1
The effects of Online Personalization and Gender:
How Online Personalization Affects Individuals’ Overall Evaluations of Male and Female Candidates
Isabel C.M. Sol Student number 5974763
Master’s Thesis Political Communication
University of Amsterdam, Graduate School of Communication Supervised by Sanne Kruikemeier
Abstract
This study examines to what extent online personalization (i.e., a focus on individual politicians instead of parties) affects individuals. The purpose of this study is threefold: first, it determines the effects of different levels of online personalization used by political actors on individuals’ evaluations of competence qualities, trustworthiness perceptions and vote intentions. Second, this study gains insight into which processes explain the relationship between personalization (i.e., intimacy) and individuals’ overall evaluations. Third, this study tests for whom these effects are most present, female or male politicians. A total of 269 participants (68.2% female) participated in the online experiment (Mage = 35.57 SDage = 14.44) which uses a 2 (individualized vs. privatized) x 2 (male vs. female) between
subjects design. A control condition was included in which depersonalization (i.e., focus on party) was manipulated. Results show that personalization positively influences closeness which in turn positively affects voters overall evaluations. Moreover, privatization (vs. individualization) leads to greater feelings of conversational human voice which positively influence trustworthiness and vote intention. Contrary to expectations, no significant moderating mediation effects were found for gender on the relationship between personalization and intimacy.
Keywords: Experiment, personalization, individualization, privatization, social media, gender,
New media are important in political communication for citizens, marketeers and politicians (Hosch-Dayican, Amrit, Aarts & Dassen, 2014; Kim, 2011). Nowadays, most politicians have either an official blog, Facebook, Twitter and/or other social media channels to communicate directly with citizens (Hosch-Dayican et al., 2014). Social media gives politicians the opportunity to interact with the public in a manner that is more personal than ever before. The use of social media is especially meaningful for female politicians as research shows women have a harder time controlling traditional media (Larson, 2001) and are differently assessed by traditional media than male politicians (Dolan & Lynch, 2014; Hayes, Lawless & Baitinger, 2014; Kahn, 1992, 1994; Krogstad & Storvik, 2012). Women are mostly associated with what are stereotypically considered female traits and policy issues as compassion and education (Dolan, 2013; Dolan & Lynch, 2014). However male traits and policy issues, as intelligence and economy (Dolan, 2013), are valued as more important for politics and are more attributed to male politicians (Falk and Kenski, 2006; Fox and Oxley, 2003; Kahn, 1996; Kittilson and Fridkin, 2008; Lawless, 2004). Through the use of social media, politicians might bypass these gender stereotypes and directly connect with the public while having more control over their image and message.
Moreover recent developments such as the rise of televised politics, the Internet (Van Aelst, Sheafer & Stanyer 2012) and the steady decline of traditional political engagement of the electorate (Hermans & Vergeer, 2012), have driven parties and politicians to search and develop new ways to persuade voters. This has led to the rise of a 'candidate-centered politics': personalization of politics. Instead of focusing on the party, candidates are predominantly selected as the face and voice of a party (Garzia, 2011; Vergeer, Hermans and Sams, 2011) This personalization trend is even more apparent since the emergence of digital social networks (Hosch-Dayican et al., 2014). Through such networks, politicians can communicate directly with citizens and increase the possibilities of interactive communication between and among parties, candidates and citizens (Hermans & Vergeer, 2012;
Hosch-Dayican et al., 2014). 'In doing so, the psychological distance between politicians and citizens is said to decrease, resulting in more favorable attitudes towards the politician' (Utz, 2009 in: Hermans & Vergeer, 2012, p. 74).
Research has been conducted on the developments of personalization, mostly in relation to media coverage, examining how politicians themselves use a more personalized campaign style (Hermans & Vergeer, 2012). There is a gap in the literature regarding the effects of personalization on voters' evaluations and behavior. For example, personalization could mobilize new and less politically interested individuals who become intrigued by the more personal thematic frame (Jebril, Albæk and De Vreese, 2012). In the political choices of this particular group, predominantly personal characteristics of politicians are important. From this perspective, examining the effects of personalization is important because it may engage that part of the electorate that is normally hard to reach. Lastly, there is no research at all that takes gender into consideration. The few studies that are related to gender issues, media and politics, predominantly focus on coverage difference without asking what this means for voters' perceptions.
The aim of the current study is to understand the effects of political personalization through the use of Facebook, the most popular social medium among politicians and voters (Boulianne, 2015), by politicians themselves. Research on the effects related to Twitter use in the Netherlands shows there is a positive correlation between personalization and preferential votes (Kruikemeier, 2014). This study aims to see if we can find similar effects, and if so, to take this one step further by examining the effects of different levels of online personalization (individualization vs. privatization) on individuals’ evaluations of competence qualities, trustworthiness perceptions and vote intentions. Second, explains this study not just what the effects are, but also how personalization processes work by looking at mediating factors (i.e. intimacy). Furthermore, examines this study for whom effects are most present by taking gender into consideration. In this way, this study will fill the current gap existing in the
theory related to personalization and start as a building block for future research.
The most important outcome measure of this study is voters' evaluations. Voters’ evaluations are, in this study, operationalized as follows: i) evaluations of competence qualities, ii) trustworthiness and iii) intention to vote for the politician/party. According to Funk (1996) candidate qualities are a crucial determinant for how we feel about politicians. Of all of the traits (i.e. reliability, competence and personal appearance) trustworthiness is deemed the most important (Garzia, 2011). Lastly is the ultimate goal of politicians to get people to vote for them to win elections; examining vote intention would prove that personalization not only changes how we think or feel about politicians, it can also change actual behavior.
Personalization
The process of political personalization has become a central concept in the field of political communication (e.g. Balmas, Rahat, Sheafer & Shenhav, 2013; Garzia, 2011; Kruikemeier, Vliegenthart & de Vreese, 2012; Lee & Oh, 2012; Rahat & Sheafer, 2007; Vos, 2014) however no consensus has been reached on neither the conceptualization (Rahat & Sheafer, 2007; Van Aelst et al., 2011) nor what the effects are (Jebril et al., 2012). In general personalization is constituted as a trend or 'a process of change over time' (Van Aelst et al., 2011, p. 205) that indicates the uprise of candidate-centered politics (Karvonen, 2010; Balmas et al. 2013). According to Karvonen (2010) 'it may be viewed as a part of an overall process of individualization of social life' (p. 4) and is related to what Swanson and Mancini type as the underlying development of socio-economic and technological modernization (Karvonen, 2010). What the outcomes of political personalization are varies widely per study. One side believes that personalization will increase cynicism, depoliticizes the political discourse, impairs the legitimacy of politics (Jebril et al., 2012) and gives too much power to one dominant leader (Balmas et al., 2013). The other side believes it will lead to heightening of political engagement, trust, interest (Kruikemeier, 2014; Van Zoonen, 2005) and is actually very democratic
because '[p]ower is now spread more evenly among leaders, elites and individual politicians, and this allows them to check and balance each other' (Balmas et al., 2013, p. 47). In contrast to other research areas, the outcomes in the field of online social media do seem to largely lean towards one direction. Effects of personalization through the use of websites are mainly positive (Enli & Skogerbø, 2013; Kruikemeier, 2014; Kruikemeier et al., 2012; Vergeer et al. 2011). Nevertheless, the field of online political personalization is still in its infants shoes and thus, here as well, further research is much needed.
Defining the concept
Seeing that the lack of clear conceptualization has led to confusing and contradictory outcomes, the first step is to clarify what political personalization exactly entails (Hotz-Bacha, et al. 2014). It is important to not tread personalization of political news as a single concept (Hotz-Bacha, et al. 2014), but to separate it to multiple dimensions (Van Aelst et al., 2011). For example, Rahat and Sheafer (2007) distinguish, by thoroughly studying personalization empirically and theoretically, between institutional personalization, media personalization and behavioral personalization. The first relates to how institutional organization can give an increasing amount of power to individuals, the second is related to a heightening of attention in and from the media to politicians and the last concentrates on personal efforts of politicians to increase attention to themselves as an individual. As well Balmas et al. (2013) use this division in a similar study on personalization. In this experimental study however, the sub-dimensions of personalization posed by Van Aelst, et al. (2011) are maintained as the central conceptual framework in this study. Based on an extensive review of the most relevant literature they operationalize political personalization by dividing it in two sub-dimensions, individualization and
privatization.
Individualization
political leaders (Balmas et al., 2013; Karvonen, 2010; Vliegenthart, Boomgaarden & Boumans, 2011), relates to a heightened visibility of the individual politicians (instead of a focus on parties) in combination with a more prominent role of that politician on the political stage (Holtz-Bacha et al., 2014), thus 'including their ideas, capacities and policies' (Van Aelst et al., 2011, p. 204). Previous research noted that individualization of online political communication positively affects feelings of political involvement, interest and behavior among the electorate (Enli & Skogerbø, 2013; Van Zoonen, 2005). The logic of why individualization affects involvement, stems from the interpersonal communication literature which refers to dyadic interaction; the verbal and nonverbal interaction between individuals operating simultaneously the sender- and receiver role. Interpersonal communication is often defined as communication between people who have some kind of dependency or knowledge of each other (Cathcart & Gumpert, 1983). This is increased by personalization because personalization is related to the key property of interpersonal communication namely intimacy.
Intimacy
Intimacy relates to proximity and personal topics of conversation (Biocca, Harms & Burgoon, 2003), predominantly exists between individuals and can be divided in intimate behavior and intimate experience. 'Intimate behavior refers to intimate sharing that involves the disclosure of personal information, preferences, and so on, while intimate experience refers to the positive feelings and perceptions resulting from partner’s responsive behavior' (Liang, Li & Turban, 2009, p. 5). It is a cognitive state wherein individuals feel directly “present” in the communication and within this process a relationship is created (Palmer, 1995). Intimacy as an overarching concept I specify in this thesis as 1) closeness, 2) conversational human voice (CHV) and 3) social presence.
Closeness is defined as how close an individual feels to politics or a politician (Kruikemeier et al., 2014). According to Garzia (2011) it is a necessary condition for electoral success of politicians
hence it would lead to greater believes of trustworthiness and responsive leadership. Research in the field of cyber-psychology shows that a more personalized website, with a focus on a person, increases the sense of closeness visitors feel in relation to the organization (Park & Lee, 2013). Also Vergeer et al. (2011) state that personal 'online networks allow regular people to become a member of the inner social circle of the political candidate. As such citizens’ proximity to candidates decreases, leading to closeness – although virtual – and might lead to more engagement on the part of the citizen' (p. 497).
CHV is focused on building and improving the online relationship (Park & Lee, 2013). It 'illustrates an engaging and natural style of organizational communication as perceived by an organization's publics based on interactions between individuals in the organization and individuals in publics' (Park & Lee, 2013, p. 265). It basically deals with how strong the perception of people is that they are communicating with other individuals, instead of lifeless organizations. Kelleher and Miller (2006) found in their study on the differences of personal blogs and organizational blogs that CHV significantly correlates with i.e. trust, satisfaction and commitment. It rises when website are created in a more personal form in which for example 'authors report on their lives and inner thoughts and feelings' (Kelleher and Miller, 2006, p. 396). Also CHV is thus an important factor in developing good relationships with publics and leads to more favorable perceptions (Park and Lee, 2013).
CHV is linked to the Social Presence Theory (Short, William & Christie, 1976) which 'has been defined as the extent to which a medium allows users to experience others as being psychologically present' (Hassanein & Head, 2007, p. 690), or feeling like being in the same room as the other (Kruikemeier et al., 2014). Social presence is constructed in the mind during interaction, or its representation, via a communication medium (Gooch & Watts, 2014). For example, followers of a Twitter page of an organization feel a stronger perception of social presence when the networking page has a human presence instead of an organizational presence and is an important factor in the development of a good relationship with the public (Park & Lee, 2013).
Based on these theoretical concepts of intimacy, it is expected that if the communication is more personalized this will lead to higher levels of intimacy, because it leads to higher feelings of closeness due to a decrease of psychological distance, higher levels of CHV due to a more engaging personal style of communication and greater feelings of social presence due to the individualized representation of the candidates (see figure 1). Based on this the first hypothesis is formulated:
H1: Individualization (vs. depersonalization) leads to leads to greater feelings of intimacy (i.e. (a) CHV
(b) social presence, (c) closeness and has subsequently a positive effect on (1) voters' evaluations of qualities, (2) trustworthiness of politicians and (3) vote intention.
Figure 1. The mediation effects of personalization (depersonalization vs. individualization vs.
privatization) on overall political evaluations via social presence, CHV and closeness. Privatization
The second dimension, privatization, refers to the increase in focus on personal traits and private life of politicians (Holtz-Bacha et al., 2014) and 'rising importance as the politician as 'ordinary' person' (Van Aelst et al., 2011, p. 206). Privatization goes beyond the question if there is a greater emphasis on the individual and also asks what kind of focus this is (Holtz-Bacha et al., 2014). By exhibiting parts of their private lives, politicians try to show the electorate that they care, try to generate sympathy and intimacy with the people to win their votes (Holtz-Bacha, 2004). According to Garzia
(2011) the electorate increasingly develops their vote on the basis of “a mental image of political leaders as persons' and 'candidate's personality traits are of crucial importance to voters for they offer an appealing shortcut to infer what sort of leader that candidate will eventually be” (p. 700).
Research is, however, not uniform considering the effects of privatization. For example, Jebril et al. (2012) found in their study on media effects that the focus on non-political traits of politicians in the news generates cynicism among the electorate. This is probably due to '[t]he closer we come to our politicians, the more we will be disappointed when they betray us' (Van Aelst et al., 2012, p. 206). On the other hand, Kruikemeier et al. (2012) finds that Dutch political candidates who used the personalized tool Twitter, and predominantly spoke about their private persona during campaigning, got more votes compared to politicians who did not. This is probably because by showing parts of their personal life, the distance between politicians and the electorate is reduced even more and the perception of intimacy rises. Based on this the second hypothesis is formulated:
H2: Privatization (vs. individualization) leads to greater feelings of intimacy (i.e. (a) CHV (b) social
presence, (c) closeness and has, subsequently, a positive effect on (1) voters' evaluations of qualities, (2) trustworthiness of politicians and (3) vote intention.
Women in Politics
A follow-up question relates to the differences between male and female politicians. Before I will discuss the theory of personalization in relation to gender, I will give an overview on the literature of women in politics and the media in general and what differs in comparison to their male counterparts.
Women from all over the world, even in conservative countries, are finally reaching the highest possible positions in politics. Examples are Benazir Butto in Pakistan, Margaret Thatcher in England and Dilma Rousseff in Brazil. Nonetheless, even though a growing amount of women are breaking
through the glass ceiling and gender diversity at the political top has grown, women still are very much underrepresented (López-Hermida Russo & Cerda Diez, 2012, p. 199). One of the main reasons given for this is that citizens and the media have the tendency to apply gender stereotypes when assessing politicians (e.g. Kahn, 1992, 1994). According to Cook (1998) '[v]oters use a candidate's gender as a cure to make inferences about the candidate's character traits, beliefs, and position on gender-related policies and to choose their representatives' (Schneider, 2014, p. 56). Political gender stereotypes apply to, for example, character, when questioning leadership, to the concern of certain issues and strength in policy, and are most importantly in favor of male candidates (López-Hermida Russo & Cerda Diez, 2012). Stereotypes are used whenever voters lack information to “fill in the blanks” in combination with current political knowledge as the basis for political evaluation. The less information we have about a political candidate, the more extensively we make use of these so called shortcuts (Aalberg & Jenssen, 2007; Kahn, 1996).
A second reason why women are still underrepresented is related to “gender blindness” among women and gender favoritism among men. Aalberg & Jenssen (2007) found, by doing an experiment, that when female and male political candidates presented the same speech, overall the men were perceived more knowledgeable, trustworthy and convincing. This was because even though women in the sample did not favor candidates of their own gender, the men most definitely did. Furthermore, did they rate as well the political party they represented, as the candidate more popular when it was represented by a male candidate. Females only scored higher among men on the negative traits, namely the speech was considered more boring and irrelevant when delivered by female candidates (Aalberg & Jenssen, 2007). This is related to the fact that the assumption 'think power, think male' still is very much rooted in politics and society in general (Campus, 2013, p. 11). What is considered as male traits is also considered as the positive norm; consequently male voters consistently prefer male politicians over female politicians. Thus even when women (re)present the same message female candidates are
still continuously put in an unfavorable condition, compared to male candidates. Personalization and Gender
In general scholars have studied many aspects of personalization, also in relation to social media, but they tend to ignore the differences between politicians of opposing gender. Even though studies have been done on issue and campaign strategies, media coverage and experiences of female candidates (Van Zoonen, 2006), a part from Van Zoonen in small detail (1998), scholars have not looked at the differences that political personalization produces for male and female candidates. No study has asked to what extent gender moderates the relation of personalization, intimacy and overall political evaluations. Therefore, I will expand my focus to related research fields.
For instance, Van Zoonen (2006) found examining the increase of celebrity politics, which is constituted by personalization and popularization, 'in politics presents a complex and often unfavorable arena to women because of its inbuilt and extreme polarization of femininity in politics' (p. 287). She states that the convergence of personal and political works better for men because female candidates could attract attention to their “abnormal” gender choices in their private lives (i.e. not being married) or showing the tension between their public and stereotypical private duties. Male candidates, on the other hand show, by introducing voters to their private lives, 'that they are complete human beings combining caring and working responsibilities (p. 298). Angela Merkel and Tarja Halonen, the Finish President, are two examples of successful highly positioned candidates whom very much kept their private lives private.
On the other side, emphasizing ones gender and reinforcing these personal stereotypical duties is not necessarily a bad decision. Both Kahn (1992) and Sanbonmatsu (2002) state that stereotypes can also lead to positive evaluations of female candidates because they are assigned specific expertise in handling stereotypically female issues. Female candidates can make use of these stereotypes by, for example, emphasizing their strength during campaigning (Kahn, 1992). Furthermore can female
candidates stand for different issues, which can attract new or less mobilized voters (Aalberg & Jenssen, 2007). Female candidates should therefore, according to Bystrom (2006), 'develop sophisticated web sites that provide more specialized messages to specific groups, (…) and generate a more personalized presence with voters (e.g., through audio/visual presentations by the candidate)' (p. 23).
Opposed to Halonen and Merkel who kept private (Van Zoonen, 2006), Hilary Clinton (Campus, 2013) and Michelle Bachelet (López-Hermida Russo & Cerda Diez 2012), for example, have successfully chosen a much more personalized strategy. According to López-Hermida Russo and Cerda Diez (2012), Bachelet and the media emphasized private aspects (i.e. being a daughter and a mother which are highly intimate discourses), both as a campaign strategy and as an editorial choice. Personalization can thus pose an advantage for women because, according to linguist Deborah Tannen (1990), 'women speak and hear the language of connection an intimacy' (Campus, 2013, p. 98). As noted, it is assumed that intimacy positively mediates the relationship between personalization and overall evaluations (i.e. Enli & Skogerbø, 2013). For women it is more natural than for men to bring their intimate and private selves into the public domain. Especially the rise of the television has brought an advantage because it is considered as an intimate medium and thus is more suited to female politicians (Jamieson, 1988 in: Campus, 2013); the same can be said about SNS and Facebook in particular. Through this medium voters can become acquainted with the politician from a distance whilst she can simultaneously constructs a personal image of oneself circumventing any interpretation of the media. In addition, the more information (female) politicians decide to share themselves, the more control they have over the way they are portrayed in the media to the public (Campus, 2013). Based on this, we would expect that female politicians have the upper hand in relation to personalization; it is for women more natural to create feelings of intimacy which could have a positive moderating effect.
However, linking back to the idea of the male norm (Campus, 2013), the question remains if we value female and male gestures of intimacy the same. In theory, personalization leads to the more of an idea of “commander in chief” as suppose to “first among equals”. Therefore Sykes (2008) states that personalization (also in parliamentary systems) 'threaten(s) to make it increasingly difficult for a woman to become leader. As a general rule, the stronger the executive office, the harder it is for a woman to win it, unless she emulates the Iron Lady' (p. 762). Thus, even though intimacy is something considered naturally feminine, the validation may not be the same as for men in the political field (Van Zoonen, 2006). Consequently this will have a different effect on the mediating relationship between privatization and overall evaluations (see figure 2). Based on these latter findings the third hypotheses is formulated: H3: Gender moderates the indirect effect of personalization on intimacy (i.e. (a) CHV (b) social presence, (c) closeness) and has subsequently a positive effect on (1) voters' evaluations of qualities, (2) trustworthiness of politicians and (3) vote intention.
Figure 2. The moderated mediation effects of privatization (vs. individualization) on overall political
evaluations via social presence, CHV and Closeness, moderated by gender. Method
Participants
To empirically test the posed hypotheses and investigate the research question, an online experiment was conducted with a 2 (individualized vs. privatized) x 2 (male vs. female) between
subjects design. A control condition was included in which depersonalization (i.e., focus on party) was manipulated. An experimental design is most suitable to “establish the existence of a cause-and-effect relationship between two variables” (Gravetter & Forzano, 2010, p. 198). Respondents were retrieved through Facebook and by email. In total 269 participants (68.2% female) participated in the online experiment (Mage= 35.57 SDage = 14.44), who were overall highly educated (M = 7.12, SD = 1.82).
Materials
The stimulus material was developed using the existing Facebook page of the Dutch political party D66 because this makes the stimulus material more realistic. D66 was chosen because it is considered as a party that is centered in the middle of the political specter of left and right in the
Netherlands (Parlement, 2015). This way the political bias was reduced as much as possible. Next, four additional political Facebook pages were developed for the individualized and privatized male and female fictional candidates; see Appendix B. The pages showed manipulated messages of the general D66 page, other than the fact that these were more individualized and privatized, the text was kept the same. Each page contained three messages with a different amount of Facebook “likes” and shares per message, but identical over the three pages. The content was about a new nature reserve, the refugee crisis in Europe and a party conference of the D66. The individualized messages were written from an 'I' standpoint, the privatized messages also had information about non-political life incorporated. The privatized pages furthermore included some additional background information that is their place of birth and age. All pages were created by the use of Photoshop.
Manipulation check
To insure the stimulus materials of the experiment had the intended effects I ran two
manipulation checks. The first manipulation check, a pretest, was conducted among 19 people (58.8% female, Mage = 30.41, SDage = 10.47). Each respondent was assigned to one of the five conditions. To
if they perceived the Facebook messages on the page to be written by an individual or party, on a seven-point Likert scale (1 = party, 7 = individual). By conducting an independent samples t-test it showed that there was a significant difference in the scores for the personalized material (M = 5.85, SD = .619) and of the depersonalized stimulus material (M = 1.00, SD = .000); t(15) = 4.25, p = .001. Thus the personalized stimulus material was effective. Privatization (vs. individualization) was measured by three questions and ten items in total on a seven-point Likert scale (i.e. 'the messages I just saw were personal or intimate or private'). However, an independent samples t-test showed that there was no significant difference in scores for the privatized material level (M = 2.86, SD = 1.11) and the non-privatized stimulus material (M = 3.55, SD = 1.51); t(11) = -.879, p = .389. Also, the individualized material even scored higher on levels of privatization. Therefore, the privatized messages were adjusted for the final experiment and made more privatized.
In the finalized questionnaire the second manipulation check was implemented. An independent samples t-test showed again that there was a very significant difference in the scores for the
personalized material level (M = 5.00, SD = 1.82) and the level of the depersonalized stimulus material
(M = 2.63, SD = 1.91) conditions; t(188) = 6.90, p = .000. Privatization (vs. individualization) was
measured this time by using three items on a seven-point Likert scale (i.e. 1 = the messages were not
about private life, 7 = the messages were about private life). This time an independent-samples t-test
showed that there was a very significant difference in scores for the privatized material level (M = 4.57,
SD = 1.45) and the individualized stimulus material (M = 2.5, SD = 1.20) conditions; t(153) = 9.49, p =
.000. The stimulus material was thus successfully in manipulated in the final questionnaire. Procedure
Respondents were randomly assigned to one of the five conditions. Before beginning the experiment a general opening statement was given, their anonymity was insured and respondents were asked for their agreement. Next some questions about political preferences were asked, followed by
questioning about their Facebook usage. The latter questions were purely implemented to reduce as much as possible any bias regarding personalization and gender. Next, one of the five manipulated Facebook accounts was shown and the respondents were asked to imagine as if they were visiting the Facebook pages from their own personal accounts. To ensure that they would read everything, the continue button did not appear until after 20 seconds. Also incorporated was a debrief immediately after the questionnaire, explaining what was manipulated and with a contact opportunity and
information for questions or remarks. It furthermore included some additional information about the study, the call for discretion about the content and they were thanked for their participation.
Measures
The following measures were included:
Evaluations. First, based on Aalberg & Jenssen (2007), the qualities were measured by seven items on a seven-point Likert scale (1= totally disagree, 7 = totally agree). Examples of the items are how truthful and convincing the party or candidates were perceived by the respondents. This scale showed to be reliable α = .90, items load on one scale (Eigenvalue = 4.46 , Explained variance = 63.72%, M = 4.11, SD = 1,01).
Trustworthiness. Trustworthiness was measured by a total of six items adopted from Carlin (2013) on a seven-point Likert scale (1= totally disagree, 7 = totally agree). Questions entailed for example the party or candidate 'will keep their promises' and 'will usually make the right policy decisions' Each variable was measured by three items on a seven-point Likert scale. Also this scale proved to be very reliable α =.93, items load on one scale (Eigenvalue = 4.48, Explained variance = 74.72%, M = 3,85, SD = 1,09).
Vote intention. Vote intention was measured by asking the respondents if they would vote for the party or candidate after reading the page on a seven-point Likert scale (1 = totally disagree, 7 =
Social presence. Social presence was measured by using three items on a seven-point Likert scale adopted from Lee and Oh (2012) (1 = totally disagree, 7 = totally agree). Questions included 'I felt as if I were engaging in an actual conversation with the party/candidate' and 'I felt as if the
party/candidate was speaking directly to me'. The scale proved to be reliable α =.94, items load on one scale (Eigenvalue = 2,67, Explained variance = 88,99%, M = 2,35, SD =1,25).
Closeness. Closeness was measured using four items which were adopted from Kruikemeier et al. (2012) and were measured on a seven-point Likert scale (1 = totally disagree, 7 = totally agree). Questions included 'This website gives me the feeling that I am closer to politics' and 'This website gives me the feeling that politicians are more involved with their electorate'. This scale showed to be reliable α =.92, items load on one scale (Eigenvalue = 2,59, Explained variance = 86,28%, M = 3,43,
SD =1,44).
Conversational human voice. Conversational human voice was measured using seven items also on a seven-point Likert-scale (1 = totally disagree, 7 = totally agree). The items were retrieved from Kelleher & Miller (2006) and included questions as the party or candidate 'uses an open style of communication' and 'is open to conversation'. This scale showed to be reliable α =.88, items load on one scale (Eigenvalue = 4.08, Explained variance = 58,23%, M = 4.42, SD = .90).
Control variables. Besides these variables multiple control variables were included and measured. Control variables included in this study were, political interest which was measured using three items (political interest in general, on a local level and national level) on a seven-point Likert scale (1 = not interested at all, 7 = very interested) (M = 4.30, SD = 1.42); political preference, measured on a eleven-point Likert scale (1 = left, 11= right) (M = 4.26, SD = 2.06); knowledge of the party D66 (96.3% said yes); if they visited the D66 Facebook page before (8.9% said yes); and if they “liked” the D66 Facebook page (8.4% said yes). Lastly, gender (68.2%), age (Mage= 35.57 SDage =
found the fictitious male and female candidate, measured by three items (handsome, attractive, niceness) on a seven-point Likert scale (1 = not at all, 7 = very much) (M = 4.53, SD = 0.89). Analyses
To test if there are significant direct effects of personalization and privatization we conduct multiple multivariate analyses of variance (MANOVA's) and multivariate analyses of covariance (MANCOVA's) to compare the multivariate sample means. During the MANCOVA we control for attractiveness. This can only be done when comparing the personalized manipulations and leaving out the control group since it was not tested for attractiveness. To test the mediation effects for the hypotheses we use Hayes' PROCESS macro (model 4). 'This macro uses an ordinary least squares or logistic regression-based path analytical framework to estimate the direct and indirect effects in mediator models' (Boerman, 2014, p. 117). In contrast to Baron and Kenny's method (1986), with Hayes' macro we can test if there is an indirect relationship between the independent variable and dependent variable via a mediation variable, without any direct effect. That makes this model suited for this study because no direct effects are assumed. To estimate the bias-corrected bootstrap confidence intervals (BCBCI) and thus the indirect effects, 10.000 bootstrap sample are used (Hayes, 2013). To give a clear overview, each dependent variable will be measured separately; see Appendix A for an overview of all mediation effects.
Results Randomization
Based on a Chi-squared test, the groups did not differ significantly regarding their knowledge of the D66 X2 (2) = 1.03, p = .905; if they visited the D66 Facebook page X2 (2) = 2.60, p = .626; or if they “liked” the D66 Facebook page X2 (2) = 6.49, p = .165. Neither did gender X2 (2) = .10, p = .999. A one-way between subjects ANOVA was conducted to test if the conditions differed significantly concerning political interest, political preference and education and age. There was no significant
difference found for political interest, F(4, 248) = 1.621, p = .169; neither for political preference, F(4,
229) = .56, p = .692; nor education F(4, 266) = .174, p = .951; or age, F(4, 262) = .850, p = .494. To test if there was a difference in how attractive the respondents rated the male and female candidate, an independent samples t-test was conducted. It showed that there was a slight significant difference in the levels of the female candidate (M = 4.74, SD = .80) and the level of the male candidate (M = 4,34, SD = .94) conditions; t(174) = -3,03 p = .003. Because this variable is not completely randomized, for internal validity purposes, we control for the variable attractiveness in all mediation models
Main effects the independent variables
Several MANOVA's showed that the factor “personalization” has no significant direct effect on vote intention F(1, 196) = 2.29, p = .132, evaluations of candidates F(1, 196) = 3.02, p = .084, trustworthiness F(1, 196) = 1.25, p = .264, or social presence F(1, 196) = 1.44, p = .232. However it does have a significant effect on closeness F(1, 196) = 7.28, p = .008; see Table 1.
Table 1
Mean +/- SD Scores of Depersonalization vs. Personalization
Dependent variable Depersonalization Personalization
M (SD) M (SD) Vote intention 2.11a (1.32) 2.49a (1.40) Evaluation 3.87a (1.02) 4.15a (1.03) Trustworthiness 3.72a (1.32) 3.85a (1.03) Closeness 2.85a (1.57) 3.57b (1.37) CHV 4.36a (0.94) 4.43a (0.90) Social presence 2.12a (1.12) 2.40a (1.28)
Note. Rows with a different subscript differ significantly from each other at the p < 0.05 level.
a refers to no statistical significant difference between two groups. b refers to statistical significant difference between two groups.
The same procedure was executed for the factor “privatization”; see Table 2. Privatization has no significant effect on vote intention F(1, 161) = .675, p = .412, evaluations of qualities F(1, 161) = .05, p = .818 trustworthiness F(1, 161) = .522, p = .471, and conversational human voice is marginally significant F(1, 161) = 2.75, p = .100.
Table 2
Mean +/- SD Scores of Individualization vs. Privatization
Dependent variable Individualization Privatization
M (SD) M (SD) Vote intention 2.37a (1.35) 2.62a (1.44) Evaluation 4.08a (0.97) 4.22a (1.08) Trustworthiness 3.73a (1.05) 3.97a (1.02) Closeness 3.47a (1.36) 3.68a (1.39) CHV 4.26a (0.88) 4.60a (0.89) Social presence 3.37a (1.24) 2.43a (1.33)
Note. Rows with a different subscript differ significantly from each other at the p < 0.05 level.
a refers to no statistical significant difference between two groups.
Individualization: mediating effect of CHV
H1a assumes that individualization leads to CHV, which in turn has a positive effect on political engagement and perception of voters. As stated before, politicians try to diminish the space between them and the electorate by presenting themselves as 'ordinary' people and blurring the lines between public and private (Garzia, 2011; Holtz-Bacha, 2004). When we feel we are talking to a person instead of organization in combination with personal feeling and thought, this has a positive effect the overall evaluation (Kelleher and Miller, 2006; Park and Lee, 2013). Therefor H1a assumes CHV is an important mediator. Considering that no relationship was found between individualization and CHV we can assume that there is no mediation. After checking with PROCESS (model 4) this indeed proved to be true. No mediation effect was found for evaluations of qualities, total effect b = 0.05 , p = .783, direct effect b = 0.23, p = .099; indirect effect 0.03, boot SE = .12, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.20, .28]. Neither for trustworthiness total effect b = 0.05 , p = .783, direct effect b = 0.04, p = .797; indirect effect 0.03, boot SE = .13, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.20, .28]. Nor for vote intention total effect b = 0.07 , p = .674, direct effect b = 0.35, p = .145; indirect effect 0.04, boot SE = .10, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.16, .25]. Therefore, H1a is not supported with regard to CHV: individualization
does not lead to greater feelings of CHV and has no positive effect on (a) voters' evaluations of qualities, (b) trustworthiness of politicians or (c) vote intention.
Individualization: mediating effect of social presence
H1b assumes that individualization has a positive effect on the overall evaluations of voters because it leads to a greater feeling of social presence (Hassanein & Head, 2007; Kruikemeier et al., 2012; Vergeer et al., 2011). Considering that no relationship is found between individualization and social presence we can assume that there is no mediation effect. After checking with PROCESS (model 4) this was proven to be true. No mediation effect was found for evaluations of qualities, total effect b = 0.28 , p = .228, direct effect b = 0.18, p = .293; indirect effect 0.10, boot SE = .08, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.06, .26]. Neither for trustworthiness total effect b = 0.28 , p = .23, direct effect b = 0.02, p = .917; indirect effect 0.11, boot SE = .09, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.07, .28]. Nor for vote intention, total effect b = 0.28 , p = .228, direct effect b = 0.24, p = .288; indirect effect 0.15, boot SE = .12, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.09, .38]. Therefore, H1b is not supported with regard to social presence (when it is used as a mediator), individualization does not lead to a greater feelings of social presence and has no positive effect on (a) voters' evaluations of qualities, (b) trustworthiness of politicians or (c) vote intention.
Individualization: mediating effect closeness
H1c assumes there is a positive relationship between individualization, feelings of closeness and positive perceptions of political candidates and vote intention. As stated before, research in the field of cyber-psychology has proven that an individualized website increases the feeling of closeness visitors feel (Park & Lee, 2013). Virtual feelings of closeness reduce the psychological distance and in turn may have a positive influence political engagement (Vergeer et al., 2011). H1c thus assumes that closeness is an important mediator which is measured by using PROCESS (model 4). After calculation
the results show a significant mediation effect for evaluation of competence qualities, total effect b = 0.72 , p = .005, direct effect b = -0.07, p = .629; indirect effect 0.35, boot SE = .14, 95% bias-corrected bootstrap confidence interval (BCBCI) [.10, .58]. As we can see the direct effect is not significant which suggest full mediation. Second, we test the mediation effect of closeness on the trustworthiness of candidates. After calculation the results show again a significant mediation effect, total effect b = 0.72 , p = .005, direct effect b = -0.21, p = .206; indirect effect 0.34, boot SE = .13, 95% bias-corrected bootstrap confidence interval (BCBCI) [.10, .59]. Here as well we see that the direct effect is not significant, which suggests full mediation. Lastly, results show a significant full mediation effect on vote intention, total effect b = 0.72 , p = .005, direct effect b = -0.05, p = .820; indirect effect 0.43, boot SE = .17, 95% bias-corrected bootstrap confidence interval (BCBCI) [.10, .77]. Based on these findings H1c is fully supported when closeness is measured as a mediator, individualization does lead to greater feelings of closeness and has a positive effect on (a) voters' evaluations of qualities, (b) trustworthiness of politicians or (c) vote intention.
Privatization: mediating effect of CHV
H2a assumes that privatization leads to CHV, which in turn has a positive effect on political engagement and perception of voters. Even though no significant relationship is found between privatization and conversational human voice after doing a MANOVA we will still test for the mediation effect. First we want to see if privatization has a mediated effect via CHV on the evaluation of qualities. For this macro PROCESS (model 4) is implemented. The results show no significant mediation effect, total effect b = 0.21 , p = .089, direct effect b = -0.15, p = .204; indirect effect 0.3, boot SE = .07, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.01, .29]. Second we test for trustworthiness. The results show a significant mediation effect total effect, b = 0.21 , p = .089, direct effect b = -0.12, p = .827; indirect effect 0.12, boot SE = .07, 95% bias-corrected bootstrap confidence interval (BCBCI) [.00, .27]. However, as we can see, is the effect marginally significant.
Lastly we test for vote intention. After calculation the results show again a significant mediation effect,
b = 0.7, p = .735, direct effect b = -0.07, p = .735; indirect effect 0.09, boot SE = .05, 95%
bias-corrected bootstrap confidence interval (BCBCI) [.00, .23]. However, as we can see is the effect marginally significant as well. In sum, privatization has a small effect on vote intention and trustworthiness of the candidate via CHV, however does not affect the voters' evaluation of politicians. H2a is thus marginally supported.
Privatization: mediating effect of social presence
H2b assumes that privatization leads to social presence, which in turn has a positive effect on overall evaluations of politicians. Even though no significant relationship is found between privatization and social presence, we will still test for the mediation effect. No significant mediation effect is found for evaluations of qualities, total effect b = 0.03 , p = .877, direct effect b = -0.03, p = .820; indirect effect 0.01, boot SE = .06, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.11, .12]. Neither for trustworthiness, total effect b = 0.03 , p = .877, direct effect b = 0.08, p = .525; indirect effect 0.01, boot SE = .06, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.12, .13]. Nor for vote intention, total effect b = 0.03 , p = .877, direct effect b = 0.14, p = .466; indirect effect 0.01, boot SE = .09, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.16, .19]. Therefor H2b not supported, privatization does not lead to a greater feeling of social presence and has no positive effect on voters' evaluations of (a) qualities, (b) trustworthiness or (c) vote intention.
Privatization: mediating effect of closeness
H3c assumes that privatization leads to closeness, which in turn has a positive effect on political engagement and perception of voters. Even though again no significant relationship is found between privatization and closeness, we will still test for the mediation effect. No significant mediation effect is found for evaluations of qualities, total effect b = 0.06, p = .763, direct effect b = -0.05, p = .690; indirect effect 0.03, boot SE = .08, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.13,
.19]. Neither for trustworthiness, total effect b = 0.06 , p = .763, direct effect b = 0.07, p = .57; indirect effect 0.02, boot SE = .08, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.12, .17]. Nor for vote intention, total effect b = 0.6 , p = .763, direct effect b = 0.12, p = .50; indirect effect 0.04, boot SE = .12, 95% bias-corrected bootstrap confidence interval (BCBCI) [-.19, .28]. Considering no significant mediation effects were found, H3c is not supported; privatization does not lead to greater feelings of closeness and has no positive effect on voters' evaluations of (a) qualities, (b) trustworthiness or (c) vote intention.
The moderated mediating role of gender
Next step is to see if the relation between privatization and intimacy changes when it is
moderated by gender, and if this in turn changes the effects on the overall evaluations. First we conduct a MANCOVA to compare the multivariate sample means; see Table 3. As we can see do male candidates consequently score higher than their female counterparts. However not one of these difference are significant, CHV, F(1, 160) = .00, p = .995, social presence F(1, 160) = .07, p = .793 or closeness F(1, 160) = .28, p = .600.
Table 3
Mean +/- SD Scores of Male Privatization vs. Female Privatization
Dependent variable Privatization*Male Privatization *Female
M (SD) M (SD) Vote intention 2.78a (1.47) 2.45a (1.40) Evaluation 4.27a (1.17) 4.17a (0.99) Trustworthiness 4.03a (1.06) 3.91a (0.99) Closeness 3.72a (1.58) 3.63a (1.19) CHV 4.64a (1.00) 4.57a (0.76) Social presence 2.75a (1.41) 2.11a (1.17)
Note. Rows with a different subscript differ significantly from each other at the p < 0.05 level.
a refers to no statistical significant difference between two groups.
Based on the literature (Aalberg & Jenssen, 2007; Campus, 2013; Van Zoonen, 2006) this study presupposes that gender influences the relation between privatization and intimacy differently and therefore will change how respondents evaluate them. H4 thus assumes that gender moderates the
mediation effect of intimacy. To test the hypothesis, an SPSS macro PROCESS (model 7) is conducted for each mediation variable and dependent variable; see Table 4. However, no significant moderated mediations were found therefore H4 is not supported.
Table 4
The Moderated Mediation Effect of Privatization (vs. Individualization) on Overall Political Evaluations via Social Presence, CHV and Closeness, Moderated by Gender
95% BCACI
Variable Indirect Effect SE LL UL
Mediating variable Social presence
Vote intention: Male 0.03 0.13 -0.23 0.29
Vote intention: Female -0.02 0.12 -0.26 0.19
Evaluation: Male 0.02 0.09 -0.16 0.19
Evaluation:Female -0.01 0.07 -0.19 0.13
Trustworthiness: Male 0.02 0.10 -0.17 0.20
Trustworthiness: Female -0.01 0.08 -0.20 0.15
Mediating variable: CHV
Vote intention: Male 0.08 0.07 -0.02 0.28
Vote intention: Female 0.08 0.08 -0.05 0.27
Evaluation: Male 0.13 0.10 -0.05 0.35
Evaluation:Female 0.13 0.11 -0.11 0.34
Trustworthiness: Male 0.12 0.09 -0.05 0.33
Trustworthiness: Female 0.12 0.11 -0.09 0.34
Mediating variable: Closeness
Vote intention: Male -0.03 0.17 -0.39 0.30
Vote intention: Female 0.09 0.16 -0.22 0.42
Evaluation: Male -0.02 0.13 -0.28 0.21
Evaluation:Female 0.07 0.12 -0.17 0.29
Trustworthiness: Male -0.02 0.11 -0.25 0.18
Trustworthiness: Female 0.06 0.10 -0.16 0.26
Note. BCACI = bias corrected and accelerated confidence interval; LL = lower limit; UL = upper limit.
Boldface highlights a significant effect as determined by the 95% bias-corrected and accelerated confidence interval based on 10.000 bootstrap samples.
Discussion
The purpose of this study is threefold: first, it determines the effects of different levels of online personalization used by political actors on individuals’ evaluations of competence qualities, trustworthiness perceptions and vote intentions. Second, this study gains insight which processes explain the relationship between personalization (i.e., intimacy) and individuals’ overall evaluations. Third, this study tests for whom these effects are most present, female or male politicians.
Findings
The first major finding is that personalization (vs. depersonalization) leads to greater feelings of closeness and therefore positively influences how people evaluate politicians competence qualities, their trustworthiness and their vote intention (H1b). Visiting a personalized page decreases the psychological distance between visitors and politicians which generates a feeling of being a part of the political circle. This is consistent with former research (Kruikemeier, et al., 2014; Park & Lee, 2013; Vergeer et al., 2011).
Contrary to what was expected, personalization does not lead to greater feelings of CHV (H1a). This can be explained by the fact that, even though on the specified page all content was written by the candidates themselves, the candidate still represents his or her political party. Thus, even when visitors do believe that the messages were personally written, these writings are still in favor of and serving the party’s ideology. Furthermore, personalization does not lead to greater feelings of social presence (H1c). This may be due to the fact that the presented page in this experiment is too one-dimensional (a print-screen without comments or possible interaction). Social presence is strongly related to interaction (Gooch & Watts, 2014) and the feeling you are in a (virtual) space together (Kruikemeier et al., 2014). It could be that if comments were included, or respondents could have scrolled and clicked on the page, social presence would have been affected more positively.
CHV (H2a), which positively affects trustworthiness and vote intentions. It is not surprising that preferential voting is positively affected together with trustworthiness; trustworthiness is the most important trait for citizens to evaluate their vote on (Garzia, 2011). By sharing more private non-political information, people feel like they are talking to an ordinary person instead of a representation of a political party (Holtz-Bacha, 2004; Kelleher and Miller, 2006). However, the evaluations of competence qualities did not significantly change. A simple explanation for this is that not enough information is given on the Facebook page to assess any competencies. Ultimately, candidate assessment is a rational process predominantly based on how candidates would conduct governmental tasks (Garzia, 2011); the three conditions (depersonalization, individualization and privatization) did not differ in the amount of information given in direct relation to this.
Furthermore, privatization did not have a significant effect on either social presence or closeness (H1b and H1c). The latter could be explained by the fact that too much non-political information leads to cynicism (Jebril et al., 2012; Van Aelst et al., 2012). Moreover, people may not believe the candidate is truthful and merely perceive the sharing of private information as a marketing tool. Additionally, privatization does not affect social presence positively. This could be explained by the fact that even though the page contains more personal information, the medium remains the same; it still lacks the ability to interact and does not contain any more richness (i.e., the option to click or comment on certain content).
Lastly, in contrast to the suggested expectation, gender does not moderate the mediating effect of intimacy of privatization on individuals’ evaluations of competence qualities, trustworthiness perceptions and vote intentions; H3 is thus not supported. Male candidates did score higher on average when comparing the mean scores of each variable, although not significantly. The higher mean scores are in line with research conducted by Van Zoonen (2006), personalization seems to be more in favor for male candidates. A possible explanation for this is that intimacy is considered more feminine (2013)
and underscoring this works unfavorably for women because it reinforces gender stereotypes (Van Zoonen, 2006).
Limitations
The most important limitation of this study is the fact that the pictures used for the Facebook profiles of the political candidates were not identical. Even though men and women are never identical, future research should definitely take this into consideration. A second limitation is related to the sample. The amount of male and female respondents was not evenly distributed (68,2% female) which could have affected the outcome. Additionally, the level of personal affinity with the political party D66 may have intervened with the outcome. Future research could, for example, focus on samples with a like-minded political preference and adjust their party and candidate to this, to avoid this particular problem.
Theoretical implications and future research
Online personalization showed to have different effects on closeness, CHV and social presence. This suggests that maybe not all three concepts fit under one umbrella term of intimacy. Future research should take this into consideration. Furthermore, as mentioned, no effect was found for social presence at all. Based on this it seems social presence has more to do with levels of interaction possible via the medium, than just the levels of personalization. It thus seems plausible that social presence only rises when personalization is combined with higher levels of interaction, as Kruikemeier et al. (2014) showed in their study. Future research should take this into consideration.
Furthermore, future research should focus on the role of stereotypes related to personalization-research and gender effects. Stereotypes have an important influence on how we think about male and female politicians (e.g. Kahn, 1992, 1994, 1996; Schneider, 2014) and tend to cause a favor for male politicians (López-Hermida Russo & Cerda Diez, 2012). This is especially important since political knowledge and interest seem to be declining (Hermans & Vergeer, 2012) and consequently stereotypes
will play a bigger role in candidate assessment (Aalberg & Jenssen, 2007; Kahn, 1996). Lastly, future research should examine whether stereotypes have a bigger moderating effect in combination with gender on personalization effects for male citizens, considering male gender favoritism (Aalberg & Jenssen, 2007).
Practical implications
Based on these findings, it is recommendable for politicians to maintain their own personal page, besides the formal political party page. Firstly, personalized online communication leads to more closeness between the candidates and citizens which positively influences evaluations of their competence qualities, their trustworthiness and leads to more preferential. Secondly, if politicians share more about their private life, CHV goes up which again positively influences feelings of trustworthiness and vote intentions. Furthermore, privatization can attract citizens who are not normally that politically interested, by focusing on the personal instead of the political. The results suggest that both male and female candidates benefit from a more personalized strategy just the same, considering no significant differences were found.
This study contributes to the field of online political personalization by gaining knowledge in how online personalization processes affect individuals’ evaluations of competence qualities,
trustworthiness perceptions and vote intentions. Now we have a greater understanding of different levels of personalization (depersonalization vs. individualization vs. privatization) and the effects on overall evaluations. Moreover, we have greater understanding of how these personalization processes occur (i.e. individualization leads to greater feelings of closeness and privatization positively affects feelings of CHV). Lastly, this study can function as a building block in examining in how online personalization processes work similarly and differently for male and female candidates.
References
Aalberg, T., & Jenssen, A. T. (2007). Gender stereotyping of political candidates. Nordicom Review,
28(1), 17-32.
Balmas, M., Rahat, G., Sheafer, T., & Shenhav, S. R. (2014). Two routes to personalized politics Centralized and decentralized personalization. Party Politics, 20(1), 37-51.
Biocca, F., Harms, C., & Burgoon, J. K. (2003). Toward a more robust theory and measure of social presence: Review and suggested criteria. Presence,12(5), 456-480.
Boerman, S. C. (2014). " This program contains product placement": Effects of sponsorship disclosure on television viewers’ responses.
Boulianne, S. (2015). Social media use and participation: a meta-analysis of current research.
Information, Communication & Society, 18(5), 524-538.
Bystrom, D. (2006). Gender and campaign communication: TV ads, web sites, and media coverage.
Institute of Governmental Studies.
Campus, D. (2013). Women political leaders and the media. Palgrave Macmillan.
Carlin, R. E. (2014). What’s not to trust? Rubrics of political party trustworthiness in Chile and Argentina. Party Politics, 20(1), 63-77.
Cathcart, R., & Gumpert, G. (1983). Mediated interpersonal communication: Toward a new typology.
Quarterly Journal of Speech, 69(3), 267-277.
Dolan, K., & Lynch, T. (2013). It Takes a Survey: Understanding Gender Stereotypes, Abstract Attitudes, and Voting for Women Candidates. American Politics Research,
1532673X13503034.
Dolan, K., & Lynch, T. (2013). It Takes a Survey: Understanding Gender Stereotypes, Abstract Attitudes, and Voting for Women Candidates. American Politics Research,
1532673X13503034.
Enli, G. S., & Skogerbø, E. (2013). Personalized campaigns in party-centred politics: twitter and Facebook as arenas for political communication.Information, Communication & Society, 16(5), 757-774.
Falk, E., & Kenski, K. (2006). Issue Saliency and Gender Stereotypes: Support for Women as Presidents in Times of War and Terrorism*. Social Science Quarterly, 87(1), 1-18. Fox, R. L., & Oxley, Z. M. (2003). Gender stereotyping in state executive elections: Candidate
selection and success. Journal of Politics, 65(3), 833-850.
Funk, C. L. (1996). The impact of scandal on candidate evaluations: An experimental test of the role of candidate traits. Political Behavior, 18(1), 1-24.
Garzia, D. (2011). The personalization of politics in Western democracies: Causes and consequences on leader–follower relationships. The Leadership Quarterly, 22(4), 697-709.
Gooch, D., & Watts, L. (2014). The Impact of Social Presence on Feelings of Closeness in Personal Relationships. Interacting with Computers, iwu020.
Gravetter, F. J., & Forzano, L. B. (2010). Research methods for the behavioral sciences (Gravetter).
Belmont: Cengage Learning.
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A
regression-based approach. Guilford Press.
Hayes, D., Lawless, J. L., & Baitinger, G. (2014). Who cares what they wear? Media, gender, and the influence of candidate appearance. Social Science Quarterly, 95(5), 1194-1212.
Hassanein, K., & Head, M. (2007). Manipulating perceived social presence through the web interface and its impact on attitude towards online shopping.International Journal of Human-Computer
Studies, 65(8), 689-708.
personalization strategies used on candidate websites of 17 countries in EP elections 2009. New
Media & Society, 1461444812457333.
Hosch-Dayican, B., Amrit, C., Aarts, K., & Dassen, A. (2014). How Do Online Citizens Persuade Fellow Voters? Using Twitter During the 2012 Dutch Parliamentary Election Campaign. Social
science computer review, 0894439314558200.
Jebril, N., Albæk, E., & De Vreese, C. H. (2013). Infotainment, cynicism and democracy: The effects of privatization vs personalization in the news.European Journal of Communication,
0267323112468683.
Kahn, K. F. (1992). Does being male help? An investigation of the effects of candidate gender and campaign coverage on evaluations of US Senate candidates. The Journal of Politics, 54(02), 497-517.
Kahn, K. F. (1994). Does gender make a difference? An experimental examination of sex stereotypes and press patterns in statewide campaigns. American Journal of Political Science, 162-195. Kahn, K. F., & Fridkin, K. (1996). The political consequences of being a woman: How stereotypes
influence the conduct and consequences of political campaigns. Columbia University Press.
Karvonen, L. (2010). The personalisation of politics: a study of parliamentary democracies. Ecpr Press.
Kelleher, T., & Miller, B. M. (2006). Organizational blogs and the human voice: Relational strategies and relational outcomes. Journal of Computer‐Mediated Communication, 11(2), 395-414. Kim, Y. (2011). The contribution of social network sites to exposure to political difference: The
relationships among SNSs, online political messaging, and exposure to cross-cutting perspectives. Computers in Human Behavior, 27(2), 971-977.
Kittilson, M. C., & Fridkin, K. (2008). Gender, candidate portrayals and election campaigns: A comparative perspective. Politics & Gender, 4(03), 371-392.
Krogstad, A., & Storvik, A. (2012). Picturing politics: Female political leaders in France and Norway. Historical Reflections, 38(3), 129-152.
Kruikemeier, S. (2014). How political candidates use Twitter and the impact on votes. Computers in
Human Behavior, 34, 131-139.
Kruikemeier, S., van Noort, G., Vliegenthart, R., & de Vreese, C. H. (2012). Political communication in a digital world: The effects of personalization and interactive communication on citizens’ political involvement.
Larson, S. G. (2001). American women and politics in the media: A review essay. Political Science &
Politics, 34(02), 227-230.
Lawless, J. L. (2009). Sexism and gender bias in election 2008: A more complex path for women in politics. Politics & Gender, 5(01), 70-80.
Lee, E. J., & Oh, S. Y. (2012). To personalize or depersonalize? When and how politicians'
personalized tweets affect the public's reactions. Journal of communication, 62(6), 932-949. Liang, T. P., Li, Y. W., & Turban, E. (2009). Personalized Services as Empathic Responses: the Role
of intimacy. PACIS 2009 Proceedings, 73.
López-Hermida Russo, A. P., & Cerda Diez, M. F. (2012). Women and politics: The privacy without Necktie. Revista de Comunicación, 11.
Palmer, M. T. (1995). Interpersonal communication and virtual reality: Mediating interpersonal relationships. Communication in the age of virtual reality, 277-299.
Park, H., & Lee, H. (2013). Show us you are real: the effect of human-versus-organizational presence on online relationship building through social networking sites. Cyberpsychology, Behavior,
and Social Networking, 16(4), 265-271.
http://www.parlement.com/id/vh8lnhrp8wsy/links_en_rechts
Rahat, G., & Sheafer, T. (2007). The personalization (s) of politics: Israel, 1949–2003. Political
Communication, 24(1), 65-80.
Sanbonmatsu, K. (2002). Gender stereotypes and vote choice. American Journal of Political Science, 20-34.
Schneider, M. C. (2014). The effects of gender-bending on candidate evaluations. Journal of Women,
Politics & Policy, 35(1), 55-77.
Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications.
Sykes, P. L. (2008). Gender in the 2008 presidential election: Two types of time collide. PS: Political
Science & Politics, 41(04), 761-764.
Van Aelst, P., Sheafer, T., & Stanyer, J. (2012). The personalization of mediated political
communication: A review of concepts, operationalizations and key findings. Journalism, 13(2), 203-220.
Van Zoonen, L. (1998). “Finally, I Have My Mother Back” Politicians and Their Families in Popular Culture. The Harvard International Journal of Press/Politics,3(1), 48-64.
Van Zoonen, L. (2005). Entertaining the citizen: When politics and popular culture converge. Rowman & Littlefield.
Van Zoonen, L. (2006). The personal, the political and the popular A woman’s guide to celebrity politics. European journal of cultural studies, 9(3), 287-301.
Vergeer, M., Hermans, L., & Sams, S. (2011). Is the voter only a tweet away? Micro blogging during the 2009 European Parliament election campaign in the Netherlands. First Monday, 16(8). Vliegenthart, R., Boomgaarden, H. G., & Boumans, J. W. (2011). Changes in political news coverage:
communication in postmodern democracy, 92.
Vos, D. (2014). Which politicians pass the news gates and why? Explaining inconsistencies in research on news coverage of individual politicians. International Journal of Communication, 8, 24.
Appendix A Table 5
The Indirect Effect of Social presence, CHV and Closeness on Overall political Evaluations, for Personalized Condition 95% BCACI Variable Indirect Effect SE LL UL
Condition: Individualization Social presence
Evaluation 0.10 0.08 -0.06 0.26 Vote intention 0.15 0.12 -0.09 0.38 Trustworthiness 0.11 0.09 -0.07 0.28 Condition: Individualization CHV Evaluation 0.03 0.12 -0.20 0.28 Vote intention 0.04 0.13 -0.16 0.25 Trustworthiness 0.03 0.13 -0.20 0.28
Condition: Individualization Closeness
Evaluation 0.34 0.13 0.10 0.59
Vote intention 0.43 0.17 0.10 0.77
Trustworthiness 0.35 0.14 0.10 0.58
Note. BCACI = bias corrected and accelerated confidence interval; LL = lower limit; UL = upper limit.
Boldface highlights a significant effect as determined by the 95% bias-corrected and accelerated confidence interval based on 10.000 bootstrap samples.
Table 6
The Indirect Effect of Social presence, CHV and Closeness on Overall Political Evaluations, for Privatized Condition
95% BCACI
Variable Indirect
Effect
SE LL UL
Condition: Privatization Social presence
Evaluation 0.01 0.06 -0.11 0.12
Vote intention 0.01 0.09 -0.16 0.19
Trustworthiness 0.01 0.06 -0.12 0.13
Condition: Privatization CHV
