in online environments is located in US-American or Dutch contexts, this study considers Germany, widening the cases under investigation. While both US and Dutch politicians have been shown to be very social media oriented (Netherlands: Graham, Jackson, & Broersma, 2016; Hermans & Vergeer, 2013; US: Geber & Scherer, 2015) and particularly in the US the publication of personal and especially private content is more normalized (Geber & Scherer, 2015), the present study examines a political communication culture in which the adaption of social media is not as widespread and in which sharing private perspectives is not as com- mon (Geber & Scherer, 2015; Hermans & Vergeer, 2013). A second distinction from pre- vious literature is the study’s timeframe. Studies discussing personalization focus almost exclusively on immediate campaign periods (Sörensen, 2016), ignoring the potential to shape impressions outside of election campaigns. We widen the perspective, acknowled- ging that politicians today are under permanent campaign pressure (Larsson, 2015, 2016). Thus, individual politicians also proﬁle themselves in their work as governing or opposing parliament members (Giasson & Small, 2017), shaping impressions that may impact citi- zens’ voting decisions in the long run. Last, by examining Facebook, we focus on the plat- form that has the highest adoption rates in Germany among private users (Tippelt & Kupferschmitt, 2015) and politicians (Oelsner & Heimrich, 2015; Quinlan et al., 2017). Sample To select a subsample of all members of parliament, the ‘Spitzenkandidaten’2 of the sixteen federal states were chosen. This approach was based upon two premises, conﬁrmed in multi-party systems: First, Lev-On and Haleva-Amir (2018) have shown that higher- ranked candidates were more active on Facebook than politicians with lower ranks in Israel. Second, Van Aelst, van Erkel, D’heer, and Harder (2017) showed that more power- ful politicians in Belgium, such as party leaders and ministers, were most popular on Twit- ter, and this popularity was mirrored in traditional mass media. Hence, we assume that Spitzenkandidaten are more visible on Facebook, and thus reach a broader audience with their online communication. INFORMATION, COMMUNICATION & SOCIETY 1487 These politicians were then identiﬁed by extracting the ‘Landeslisten’3 of the national elections in 2013 from those four parties represented in the 17th national parliament. A total of 64 Spitzenkandidaten were identiﬁed, of which 38 were included in the ﬁnal sample.4 All Facebook posts were automatically extracted using the tool Facepager (Keyl- ing & Jünger, 2017). Using Facepager, all data were collected simultaneously in order to guarantee that potential changes in metrics over time (e.g., number of posts) could not aﬀect the results. In sum, 435 Facebook posts were collected and coded. For each post, we collected sentiments (i.e., likes and emojis), shares and comments. Moreover, we included a number of control variables that have been shown to contribute to Facebook posts’ popularity (i.e., gender, age, party aﬃliation of the individual politicians and the number of overall Facebook fans). Operationalization The self-personalization indicators were derived from Van Santen and Van Zoonen (2010) and adapted to suit a social media context based on Filimonov et al. (2016), Gerodimos and Justinussen (2015), Kruikemeier (2014), and Otto and Maier (2016). Self-personalization levels were considered present when the post featured the individ- ual politician explicitly (i.e., I-form, name or image of the politician). This approach assured that party-centered communication (i.e., we-form or in the name of the associated party) was not considered self-personalization. Text and visuals were coded simultaneously. Due to the higher complexity of coding visuals, the codebook included several instructions and examples for visual coding.5 A brief overview of the coding scheme is presented in Table 1. Each indicator was ﬁrst coded as present (1) or absent (0). Scores of all indicators were summed and subsequently turned into binary self-personalization indices (0 = self-personalization level not present, 1 = self- personalization level present): Professional self-personalization (M =.33, SD =.47), emotional self-personalization (M =.33, SD =.47) and private self-personalization (M =.16, SD =.37). Table 1. Self-personalization levels, indicators and operationalization. Self- Personalization Indicators Operationalization Professional Reference to professional The politician themselves is shown performing activities related to the activities political function as a member of parliament (e.g., participation in panel discussions regarding political topics). Reference to professional Professional qualities related to the politician’s role as a member of qualities parliament (e.g., honesty, experience, integrity) are shown. Professional audience The politician acts as a professional on social media directly speaking to the appeal audience (e.g., ‘ihr’, ‘follower’, ‘fans’) or using the imperative mode (implying a special status comparable to celebrities on social media). Emotional Emotional expression Emotions of politician (e.g., joy, astonishment, anger, grief) are displayed. Emotional appeal Emotional content (e.g., emojis, memes, humor) is used by the politician. Private Personal thought Politician expresses thoughts using opinion verbs (e.g., ‘denken’, ‘glauben’, ‘ﬁnden’). Private information Information that is not related to the politician’s role as a member of parliament is shared. This information is considered intimate and concerns the person behind the oﬃcial oﬃce (e.g., marital status, children, favorite football club). Private impression Intimate impression of the politician’s surrounding (i.e., use of selﬁe- perspective) are shown. 1488 M. METZ ET AL. The audience engagement indices shares (M = 67.93, SD = 322.12) and comments (M = 34.41, SD = 99.26) were directly collected via Facepager, likes and emojis were added to create the sentiments index (M = 568.27, SD = 2388.56). Reliability The coding of the self-personalization levels was performed by the ﬁrst author of the study, who trained one additional coder. A subsample of approximately ten percent from the complete sample was drawn randomly to test the quality of the coding instrument. Inter-coder reliability was assessed with two measures. Both Krippendorﬀ’s Alpha (Kal- pha) and the standardized Lotus Gold Standard (S-GS) by Fretwurst (2013)6 were calcu- lated using SPSS version 23.0. Due to the explorative nature of the study and the skewed dichotomous measurements, we consider the S-GS a more adequate measures of inter- coder reliability for the present dataset, as more conservative measures are known for pun- ishing data structures like ours (Aaldering & Vliegenthart, 2016). The results for pro- fessional (Kalpha =.72, S-GS =.82), emotional (Kalpha =.70, S-GS =.82) and private self-personalization (Kalpha =.69, std. S-GS =.89) are considered reliable. Given that the S-GS is above.80 for all self-personalization levels and even the conservative Kalpha values exceed.67 (Krippendorﬀ, 2012), inter-coder reliability is considered acceptable. Sample description The median age in the sample is approximately 54: The oldest candidate in the sample was 73, the youngest 37 years old. Due to the aforementioned approach, there is a bias towards more popular candidates. The distribution among the four parties represented in parlia- ment is rather equal: Nine politicians from ‘CDU/CSU’, ‘SPD’ and ‘Gruene’ and with ele- ven politicians slightly more from the ‘Linke’. The sample contains 19 male and 19 female politicians. The distribution of female and male candidates is unequal across the four par- ties. Strongest diﬀerences are found between ‘Gruene’ and ‘CDU/CSU’: the sample of the Green party includes majorly female politicians, the Union parties ‘CDU/CSU’ on the con- trary incorporate more male politicians. Results First, the levels of self-personalization in the collected Facebook posts were examined. Table 2 shows qualitative examples of posts including either one of the personalization levels. In line with H1, it was found that posts with professional self-personalization are most common in the sample (33.3%, N = 145). Interestingly, the share of posts containing emotional self-personalization is equally high (33.1%, N = 144).7 Posts with private self- personalization are less represented in the sample (16.3%, N = 71).8 Given these results, H1 can be conﬁrmed: politicians’ online communication prominently features self-per- sonalized elements, and this is true for both professional and emotional self-personaliza- tion. In contrast, in our sample, politicians seem to be rather hesitant to share private impressions and information on Facebook. This may be partly in line with previous work which showed that German political websites were less personalized compared to INFORMATION, COMMUNICATION & SOCIETY 1489 Table 2. Qualitative Coding Examples. Professional Emotional Private ‘Today, I am participating in a discussion ‘My deepest condolence goes to his ‘It says “See you again” on the cross of at the IT-summit of the government in family. I knew Peter Hintze since my parents’ grave in Bosau at the Saarbrücken about the advanced we entered parliament together. Pläuner See. The cross stems from training in the digital era. In the course He had clear visions and fought for the family graveyard in Kurland/ of the platform “Digital Work them. We will miss him dearly.’b Lettland. Today, candles were lighted Environment” we have collected many at the Bosauer church for the dead of good examples in the ministry of the last year – also for my mother, work. The respective leaﬂet can, of who died in July. What a nice course, be downloaded at www.bmas. custom.’c de or ordered – rather traditionally – as a print product. Have fun reading!’a ‘The StartGreen Award honours ‘Humor is when you laugh despite : ‘Coming out of my small Prenzlauer innovative Start-Up Ideas and young D’d neighborhood and seeing a beautiful entrepreneurs. Today, I had the sunrise http://ift.tt/2gfoJ1h’ chance to welcome the ﬁnalists to the award ceremony at the ministry of environment and urban development: https://start-green.net/award/ ﬁnalisten/’ a Andrea Nahles (SPD), 17.11.2016 [Post included photo of two leaﬂets]. b Maria Böhmer (CDU), 28.11.2016 [Post included link to an obituary]. c Elisabeth Motschmann (CDU), 20.11.2016 [Post included three photos of the cross, the graveyard and the church]. d Elke Ferner (SPD), 18.11.2016 [Post included a shared post illustrating Michelle Obama ironically holding a sign ‘An immi- grant is taking my job’]. other (West-European) countries (Kruikemeier, Aparaschivei, Boomgaarden, van Noort, & Vliegenthart, 2015), indicating that this ﬁnding might be context dependent. In response to RQ1, Chi-square tests of independence were calculated between post type (i.e., text vs. text-visual) and overall self-personalization as well as each self-persona- lization level. The relation between post type and self-personalization is signiﬁcant, X 2(1, N = 435) = 45.19, p <.001. Self-personalized posts are not equally distributed across posts with and without visuals (M =.61, SD =.49). While in text posts only 39.2% contained a self-personalized style, 60.8% of the text-visual posts embedded self-personalized elements. The diﬀerence between post types (i.e., text vs. text-visual) and the use of self-persona- lization were signiﬁcant for all three types of self-personalization: We found that the use of professional self-personalization was higher in a post containing a visual (Mprofessional =.61, SDprofessional =.49) compared to a post without a visual (Mprofessional =.41, SDprofessional =.49), X 2 (1, N = 435) = 14.45, p <.001. Additionally, also the use of emotional self-personalization was higher in a post containing a visual (Memotional =.70 SDemotional =.46) than in a post without a visual (Memotional =.37, SDemotional = 48), X 2 (1, N = 435) = 42.99, p <.001. Lastly, the use of private self-personalization was also higher in a text-visual post (Mprivate =.66, SDprivate =.48) compared to a text-post (Mprivate =.44, SDprivate =.50), X 2 (1, N = 435) = 11.49, p <.01. Thus, self-personalization is more present in posts including visuals than in posts without visuals and this is true for emotional and private self-personalization, but also for professional self-personalization (see Figure 1). Last, we performed a multiple regression analysis to examine the impact of self-perso- nalization on audience engagement. Because observations are not independent and because the posts (which is our unit of analysis) are nested in politicians, we used clustered standard errors. Table 3 presents the ﬁndings for all three audience engagement 1490 M. METZ ET AL. Figure 1. Percent comparisons for professional (N = 145), emotional (N = 144) and private self-perso- nalization (N = 71) on posts without and with visuals. Table 3. Direct eﬀects of self-personalization levels on sentiments, shares and comments. Sentiments Shares Comments Constant 4.04 (1.41) 1.06 (1.09) 2.23 (1.62) Professional −.24 (.02) −.22 (.20).04 (.26) Emotional.79*** (.17).31# (.16).51** (.16) Private.57* (.27).47 (.29).33 (.24) Control variables Age −.02 (.02) −.01 (.02) −.02 (.03) Gender.27 (.43) −.19 (.38) −.12 (.50) Linke.75 (.58) 1.03 (.54).63 (.66) Gruene.73 (.55).38 (.41).96 (.69) CDU/CSU −.26 (.46) −.47 (.35) −.60 (.58) Fanpage Likes.00** (.00).00*** (.00).00** (.00) N 435 435 435 R2.38.39.31 Note: Signiﬁcance levels based on two-tailed signiﬁcance tests. # p <.10, *p <.05, **p <.01, ***p <.001. a For the dependent continuous variables, the natural logarithm was used due to non-normal distribution. b The eﬀects of emotional and private self-personalization become stronger when tested without control variables. INFORMATION, COMMUNICATION & SOCIETY 1491 indicators.9 Partially conﬁrming H2, we ﬁnd that both emotional (b =.79, t(425) = 4.50, p <.001) and private self-personalization (b =.57, t(425) = 2.10, p <.05) positively and sig- niﬁcantly impact sentiments. Moreover, emotional self-personalization exerts additional positive eﬀects on shares (b =.31, t(425) = 1.87, p <.10) and comments (b =.51, t(425) = 3.21, p <.01). In line with H2, professional self-personalization does not aﬀect audience engagement. None of the control variables10 exert a signiﬁcant eﬀect on audience engage- ment with the exception of the overall amount of Facebook Fanpage likes.11 Subsequently, we tested for interaction eﬀects between self-personalization and post type (i.e., text vs. text-visual). Our results indicate a clear pattern: Private self-personaliza- tion negatively interacts with visuals on all audience engagement variables (bsentiments = −1.26, t(424) = −2.67, p <.05; bshares = −1.15, t(424) = −1.75, p <.10; bcomments = −1.67, t (424) = −3.29, p <.01). This negative interaction eﬀect suggests that for posts not contain- ing a visual, the inclusion of private elements induces more audience engagement, while there is no interaction eﬀect present if the post contains a visual. In sum, it seems that poli- ticians’ impression management via Facebook positively aﬀect audience online engage- ment, if politicians are willing to present themselves in a more emotional and intimate fashion. Conclusion and discussion Social media fundamentally changed how politicians communicate to the public. Poli- ticians do not only increasingly use online platforms to receive support in election times (e.g., Gulati & Williams, 2007; Zittel, 2009), but today they are in a state of perma- nent campaigning on the various social media channels available (e.g., Larsson, 2015, 2016). Drawing on our manual content analysis, this study is one of the ﬁrst to scrutinize the extent and eﬀects of such self-personalization patterns on Facebook. Applying a com- prehensive, conceptual model of diﬀerent types of self-personalization, the study was able to show that politicians frequently employ professional and emotional self-personaliza- tion, and in turn, emotional and private self-personalization positively aﬀect audience engagement. This audience demand for more private impressions of public ﬁgures is in line with a documented rise in journalism’s interest in the private life of politicians (Ekman & Widholm, 2015) and reported positive outcomes of private self-personalization strategies in experiments (e.g., Colliander et al., 2017; Meeks, 2017). However, expanding previous ﬁndings, our study implies that audiences also strive for emotional self-presenta- tions of politicians. Overall, we conclude that these softer self-personalization styles can be beneﬁcial tools in politicians’ impression management. This study holds several theoretical and societal implications. First, it conﬁrms that self- personalization is a multi-layered concept on social media (Van Aelst et al., 2012; Van Santen & Van Zoonen, 2010). The distinction between professional, emotional, and pri- vate self-personalization provided an adequate framework for examining political perso- nalization online. We found that all three types of self-personalization play an important role in politicians’ social media performance. Still, the results revealed that it is valuable to look at the extent and eﬀects of each layer individually. Hereby, we detected a paradox: On the one hand, professional self-personalization is one of the most used self-personalization strategies online, while it does not exert any eﬀect on audience engagement. On the other hand, private self-personalization can positively stimulate audience engagement, but is 1492 M. METZ ET AL. rarely used. Future work should examine why emotional and private self-personalization trigger audience engagement. One possibility may be that an increase in staged and pro- fessionally created images online rises the demand for more natural images of politicians as a counter-tactic (Enli, 2016). Thus, uploading a selﬁe of one’s morning run or speaking from personal experience could contrast this highly professional environment by persuad- ing audiences with authenticity (Enli & Skogerbø, 2013; Helms, 2012; Parry, 2015). Second, as one of the ﬁrst studies considering visual elements in politicians’ social media use, this analysis shed light on the importance of visuals to promote personalized communication styles. In the present sample, self-personalization was much more promi- nently featured in posts including photos or videos, and this diﬀerence was present for all three self-personalization types. Last, it is important to consider regular terms of oﬃce more rigorously when discussing personalization, as voters do not only evaluate politicians during the relatively short elec- tion period, but also throughout their term of oﬃce (Giasson & Small, 2017). The study has shown that self-personalization is much applied in regular government periods, suggesting that self-personalization is more than an election campaign strategy. Beside these theoretical and societal implications, our ﬁndings oﬀer insights for political marketers and politicians. Our analysis revealed that roughly 40% of the sample did not oﬀer a public Facebook account. These politicians miss out on the opportunity to gain more control over their personal narratives. Several studies have shown that voters’ evalu- ations of politicians’ characters impact their voting decisions (e.g., Funk, 1999; Hayes, 2005) and that also non-political character traits such as integrity play a role in the decision-making process (e.g., Lavine & Gschwend, 2007; Olivola & Todorov, 2010). In addition, recent political elections have highlighted the appeal of candidates such as Justin Trudeau or Donald Trump who oﬀer a more personal and ordinary and, therefore, auth- entic image through social media (Enli, 2017; Lalancette & Raynauld, 2017). Taken together, impression management via social media can help politicians to connect to citi- zens on a more emotional and intimate level which may satisfy a growing demand for poli- ticians to be ‘one of us’. Moreover, candidates perceived less newsworthy by traditional media could beneﬁt from the incorporation of social media by generating greater visibility (Skovsgaard & Van Dalen, 2013). However, adapting to a rhetoric in which ‘style, looks and visual performativity are central features’ (Ekman & Widholm, 2017) may also thrive depoliticization. While many politicians today discuss and present their work on Facebook sparking at least some debate about political issues, a shift towards more private and emotional content could further prevent people from substantial political exchange in the digital sphere. Still, self-personalization is an indispensable communication strategy for politicians nowadays. Limitations and suggestions for future research It is important to keep in mind that the study did not explore communication content beyond the type of self-personalization. Although the results promise positive eﬀects from the use of social media for political communication, one can expect that content must ﬁt the politician’s and party’s proﬁle, be ethically correct and appealing in order to cre- ate these eﬀects. Hence, the relationship between actual (political) content and audience engagement should be addressed in future research. Moreover, we direct future research INFORMATION, COMMUNICATION & SOCIETY 1493 to examine self-personalization patterns beyond prominent politicians included in the pre- sent, and most datasets on online personalization. One could expect that communication patterns of second-row or local politicians diverge signiﬁcantly from top politicians, who may be less willing to share private information, given their greater exposure in the news media (Van Aelst et al., 2017). In addition, we suggest looking at the present phenomenon comparatively to detect potential similarities and diﬀerences across countries. Since pre- vious research has stated that the use of social media in Germany is rather limited, our ﬁndings could be the result of a more discreet German communication culture. Last, although a personalized way of communication on social media potentially con- tributes to a politician’s image, for most accounts it seems logical that the audience is lar- gely composed of politically interested and even ideologically like-minded people. Thus, we should be cautious not to overestimate the eﬀects found in our analysis. Still, social media provide a promising venue for politicians who intend to gain more control over their narrative, as this study has shown. Notes 1. It should be noted that although these studies clarify the conceptual deﬁnition of personali- zation, they do not measure the actual use by examining the actual presence of these types of personalization. 2. The political candidates placed on top of the party lists (rank 1). 3. ‘Landeslisten’ are tied to the second vote in the German electoral system. In each federal state, political parties select and place candidates in order on these party lists. German voters then vote for a party list in their respective federal state (‘Landeslisten’, n.d.). 4. 5 politicians had left parliament by the time of the data conduction; 2 politicians did not make it to parliament in the 2013 elections; 18 politicians did either not have a Facebook account or only a private account and 1 politician was excluded due to posting inactivity in the study’s timeframe. In sum, 26 out of 64 politicians were excluded from the sample. 5. First, to identify the politicians visually, an image of each politician was included in the codebook. Second, using several examples we illustrated how to code visual posts. For example, an emotional expression in an image or video was deﬁned as a visible emotion (e.g., strong laugh, cry). 6. The gold standard instead of the Lotus was reported because the ﬁrst author of the study embodies the main coder, thus the second coder is compared to the gold standard set by the ﬁrst coder. 7. A Chi-square test of independence was performed on the binary variable professional and emotional self-personalization. The relationship is insigniﬁcant X 2 (1, N = 435) = 2.29, p =.130. 8. Bivariate correlations show that none of the three self-personalization levels correlate signiﬁ- cantly with each other, indicating that professional, emotional and private self-personaliza- tion are distinct forms of self-personalization. 9. Sentiments, shares and comments are strongly correlated: Sentiments and shares (r =.81, p <.001), sentiments and comments (r =.876, p <.001) and shares and comments (r =.65, p <.001). Results present variables without natural logarithms. 10. For the party dummies, the SPD served as the reference category in the analyses. 11. The eﬀects of Facebook likes are statistically signiﬁcant, but very small. Notes on contributors Manon Metz is a PhD Student at the Department of Political Sociology at the University of Man- nheim, Germany. Her research interests include online political communication, interpersonal communication and political polarization [email: [email protected]].

Personalization of Politics on Facebook: Examining Content & Effects (PDF)

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue