MASTER’S THESIS
Emoticons as Social Influence: The Effects of Valence and Quantity of Emoticon Reactions on Message Credibility and Brand Attitude
Mette Hejlskov Student ID: 11117842
Graduate School of Communication Master’s programme of Communication Science
Supervisor: Dr. Young-Shin Lim
Abstract
This paper explores the interpersonal effects of the emoticon reactions recently introduced on Facebook. Specifically, it explores the effects of the quantity and valence of the reactions on message credibility and brand attitude. One hundred and fifty women and 61 men (Mage = 25.95, SDage = 6.59) were randomly assigned to an online two by two experimental design and shown a fictitious brand-generated Facebook post with either few or many positive (Like) or negative (Angry) emoticon reactions. An Analysis of Covariance (ANCOVA) and a
mediation analysis using PROCESS macro were performed in order to test the effects of the valence and quantity of the reactions on message credibility and brand attitude. Findings from the analyses show that when people are exposed to the Like reactions they perceive the message to be more credible than when they are exposed to the Angry reactions (p < .001). The quantity of the emoticon reactions, however, does not have any impact. Furthermore, the valence of the reactions does not influence brand attitude directly. However, the effect is mediated through message credibility (95% bootstrap CI = .1500 to .5252). Moreover, message credibility affects brand attitude directly (p < .001). Marketers can benefit from this study because it demonstrates the relative importance of understanding the effects of the emoticon reactions on Facebook. The study extends literature on the emotions-as-social-information (EASI) model because it shows that emoticon reactions carry social emotions-as-social-information that can be used by observers to judge the credibility of a message and subsequently infer their opinions about the brand. Moreover, future directions for research is suggested to extend the knowledge gained from this study.
Keywords: Emoticons, emotional expressions, emotions-as-social-information, social media advertising, social recommendations, message credibility, brand attitude
Emoticons as Social Influence: The Effects of Valence and Quantity of Emoticon Reactions on Message Credibility and Brand Attitude
Social media has altered the way consumers interact with each other and organizations (Duffett, 2015) and has given consumers new possibilities to create, publish, and share
content and opinions with others (Ghosh, Varshney, & Venugopal, 2014). Due to a rapid growth in social media usage, Facebook has become the largest social networking site. With an average of over one billion active users, Facebook has thereby exceeded Twitter,
Instagram, and Snapchat combined (Statista, 2016). Social media provides marketers with an opportunity to reach a large number of consumers in an easy and cost-effective way and research has shown that brands may benefit from social media marketing because it is able to influence consumers’ behavioral attitudes positively (Duffett, 2015). Therefore, marketers have deemed it necessary to integrate social media in their marketing efforts and to be present on various social networking sites, including Facebook.
It can be difficult for marketers to determine if their social media efforts have any impact on their brand and if their efforts are paying off. Most social networking sites, however, offer consumers ways to express their opinions about content and brands, for instance, with the Like reaction on Facebook. By using the Like reaction, consumers let their Facebook friends know what content they are engaging with, and which brands and products they like. Thus, the Like reaction provides marketers with an indication of how successful their social media marketing efforts are. In February 2016, Facebook launched five additional emoticon reactions (Love, Happy, Haha, Angry, and Sad) that users could use to react to content, posts, and pictures (Krug, 2016). The introduction of these emoticon reactions does not only give consumers new ways to show their appreciation about brand posts, but also makes it easier for consumers to show their disapproval about them. It is therefore crucial for
brands and marketers to understand how these reactions can influence brand evaluations. Research has shown that the Like reaction has positive effects on brand attitude (Phua & Ahn, 2016) and leads to higher sales (Johannes Gutenberg University Mainz, 2013), however, no research so far has explored the interpersonal effects of the new emoticon reactions on Facebook.
An emoticon is a symbol used to express emotions and bodily gestures, and is therefore thought to be able to replace certain parts of standard language (Vidal, Ares, & Jaeger, 2016). According to the emotions-as-social-information (EASI) model, emotions are thought to play a regulating role in social interactions; meaning emotions are able to influence others (Van Kleef, 2007). A study by Van Kleef, Van den Berg, and Heerdink (2015)
demonstrated that emotional expressions have interpersonal effects on attitude formation and change, regardless of whether they are expressed through written words, pictures, film clips, or emoticons. Thus, when people are exposed to negative emotional expressions, they form a negative attitude towards the topic and when they are exposed to positive emotional
expressions they form a positive attitude towards the topic.
The Like reaction on Facebook is considered a way to express an opinion about something and an effective way to share or promote, for instance, brands on social media (Jin, Wang, Luo, Yu, & Han, 2011). The new emoticon reactions may be considered the same because their purpose is to give Facebook users new ways to express either their positive or negative opinions about something (Krug, 2016). Jiménez and Mendoza (2013) argue that a social endorsement, such as a Like on Facebook, can serve as a heuristic cue for the
credibility and popularity of a brand and thereby influence consumers’ brand evaluations and product choices. This idea may be explained by the bandwagon effect, which posits that people tend to conform with the opinion of the majority because that opinion is perceived to
be the norm and is thus judged as the correct one (Lee & Hong, 2016). A study by Egebark and Ekström (2011) demonstrated that people only conform with the opinion of others when there are sufficiently many people influencing them. Furthermore, results from a study by Sherman, Payton, Hernandez, Greenfield, and Dapretto (2016) show that adolescents are more likely to “like” a photo on Instagram if that photo has been “liked” by many (rather than few) others, which illustrates the influence of the quantity of endorsements. Previous research has shown that both negative and positive opinions of others can influence credibility and brand attitudes (Ghosh, Varshney, & Venugopal, 2014; Lee, Park, & Han, 2008), however, none so far has explored the interpersonal effects of the emoticon reactions on Facebook. Therefore, the present study aims to explore if and how the different emoticon reactions can influence credibility and brand attitudes.
For the purpose of this study, I have chosen to focus only on the Like and Angry reactions because they represent positive and negative emotions and can therefore provide an understanding of how positive and negative emotional expressions towards content on Facebook can influence observers’ evaluations. The aim of this study is to explore how the emotional valence of emoticon reactions to brand-generated content on Facebook influences people’s perception of message credibility and their brand attitude. Moreover, the study seeks to investigate the moderating effect of the number of reactions on message credibility and the mediating role of message credibility on the relationship between the emotional valence of the reactions and brand attitude. The remainder of this paper is organized as follows. First, hypotheses are developed from a theoretical framework that provides an understanding of the underlying concepts. Next, the method, development of the research model, and
Finally, practical and theoretical implications are provided as well as limitations and future research ideas.
Theoretical Framework Emotions as Social Information
Research has often treated emotions as intrapersonal; meaning they are something people feel and something that arise as a result of an individual’s evaluation of events, thus they are thought to occur within the individual’s mind (Van Kleef, 2009). However, the emotions-as-social-information (EASI) model suggests that emotions also have interpersonal functions, in that they can influence, for instance, attitudes and behaviors of the observers of an emotional expression. The fact that emotions are expressed in social interactions and show on people’s faces indicates that the purpose of an emotional expression is to influence others, hence serve as social influence (Van Kleef, Van Doorn, Heerdink, & Koning, 2011).
Furthermore, Côté and Hideg (2011) posit that people can identify and understand emotions in facial expressions, pictures, and voices and use them to process information and infer their own behavioral decisions.
The EASI model suggests that the interpersonal effects of emotional expressions trigger inferential processing and/or elicit affective reactions in the observers (Van Kleef, 2009), meaning that emotional expressions provide information to observes about the expresser’s feelings, attitudes, and behavioral intentions, which in turn may trigger certain feelings in the observer and influence the observer’s behavior.
While no research has explored the interpersonal effects of emotional expressions directed at brand-generated content on Facebook, research indeed shows that emotions can play a regulating role in social interactions, in both offline and online contexts. The social
information processing (SIP) theory postulates that computer-mediated communication can convey as much information as face-to-face communication, yet, online communication is slower, hence exchanging information in online settings takes longer than in offline settings (Walther & Parks, 2002). The SIP theory suggests that online communicators use the content, style, and timing of verbal messages to exchange information, thus they make use of the cues available on the given medium, which, for instance, could be emoticons on Facebook.
Research shows that emotions in tweets can shape food choice, food intake, and liking (Vidal, Ares, & Jaeger, 2016) and that expressing emotions in both offline and online negotiations is very powerful because it enhances trust (Van Kleef, Van Doorn, Heerdink, & Koning, 2011).
The presented findings suggest that emotional expressions affect others’ behaviors and attitudes. Although, whether the effects are negative or positive depend on the emotion expressed and the situation or context it is expressed in (Hareli & Rafaeli, 2008). To extend the knowledge about the EASI model and the SIP theory and to understand the effects of the new emoticon reactions on Facebook, more research is deemed necessary.
Valence as Social Information
The introduction of the new emoticon reactions on Facebook has provided users with a shortcut to express their opinions online and users now have the opportunity to express non-verbal opinions about branded content that are either positive or negative. Before the
introduction, the only non-verbal reaction available was the Like reaction, which has positive valence. It has served as a useful tool for marketers, because when users “like” branded content on Facebook, they simultaneously share the content with friends, and thereby promote and spread awareness about the brand (Zhou, 2013). However, marketers now face new challenges because the new emoticons include negative reactions too, thus users also share their negative opinions with friends, which may generate additional negative opinions.
Communication on social media is mainly text-based and it has therefore previously been postulated that emotions are absent or at least very difficult to communicate in online settings (Huang, Yen, & Zhang, 2008). However, research on the role of emotions in computer-mediated communication suggests that textual communication can convey emotional valence through non-verbal displays such as emoticons (Derks, Fischer, & Bos, 2008). Huang, Yen, and Zhang (2008) furthermore argue that emoticons can replace certain parts of face-to-face communication because they represent emotional expressions, such as bodily gestures and facial expressions. These findings support the SIP theory because they demonstrate that communication can be affective in both online and offline settings (Walther & Parks, 2002). Considering these findings, the new emoticon reactions on Facebook may be seen as emotional expressions that, in line with the EASI model and SIP theory, convey interpersonal effects on the observers of the emoticons.
Interpersonal effects on message credibility. Message credibility is usually determined by the perceived credibility of the source of a message, however, in online settings the source is often unknown to the receiver, thus the credibility can be difficult to assess. It is therefore assumed that people use heuristic cues, such as other people’s opinions, to evaluate the source’s credibility (O’Reilly, MacMillan, Mumuni, & Lancendorfer, 2016).
Literature suggests that emotional expressions are important non-verbal signs used to judge the credibility of others (Hareli & Rafaeli, 2008; Kaufmann, Drevland, Wessel,
Overskeid, & Magnussen, 2003). For instance, a study on witness credibility showed that the emotions displayed during a testimony partly determines how credible a rape victim is
perceived and that displaying neutral or incongruent emotions reduce the perceived credibility (Kaufmann et al., 2003).
Moreover, literature on word-of-mouth (WOM) claims that other consumers’ opinions are perceived as more credible than the opinions of marketers (Lee, 2009). WOM
communication is defined as either face-to-face communication or online communication between consumers, regarding opinions and information about products or brands (Ghosh, Varshney, & Venugopal, 2014). Information about other people’s opinions is considered very credible because, in contrast to marketer-generated information, it is not seen as an attempt at persuasion, therefore people may use others’ opinions to make up their own (Lee, 2009).
Considering these findings, it is likely that when consumers are faced with brand-generated posts on Facebook, they take the cues on Facebook into consideration to evaluate whether the content is credible or not. Such a cue might be the opinion of others; thus, if people’s reactions to the brand-generated content are negative, it might reduce the perceived credibility of the content. On the other hand, if people react in a positive way about the content, it might signal that the content is credible. The following hypothesis is postulated. H1: The valence of the emoticon reactions influences message credibility, such that the Like reaction has a positive effect on message credibility, and the Angry reaction has a negative effect on message credibility.
Moderating role of quantity. The negative or positive reactions to brand-generated content on Facebook mostly come from sources unknown to the observer. Social media sites, such as Facebook, are based on the aspect of many-to-many communication, meaning senders and recipients are often unknown to each other and many people can both post and receive messages (Khosrowpour, 2012). This indicates that brands on Facebook can accrue a large number of reactions to their posts from a wide variety of people. Although it is possible for observers to see the names and profile pictures of people reacting to the content on Facebook, they often do not make an effort to do so and therefore merely see a summary of the
reactions. According to the elaboration likelihood model (ELM) and literature on WOM, people rely on heuristic cues to form their attitudes and evaluate the credibility of a message and message source (Lee, 2009; Zhou, Lu, & Wang, 2016), thus the number of reactions to a Facebook post could be a heuristic cue used to infer the credibility of the message. For instance, a study on the perceived credibility of a news feed demonstrated that when a news feed is endorsed by a lot of people (rather than few) it is perceived as more credible, which indicates that the number of non-verbal reactions may serve as a cue for the message credibility (Xu, 2013).
The effects of the number of reactions on message credibility may furthermore be explained by conformity (Cialdini & Goldstein, 2004) or bandwagon effects (Lee & Hong, 2016). Conformity refers to the fact that people are influenced by others; meaning they change their opinion about something to match the opinion of others and thereby follow the most popular opinion (Chin, Lu, & Wu, 2015). In a similar vein, the bandwagon effect suggests that people rely on the “wisdom of the crowd” to infer judgments and are therefore more likely to adopt certain beliefs when those beliefs are supported by others too (Lee & Hong, 2016). Lee and Hong (2016) studied the roles of emotional appeal, informativeness, and creativity in social media advertising and found that when users are exposed to brand-generated content that is “liked” by many others, they are more likely to click the Like reaction as well and comply with the expectations of others. Furthermore, Egebark and Ekström (2011) argue that the number of “likes” on a Facebook status update influences people’s opinions, such that the more “likes” the status update has, the more likely it is that the observer expresses a positive opinion about the post. The authors found that the effects were strongest when people observed the actions of their peers, however, significant effects were also found when the “likes” were expressed by unknown others.
As previously mentioned, emotional expressions, hence the negative and positive opinions of others, may affect message credibility. However, research suggests that negative information influences evaluations more strongly than positive information, which is referred to as the negativity bias (Ito, Larsen, Smith, & Cacioppo, 1998). In line with this, Baumeister, Bratslavsky, Finkenauer, and Vohs (2001) suggest that bad emotions have more impact than good emotions and that bad information is processed more thoroughly. In a social context, this means that when a group of people dislikes something or expresses negative opinions about something, the effect is more pronounced than if positive opinions are expressed. Ghosh, Varshney, and Venugopal (2014) illustrated this idea in a conceptual model about how social media WOM influences consumer decision making, by postulating that people give more attention and importance to negative WOM than positive WOM, thus negative opinions may be assumed to have a stronger impact.
Based on research that suggests that quantity affects message credibility, it may be assumed that the higher the quantity of emoticon reactions, the stronger the effect on message credibility is. However, based on the negativity bias, the Angry emoticon reactions may be assumed to have a stronger impact on message credibility than the Like reactions and even a few Angry reactions may have a strong impact on message credibility. The difference
between the negative effect of many and few Angry emoticon reactions is thus expected to be less than the difference between many and few Like reactions. This assumption is postulated in the following hypothesis.
H2: The quantity of emoticon reactions moderates the effect of the valence of the emoticon reactions on message credibility, so that the difference between the effect of many and few emoticon reactions is greater for the Like reactions than the Angry reactions.
Interpersonal effects on brand attitude. Brand attitude is an overall favorable or unfavorable evaluation of a brand (Wu & Wang, 2011) that is constructed of stored representations about the brand and information that is currently available about the brand (Van Kleef, Van den Berg, & Heerdink, 2015). In line with the assumption that consumers seek others’ opinions about brands to infer their own, it is suggested that the opinion of others also serve as a cue for forming brand attitudes (Chen, Kim, & Lin, 2015). Phua and Ahn (2016) studied the interpersonal effects of liking a brand on Facebook on consumers’ brand evaluations and found that it positively affects consumers’ brand trust, brand attitude, and purchase intention. It could therefore be assumed that reacting with a Like on brand-generated content would generate similar results.
Ghosh, Varshney, and Venugopal (2014) studied the effects of social media WOM and found that when consumers are exposed to positive WOM on social media, they are more likely to build a positive attitude towards the brand and conversely more likely to form a negative attitude when they are exposed to negative WOM. This idea is furthermore
supported by Van Kleef, Van den Berg, and Heerdink (2015) in their study on the effects of emotional expressions on attitude formation and change in various situations. The authors found that people’s attitudes are more positive when they are exposed to positive (rather than negative) expressions, and more negative when they are exposed to negative (rather than positive) expressions. The authors also found that emotional expressions may change the observer’s pre-existing attitude, thus the study further demonstrates the influence of
emotional expressions in social interactions. Moreover, it was found that the effects occurred regardless of the format in which the emotion was expressed (i.e., through written words, pictures, or emoticons). It can therefore be predicted that the emoticon reactions on Facebook may influence the brand attitude of observers, thus the following hypothesis is postulated.
H3: The valence of the emoticon reactions influences brand attitude, such that the Like reaction has a positive effect on brand attitude, and the Angry reaction has a negative effect on brand attitude.
Mediating role of message credibility. People reacting to brand-generated content on Facebook are sources of opinions that can be valued credible or not. However, as previously mentioned, their reactions can also be used to infer the credibility of a brand-generated message because observers may rely on the quantity of reactions as a heuristic cue to judge the credibility of the message. Moreover, people use others’ opinions as information to infer their own, thus to infer their attitude towards a brand (Chen, Kim, & Lin, 2015).
If the credibility of a message is perceived to be low, it is unlikely that the consumer’s brand attitude is as favorable as if the message credibility is perceived to be high. This
assumption is illustrated in a study by Esmaeilpour and Aram (2016) who explored the impact of message appeal and message source credibility on consumer brand attitude with the source of the message being the company behind the brand. The authors found that message source credibility positively affects consumer attitudes, so that the more credible the source of the message is perceived, the more favorable the consumer’s brand attitude is. Moreover, a study by Wu and Wang (2011) demonstrated that a highly credible WOM message source leads to a more favorable brand attitude than a less credible WOM message source. Although these studies examine the effects of the credibility of the message source, it is to be expected that message credibility also influences brand attitude because people may evaluate the credibility of a message and thereby form their attitudes towards a brand based on other people’s reactions. It is therefore assumed that the perceived message credibility does
attitude, so that when message credibility increases or decreases so does brand attitude. In light of this, the following hypotheses are posited.
H4: Message credibility influences brand attitude, so that when message credibility is high it affects brand attitude positively, and when message credibility is low it affects brand
attitude negatively.
H5: Message credibility mediates the effects of the valence of the emoticon reactions on brand attitude.
H6: Message credibility mediates the effects of the interaction between the valence and quantity of the emoticon reactions on brand attitude.
Figure 1. Conceptual model
Method Participants
Participants were approached and recruited through various groups on Facebook, which ensured that they were familiar with Facebook and possibly with the new emoticon reactions. In total the sample consisted of 211 participants (71.1% female, 28.9% male) between the age of 18 and 68 (M = 25.95, SD = 6.59). Out of the participants, 42.2% were
Valence of reactions (Like vs Angry)
Quantity of reactions (many vs few)
Message
credibility Brand attitude
H5: Valence of reactions → Message credibility → Brand attitude
H6: Valence x quantity of reactions → Message credibility → Brand attitude H1
H3
H2
Danish, 19.0% were Dutch, 7.6% were German, and 31.2% were other nationalities. As a compensation for their participation they were offered to enter a prize draw of a €25 gift voucher.
Design and Material
Participants were randomly assigned to one of four conditions in an online 2 (valence of reactions: negative vs positive) x 2 (quantity of reactions: few vs many) between-subjects factorial experiment assessing the main effects and possible interaction effects of the
emotional valence and number of reactions on a brand-generated Facebook post, and the indirect effects of message credibility on brand attitude.
Four brand-generated Facebook posts were created corresponding to each of four experimental conditions (negative valence/many reactions, positive valence/many reactions, negative valence/few reactions, positive valence/few reactions). The brand behind the Facebook posts was called Techtiq Audio and was developed only for the purpose of this study. The brand was positioned as a brand producing and selling audio and sound equipment and as a relatively new brand. The text developed for the Facebook post was kept simple to minimize possible effects of message length and appeal, however, the text had a slightly positive tone of voice to ensure ecological validity of the design, as brand-generated
Facebook posts usually have a positive tone of voice. The message was held constant across all four conditions.
Two of the Facebook posts featured Angry emoticon reactions and the other two featured Like reactions. Additionally, the Facebook posts differed in terms of the number of reactions being featured, such that there was either a few reactions or many reactions. The number of reactions representing few and many was determined by assessing similar brands
on Facebook and the numbers two and 336 were eventually chosen to represent few and many reactions. The experimental material can be found in Appendix A.
Procedure
The experiment was conducted using an online questionnaire, in which participants were asked to view the corresponding experimental material, which was a screenshot of the brand-generated Facebook post by the fictitious brand Techtiq Audio. Participants were provided with a brief description of the brand and were then instructed to take a close look at the Facebook post. They then completed a series of questions assessing their perceived message credibility, brand attitude, product involvement, and demographic information.
Prior to the actual experiment, a pilot test was conducted to check whether the manipulations were successful or not. The pilot test resulted in a few regulations in terms of how questions were phrased and placed, which was corrected before the actual experiment was carried out.
Measures
Message credibility. Message credibility was measured by six items on a seven-point Likert scale (1= very poorly, 7= very well) adapted from Xu (2013), and Appelman and Sundar (2016). Participants were asked to indicate how well the six items (‘credible’, ‘accurate’, ‘trustworthy’, ‘believable’, ‘fair’, and ‘authentic’) described the corresponding Facebook post. The six items were averaged to create a scale of message credibility (Cronbach’s α = .93, M = 3.46, SD = 1.23).
Brand attitude. Brand attitude was measured by six items from Phua and Ahn (2016) on a seven-point semantic differential scale, asking participants how they felt about the brand (‘unappealing/appealing’, ‘unpleasant/pleasant’, ‘boring/interesting’, ‘dislike/like’,
‘negative/positive’, and ‘bad/good’). The six items were averaged to create a scale for brand attitude (Cronbach’s α = .93, M = 3.51, SD = 1.18).
Manipulation checks. To check whether the manipulations of valence and quantity of reactions worked, participants were asked whether they agreed or disagreed with the
following four statements: ‘many people reacted to the brand post’, ‘few people reacted to the brand post’, ‘people reacted negatively to the brand post’, and ‘people reacted positively to the brand post’. This was measured on a seven-point Likert scale ranging from one to seven (1= strongly disagree, 7= strongly agree). Two scales were reversed, namely ‘people reacted negatively to the brand post’ and ‘few people reacted to the brand post’, and the four items were then averaged to create two scales; one for the manipulation check of the valence of reactions (Cronbach’s α = .94, M = 3.39, SD = 2.05) and one for the manipulation check of the quantity of reactions (Cronbach’s α = .91, M = 3.42, SD = 1.96).
Control variables. Product category involvement and message valence were controlled for in order to isolate the effects of the valence and quantity of the emoticon reactions. Product category involvement normally refers to a person’s perception of the importance and relevance of an object based on that person’s needs and values (Dens & De Pelsmacker, 2010). According to the ELM, people that are highly involved with the product category process and evaluate information about the product differently than those that are less involved with the product category. Therefore, it seems likely that people’s evaluation of the message credibility and their brand attitude may be influenced by their level of
involvement with the product category. The valence of the brand-generated message may also influence people’s evaluation of the message credibility and their brand attitude. For instance, if the valence of the brand-generated message is highly negative, it might influence the perceived message credibility and brand attitude negatively and vice versa.
Product category involvement. Product category involvement was measured on a seven-point semantic differential scale and was assessed using 10 items adopted from Phua and Ahn (2016). The items were ‘unimportant/important’, ‘boring/interesting’,
‘irrelevant/relevant’, ‘unexciting/exciting’, ‘means nothing to me/means a lot to me’, ‘unappealing/appealing’, ‘mundane/fascinating’, ‘worthless/valuable’,
‘uninvolving/involving’, and ‘not needed/needed’. The items were averaged to create a product category involvement scale (Cronbach’s α = .96, M = 4.36, SD = 1.38).
Message valence. To assess how the participants perceived the valence of the message in the Facebook post, they were asked to rate the valence of the message on a seven-point Likert scale ranging from one to seven (1= negative, 7= positive). The participants perceived the valence of the message to be slightly positive (M = 4.66, SD = 1.33). Complete scales for all measurements can be found in Appendix B.
Table 1
Correlations, means, and standard deviations for model variables
Measure M SD 1 2 3
1. Message credibility 3.46 1.23 1
2. Brand attitude 3.51 1.19 .64***
3. Product category involvement 4.36 1.38 .12 .29***
4. Message valence 4.66 1.33 .32*** .38*** .08
***
p < .001
Results
A correlation analysis was performed to assess if product category involvement and message valence correlate with message credibility and brand attitude (see results in Table 1). The correlation analysis shows that both product category involvement, r = .29, p < .001, and
message valence, r = .38, p < .001, significantly correlate with brand attitude, and that message valence, r = .32, p < .001, significantly correlates with message credibility, thus these two variables are included as control variables in all main analyses.
Manipulation Checks
Before testing the main model independent samples t-tests were performed to determine if the manipulation of valence and quantity of reactions was successful. The independent samples t-test with valence as the independent variable and the manipulation check of valence as the dependent variable revealed that valence was successfully
manipulated, t (204) = -20.56, p = .006, meaning participants exposed to the Angry emoticon reactions evaluated the reactions to be significantly more negative (M = 1.71, SD = 1.08) than those exposed to the Like reactions (M = 5.06, SD = 1.28). Additionally, the independent samples t-test with quantity as the independent variable and the manipulation check of
quantity as the dependent variable showed that quantity was successfully manipulated, t (186) = -13.83, p = .001. The participants exposed to few reactions evaluated the quantity of
reactions to be significantly lower (M = 2.10, SD = 1.19) than those exposed to many reactions (M = 4.81, SD = 1.62). These tests show that the manipulation of valence and quantity of reactions were both successful and can thus be used as predictors in the model and in further analyses.
Effects of Valence and Quantity of Reactions on Message Credibility
In order to test the direct effect of valence of the reactions and the interaction effect of valence and quantity of the reactions on message credibility, a two-way Analysis of
Covariance (ANCOVA) with valence and quantity of reactions as independent variables and message credibility as the dependent variable was performed.
Hypothesis 1 predicted that the valence of the emoticon reactions to the brand-generated Facebook post would affect the message credibility, thus the message would be perceived as more credible by the participants exposed to the Like reactions and less credible by those exposed to the Angry emoticon reactions. The analysis revealed that the message credibility was indeed influenced by the valence of the emoticon reactions and with that hypothesis 1 was supported. The analysis showed a significant, moderate effect of the valence of reactions on message credibility, F (1, 205) = 15.43, p < .001, ηp2 = .07. Thus, participants
who were exposed to the Like reactions perceived the message credibility to be significantly higher (M = 3.83, SD = 1.03) than those exposed to the Angry emoticon reactions (M = 3.10, SD = 1.27).
Hypothesis 2 predicted that the effect of the valence of the emoticon reactions on message credibility would be moderated by the quantity of the reactions, so that the difference between the effect of many and few Like reactions would be bigger than the difference between the effect of many and few Angry reactions. The quantity of the reactions did not moderate the effect of the valence of the reactions to the brand-generated Facebook post, so hypothesis 2 was not confirmed. The analysis revealed no significant interaction effect between valence and quantity of reactions on message credibility, F (1, 205) = 2.42, p = .122, ηp2 = .01, thus participants exposed to many Angry reactions (M = 3.05, SD = 1.20)
did not perceive the message credibility significantly differently than those exposed to few Angry reactions (M = 3.14, SD = 1.34), and those exposed to many Like reactions (M = 3.98, SD = 1.03) did not perceive the message credibility significantly differently than those exposed to few Like reactions (M = 3.67, SD = 1.12).
Additionally, the ANCOVA revealed that there was no significant main effect of the quantity of the reactions on message credibility, F (1, 205) = 1.07, p = .358, ηp2 = .00,
meaning message credibility did not differ between those exposed to few (M = 3.40, SD = 1.26) and many (M = 3.52, SD = 1.20) reactions. Moreover, the analysis showed that the assumption of equal variances in the population had not been violated, Levene’s F (3, 207) = 1.02, p = .384.
Effects on Brand Attitude: Mediation Through Message Credibility
To test the direct effect of the valence of the reactions and the indirect effect of message credibility on brand attitude, a mediation analysis using PROCESS macro based on Ordinary Least Squares (OLS) regression was performed (Hayes, 2012). Given the fact that the ANCOVA showed no significant interaction effect between valence and quantity of reactions on message credibility, this meant that quantity also did not moderate the indirect effect of message credibility on brand attitude, thus moderated mediation did not occur. Therefore, quantity was left out of the analysis and a simple mediation analysis using model 4 with 5,000 bootstrapped samples was performed instead.
Hypothesis 3 predicted that the valence of the emoticon reactions to the
brand-generated Facebook post would influence people’s brand attitude, so that participants exposed to the Angry emoticon reactions would have a more negative brand attitude and participants exposed to the Like reactions would have a more positive brand attitude. The PROCESS analysis revealed that the valence of the reactions did not influence brand attitude directly, thus hypothesis 3 was not confirmed. The analysis showed no significant effect of valence of the reactions on brand attitude (point estimate = .1069, SE = .1249, 95% CI = -.1394 to .3532), meaning participants exposed to the Angry emoticon reactions (M = 3.22, SD = 1.09) did not differ in terms of brand attitude from those exposed to the Like reactions (M = 3.80, SD = 1.19).
Hypothesis 4 predicted that message credibility would influence brand attitude, so that if message credibility was low brand attitude would be more negative than if message
credibility was high, in which case brand attitude would be more positive. This hypothesis was confirmed by the PROCESS analysis because it revealed a significant main effect of message credibility on brand attitude (point estimate = .5177, SE = .0654, 95% CI = .3888 to .6466). Thus, the more credible the message is perceived, the more positive the brand attitude is and vice versa.
Hypothesis 5 predicted that message credibility would mediate the relation between the valence of the reactions and brand attitude. The analysis supported this hypothesis because reactions with a positive valence increased the message credibility, which in turn increased the brand attitude, and reactions with a negative valence decreased the message credibility, which in turn decreased the brand attitude. Thus, the analysis revealed a significant positive indirect effect of the valence of the reactions on brand attitude through message credibility (point estimate = .3181, SE = .0949, 95% bootstrap CI = .1500 to .5262).
Lastly, hypothesis 6 predicted that message credibility would also mediate the effects of the interaction between the valence and quantity of the reactions on brand attitude.
However, as mentioned, quantity was not included in the analysis because moderation was not present. Moderated mediation did therefore not occur either and the hypothesis was not confirmed.
Discussion and Conclusion
The present study aimed to examine how exposure to emoticon reactions to brand-generated posts on Facebook influences people’s perception of the credibility of the message
and their attitude towards the brand. Specifically, it focused on the individual and
combinatory influences of the valence and quantity of the reactions and message credibility. Overall, the results of the study show that the emoticon reactions on Facebook, hence the Like and the Angry reactions, do have effects on message credibility and brand attitude, thus it supports and extends literature on the EASI model. The results indicate that the valence of the reactions to a brand-generated Facebook post influences how credible the message is perceived, but does not influence people’s attitude towards the brand. The quantity of the reactions does not have any impact on the message credibility, nor does it moderate the relationship between the valence of the reactions and message credibility. Therefore, quantity is not expected to influence brand attitude either, hence moderated mediation is not present. However, message credibility has a direct effect on brand attitude and it mediates the
relationship between the valence of the reactions and people’s attitude towards the brand, thus the results reveal a full mediation. Therefore, hypotheses 1, 4, and 5 are confirmed and
hypotheses 2, 3, and 6 are rejected.
This study reveals that positive reactions have positive effects on the perceived
message credibility and that negative reactions have negative effects. If people are exposed to the Like reaction, that is a positive emotional expression, they perceive the credibility of the brand-generated Facebook post to be significantly higher than if they are exposed to the Angry reaction, that is a negative emotional expression. It can therefore be assumed that emoticon reactions on Facebook carry social information as non-verbal emotional expressions and that the interpersonal effects of emoticon reactions differ depending on the emotional valence of the reactions. Thus, this study extends the literature on the EASI model and SIP theory by demonstrating that emotions expressed through emoticon reactions on Facebook have effects on observers.
However, the results indicate that the interpersonal effects of the emoticon reactions are restricted because the results show that the emoticon reactions do not affect observers’ brand attitude. Thus, whether the reactions are positive or negative do not influence people’s attitude towards the brand. While Van Kleef, Van den Berg, and Heerdink (2015) found that emotional expressions are used by others to infer their own attitudes, this study challenges that view. According to the EASI model, people need to be sufficiently motivated and able to process the emotional expressions in order for the emotions to have any interpersonal effects (Van Kleef, 2009). Therefore, it might be postulated that the participants of this study were not sufficiently motivated or able to process the information, which could explain why there were no effects of the valence of the reactions on brand attitude.
Although it has frequently been suggested that quantity plays a significant role in social recommendations and WOM (Egebark & Ekström, 2011; Xu, 2013), the results from this study propose otherwise. Whether there are many or few reactions to the brand-generated Facebook post does not affect how credible people perceive the message to be, thus the study reveals that the quantity of the reactions has no effects. Moreover, the results did not support the notion that even when emotions are of equal intensity, those of a negative nature have a greater effect than those of a neutral or positive nature, hence there was no negativity bias. Despite the fact that the manipulation of quantity was successful in the study, it is possible that the participants only noticed the number of reactions they were exposed to on a
subconscious level until they were explicitly asked about it. Questions assessing the perceived message credibility and brand attitude were asked before the manipulation check and it is therefore possible that the manipulation was not successful enough to have any effect on a subconscious level. Much literature has demonstrated that quantity matters in social
determine the credibility for experience products (i.e., hotels) and not search products, hence products that possess attributes that can be assessed before a potential purchase. In this study, the product category was headphones, which is defined as a search product. This could explain why the quantity of reactions to the brand-generated post did not have any influence. Lastly, it is likely that the power of the effects of emotional valence was simply so strong that it overshadowed any effect of quantity.
The study furthermore reveals that message credibility influences people’s attitude towards the brand and that message credibility fully mediates the effect of the valence of the reactions on brand attitude. This indicates that any effect on brand attitude is conditional on message credibility, so that brand attitude is only affected when message credibility is
lessened or strengthened. When people perceive the message to be highly credible their brand attitude subsequently becomes more favorable and the opposite follows when they perceive the message credibility to be low. The valence of the reactions does not have a direct effect on brand attitude, but the study shows that the valence has an indirect effect on brand attitude through message credibility. Thus, people form their brand attitudes based on the perceived credibility of the message they are exposed to rather than on the emotional expression (cf., emoticons), nevertheless, message credibility is determined by the valence of the emotional expressions.
While some of the limitations of this study have already been discussed, it is also likely that the fact that the study operates in an online environment could result in people being less affected by the bandwagon effect and conformity. In online settings people are relatively anonymous and can easily hide their identity, therefore they might not feel as much of a pressure to conform with the opinion of the majority. Egebark and Ekström (2011) found that the effects of the number of “likes” on a Facebook post were stronger if the “likes” came
from the observer’s peers than if they came from unknown others, thus it is possible that people would be more likely to conform with the majority, if that majority included peers of the observer. Moreover, the sample of the study is not representative for the general
population because it mainly consists of women and most participants are between the age of 20 and 30, thus it is likely that the results would differ if the sample was bigger and more representative. To be able to apply the findings to the general population a larger exploration is therefore deemed necessary.
However, although the study has its limitations, it also shows promising results that can be related to the EASI model, the SIP theory, and literature on social recommendations, because it indicates that the new emoticon reactions on Facebook have interpersonal effects on message credibility and indirect effects on brand attitude. The study thus demonstrates that the emoticon reactions can serve as online social recommendations.
Previous research has explored the Like reaction on Facebook and found that the Like reaction has both intrapersonal and interpersonal effects on people. However, this study contributes to these findings by demonstrating interpersonal effects of the Angry reaction too. This study offers some practical implications for marketers because it shows them that not only the Like reaction influences how people perceive a brand on Facebook, but that the Angry reaction, and possibly the other new emoticon reactions, influence how people perceive a brand. The introduction of the new emoticon reactions on Facebook has given consumers new and easier ways to express their positive and negative opinions about brands and brand-generated posts. In order for marketers to ensure successful social media
marketing, it is therefore crucial that they understand how these emoticon reactions affect their consumers and potential consumers. This study does not provide marketers with answers on how to write Facebook posts or what to include in these posts to generate positive
reactions. The study does, however, illustrate the importance of carefully considering the Facebook posts because it shows that negative reactions can damage the brand. The emoticon reactions on Facebook are normally shown in combination and only the three most used ones will appear if more than three different kinds of emoticon reactions are given, thus marketers should aim to produce content that will generate mainly positive reactions, so any potential negative ones do not appear. Therefore, it might make sense for marketers to produce content that appeals to all the positive emoticon reactions (cf., Happy, Haha, Love, and Like) to generate more credible Facebook messages and build a more favorable brand attitude.
However, the effects of the other emoticon reactions on Facebook are not yet known, as no research so far has examined them. For future research, it would therefore be interesting to explore if the interpersonal effects of the positive emoticon reactions differ from each other and correspondingly if the effects of the negative emoticon reactions differ from each other. The psychological differences between emotions have been extensively researched, but far less research has been conducted exploring the interpersonal effects of different emotions in online settings, which thereby justifies more research in this area.
Moreover, this study only explored the emoticon reactions’ interpersonal influence on the affective reactions in the observer, it would therefore be interesting to also explore their influence on the observer’s actual behavior. The EASI model postulates that emotional expressions can influence observers’ behavioral intentions and actions (Van Kleef, 2010), thus it is likely that the emoticon reactions on Facebook can influence people’s purchase intentions and actual purchase behavior, however, more research is deemed necessary to conclude upon this.
Overall, this study shows that the emotional valence of the emoticon reactions on Facebook has an impact on people’s evaluations of brands and brand-generated Facebook
messages. This is an important finding because it extends current literature and provides insights for marketers into the field of social media marketing. From this study, it becomes clear that marketers cannot and should not avoid social media if they wish to expand their businesses. Social media still alters the way consumers interact with each other and brands online, thus it is essential that marketers stay updated on changes in the online domain of social media.
References
Appelman, A., & Sundar, S. S. (2016). Measuring message credibility: Construction and validation of an exclusive scale. Journalism & Mass Communication Quarterly, 93(1), 59-79.
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5(4), 323.
Chen, K. J., Kim, J., & Lin, J. S. (2015). The effects of affective and cognitive elaborations from Facebook posts on consumer attitude formation. Journal of Consumer
Behaviour, 14(3), 208-218.
Cialdini, R. B., & Goldstein, N. J. (2004). Social influence: Compliance and conformity. Annu. Rev. Psychol., 55, 591-621.
Chin, C. Y., Lu, H. P., & Wu, C. M. (2015). Facebook Users' Motivation for Clicking the “Like” Button. Social Behavior and Personality: an international journal, 43(4), 579-592.
Côté, S., & Hideg, I. (2011). The ability to influence others via emotion displays A new dimension of emotional intelligence. Organizational Psychology Review, 1(1), 53-71. Dens, N., & De Pelsmacker, P. (2010). Consumer response to different advertising appeals
for new products: The moderating influence of branding strategy and product category involvement. Journal of Brand Management, 18(1), 50-65.
Derks, D., Fischer, A. H., & Bos, A. E. (2008). The role of emotion in computer-mediated communication: A review. Computers in Human Behavior, 24(3), 766-785.
Duffett, R. G. (2015). Facebook advertising’s influence on intention-to-purchase and purchase amongst Millennials. Internet Research, 25(4), 498-526.
Egebark, J., & Ekström, M. (2011). Like what you like or like what others like? Conformity and peer effects on Facebook.
Esmaeilpour, M., & Aram, F. (2016). Investigating the impact of viral message appeal and message credibility on consumer attitude toward the brand. Management &
Marketing, 11(2), 470-483.
Ghosh, A., Varshney, S., & Venugopal, P. (2014). Social Media WOM: Definition,
Consequences and Inter-relationships. Management and Labour Studies, 39(3), 293-308.
Hareli, S., & Rafaeli, A. (2008). Emotion cycles: On the social influence of emotion in organizations. Research in Organizational Behavior, 28, 35-59.
Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling.
Huang, A. H., Yen, D. C., & Zhang, X. (2008). Exploring the potential effects of emoticons. Information & Management, 45(7), 466-473.
Ito, T. A., Larsen, J. T., Smith, N. K., & Cacioppo, J. T. (1998). Negative information weighs more heavily on the brain: the negativity bias in evaluative categorizations. Journal of Personality and Social Psychology, 75(4), 887.
reviews on purchase intentions of search and experience products. Journal of Interactive Marketing, 27(3), 226-235.
Jin, X., Wang, C., Luo, J., Yu, X., & Han, J. (2011, August). LikeMiner: a system for mining the power of 'like' in social media networks. In Proceedings of the 17th ACM
SIGKDD international conference on Knowledge discovery and data mining (pp. 753-756). ACM.
Johannes Gutenberg University Mainz. (2013, June). Facebook recommendations could be worth millions of euros [Press release]. Retrieved from
https://www.uni-mainz.de/presse/16465_ENG_HTML.php
Kaufmann, G., Drevland, G. C., Wessel, E., Overskeid, G., & Magnussen, S. (2003). The importance of being earnest: Displayed emotions and witness credibility. Applied Cognitive Psychology, 17(1), 21-34.
Khosrowpour, M. (Ed.). (2012). Dictionary of Information Science and Technology. IGI Global.
Krug, S. (2016, February). Reactions now available globally. Retrieved from http://newsroom.fb.com/news/2016/02/reactions-now-available-globally/
Lee, S. H. (2009). How do online reviews affect purchasing intention?. African Journal of Business Management, 3(10), 576.
Lee, J., & Hong, I. B. (2016). Predicting positive user responses to social media advertising: The roles of emotional appeal, informativeness, and creativity. International Journal of Information Management, 36(3), 360-373.
Lee, J., Park, D. H., & Han, I. (2008). The effect of negative online consumer reviews on product attitude: An information processing view. Electronic Commerce Research and Applications, 7(3), 341-352.
O’Reilly, K., MacMillan, A., Mumuni, A. G., & Lancendorfer, K. M. (2016). Extending our understanding of eWOM impact: The role of source credibility and message
relevance. Journal of Internet Commerce, 15(2), 77-96.
Phua, J., & Ahn, S. J. (2016). Explicating the ‘like’ on Facebook brand pages: The effect of intensity of Facebook use, number of overall ‘likes’, and number of friends' ‘likes’ on consumers' brand outcomes. Journal of Marketing Communications, 22(5), 544-559. Sherman, L. E., Payton, A. A., Hernandez, L. M., Greenfield, P. M., & Dapretto, M. (2016).
The power of the like in adolescence: Effects of peer influence on neural and behavioral responses to social media. Psychological Science, 27(7), 1027-1035. Statista. (2016, September). Most famous social network sites worldwide as of September
2016, ranked by number of active users (in millions). Retrieved from
https://www.statista.com/statistics/272014/global-social-networks-ranked-by-number-of-users/
Van Kleef, G. A. (2007). Emotion in conflict and negotiation: Introducing the emotions as social information (EASI) model. In IACM 2007 Meetings Paper.
Van Kleef, G. A. (2009). How emotions regulate social life the emotions as social
information (EASI) model. Current Directions in Psychological Science, 18(3), 184-188.
Van Kleef, G. A. (2010). The emerging view of emotion as social information. Social and Personality Psychology Compass, 4(5), 331-343.
Van Kleef, G. A., Van den Berg, H., & Heerdink, M. W. (2015). The persuasive power of emotions: Effects of emotional expressions on attitude formation and change. Journal of Applied Psychology, 100(4), 1124.
for influence. European Review of Social Psychology, 22(1), 114-163.
Vidal, L., Ares, G., & Jaeger, S. R. (2016). Use of emoticon and emoji in tweets for food- related emotional expression. Food Quality and Preference, 49, 119-128.
Walther, J. B., & Parks, M. R. (2002). Cues filtered out, cues filtered in. Handbook of Interpersonal Communication, 3, 529-563.
Wu, P. C., & Wang, Y. C. (2011). The influences of electronic word-of-mouth message appeal and message source credibility on brand attitude. Asia Pacific Journal of Marketing and Logistics, 23(4), 448-472.
Xu, Q. (2013). Social recommendation, source credibility, and recency: Effects of news cues in a social bookmarking website. Journalism & Mass Communication Quarterly, 90(4), 757-775.
Zhou, C. (2013). Consumers as Marketers: An Analysis of the Facebook Like Feature as an Endorsement. W. St. UL Rev., 41, 115.
Zhou, T., Lu, Y., & Wang, B. (2016). Examining online consumers’ initial trust building from an elaboration likelihood model perspective. Information Systems Frontiers, 18(2), 265-275.
Appendix A Experimental Stimuli per Condition
Negative valence x Many reactions
Negative valence x Few reactions
Appendix B Measurements
Message credibility. How well do the following adjectives describe the Facebook post by Techtiq Audio? (‘very poorly/very well’)
Credible Accurate Trustworthy Believable Fair Authentic
Brand attitude. On the scales below, indicate your feelings about the brand Techtiq Audio. Unappealing/appealing Unpleasant/pleasant Boring/interesting Dislike/like Negative/positive Bad/good
Message valence. How would you rate the valence (tone of voice) of the brand post? (‘negative/positive’)
Product category involvement. On the scales below, indicate your feelings about headphones in general
Unimportant/important Boring/interesting Irrelevant/relevant Unexciting/exciting
Means nothing to me/means a lot to me Unappealing/appealing
Mundane/fascinating Worthless/valuable Uninvolving/involving Not needed/needed
Manipulation checks. How much do you agree with the following statements? (1= strongly disagree, 7= strongly agree)
Many people reacted to the brand post Few people reacted to the brand post People reacted negatively to the brand post People reacted positively to the brand post
