Celebrities' Social Influence on Twitter:
Changing Attitudes in 140 Characters or Less?
Nicole ten Velde
11093730
Master's Thesis
Graduate School of Communication
Entertainment Communicaton
Dr. M. Timmers
June 24th, 2016
2
Abstract
The distinction between entertainment and political messages seems to be fading. Celebrities are becoming more vocal regarding social issues, using their fame as a platform to spread information about what they think is important. To share their opinion, social media are often used, enabling two-way communication between celebrities and their fans. The question arises in what way fans and online followers are affected by celebrities sharing their opinions on social issues.
Psychological involvement with celebrities is a key predictor of celebrity influence and can be conceptualized by two main concepts: parasocial interaction and identification. These concepts are hypothesized to affect perceived source credibility and elaboration of the message, which in turn, affect celebrity influence. Experimental survey research was
conducted to test the effects of parasocial interaction and identification.
135 fans were randomly assigned to three conditions. In condition A, participants were exposed to a tweet supporting the feminist campaign HeForShe, sent by a celebrity they consider themselves to be fans of. In condition B, a similar tweet was sent by a "random" celebrity. Condition C was a control group.
Results show that fans perceive higher levels of parasocial interaction and
identification with the sender of the message than non-fans. Fans also perceive the celebrity they are fans of as credible sources and process messages coming from them more
elaborately. However, these factors do not affect celebrities' influence on their fans' attitudes towards feminism. No significant differences were found between the attitudes of fans, non-fans, and participants in the control condition. Therefore, this research shows that fans and followers are unaffected by celebrities sharing opinions about social issues online. Regarding participants' personality, findings suggest that openness positively correlates with feminist attitudes, whereas neuroticism negatively correlates with feminist attitudes.
3
Introduction
In recent years, it seems like there is no longer a clear distinction between entertainment and political messages. Celebrities are becoming more vocal regarding social issues, using their fame as a platform to spread information about what they think is important (e.g., Morin, Ivory, & Tubbs, 2012). For example, during the 2015 MTV Europe Music Awards, singer Pharrell Williams dedicated his performance of his song "Freedom" to the refugee crisis in Europe. At the 2016 Academy Awards, actor Leonardo DiCaprio devoted his first time Oscar winning speech to address climate change. In 2014, actress Emma Watson became the public face of the HeForShe movement, a campaign for gender equality initiated by UN Women.
Another aspect of modern-day entertainment which has developed over the years concerns social media. Nowadays, social media are widely used by celebrities to interact with their fans. Especially on Twitter, a micro-blogging platform that enables users to
communicate with each other through short messages known as tweets, celebrities often reach millions of followers, informing and entertaining their fan base with messages that are limited to a maximum of only 140 characters (e.g., Kaplan & Haenlein, 2012). With that, the presence of celebrities on social media contributes to fans' and followers' perceived sense of closeness and familiarity to them (Click, Lee, & Holladay, 2013). In recent years, it has become
possible to interact with celebrities by sending them a tweet and possibly even getting a reply. This shift from one-way to two-way communication with media figures changes the
relationships fans and followers have with them (Click et al., 2013).
While celebrities are becoming more vocal about their personal opinions on world matters online, at the same time, it seems like many young people might not be politically interested and feel alienated from existing political structures (Inthorn & Street, 2011). However, they do appreciate celebrities. Especially when people idolize a celebrity and consider themselves as fans, they might be particularly interested in what their idol has to say
4
and therefore become more susceptible to their idol's views.
With the two developments regarding celebrities, entertainment and politics in mind, the question arises in what way fans and online followers are affected by them. When celebrities' expression of opinions about social issues and their use of social media come together, followers are exposed to their idol's personal views, which could potentially influence their own opinions on the matter.
It is important to inquire in what way celebrities are able to influence their fans' attitudes. Not only because celebrities seem to become more expressive about social issues, but also because social media have decreased the distance between idol and fan and
subsequently provides a platform that comes across as personal. Moreover, since the current (online) entertainment industry allows common people to build up their own network with followers and become idolized, celebrity influence is a relevant topic to explore. This research will therefore look into celebrity influence via Twitter, focusing on factors that are relevant in this process, answering the following research question:
In what way are celebrities able to influence fans' attitudes about social issues by sharing personal opinions via Twitter?
Theoretical Background
In order to gain more knowledge about celebrity influence through a fan-perspective, it is important to look into the practices relevant to fans processing celebrity information. Research on celebrity influence shows that a concept relating to celebrity influence is involvement, which can be described as the cognitive, affective and behavioral participation prompted by the media throughout the time of media exposure (e.g., Brown, Basil, &
5
information is processed. Highly involved people tend to expose themselves to more media content relating to the media character. They are also more likely to go through attitudinal or behavioral changes caused by exposure to the media character because they are more
susceptible to persuasive influence. Therefore, involvement does not only influence media exposure, but also media effects (Brown et al., 2003).
As previously discussed, social media platform Twitter is increasingly used by celebrities to vocalize political views. It seems like the perfect platform to reach their fans. Research shows that emerging adults spend more time using media than they spend on any other activity (Coyne, Padilla-Walker, & Howard, 2013). According to Luchman, Bergstrom, and Krulikowski (2014) social media are increasingly becoming the medium of choice for collecting information and interacting with others among young adults. Twitter also seems to be a key social platform used to engage in fan-related behavior, such as following favorite celebrities and getting into contact with other fans (Arvidsson, Caliandro, Airoldi, & Barina, 2016).
Furthermore, whereas in the past fans mostly only consumed celebrity content, it seems like modern-day fandom is focused on consuming as well as producing content, which can only intensify the processes of involvement (Korobkova, 2014). Twitter allows users to add their own content to the content that already exists, for example by re-tweeting tweets by means of positive evaluation, or by replying to it.
Brown et al. (2003) discuss how psychological involvement with celebrities can be conceptualized by two main concepts that therefore predict celebrity influence: parasocial interaction and identification.
Parasocial Interaction
6
seemingly face-to-face relationship between a media performer and a spectator. The
interaction is more or less one-sided and controlled by the performer (Horton & Wohl, 1956). Originally, research on parasocial interaction focused on the connection between television personalities and viewers only. In the past few decades, however, research expanded to broader topics, focusing on fans' connection with celebrities, as well.
Fans often describe their relationship with their favorite celebrity as if it is an actual relationship, speaking of their idol as a friend, sibling, guide or mentor (Click et al., 2013) and therefore experiencing parasocial interaction. Moreover, Brown and De Matviuk (2010) showed that selective exposure to content involving the celebrity is positively associated with parasocial interaction. This implies that fans who expose themselves to content that matches their interests, e.g., content involving their favorite celebrity, only results into their perceived connection to the celebrity intensifying.
Regarding effects of experiencing parasocial interaction, Kosenko, Binder, and Hurley (2016) argue that because fans treat parasocial relationships like interpersonal ones, beloved media personalities can influence their fans in a way that is normally reserved for real-life relational partners, functioning as behavioral models. Numerous studies on various topics like health, marketing and prosocial behavior show that parasocial interaction affects the response to messages (Kosenko et al., 2016). Even the level of parasocial interaction with celebrity is relevant, as it predicts message effectiveness (Morin et al., 2012).
The question arises if parasocial interaction also affects fans' response to messages spread by a celebrity via Twitter and if fans are more likely to be influenced by whatever their favorite celebrity tweets about than non-fans. Considering the fact that parasocial interaction deals with the connection between the receiver and the sender of the message only, it is likely that messages that celebrities send out are still able to influence fans, even though the subject of the message (e.g., a political view) is not something the celebrity is known for.
7
Identification
Identification with celebrities has first been described by Kelman (1958) as a persuasion process that occurs when "an individual accepts influence because he wants to establish or maintain a satisfying well-defining relationship to another person or a group" (Kelman, 1958, p. 53). Imitation occurs because the fan believes in the celebrity, and the fan does not
necessarily have to be aware of the identification process (Brown, 2009).
Identification and parasocial interaction seem similar, but are distinct concepts. Research shows that parasocial interaction can be a predictor of identification (Brown, 2009). Identification, however, does not always lead to parasocial interaction (Brown & De Matviuk, 2010). Explaining the difference between the two concepts, parasocial interaction focuses on fans' perceived relationships with celebrities, whereas in the process of identification, the emphasis is on imitation (Brown, 2009) and social influence (Fraser & Brown, 2002).
Regarding effects of identification, Click et al. (2013) found that over time,
identification can lead to the fan incorporating their favorite celebrity's values into their own lifestyle. This can therefore lead to attitude and behavioral changes. They also state that for social and political cues specifically, audiences often look to celebrities through identification (Click, Lee, & Holladay, 2015).
Cohen (2001) states that identification with media characters seen on television or read about in books increases enjoyment, involvement and intense emotional responses, but
decreases the likelihood of critical interpretation of what is said. This also implies that fans who identify with their favorite celebrity are more likely to be influenced by their idol's views, potentially also nowadays, when views are spread via social media platforms like Twitter. Therefore, based on current knowledge about parasocial interaction and
8
H1: Fans' parasocial interactions and identification with celebrities positively contribute to celebrity influence on fans' attitudes about social issues.
Information Processing Routes
When it comes to information processing, one of the traditional often used models is the Elaboration Likelihood Model (ELM). ELM describes the route that recipients of information take during their processing of the message. The peripheral route is taken when recipients are not motivated or able to think about the information they are being exposed to. When people are motivated and able to process a message, the central route is taken, in which careful consideration of information takes place and attitudinal or behavioral change as a response is more likely (Petty & Cacioppo, 1984).
Looking into fans' processing of information sent by their favorite celebrity, it seems like the processes of parasocial interaction and identification discussed earlier are affecting their processing. Petty and Cacioppo (1984) showed that involvement with messages
increases the elaboration of messages. The way recipients respond to a message relating to a celebrity also depends on the level of involvement (Karasiewicz & Kowalczuk, 2014). When people are highly involved, the processing of the message is elaborate and systematic. In case of low involvement, the processing is peripheral and heuristic, weakening the influence on attitude change. Thus, highly involved people take the central route in information processing and lowly involved people take the peripheral route.
Connecting ELM to fans' processes of parasocial interaction and identification, Hung (2014) states that strong parasocial bonds and high levels of identification are relevant to fans, and also lead to fans being highly involved with messages relating to their favorite celebrity. Therefore, fans are likely to take the central route of information processing. Hung (2014) also argues that processes of parasocial interaction and identification with characters are key
9
in a message's persuasive outcome, because they both reduce fans' reactance and
counterargument. This also implies excessive susceptibility to celebrity influence among fans. Moreover, studies show that whenever a persuasive message involves a narrative, therefore offering information about a character, it positively affects the persuasive outcome (e.g., McQueen, Kreuter, Kalesan, & Alcaraz, 2011). Furthermore, Igartua (2010) showed that identification with film characters leads to greater cognitive elaboration and development of reflective processes that were of greater quality. His results also show that involvement with film characters affects the persuasive impact caused by viewing the film (Igartua, 2010). The high level of fans' message elaboration could therefore be caused by simulated social interaction that requires more cognitive effort.
H2: Fans who perceive higher levels of identification and parasocial interactions with celebrities are more likely to take the central route in information elaboration, which positively contributes to celebrity influence on fans' attitudes about social issues.
Source Credibility
Another aspect that is essential in persuasive outcomes of messages according to the ELM is source credibility. Ohanian (1990, p. 40) described this concept as "the term commonly used to imply a communicator's positive characteristics that affect the receiver's acceptance of the message". Morin et al. (2012) state that higher credibility sources are more persuasive in changing attitudes and behaviors than lower credibility sources. For example, in marketing, if recipients of a message perceive the sender to be trustworthy, their evaluation of the product endorsed is more likely to be positive (Li, 2013). Studies show that when celebrities are used to endorse products, the same process is relevant (e.g., Sertoglu, Catl, & Korkmaz, 2014).
10
and state that the perceived parasocial interaction between reader and blogger contributes to bloggers' perceived credibility. Regarding celebrities on Twitter, Jin and Phua (2014) also state that because followers feel closely connected to their favorite celebrities through social media, often viewing them as fellow social media users, the tweets they send out are seen as more credible and trustworthy. This also implies that involvement with celebrities increases the perceived source credibility of celebrities.
However, their research also shows that celebrity Twitter accounts with high numbers of followers are perceived as more credible than accounts with lower numbers of followers. A large amount of followers on the celebrity's Twitter account also significantly increased users' intent to "build an online friendship with the celebrity" (Jin & Phua, 2014). This implies that involvement processes (i.e., parasocial interaction and identification) not only contribute to source credibility, but source credibility also contributes to the involvement processes.
Therefore, there seems to be an interaction between parasocial interaction and
identification on one side, and source credibility on the other. Both positively influence each other. Because fans are likely to perceive celebrities as credible because of parasocial interaction and identification, and source credibility can lead to parasocial interaction and identification, the following hypotheses have been formed:
H3a: Fans who perceive high levels of identification and parasocial interactions with celebrities are more likely to perceive the celebrity as trustworthy, which positively contributes to celebrity influence on fans' attitudes about social issues.
H3b: Fans who perceive the source of a tweet as credible are more likely to develop
parasocial interactions and identify with the sender of the tweet, which positively contributes to celebrity influence on fans' attitudes about social issues.
11
Fans' Personality
As discussed, there is research available about celebrity influence and factors that could potentially contribute to it. However, there is not much known about when celebrity influence occurs, or why. There is little research available about individual differences among fans and why one fan is susceptible to celebrity influence and another one is not. Therefore, this research will attempt to get a clearer idea of the effect of individual differences on celebrity influence, focusing on personality.
DeYoung, Peterson, and Higgins (2002) showed that certain traits of the Big Five predict conformity. People with personalities with high levels of emotional stability (reversed neuroticism), agreeableness and conscientiousness (collectively named stability traits) are more likely to be conforming. On the other hand, openness to experience and extraversion (collectively named plasticity traits) are negatively related to conforming.
Alkiş and Temizel (2015) confirm this. According to them, closed people are more likely to follow authority without questioning, whereas open individuals are more likely to go against other (opposing) ideas due to their beliefs in their own ideas and imaginations. They state that agreeableness is the most susceptible personality trait compared to the other Big Five traits (Alkiş & Temizel, 2015).
Especially because this research focuses on fans, the topic of personality seems to be interesting. Research shows that there are a lot of stereotypes that linger around fans.
Especially female fans seem to be linked to pathological behavior and hysteria (Hermann, 2008). Whereas male fans are often viewed as violent (e.g., football hooligans), women are often considered incapable of separating fantasy from reality (Klink, 2010) and therefore seen as more likely to be influenced. About female dominated fandoms, for example fans of boy band One Direction, VanCant (2014, p. 7) states that it is assumed that "the majority of the [One Direction] fan base are wild obsessed pre-teen females, known to cry, scream and pass
12
out at any event in which they make an appearance".
However, research also shows that there seem to be certain personality traits that predict fan related processes regarding involvement. For example, Sun & Wu (2012) showed that need for material resources (materialism) positively affects development of parasocial relationships with sports celebrities. They also found that self-esteem is negatively linked to developing parasocial relationships.
Looking at the Big Five personality traits earlier discussed, it seems like introversion could be positively linked to parasocial interaction. Parasocial interaction does not require real-life communication and therefore offers shy or lonely people social connections without the discomfort they could experience with real-life interactions (Ashe & McCutcheon, 2001).
In general, evidence on the effects of personality on involvement is lacking. Regarding identification, research indicates that higher levels of agreeableness, conscientiousness and extraversion correlate with higher levels of identification (Manninen, 2016). However, this research was focused on generalized identification, i.e., identification with groups, and not with one person or celebrity in particular.
Because there is little research available on fans' personality and it is clear that there are stereotypes surrounding this group of people, this research connects personality to general conformity, measuring fans' personality to test the stereotype and looking into the effects of personality on celebrity influence. Connecting literature on the influence of personality on conformity previously discussed to the subject of this research, the following hypotheses have been formulated:
H4a: Fans with high levels of stability personality traits (emotional stability, agreeableness and conscientiousness) are more likely to be susceptible to celebrity influence.
13
H4b: Fans with high levels of plasticity traits (openness to experience and extraversion) are less likely to be susceptible to celebrity influence.
All formed hypotheses are displayed in Figure 1. Figure 1
Research model
Method
To test the hypotheses, experimental survey research was conducted. The survey, which can be found in Appendix C, was distributed online. The survey has a between-subjects design and three conditions to which participants were randomly assigned. Participants filled out the self-report survey individually and were able to finish it within approximately 10 minutes.
Participants
14
questionnaire and were therefore left out of analysis and 20 participants were underage. A number of 137 participants remained (mean age 20.10, SD = 3.860;99.3% female). However, two of them were significantly older than the others, being 37 and 54 years old. All other respondents were between the ages of 18 and 29 and therefore part of the generation labeled emerging adults. To keep the focus on this particular audience, the two older participants were left out of analysis, so 135 participants remained (mean age 19.73, SD = 2.068; 99.3%
female). 108 of them (80.0%) are still in school, college or university, of which 43 (31.9%) are enrolled in mbo education.
Design
As this research takes a fan point-of-view in learning more about celebrity influence via Twitter, the sample consists of a group of fans of the same celebrity. Popular British-Irish boy band One Direction was chosen as the celebrity, as they are highly influential on Twitter (Lazzaro, 2015; Luckerson, 2016) and seem to have a great amount of Dutch fans.
It has been a conscious decision to not reach out to people merely following One Direction members on Twitter. Instead, another phenomenon, the prevalence of Twitter
fan-accounts, has been explored. These type of accounts are created by fans to keep other fans
updated about their mutual favorite celebrity and can have up to thousands of followers. Whereas celebrities' Twitter accounts often have millions, it does not necessarily mean that all followers are actually fans. When people look for fan-accounts and make the decision to follow them, it is more likely that they are real fans of the celebrity as opposed to the celebrity's own followers.
To measure celebrity influence on fans' attitudes about social issues, participants were randomly assigned to one of three conditions. In two conditions, participants were exposed to a tweet sent by a celebrity addressing a social issue, before their attitude regarding the social
15
issue was measured. In the third condition, participants' attitude was measured without prior exposure to a tweet addressing the social issue. Participants in condition A (N = 55) were exposed to a tweet sent by One Direction member Harry Styles; participants in condition B (N = 36) were exposed to a similar tweet sent by actor Tom Hiddleston; participants in condition C (N = 44) were not exposed to any tweet.
The social issue chosen to apply to this research concerns the HeForShe campaign, initiated by UN Women. HeForShe is a campaign for gender equality and aims to engage men in achieving women's rights, encouraging them to stand up against inequalities faced by
women and girls.
In this research, the support towards the HeForShe campaign reflected in the celebrity tweets will be considered as an expression of an opinion about a social issue. Therefore, the influence celebrities can have on fans' attitudes towards feminism by tweeting about the HeForShe campaign is measured by assigning participants to different conditions. Potential between-group differences regarding attitudes toward feminism will be generated by celebrity influence.
The decision to not just measure the differences between condition A and condition C, but also add condition B and expose One Direction fans to a "random" celebrity, was made so it can be tested whether potential effects occur because of fan-related involvement processes (i.e., parasocial interaction and identification) or more general celebrity role model-effects.
To disguise the purpose of exposing participants to tweets addressing a social issue, the stimuli were delivered as a cover story about Twitter turning 10 years old and Twitter usage by celebrities. The tweets sent out by Harry Styles and Tom Hiddleston addressing the HeForShe campaign were attached as an example. After the tweets, a page followed listing the links mentioned in the tweet (e.g., #HeForShe) for the participants to click on and explore.
16
Procedure
Twitter fan-account @1DStreetteamNL was used to recruit participants, i.e., One Direction fans. @1DStreetteamNL is (one of) the biggest Netherlands-based Twitter fan accounts, updating its followers about boy band One Direction (1D). As of May 4th 2016, Twitter fan-account @1DStreetteamNL has 37,722 followers.
The owners of the Twitter account were contacted and agreed to inform their followers about the research. The link to the questionnaire was tweeted as well as mentioned in a
newsletter they regularly mail out to subscribers. Fans' gender was not specifically relevant for this research. Regarding obtaining informed consent, respondents had to be at least 18 years old. Since all participants are followers of @1DStreetteamNL, it can be assumed they are indeed fans of One Direction.
Participants were informed about the survey before they started it. It was explained to them that the research is conducted as part of the researcher's study and master's thesis. Participants were also informed that data will be processed anonymously. After an indication was given about the time it will take to fill out the survey (approximately five to ten minutes), participants had to confirm their consent by ticking a box. If they did not tick the box, they could not participate.
Starting the survey, participants were asked about their age. When participants stated they were younger than 18 years old, the survey was automatically ended. Because it is impossible to check if participants lie about their age, extra questions were added to test participants. When they earlier stated they were 18 years old or above, they were tested and still had to fill out their year of birth. When they made an error and entered a birth year implying they are underage, the survey was still automatically ended.
17
Research Instrument
Demographic information. Participants' age, sex and current or completed level of
education (vmbo, havo, vwo, mbo, hbo or wo) were measured.
Stimuli. Next, respondents were randomly assigned to the discussed conditions.
Condition A and condition B were exposed to a cover story about Twitter and an attached tweet addressing the HeForShe campaign, which can be found in Appendix A. Participants assigned to condition C were not exposed to any cover story or tweet.
Attitude towards feminism. Next, all participants' attitudes towards feminism were
measured using a scale adapted from the Attitudes Toward Feminism (FEM) Scale (Smith, Ferree, & Miller, 1975) and the more recent Liberal Feminist Attitude and Ideology Scale (LFAIS) (Morgan, 1996). The constructed scale consists of 10 items using a six-point Likert scale with possible answers ranging from strongly disagree to strongly agree. An example of an item is "A woman should be expected to change her name when she marries". After reliability analysis, three items were deleted. Seven items remained (Cronbach's alpha = .69). All used scales can be found in Appendix B.
Parasocial interaction and identification. In condition A and B, fans' perceived
parasocial interaction and identification with the source of the tweet have been measured using the Celebrity-Persona Parasocial Interaction Scale (CPPI) and the Celebrity-Persona Identification Scale (CPI), both formulated by Bocarnea and Brown (2007). Examples of these scales include "Sometimes I feel like calling or writing [celebrity]" (measuring
parasocial interaction) and "What is important to [celebrity] is important to me" (measuring identification). Participants also answered a question regarding if they perceived themselves to be a fan of the source of the tweet they were exposed to. Again, answers were given using a five-point Likert Scale. Participants assigned to condition C automatically skipped this part of the survey. Both scales appeared to be sufficiently reliable (Cronbach's alpha = .90, N = 5 for
18
parasocial interaction and Cronbach's alpha = .89, N = 5 for identification).
Perceived credibility. Fans' perceived credibility of the source of the tweet has been
measured using Ohanian's (1990) Source Credibility Scale, which focuses on three
components: perceived attractiveness, trustworthiness and expertise of the source. Participants rated the sender of the tweet they were exposed to on a six-point bipolar scale. The scale consisted of 15 items and appeared to be sufficiently reliable, Cronbach's alpha = .97.
Elaboration of the message. The extent to which participants elaborately processed
the information they were exposed to (i.e., the cover story and the tweet) was measured by Reynolds' (1997) Elaboration Measure. This scale consists of 12 items measuring participants' message elaboration using a seven-point Likert scale with answer possibilities varying from strongly disagree to strongly agree. Five items were initially chosen and used in the
questionnaire. An example of an item is "While reading the [text and the tweet] I was doing my best to think about what was written". After reliability analysis, one item ("While reading the [text and the tweet] I was searching my mind in response to the ideas") was deleted. Four items remained, proving to form a reliable scale, Cronbach's alpha = .76. For both perceived credibility as elaboration of the message, only the participants assigned to condition A and condition B answered questions. Again, participants assigned to condition C skipped these parts of the questionnaire.
Personality. Finally, all participants answered questions about their personality, using
a shortened version of the Big Five personality assessment (Rammstedt & John, 2007) which measures personality in terms of levels of openness to experience, conscientiousness,
extraversion, agreeableness and neuroticism. Eleven personality statements were presented to the respondent, e.g., "I see myself as someone who is reserved". Possible answers vary from strongly disagree to strongly agree, using a five-point Likert scale. Reliability analysis was conducted for the separate Big Five traits. Only for neuroticism (N = 2), the Cronbach's alpha
19
value appeared to be acceptable, Cronbach's alpha = .68. Extraversion (N = 2) followed with a Cronbach's alpha value of .53. The other three measured personality traits, agreeableness (N = 3), conscientiousness (N = 2) and openness (N = 2), all generated low Cronbach's alpha values (.34, .29 and .29 respectively) and therefore have to be considered unreliable. Initially, the plan was to use the Big Five personality traits to create scales measuring stability (reversed neuroticism, agreeableness and conscientiousness) and plasticity (openness and extraversion). However, because most of the scales reflecting the Big Five traits have rather low Cronbach's alpha values (< .60), the decision was made to look into the effects of neuroticism as a scale only, and focus on the effects of the 11 personality items separately.
Figure 2 shows all variables measured and the order in which they were presented to the participants.
Figure 2
20
Results
Participants and Fans
As a manipulation check, it was first tested if the participants assigned to condition A and B considered themselves as a fan of the source of the tweet they were exposed to.
Table 1
Participants' fan perception
I consider myself as a fan of... N Mean SD
Condition A: One Direction and Harry Styles 55 4.36* 1.14 Condition B: actor Tom Hiddleston 36 1.89* 1.04
* Five-point Likert scale (1 = strongly disagree, 5 = strongly agree)
An independent-samples t test was conducted to compare the means of the different
conditions. Results show that participants assigned to condition A scored significantly higher (M = 4.36, SD = 1.14) than participants assigned to condition B (M = 1.89, SD = 1.04), t(89) = 10.47, p < .001, 95% CI [2.00, 2.94]. Next, a one-sample t test was conducted to evaluate whether the mean was significantly different from test value 3, the neutral point on the Likert scale. For condition A, the mean of 4.36 was significantly different from 3, t(54) = 8.84, p < .001. For condition B, the mean of 1.89 was also significantly different from 3, t(35) = -6.44,
p < .001. It can therefore be concluded that participants assigned to condition A significantly
considered themselves to be fans of One Direction and band member Harry Styles, and that participants assigned to condition B significantly did not consider themselves to be fans of actor Tom Hiddleston.
21
Testing the Variables
Table 2 shows the means and standard deviations of all variables measured, separated by condition. First, for attitude towards feminism, one-sample t tests were conducted to test whether the means in all three conditions were significantly different from test value 3.5, the neutral point on the Likert scale for attitude towards feminism. For condition A, the mean of 4.74 was significantly different from 3.5, t(54) = 15.312, p < .001. For condition B, the mean of 4.93 was also significantly different from 3.5, t(35) = 10.971, p < .001. Finally, for
condition C, the mean of 4.60 was significantly different from 3.5 as well, t(43) = 9.244, p < .001. It can therefore be concluded that all participants' attitudes are significantly feminist. Table 2
Variables characteristics
Variables Condition A Condition B Condition C Mean SD Mean SD Mean SD
1 Attitude towards feminism 4.74 0.60 4.93 0.78 4.60 0.79 2 Parasocial interaction 3.48 0.70 1.86 0.83 - - 3 Identification 2.77 0.79 1.58 0.72 - - 4 Perceived source credibility 4.96 0.70 3.87 1.04 - - 5 Elaboration of the message 4.88 1.10 4.08 1.22 - - 6. Neuroticism 3.38 1.02 3.22 0.99 3.57 0.79 To test in what way are celebrities able to influence fans' attitudes about social issues by sharing personal opinions via Twitter, univariate analyses of variance were used. First, it was tested whether the main effect relevant in this research was significant: the effect of condition on attitude towards feminism, i.e., whether celebrity influence takes place or not. Results show that the effect of the condition on attitude towards feminism is not significant (F(2, 132) = 2.07, p = .130), indicating an insignificant difference between the conditions. This means
22
that the condition participants were randomly assigned to does not significantly affect participants' attitude towards feminism that was measured.
However, univariate analysis of variance shows that the assigned condition affects both parasocial interaction (F(1, 88) = 99.77, p < .001) and identification (F(1, 89) = 53.22, p < .001) significantly. As shown in Table 2, participants significantly experienced higher levels of parasocial interaction (M = 3.48, SD = 0.70) in the experimental condition A, than participants assigned to condition B (M = 1.86, SD = 0.83). This also applies to identification (M = 2.77, SD = 0.79 in condition A; M = 1.58, SD = 0.72 in condition B). As participants assigned to condition A significantly consider themselves as fans and those assigned to condition B do not, it can be stated that being a fan of a celebrity elicits significantly higher levels of parasocial interaction and identification.
However, because results showed that condition does not significantly affect participants' attitude, even though condition does significantly affect parasocial interaction and identification, higher levels of parasocial interaction and identification do not affect participants' attitude towards feminism. Testing the effects of parasocial interaction and identification on attitude towards feminism, results show no significant correlations, r(88) = -.08, p = .455 for parasocial interaction and r(89) = -.05, p = .636 for identification.
Besides affecting parasocial interaction and identification, univariate analysis of variance showed that the condition participants were assigned to also significantly affects the elaboration of the message participants were exposed to, F(1, 89) = 10.53, p = .002.
Participants assigned to condition A processed the message more elaborately (M = 4.88, SD = 1.10) than participants assigned to condition B (M = 4.08, SD = 1.22), indicating that they were more likely to take the central route in information elaboration. Taking a closer look at the significant difference in elaboration between the two conditions, it appears that parasocial interaction and identification are associated with the effect. As shown in Table 3, both show a
23
significant, weak positive correlation, r(88) = .263, p = .012 for parasocial interaction and
r(89) = .220, p = .036 for identification.
Table 3
Bivariate correlations among variables
Variables 1 2 3 4 5 6
1 Attitude towards feminism 1 -.080 -.050 -.075 .121 -.164 2 Parasocial interaction -.080 1 .785** .652** .263* .037 3 Identification -.050 .785** 1 .571** .220* .110 4 Perceived source credibility -.075 .652** .571** 1 .172 .060 5 Elaboration of the message .121 .263* .220* .172 1 .023 6 Neuroticism -.164 .037 .110 .060 .023 1
** correlation is significant at the .01 level (2-tailed) * correlation is significant at the .05 level (2-tailed)
Next, the effects of parasocial interaction and identification on perceived source credibility were tested. Firstly, univariate analysis of variance shows that condition significantly affects source credibility (F(1, 89) = 35.56, p < .001). Looking at parasocial interaction and
identification specifically, both appear to be associated with the condition's effect. Parasocial interaction and perceived source credibility are strongly positively correlated, r(88) = .652, p < .000, and identification and source credibility are, as well, r(89) = .571, p < .000.As shown in Table 2, the perceived source credibility is significantly higher in condition A (M = 4.96,
SD = 0.70) compared to condition B (M = 3.87, SD = 1.04). Results above indicate that this
effect is related to parasocial interaction and identification. Next, it was tested if perceived source credibility affects participants' attitudes towards feminism. As earlier results showed no between-condition differences, the correlation does not appear to be significant,r(89) =
-.075, p = .478.
24
alpha values of four of the Big Five scales were lower than .60, the decision was made to not look into the stability and plasticity scales. Instead, the effects of the only reliable scale (neuroticism, N = 2) and all 11 separate personality items have been analyzed.
As shown in Table 3, neuroticism as a scale does not significantly correlate with any of the other measured variables. Looking into the 11 personality items separately, there is a significant, weak negative correlation between item 9 ("I see myself as someone who gets nervous easily"), measuring neuroticism, and attitude towards feminism (r(133) = -.198, p = .022). There is also a significant, weak positive correlation between item 10 ("I see myself as someone whohas an active imagination"), measuring openness, and attitude towards attitude (r(133) = .278, p = .001). Also, item 3 ("I see myself as someone who tends to be lazy"), a reversed item measuring conscientiousness, shows a significant, negative weak correlation with identification, r(89) = -.218, p = .038. All other separate items measuring personality traits do not significantly correlate with any of the other constructs measured
Conclusion and Discussion Conclusion
This research shows that fans experience higher levels of parasocial interaction and identification with the celebrity they are fans of than with a celebrity they do not consider themselves to be fans of. Related to parasocial interaction and identification, fans also
perceive the celebrity they are fans of as credible sources and process messages coming from them more elaborately. However, these factors do not affect the influence celebrities can have on their fans' attitudes towards social issues. No significant differences were found between the attitudes towards feminism of fans, non-fans and participants that were not exposed to a tweet. This means that, in contrast to theory previously discussed, involvement with the celebrity in terms of experiencing parasocial interaction and identification does not positively
25
contribute to participants' attitudes towards social issues. Furthermore, higher elaboration of the message among fans does not affect attitude either. Also, whereas parasocial interaction and identification are related to perceived source credibility as hypothesized, source
credibility also does not affect attitude.
As for participants' personality, findings suggest that the personality trait openness positively correlates with attitude towards feminism. In contrast, neuroticism negatively correlates with attitude towards feminism. However, both significant correlations are weak and no statements can be made about the effects of personality on celebrity influence, as no between-condition differences were found for attitude towards feminism.
Discussion
This research shows that participants' attitudes towards feminism are unaffected by celebrity influence. As discussed, research (e.g., Kosenko et al., 2016; Click et al., 2013) implies that fans' parasocial interaction and identification with celebrities leads to celebrities being able to influence their fans' attitudes, whether or not via elaboration of the message or through effects of perceived source credibility. However, this research shows that experiencing parasocial interaction and identification with a celebrity does not affect fans' attitude.
Explaining this finding, in general, it is known that attitude is a complex concept. Research on general persuasion shows that different variables affect attitudes through different processes, in which different situations or moods also matter, making attitude seem like a difficult concept to change (Petty & Brinol, 2010). Moreover, the design of this study could be the reason of biased results. Examples of studies previously discussed focused on the effects of a celebrity's personal, detailed health story (Kosenko et al., 2016) and fan
involvement on social media over an extended period of time (Click et al., 2013). In this research, however, participants were exposed to only one tweet for a short amount of time.
26
This only allows measurement of very short term effects. Also, the tweets shown to the participants assigned to condition A and B only included a short statement of support towards the HeForShe campaign. Neither Harry Styles nor Tom Hiddleston used methods such as vivid information, persuasive discounting appeals, fear appeals or humor, which are all known to facilitate attitude change among readers of the message (Olson & Zanna, 1993). According to the Elaboration Likelihood Model, it is important to use clear arguments in order to
facilitate attitudinal change (Petty & Cacioppo, 1986). Again, in short tweets like the ones shown to the participants, this is not the case. Perhaps the combination of using sources with whom readers are able to parasocially interact and identify on one hand and using other persuasive methods on the other is the key to attitudinal change, and using parasocial interaction and identification without other persuasive methods is not enough.
Moreover, in this research, the sample consisted of One Direction fans only. Results show that participants' attitudes towards feminism are not influenced by celebrities. However, it could also imply that One Direction fans are already feminist oriented and are therefore unaffected by being exposed to a tweet addressing this issue, or not. As previously discussed, participants score significantly high on the attitude towards feminism scale in all three
condition, indicating that the control group, as well, is feminist. Furthermore, the One Direction fandom is known as a very participatory fandom (Korobkova, 2014). It also seems like One Direction and feminism have been a topic of discussion among fans. VanCant (2014) talks about how One Direction fans are actually aware of the critique they receive from the general public, being labeled as "wild obsessed pre-teen females, known to cry, scream and pass out" and therefore not being taken seriously, and how this reputation also leads to criticism towards their favorite band. Perhaps for abovementioned reasons, One Direction fans might be more aware of the feminist agenda than the average group of young people. It is also possible that the sample has already been exposed to messages about the
27
HeForShe campaign prior to the research. The tweets used are existing tweets, sent out by Harry Styles and Tom Hiddleston in late September 2014. It is possible that One Direction fans following Harry Styles on Twitter have been exposed to the tweet before. Prior exposure to any tweet addressing the HeForShe campaign could potentially lead to different reactions to feminism in this research. Participants could have already been influenced by the tweets before they took part in the research. Furthermore, it is known that repeated exposure to a message positively influences the evaluation of the message. This phenomenon is known as the mere exposure effect (Heuvelman, Fennis, & Peters, 2011) and, although to a lesser degree relevant to the control group, could affect the results of this research.
However, even though fans' attitude was unaffected in this research, all other findings do correspond with past research, proving that being a fan of a celebrity does affect
psychological processes. As predicted, participants' levels of parasocial interaction and identification, perceived source credibility and elaboration of the message were significantly higher for fans than for non-fans.
Limitations
A limitation of this study is the fact that it is cross-sectional. This way, no conclusions can be drawn about causality or long term effects. Also, the method of this study could be improved by revising the scales used in the questionnaire to measure the constructs. For attitude towards feminism, three items were deleted after reliability analysis. A more valid scale, perhaps focusing on a specific aspect of feminism, could be used in future research. Creating a new scale dealing with modern-day issues regarding gender roles would be desirable. The current scale is dated and perhaps no longer shows diverse results, as times have changed and views on gender equality in general have become less conservative.
28
more general Big Five traits would be beneficial. This way, higher validity is more likely, as the scales that consisted of items measuring one of the Big Five traits appeared to be
unreliable and further conclusions about the effects of the scales could therefore not be drawn. Future research could focus on the specific content of the message spreading a social statement and the effects of frequency of exposure. Future research could also focus on differences in attitudes between fans and non-fans of a specific celebrity. As discussed, a particular audience could already be more focused on certain social issues than others.
Looking further into the effects of all measured variables could explain why in some cases, fans experiencing higher levels of parasocial interaction and identification, as well as perceiving the source as more credible and processing the message more elaborately, still does not lead to celebrity influence. This would provide more insight into the interesting development in entertainment in which celebrities are more vocal about social issues on social media and will provide more answers to the question in what cases they are able to influence fans' attitudes.
References
Alkiş, N., & Temizel, T. T. (2015). The impact of individual differences on influence strategies. Personality and Individual Differences, 87(2015), 147-152.
Arvidsson, A., Caliandro, A., Airoldi, M., & Barina, S. (2015). Crowds and value. Italian Directioners on Twitter. Information, Communication & Society, 19(7), 921-939. Ashe, D. D., & McCutcheon, L. E. (2001). Shyness, loneliness, and attitude toward
celebrities. Current Research in Social Psychology, 6(9), 124-133.
Bocarnea, M. C., & Brown, W. J. (2007). Celebrity-persona parasocial interaction
29
Brown, W. J. (2009). Mediated influence of Pope John Paul II. Journal of Communication &
Religion, 32(1), 33-62.
Brown, W. J., Basil, M. D., & Bocarnea, M. C. (2003). Social influence of an international celebrity: Responses to the death of Princess Diana. Journal of Communication, 53(4), 587-605.
Brown, W. J., & De Matviuk, M. A. C. (2010). Sports celebrities and public health: Diego Maradona's influence on drug use prevention. Journal of Health
Communication, 15(4), 358-373.
Click, M. A., Lee, H., & Holladay, H. W. (2013). Making monsters: Lady Gaga, fan identification, and social media. Popular Music and Society, 36(3), 360-379.
Click, M. A., Lee, H., & Holladay, H. W. (2015). ‘You’re born to be brave’: Lady Gaga’s use of social media to inspire fans’ political awareness. International Journal of Cultural
Studies, 1-17.
Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences with media characters. Mass Communication & Society, 4(3), 245-264.
Colliander, J., & Dahlén, M. (2011). Following the fashionable friend: The power of social media. Journal of Advertising Research, 51(1), 313-320.
Coyne, S. M., Padilla-Walker, L. M., & Howard, E. (2013). Emerging in a digital world a decade review of media use, effects, and gratifications in emerging
adulthood. Emerging Adulthood, 1(2), 125-137.
DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2002). Higher-order factors of the Big Five predict conformity: Are there neuroses of health? Personality and Individual
Differences, 33(4), 533-552.
Fraser, B. P., & Brown, W. J. (2002). Media, celebrities, and social influence: Identification with Elvis Presley. Mass Communication & Society, 5(2), 183-206.
30
Hermann, M. (2008). Not one of those screaming girls: Representing female fans in 1990s' Germany. Women's Studies in Communication, 31(1), 79-103.
Heuvelman, A., Fennis, B., & Peters, O. (2011). Mediapsychologie. Den Haag, Nederland: Boom Lemma.
Horton, D., & Wohl, R. (1956). Mass communication and para-social interaction: Observations on intimacy at a distance. Psychiatry, 19(3), 215-229. Hung, K. (2014). Why celebrity sells: A dual entertainment path model of brand
endorsement. Journal of Advertising, 43(2), 155-166.
Igartua, J. J. (2010). Identification with characters and narrative persuasion through fictional feature films. Communications, 35(4), 347-373.
Inthorn, S., & Street, J. (2011). 'Simon Cowell for prime minister'? Young citizens' attitudes towards celebrity politics. Media Culture and Society, 33(3), 479-489.
Jin, S.-A. A., & Phua, J. (2014). Following celebrities' tweets about brands: The impact of Twitter-based electronic word-of-mouth on consumers' source credibility perception, buying intention, and social identification with celebrities. Journal of Advertising,
43(2), 181-195.
Kaplan, A. M., & Haenlein, M. (2012). The Britney Spears universe: Social media and viral marketing at its best. Business Horizons, 55(1), 27-31.
Karasiewicz, G., & Kowalczuk, M. (2014). Effect of celebrity endorsement in advertising activities by product type. International Journal of Management and
Economics, 44(1), 74-91.
Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2(1), 51-60.
Klink, M. L. (2010). Laugh out loud in real life: Women's humor and fan identity
31
USA.
Korobkova, K. (2014). Schooling the Directioners: Connected learning and identity-making
in the One Direction fandom. Irvine, CA: Digital Media and Learning Research Hub.
Kosenko, K. A., Binder, A. R., & Hurley, R. (2016). Celebrity influence and identification: A test of the Angelina effect. Journal of Health Communication, 21(3), 1-9.
Lazzaro, S. (2015, December 7). The most retweeted Tweets of 2015 were all by One
Direction members. Retrieved March 14, 2016, from http://observer.com/2015/12/the-most-retweeted-tweets-of-2015-were-all-by-one-direction-members/
Li, C. Y. (2013). Persuasive messages on information system acceptance: A theoretical extension of elaboration likelihood model and social influence theory. Computers in
Human Behavior, 29(1), 264-275.
Luchman, J. N., Bergstrom, J., & Krulikowski, C. (2014). A motives framework of social media website use: a survey of young Americans. Computers in Human Behavior,
38(2014), 136-141.
Luckerson, V. (2016, March 20). These are the 10 most popular Tweets of all time - A certain music group dominates the list. Retrieved March 31st, 2016, from
http://time.com/4263227/most-popular-tweets/
Manninen, E. (2016). Generalized identification: Individuals' levels of identification with
groups and its relation to personality. (Unpublished bachelor's thesis). Uppsala
University, Uppsala, Sweden.
McQueen, A., Kreuter, M. W., Kalesan, B., & Alcaraz, K. I. (2011). Understanding narrative effects: the impact of breast cancer survivor stories on message processing, attitudes, and beliefs among African American women. Health Psychology, 30(6), 674-682. Morgan, B. L. (1996). Putting the feminism into feminism scales: Introduction of a Liberal
32
Morin, D. T., Ivory, J. D., & Tubbs, M. (2012). Celebrity and politics: Effects of endorser credibility and sex on voter attitudes, perceptions, and behaviors. The Social Science
Journal, 49(4), 413-420.
Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers' perceived expertise, trustworthiness, and attractiveness. Journal of Advertising, 19(3), 39-52.
Olson, J. M., & Zanna, M. P. (1993). Attitudes and attitude change. Annual Review of
Psychology, 44(1), 117-154.
Petty, R. E., & Brinol, P. (2010). Attitude change. Advanced social psychology: The state of
the science, 217-259.
Petty, R. E., & Cacioppo, J. T. (1984). The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion. Journal of
Personality and Social Psychology, 46(1), 69-81.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (123-205). New York: Academic Press.
Rammstedt, B., & John, O. P. (2007). Measuring personality in one minute or less: A 10-item short version of the Big Five Inventory in English and German. Journal of Research in
Personality, 41(1), 203-212.
Reynolds, R. A. (1997). A validation test of a message elaboration measure. Communication
Research Reports, 14(3), 269-278.
Sertoglu, A. E., Catl, O., & Korkmaz, S. (2014). Examining the effect of endorser credibility on the consumers' buying intentions: An empirical study in Turkey. International
33
Smith, E. R., Ferree, M. M., & Miller, F. D. (1975). A short scale of attitudes toward feminism. Representative Research in Social Psychology, 6, 51-56.
Sun, T., & Wu, G. (2012). Influence of personality traits on parasocial relationship with sports celebrities: A hierarchical approach. Journal of Consumer Behaviour, 11(2), 136-146. VanCant, B. (2014). Whovians and Directioners: Challenging the fangirl identity. (Bachelor's
thesis, Columbia College Chicago). Retrieved from
http://digitalcommons.colum.edu/cgi/viewcontent.cgi?article=1004&context=cultural_ studies
34
Appendices
Appendix A
Figure A1
35
Figure A2
36
Figure A3
37
Figure A4
38
Appendix B
Table B1
Attitude towards feminism assessment
Retrieved from Attitude Towards Feminism Scale (FEM) (Smith, Ferree, & Miller 1975) 1 A woman should be expected to change her name when she marries
2 Profanity sounds worse generally coming from a woman
Retrieved from The Liberal Feminist Attitude and Ideology Scale (LFAIS) (Morgan, 1996). 3** A man who has chosen to stay at home and be a house-husband is not less masculine
than a man who is employed full-time
4** A woman should not let bearing and rearing of children stand in the way of a career if she wants it
5 Women should be more concerned with clothing and appearance than men* 6 Although women can be good leaders, men make better leaders*
7** Violence against women is not taken seriously enough 8 There is no such thing as rape between a man and his wife* 9 A woman who has many sexual partners is not necessarily a slut
10 Even though some things have changed, women are still treated unfairly in today's society
* Reversed items
** Items deleted after reliability analysis
Table B2
CPPI: Celebrity-Persona Parasocial Interaction Scale (Bocarnea, & Brown, 2007)
1 [...] makes me feel as if I am with someone I know well
2 If I saw a newspaper or magazine story about [...], I would read it 3 I find myself thinking about [...] on a regular basis
4 Sometimes I feel like calling or writing [...]
39
Table B3
CPI: Celebrity-Persona Identification Scale (Bocarnea, & Brown, 2007)\
1 What is important to [...] is important to me 2 I look to [...] as a role model
3 I would like to be more like [...]
4 [...] has set an example for me of how to think and act 5 I sometimes imitate [...]
Table B4
Ohanian's (1990) Source Credibility Scale
Attractiveness Attractive Unattractive Classy Not Classy
Beautiful Ugly
Elegant Plain
Sexy Not Sexy
Trustworthiness Dependable Undependable
Honest Dishonest
Reliable Unreliable Sincere Insincere Trustworthy Untrustworthy Expertise Expert Not an expert
Experienced Inexperienced Knowledgeable Unknowledgeable Qualified Unqualified Skilled Unskilled
40
Table B5
Reynold's (1997) Message Elaboration Measure
While reading the [text and the tweet] were you... 1 Not very attentive to the ideas
2 Extending a good deal of cognitive effort
3 Distracted by other thoughts not related to the message 4 Doing your best to think about what was written 5** Searching your mind in response to the ideas
** Item deleted after reliability analysis
Table B6
Short version of the Big Five Inventory formulated by Rammstedt and John (2007)
I see myself as someone who... Big Five trait measured 1 ... is reserved* Extraversion
2 ... is generally trusting Agreeableness 3 ... tends to be lazy* Conscientiousness 4 ... is relaxed, handles stress well* Neuroticism 5 ... has few artistic interests* Openness 6 ... is outgoing, sociable Extraversion 7 ... tends to find fault with others* Agreeableness 8 ... does a thorough job Conscientiousness 9 ... gets nervous easily Neuroticism 10 ... has an active imagination Openness 11** ... is considerate and kind to almost everyone Agreeableness
* Reversed items
41
Appendix C
The following pages show the survey as presented to the participants. Notes will clarify which questions were presented to participants assigned to what condition.
43
Note. The instruction above was shown to participants assigned to condition A and B only.
Next, participants assigned to condition A were exposed to a cover story (see Figure A1) and tweet by Harry Styles (see Figure A2). Participants assigned to condition B were exposed to a cover story (see Figure A3) and tweet by Tom Hiddleston (see Figure A4).
44
Note. The text and links above were presented to participants assigned to condition A after
they were exposed to the cover story and the tweet.
Note. The text and links above were presented to participants assigned to condition B after
46
Note. The question above was presented to participants assigned to condition A. Participants
assigned to condition B were presented the same question, replacing One Direction singer Harry Styles with actor Tom Hiddleston. Participants assigned to condition C skipped this question.
47
Note. The question above was presented to participants assigned to condition A. Participants
assigned to condition B were presented the same question, replacing Harry (Styles) with Tom (Hiddleston). Participants assigned to condition C skipped this question.
48
Note. The question above was presented to participants assigned to condition A. Participants
assigned to condition B were presented the same question, replacing Harry (Styles) with Tom (Hiddleston). Participants assigned to condition C skipped this question.
49
Note. The question above was presented to participants assigned to condition A. Participants
assigned to condition B were presented the same question, replacing Harry (Styles) with Tom (Hiddleston). Participants assigned to condition C skipped this question.
50
Note. The question above was presented to participants assigned to condition A and B.
Participants assigned to condition C skipped this question, as they were not exposed to any text or tweet.
