Consumer Brand Engagement via
Instagram: The Influence of Positive
Emotions
Name author: Bo Vriend
Student ID-card number: 10581200 Name supervisor: Rinaldo Kühne University of Amsterdam
Master’s programme Entertainment Communication Science Master’s Thesis
Abstract
With the rise of social media and photos becoming prominent means of
communication online, Instagram offers new possibilities for brands and organizations to create and maintain engagement. However, little is known about how consumers engage with their content and brands. With engagement, brands wish to interact with consumers in order to build and maintain a relationship with them. From previous studies we know that human faces are powerful channels of non-verbal communication and positive emotional content drives relationships. In this paper, we study this engagement phenomena online. The study investigated the effect of positive facial and textual emotional expression on affective,
cognitive and behavioral consumer brand engagement, including 226 participants in an online survey. The participants were exposed to an Instagram post by Reebok, including a model that shows a sweater of Reebok. In the experiment, the emotional expression of the testimonial was manipulated. Also, the textual emotional expression through the caption in the Instagram post was manipulated. The results only showed a significant effect of the joyful facial
emotional expression on affective consumer brand engagement. And when only including participants who passed the manipulation checks, this effect of the joyful facial emotional expression became also significant on behavioral consumer brand engagement. Finally, a number of limitations are discussed and suggestions are made for further research.
Keywords: social media, Instagram, emotions, joyful emotion, caption, experiment, consumer brand engagement, affective, cognitive, behavioural
Introduction
Nowadays, with the rise of social media, new possibilities have arisen for brands and organizations to create and maintain engagement. Brands and marketers wish to engage with users through their organization, products and online in order to build and maintain a
relationship with them (Erkan, 2015). As a result of a good relationship, brands can create loyal consumers and therefore improve their organizational performance by increasing productivity, profitability and consumer satisfaction (Hoffman & Fodor, 2010). One way of creating engagement is the use of online photo sharing communities, which have grown at an impressive pace in social media. Take Instagram, where users upload 55 million photos a day to the site (Bakhshi, Shamma & Gilbert, 2014). This presents a key research challenge for photo sharing communities like Instagram. Since engagement is vital to photo sharing communities, it is critical to understand what form of content drives consumers to engage.
Photos are becoming prominent means of communication between brands and consumers online. Also, for companies and brands to engage with their consumers (Luarn, Lin & Chiu, 2015). Despite photos’ pervasive presence in online world, we know little about how people interact and engage with brands on Instagram. Several studies have focused on how users engage with textual content (Berger & Milkman, 2012; Burke & Kraut, 2008; Jamali & Rangwala, 2009; Millen & Patterson, 2002). These studies have shown that textual content can have a positive impact on online engagement. Therefore, the caption, the textual content of Instagram that is shown under a posted picture, is included in this study. Only a few studies did research on what makes visual content socially engaging online. By carrying out this study, this gab could be fulfilled and academic knowledge will be supplemented in areas where it is missing.
Not only visual content, but especially faces engage by attracting more likes and comments on Instagram (Bakhshi, Shamma & Gilbert, 2014). Many pictures op people’s faces are posted by brands to promote their products and are available on Instagram. In order to distinguish a photo and to get the best out of it, this paper focus on how facial emotional expression might impact consumer brand engagement (CBE).
Previous research indicates that emotional content drives relationships. Not only between people, but also between people and brands. Besides, positive emotion can increase the favourability towards a brand (Heath, Brandt & Nairn, 2006). It can also stimulate sharing content because people share content when it makes them look good. Or to create happiness and excitement (Yuki, 2015). Moving forward from this, successful social media marketing should create content, textual as well as image content, that triggers a positive emotional response in the receiving consumer. Positive emotions like joy, that are dominant in
advertisement, can improve relationships between brands and consumers (Teixeira, Wedel & Pieters, 2012). In turn, this stimulates consumers’ forwarding responds by engaging (Dobele, Lindgreen, Beverland, Vanhamme & van Wijk, 2007).
Therefore, this study takes a closer look into the effect of the presence of a joyful facial and caption emotional expression in organization’s post on Instagram on online CBE on Instagram. The following research question has been composed: To which extent do photos, made by or for an organization posted on their Instagram, with a facial emotional expression influence the online consumer brand engagement? And what effect does the caption emotional expression in the Instagram post of the brand has as a moderator? Understanding how
positive emotional content might signify engagement, can impact both science and brands in practice. Since there is little research on positive emotions and image content, this paper adds value to the academic field. As Hoffman and Fodor (2010) have shown, CBE can increase brand performance outcomes, like sales growth, cost reduction, visibility and contributed
consumers. Therefore, this paper has an important practical relevance to it and multiply a great value for organizations and brands.
The paper is structured as follows. First, we present the main concepts of Instagram, customer brand engagement, facial emotional expression, caption emotional expression and their hypothesized relationships. Second, the method is presented, followed by analyses of the findings. We conclude by discussing the results and offering managerial implications,
limitations, and future research directions.
Theoretical background
To be able to provide a clearer answer to the research question and to go deeper into this topic, the social media platform Instagram is fist discussed. Then the dependent variable, namely consumer brand engagement and then the two independent variables ‘facial emotional expression’ and ‘caption emotional expression’ will be made clear.
Why using Instagram out of all social media platforms?
Instagram is a social media platform where users can post pictures or short videos. Users can also interact with people, brands and organizations by following them or through liking, sharing, tagging, commenting and seeing their pictures. They can also send a private message to each other or save photos (de Vries, Möller, Wieringa, Eigenraam & Hamelink, 2018).
Nowadays, Instagram is a very popular social media platform and is often integrated in organization’s marketing strategy. In fact, 86% of top brands have official accounts on Instagram (Erkan, 2015). In 2018, Instagram saw the number of users grow to more than 1 billion, with more than 500 million daily users (Sherman, Greenfield, Hernandez & Dapretto,
2018). It is also widely used because of the growing popularity of visual content. Compared to other social media platforms like Facebook, Twitter or YouTube, Instagram is a very good social media platform to provide engagement. This is because images receive 22% more engagement than video post and 53% more than only textual content. On a whole, images are the most compelling and therefore an important element of the posting strategy for brands (Erkan, 2015). However, textual content is also being used on Instagram. To describe and clarify the image, users can add text under the photo, which is known as the caption. Textual content can also stimulate engagement, for example by adding a call-to-action or using hashtags (Smith & Sanderson, 2015). Therefore, this social media platform was selected for this study.
The relevance of consumer brand engagement
Consumer brand engagement refers to the consumers’ particular interactive brand relationship. It is conceptualized as consumer’s positive brand-related interactions and determined by three dimensions; cognitive, affective and behavioural investment (Hollebeek, Glynn & Brodie, 2014). It can be difficult for brands to create CBE because consumers are active participants, not only passive recipients. They will decide if they are going to engage with the brand, based on their needs, goals and motives. Therefore, it is related to consumer satisfaction (Ashley & Tuten, 2015).
Many studies only focus on the behavioural part of CBE by looking at explicit actions on the content of the brand’s social media (Bakhshi, Shamma & Gilbert, 2014; Erkan, 2015; Gummerus, Liljander, Weman & Pihlström, 2012). For example, by measuring the number of likes, shares and comments on the Instagram-post. Luarn, Lin and Chiu (2015) have shown that more likes and comments can lead to more engagement. Most importantly, it is about consumers taking action, making efforts and spending time on a brand (Hollebeek, Glynn &
Brodie, 2014). This form of engagement is very visible and can be a valuable indicator of the success of the Instagram-post (Erkan, 2015). In this experiment, actual behavioural
engagement is not possible. Therefore, we will refer to behavioural intention engagement. However, in this paper, engagement has a more extensive meaning. Namely, consumers do not take action through behavioural engagement without feeling of thinking something. Cognitive engagement is associated with logical, problem-oriented situations. It refers to “a consumer’s level of brand-related thought processing and elaboration in a particular consumer/brand interaction.” (Hollebeek, Glynn & Brodie, 2014). While affective engagement is primarily emotional, focussing on positive brand-related affect (and feelings towards the brand Dessart, Veloutsou & Morgan-Thomas, 2015).
By not only focussing on the behavioural part of engagement but also looking at cognitive and emotional engagement, the brand’s messaging has a more interactional perspective where the consumer can become more engaged with the brand. Creating
engagement with consumers is important to brands because it can retain information, helps to stay on top of mind, makes the consumer feel involved and eventually it can stimulate them to buy the product (Ashley & Tuten, 2015).
The role of facial emotional expression
The word ‘emotion’ is not very accurate, quite complex and there is little consistency among the definition in academic literature. Namely, it refers to a loose collection of
phenomena and the boundaries of these collections are hard to drawn. People experience emotions when there is a personal interest. It can be a response to internal and external stimuli that are personally relevant (Schere, 1982). Or, as Kleinginna and Kleinginna (1981)
described it “Emotion is a strong feeling deriving from one’s circumstances, mood, or relationships with others.”. Another definition of emotion by Kleinginna and Kleinginna
(1981) that is more extensive and applicable for this study because of the affective, cognitive and behavioural aspects is: “Emotion is a complex set of interactions among subjective and objective factors, mediated by neural-hormonal systems, which can (a) give rise to affective experiences such as feelings of arousal, pleasure/displeasure; (b) generate cognitive processes such as emotionally relevant perceptual effects, appraisals, labelling processes; and (c) lead to behaviour that is often, but not always, expressive, goal directed, and adaptive.”
After years of evolution, a set of basic emotions has been developed, each owning their specific content, facial expressions and visual constructions (Frijda, 2016). In this study, we will focus on the facial expression because faces have visual skills with an impressive variety of context in order to recognize the complexity of emotions (Kramer, Guillory & Hancock, 2014). From previous studies, we know that human faces are powerful channels of non-verbal communication (Bakhshi, Shamma & Gilbert, 2014). Facial expressions have the ability to communicate emotions to others. This can lead to emotional contagion; a
phenomenon that triggers emotions and behaviour by others through one person’s emotions (Kramer, Guillory & Hancock, 2014). There is a strong connection between facial expression and experienced emotions. A facial emotional expression can therefore affect others’
emotional expression and emotional state (Bakhshi, Shamma & Gilbert, 2014).
A few facial expressions are distinct and cross-cultural, for example joy. Therefore, joy is a good emotion to measure continuously and non-intrusively, which offers advantages for the validity. Positive emotional expressions in faces can activate several areas of the brain (Bakhshi, Shamma & Gilbert, 2014). As a response, people want to avoid or reject negative emotions but approach or retain positive emotions (Teixeira, Wedel & Pieters, 2012). Positive emotions like enjoyment appeared to facilitate heuristic processing which leads to greater endorsement (Griskevicius, Shiota & Neufeld, 2010). This endorsement can also be
Knobloch-Westerwick (2016) showed in their study that positive expressed emotions on Instagram can elicit positive emotions among viewers. They suggested that individuals detect emotions on social media posts and adopted the emotions themselves, for example by feeling positive about the post, thinking positively about the post or posting a positive comment. Regarding these finding, this study predicts a positive effect of a positive facial emotional expression like joy on online CBE.
H1: Cognitive (a), emotional (b) and behavioural (c) online consumer brand engagement will be higher after showing a product by a person with a joyful facial emotional expression on an organization’s Instagram post than an organization’s Instagram post showing a product by a person with a neutral facial emotional expression.
The role of caption emotional expression
Besides posting pictures or videos, users of Instagram can also put text under their picture or video to describe the content. This is called a caption. “A caption provides an illustration or a picture with a title or explanation.” (Smith & Sanderson, 2015). Using a good caption can add personality to a brand and inspire viewers to take action (Dobele, Lindgreen, Beverland, Vanhamme & van Wijk, 2007). As been told before, studies have shown that textual content can impact engagement positively and emotional contagion can play a role as well. Emotional contagion is not limited to face-to-face communication, it can also occur as a result of written messages (de Vries, Möller, Wieringa, Eigenraam & Hamelink, 2018). Receivers of written messages can successfully detect emotions as intended. As a result, consumers can also adopt the emotions of other people or brands without directly viewing them (de Vries, Möller, Wieringa, Eigenraam & Hamelink, 2018). In this study, we will refer
to this textual expression as caption emotional expression. Yuki (2015) has found that content that showed happiness is the most significant emotion to share, an element of engagement. Also, positive content is more viral then negative content. And in order to go viral, people have to engage with the content (Berger & Milkman, 2012). Regarding these finding, this study predicts a positive effect of a positive caption emotional expression like joy on online CBE.
H2: A joyful caption emotional expression that is shown on an organization’s Instagram post has a more positive effect on cognitive (a), emotional (b) and behavioural (c) online consumer brand engagement compared to a neutral caption emotional expression that is shown on an organization’s Instagram post.
Caption emotional expression as a moderator
Finally, it is examined whether the caption emotional expression has an effect on the relationship of a joyful facial emotional expression on CBE. It is suggested that consumers engage more with positive content because it reflects positively on them when they engage for self-presentation purposes or to communicate their identity (Berger & Milkman, 2012). This applies to visual content as well as for textual content. If there is positive emotion expressed through text, it will strengthen the positive visual emotion because the content would radiate positive emotion as a whole. Therefore, the least CBE will come from a neutral facial emotional expression and a caption that does not describes a joyful emotion, but the CBE would increase more when there is a joyful facial emotional expression in the Instagram post. When there is a joyful facial emotion visible, the CBE from both directions would remain more constant but the Instagram post with the joyful facial emotional expression and the
joyful caption emotional expression would be the best to obtain CBE. Therefore, the next hypotheses have been set up:
H3: Cognitive (a), emotional (b) and behavioural (c) online consumer brand engagement will be higher after showing a product with a joyful facial emotional expression on an organization’s Instagram post than after showing a product with a neutral facial emotional expression on an organization’s Instagram post, but this effect will be more pronounced if the participant saw a joyful caption emotional expression that is shown on an the Instagram post than if they saw a neutral caption emotional expression that is shown on the Instagram post.
Figure 1 Conceptualised model + + + Method
Sample and Design
All people that are familiar with Instagram, irrespective of age or gender were eligible for this study. This is because this research could be helpful for many types of
Consumer brand Engagement Caption emotional expression
organizations who wants to have more online CBE, including different kind of target groups with different kind of characteristics. A convenience sample was used to draw the sample. This means that participants were recruited from the network of the researcher who carried out the research.
In the invitation for the research, there was a link to the questionnaire in the Qualtrics research program. The participants were recruited in May 2019 via Facebook, Instagram, LinkedIn, e-mail and WhatsApp. The number of participants were n = 305 adults
and n = 226 of them were included as a sample if they were at least 18 years old. In fact, a total of 79 people was made missing, of which 64 people did not fill in the statements about the dependent variable CBE, four people were under the age of 18 and eleven were not familiar with Instagram and therefore could not empathize with this experiment. The average age of the 226 participants group was 33 years (M = 33.06, SD = 13.12). The sample
consisted of 64.2% women and 35.8% men.
For this online questionnaire, an experiment with a 2x2 factorial design was conducted. The first factor, the facial emotional expression consisted of two levels. The second factor, caption emotional expression, had also two different groups. Both factors were nominal and a between-subject design, where each participant only saw one of the four conditions. See table 1 for the demographic data of the participants per condition.
Table 1
Conditions experiment
Neutral Face Joyful Face Neutral Caption Condition 1 Condition 2 N= 52 N=55 19 men/ 33 women 29 men/ 26 women (M = 35.00, SD = 14.25) (M = 32.73, S D= 14.29)
Joyful Caption Condition 3 Condition 4 N= 55 N= 64
17 men/ 38 women 16 men/ 48 women (M = 33.03, SD= 12.29) (M = 31.78, SD= 11.89)
Experimental Stimuli
Independent variable. In this study, the focus is on a positive and joyful emotion. The independent variable was operationalized by using different facial expression, namely a neutral face without emotion and a joyful face laughing, showing positive emotion. This was implemented by having a voluntary person showing a product with a neutral face versus showing a product with a joyful face. The product that was showed to the participants is a sweater from Reebok, looking like it was posted on the Instagram of Reebok Sporstwear. Moderator. The moderator caption emotional expression was operationalized by using different sentences under the Instagram picture, namely a caption that described the joyful emotion and a caption that did not describe an emotion. This was implemented by using the sense ‘Feeling HAPPY while wearing the new Wor Double Woven Collection. #linkinbio #reebok’ for the caption that described the joyful emotion versus using the sense
‘Wearing the sweater from the new Wor Double Woven Collection. #linkinbio #reebok’ for the caption that had a neutral tone.
Stimuli besides the manipulations. In order to warrant the intern validity, the stimuli were similar in everything but the independent variable and moderator. Here are the efforts that were made for making the conditions as equal as possible besides the manipulations: The Instagram post was in all the conditions posted by ‘Reebok’, and had the same number of likes and comments. The same product (the Reebok sweater) was used and worn by the same person. The lay-out of the Instagram post was the same and the same hashtags were used (See Appendix 1).
Measures
Three 6-item scales inspired by Dessart, Veloutsou and Morgan-Thomas (2015) were used to measure affective, cognitive and behavioural online consumer brand engagement. This measurement was used because it focusses on all three parts of online CBE instead of only one part. Besides, it emphasizes engagement through online social media platforms especially with brands. A 7-point Likert-scale was used, ranking from ‘‘strongly disagree’’ to ‘‘strongly agree.’’ See Appendix 2 to see the items that were used to measure the three different components of online CBE.
Affective, cognitive as well as behavioural engagement all included six items and a 7-point Likert scale. The higher the score on the scale, the higher the degree of CBE. However, item four of the affective CBE scale was a reversed statement. This question was reversed before a resistance scale was made. Three factor analyses had indicated that the six items together formed a one-dimensional scale with an eigenvalue above one. The affective CBE scale had an eigenvalue of 3.91, the cognitive CBE scale an eigenvalue of 3.32 and an
factor loading and explained variance of 65.22%, cognitive CBE 55.34% and behavioural CBE had 69.46%. In addition, all three scales had a clear nod after component one in the scree plot.
The reliability analysis of the affective CBE scale showed a reliable scale with a Cronbach alpha of 0.887 (M = 3.52, SD = 0.87). However, the reliability analysis of cognitive CBE showed a Cronbach alpha of 0.697. If item two “I spend a lot of time thinking about the brand” would be deleted, the Cronbach alpha would increase to 0.862. Therefore, item two was deleted before making the scale for cognitive CBE (M = 2.26, SD = 1.06). Behavioural CBE showed an excellent reliable scale with a Cronbach alpha of 0.901 (M = 2.45, SD = 1.18). Because all scales were found reliable, the means of all single items per scale were computed and combined as one scale, which means three scale in total to measure CBE.
Besides, all the Skewness values and Kurtosis values were above -2 and under 2, which means that these three variables ware normally distributed. The Skewness value of affective engagement was -0.14 and the Kurtosis was 0.45. Cognitive engagement had a Skewness value 1.08 of and a Kurtosis value of 1.02. And the Skewness value of behavioural engagement was 0.95 and the Kurtosis was 0.82.
Finally, after making the three scales, a factor analysis had been conducted with all the 17 separate items that measured these three forms of engagement. This final factor analysis also indicated a three-dimensional scale. This was a good indicator, since there were three different scales in this research to measure CBE. The first component had an eigenvalue of 7.99 and an explained variance of 47.04%. The second component had an eigenvalue of 1.95 and an explained variance of 11.47%. The third and final component had an eigenvalue of 1.70 and an explained variance of 9.99%. Together, these three components had an explained variance of 68.50%.
Besides the dependent variable, the single items of attitude towards Reebok was also made into one scale. Five statements were made about attitude, measured with a 5-point Likert scale, ranging from for example (1) Bad to (5) Good or (1) Unattractive to (5)
Attractive. To construct the variable “attitude”, a principal component factor analysis with an orthogonal (Varimax) rotation was performed. The five different items were included in this analysis. In this analysis, one component was found that had an eigenvalue greater than one, namely 3.72. In addition, this component was on the left of the kink from the scree plot. The component explained in total 74.4% of the variance between the items. Cronbach’s alpha indicated a very reliable scale (α = 0.910).
Procedure
Before the participants filled out the online survey, they were asked to agree with the term of this study by means of a declaration of consent. This stated that the anonymity of the participants remains guaranteed and for which the results of the research will be used. If the participants agreed, the survey could begin.
First, some demographic data like age, degree and gender were asked. Second, the participants were asked if they were familiar with Instagram and to what extend they use it. Before seeing one of the conditions, the participants were asked to have a good look at the Instagram post they were going to see and if they could imagine seeing this when scrolling through their own timeline. To stimulate this, the stimuli was shown for ten seconds before participants could go to the next question. After seeing the Instagram post, the participants were asked about their online CBE. To operationalize online CBE on Instagram, three dimensions of engagement were measured, namely cognitive, emotional and behavioural engagement. After, participants were asked if they knew the brand Reebok and their attitude towards this brand. Finally, the online survey was closed with two questions for the
manipulation check to clear out the different conditions of the independent variable and moderator. Participation in the experiment lasted approximately ten minutes.
Before making the online survey, a preliminary research had been carried out in order to choose the right brand, namely Reebok. This pilot was carried out in order to assess whether extreme brand attitudes existed. In the preliminary research, 26 respondents were asked about their attitude towards ten different sport brands. Their attitude was measured through a 5-point Likert-scale, ranking from (1) very negative to (5) very positive, based on the study of Cacioppo and Berntson (1994). The attitude towards the brand Reebok was the least extreme (M = 3.21, SD = 0.60) which reduces the chances of different opinions about the brand and therefore chosen for this study.
Results
Analysis strategy
All analyses were conducted with IBM SPSS. Before testing the hypotheses, two manipulation checks were conducted. To test the manipulation check for facial emotional expression, a two-way variance analysis (Univariate) was used. For the second manipulation check, a chi-squared test was used to test caption emotional expression.
After testing the manipulation checks, multiple randomization checks were carried out and possibly included as a covariate when testing the hypotheses. These variables included age, the use of Instagram, the quantity of the use of Instagram, the familiarity with Reebok, attitude towards Reebok and gender. For familiarity with Reebok and gender a chi-squared test was used and the rest of the variables were tested with a two-way variance analysis (Univariate).
When analyzing the hypotheses, first the equal variances were looked at by using Levene's test for affective, cognitive and behavioral engagement. Then, hypothesis 1
examines the effect of facial emotional expression on the three types of online CBE, namely affective, cognitive and behavioral engagement. Hypothesis 2 investigates the effect of the caption emotional expression on the three types of CBE. In hypothesis 3, it was examined whether the variable caption emotional expression had a moderating effect on the relationship investigated in hypothesis 1. These three hypotheses were tested by a two-way variance analysis (Univariate).
Finally, after analyzing the hypotheses with only the dependent, independent and moderating variables, two ad hoc analyses were conducted. The first ad hoc analysis added the covariate gender with the three hypotheses and the second ad hoc analysis added the covariate gender and only included participants who passed the manipulation checks. These two ad hoc hypotheses were also tested by a two-way variance analysis (Univariate).
Manipulation and randomization checks
First manipulation check. Two manipulation checks were performed to see if the difference in the independent variables facial and caption emotional expression in the communication was clear for the participants. The first question of the manipulation check was which emotion on the face in the Instagram post was expressed. The participants had to choose between six different emotions, including the emotions disgust, fear, joy, sadness, anger and surprise. Participants had to answer if they saw these emotions expressed on the face of the model with a 7-point Likert-scale, ranking from (1) None at all to (7) Totally. For the first manipulation check, six two-way variance analyses (Univariate) were conducted for each different emotion. First, results showed that the analysis with the emotion ‘joy’ was significant; F(1, 224) = 383.89, p<0.05. The other five emotions that were not expressed in
the stimuli were also checked with Univariate analyses. The result of the emotion ‘disgust’ was: F(1, 224) = 62.27, p<0.05. The emotion ‘fear’ showed: F(1, 224) = 56.47, p<0.05. The four emotion ‘sadness’ had a result of F(1, 224) = 76.42, p<0.05. ‘Anger’ showed the
following result: F(1, 224) = 46.20, p<0.05. And finally, the emotion ‘surprise’ had a result of
F(1, 224) = 6.92, p<0.05. All these five emotions also had a significant effect. However,
when taking a look at the graph of the Post Hoc, these five emotions were negative and decreased, while the emotion joy increased. Therefore, the first manipulation check was successful.
Second manipulation check. For the second manipulation check, the participants answered the following question: ‘Did the caption (the text under the picture) include a word that described an emotion?’ The participants could choose between the answer options (1) Yes, (2) No or (3) I don’t know. The ones who responded with (3) I don’t know n = 82, were made missing for this manipulation check. The participants who completed the neutral caption condition also experienced this to a large extent, namely 71.4%. The same applies to the participants from the joyful caption condition, namely 80.6%. A correlation was used to carry out this second manipulation check. The chi-squared test shows that there was a
significant correlation between the question about the caption and the two different conditions χ2(1, N = 144) = 0.52, p<0.05. The Cramer’s V value was 0.52 which indicated a large effect size. This significant result shows that the second manipulation check was successful.
First randomization check. Before running the analyses for the randomization checks, the conditions were made into one variable with four different values for each condition. The first randomization check was for the variable age (M = 33.06, SD = 13.12). With the help of a Univariate analysis it was concluded that the averages in age did not differ significantly between the conditions: F(3, 222) = 0.59, p = 0.622.
variable. One question to measure the use was: ‘Do you use Instagram with your own Instagram account?’ The participants were able to choose between the answer options (1) Yes and (2) No. A Univariate analysis was carried out (M = 2.64, SD = 0.16). A non-significant effect was found: F(1, 224) = 0.21, p = 0.646.
Third randomization check. In addition, the second question about Instagram use was: “How often do you use Instagram?”. The scale ran from (1) A few times a year to (5) Multiple times a day. 25.2% of the participants uses Instagram on a daily basis and the largest proportion (46.5%) even indicated that they use is multiple times a day. The average use of Instagram was 4.09 (SD = 1.16). To check this variable, a Univariate analysis was conducted. Also, a non-significant effect was found: F(4, 207) = 2.15, p = 0.076.
Fourth randomization check. Participants were asked is they knew the brand Reebok before they stated the online survey. They could choose between the answer options (1) Yes, (2) No, (3) I don’t know, where 99.1% of the participants did and 0.9% did not know if they knew the brand. The chi-squared test shows that there was not a significant relationship between the conditions and familiarity with Reebok χ2(1, N = 226) = 2.25, p = 0.523. The Cramer’s V value was 0.10 which indicated a small effect size.
Fifth randomization check. The attitude towards the brand Reebok was also included for a randomization check. A Univariate analysis was used to check this variable (M = 2.54, SD = 0.16). A non-significant effect was found F(22, 198) = 0.59 p = 0.927. From this it can be concluded that age, the use and quantity of use of Instagram, the familiarity with Reebok and the attitude towards Reebok were all normally distributed over the conditions and not further included and checked in the analyzes.
Sixth randomization check. Finally, a randomization check for gender was performed. The chi-squared test shows that there is a significant relationship between the conditions and gender χ2(1, N = 226) = 10.68, p<0.05. The Cramer’s V value was 0.22 which
indicated a medium effect size. It is therefore assumed that there were significant differences between men and women in the conditions. From this, gender was included as a covariate to see what effect it has on affective, cognitive and behavioural CBE.
Analyzing the hypotheses
First, the equal variances were looked at. For this, the assumption of equal variances in the population has been assumed, Levene's F(3, 222) =.1.82, p = 0.145 for affective
engagement, but not for cognitive engagement Levene's F(3, 222) = 4.19, p<0.05 or behavioral engagement Levene's F(3, 222) = 3.51, p<0.05. This assumes that the homoscedasticity of cognitive and behavioral CBE were violated.
All hypotheses were tested with two-way variance analyses (Univariate). Hypothesis 1a assumes that the facial emotional expression has a positive effect on affective online CBE. A significant effect was found for this relationship, F(1, 222) = 15.57, p <0.05, η² = 0.07. Participants who were exposed to the joyful face therefore scored higher on the affective engagement scale (M = 3.74, SD = 0.88) than participants who were exposed to the neutral face (M = 3.29, SD = 0.79). The analysis shows that it can be assumed that a joyful facial emotional expression does indeed influence consumers’ brand affective engagement more positively compared to a neutral facial emotional expression. Hypothesis 1a is hereby accepted.
Hypothesis 1b assumes that the facial emotional expression has a positive effect on cognitive online CBE. However, there was not a significant effect: F(1, 222) = 1.79, p = .183. Participants who were exposed to the joyful face scored (M = 2.35, SD = 1.20) on the
cognitive engagement scale and the participants who were exposed to the neutral face scored (M = 2.16, SD = 0.87). Hypothesis 1b is hereby rejected. A joyful facial emotional expression
does not have an effect on consumers’ cognitive brand engagement compared to a neutral facial emotional expression.
For the final part of hypothesis 1, we take a look at hypothesis 1c, namely the positive effect of facial emotional expression on behavioral engagement. No significant effect was found F(1, 222) = 2.86, p = 0.092. Participants who were exposed to the joyful face scored (M = 2.57, SD = 1.32) on the behavioral engagement scale and the participants who were exposed to the neutral face scored (M = 2.30, SD = 0.98). A joyful facial emotional expression does not have an effect on consumers’ behavioral brand engagement more positively compared to a neutral facial emotional expression and hypothesis 1c is hereby rejected.
Hypothesis 2a assumes that the caption emotional expression has a positive effect on affective online CBE. This effect was not found significant F(1, 222) = 1.79, p = 0.182. Participants exposed to the positive caption scored (M = 3.60, SD = 0.85) on affective engagement and participants exposed to the neutral caption scored (M = 3.44, SD = 0.88). This means that it cannot be established that a joyful caption emotional expression leads to more affective engagement than an Instagram post with a neutral caption emotional
expression. This assumes hypothesis 2a is rejected.
Hypothesis 2b assumes that the caption emotional expression has a positive effect on cognitive online CBE. This result is not significant; F(1, 222) = 1.01, p = 0.317.
Participants exposed to the positive caption have scored (M = 2.33, SD = 1.08) on cognitive engagement compared to participants exposed to the neutral caption (M = 2.19, SD = 1.04). Therefore, a joyful caption emotional expression does not have an effect on consumers’ brand cognitive engagement more positively compared to a neutral caption emotional expression, by which hypothesis 2b is rejected.
With hypothesis 2c, the positive effect of caption emotional expression on
behavioral engagement is assumed. No significant effect has been found: F(1, 222) = 0.53, p = 0.468. Participants exposed to the positive caption have scored (M = 2.50, SD = 1.19) on cognitive engagement and participants exposed to the neutral caption (M = 2.38, SD = 1.17). This means that it cannot be established that a joyful caption emotional expression leads to more behavioral engagement than an Instagram post with a neutral caption emotional expression. Therefore, hypothesis 2c is rejected.
Hypothesis 3a assumes that the caption emotional expression has a moderating effect on the effect of facial emotional expression on affective CBE. The Univariate analysis shows no significant interaction effect F(1, 222) = 0.66, p = 0.417. The results show that the condition with a joyful face, joyful caption yields affective engagement of (M = 3.85, SD = 0.78) compared to the condition neutral face, neutral caption (M = 3.26, SD = 0.72). The condition with a joyful face, neutral caption scored (M = 3.61, SD = 0.98) on affective engagement. The two independent variables, facial and caption emotional expression, therefore, in combination with each other, have no influence on the dependent variable affective CBE. This does not assume hypothesis 3a.
Hypothesis 3b assumes that the caption emotional expression has a moderating effect on the effect of facial emotional expression on cognitive CBE. Cognitive engagement has been shown in the condition with a joyful face, joyful caption (M = 2.38, SD = 1.18)
compared to the condition neutral face, neutral caption (M = 2.05, SD = 0.77). The condition with a joyful face, neutral caption scored (M = 2.32, SD = 1.24) on cognitive engagement. No significant interaction effect was found F(1, 222) = 0.27, p = 0.603. The two independent variables, facial and caption emotional expression, therefore, in combination with each other, have no influence on the dependent variable cognitive CBE. Hereby, hypothesis 3b is
Finally, Hypothesis 3c assumes that the caption emotional expression has a
moderating effect on the effect of facial emotional expression on behavioral CBE. Also, no significant interaction effect was found F(1, 222) = 0.07, p = 0.795. Behavioral engagement has been shown in the condition with a joyful face, joyful caption (M = 2.65, SD = 1.35) compared to the condition neutral face, neutral caption (M = 2.27, SD = 1.01). The condition with a joyful face, neutral caption scored (M = 2.49, SD = 1.30) on behavioral engagement. The two independent variables, facial and caption emotional expression, therefore, in combination with each other, have no influence on the dependent variable behavioral CBE. This assumes hypothesis 3c is rejected.
Ad hoc analyses, including covariate and specific participants
As been shown in the randomization check, one covariate was found. Therefore, besides doing clean analysis with only the dependent (consumer brand engagement) and independent variables (emotion and caption), these two-way variance analysis (Univariate) were executed including the covariate gender. This has been done in order to see if gender has an impact on the result of the hypotheses. The results are summarized in Table 3. The
analyses indicate that hypothesis 1a, the effect of facial emotional expression on affective engagement, is still accepted. (p<0.05, η² = 0.09). The rest of the hypotheses are still rejected with the covariates included. Therefore, the results of the hypotheses are still the same.
Finally, we also tested whether the results of the hypotheses would differ when including the covariate gender and when we only select the participants who passed the manipulation checks. The results are summarized in Table 4. When including only participants who passed the manipulation checks, the sample size became much smaller, namely from N = 226 to N = 102. Hypotheses 1a is still accepted in this analysis (p<0.05, η² = 0.16) and therefor this hypothesis is very constant. An interesting finding, now that
participants who did not succeed the manipulation checks are excluded, is that facial
emotional expression now has a significant effect on behavioral CBE and almost a significant effect on cognitive CBE. The rest of the hypotheses are still rejected. Therefore, not all results of the hypotheses are the same anymore.
Table 3. Test of Between-Subjects Effects of emotion and caption on affective (a), cognitive (b) and behavioral intention (c) consumer brand engagement. Including the covariates gender, attitude and Instagram use.
G e n d e r F v a l u e p v a l u e 0 . 7 6 0 . 3 8 3 1 . 2 9 0 . 2 5 8 0 . 1 9 0 . 6 6 7 F a c i a l F v a l u e 1 6 . 2 8 < 0 2 . 3 0 0 . 1 2 . 5 2 0 . 1 Affective, Cognitive and Behavioral intention consumer brand engagement
e m o t i o n a l e x p r e s s i o n p v a l u e . 0 5 3 1 1 4 C a p t i o n e m o t i o n F v a l u e p v a l u e 1 . 6 6 0 . 1 9 9 0 . 8 8 0 . 3 4 8 0 . 5 6 0 . 4 5 5
a l e x p r e s s i o n E m o t i o n * C a p t i o n e m o t i F v a l u e p v a l u e 0 . 8 3 0 . 3 6 4 0 . 1 5 0 . 7 0 0 0 . 0 4 0 . 8 3 6
o n a l e x p r e s s i o n
Table 4. Test of Between-Subjects Effects of emotion and caption on affective (a), cognitive (b) and behavioral intention (c) consumer brand engagement. Including the covariates gender, attitude and Instagram use. Only included with participants who passed the manipulation check. G e n d e r F v a l u e 0 . 1 3 0 . 7 0 . 7 1 0 . 0 . 1 3 0 . 7 Affective, Cognitive and Behavioral intention consumer brand engagement
p v a l u e 2 1 4 0 1 1 7 F a c i a l e m o t i o n a l e x p r e s s i o n F v a l u e p v a l u e 1 9 . 9 4 < 0 . 0 5 3 . 8 0 0 . 0 5 4 4 . 3 1 < 0 . 0 5 C F 0 0 0
a p t i o n e m o t i o n a l e x p r e s s i o n v a l u e p v a l u e . 1 6 0 . 6 9 0 . 4 2 0 . 5 2 0 . 5 3 0 . 4 6 8 E m o t i o n F v a l u e p 0 . 1 1 0 . 7 3 0 . 6 7 0 . 4 1 0 . 2 1 0 . 6 4
* C a p t i o n e m o t i o n a l e x p r e s s i o n v a l u e 8 4 5 Discussion Interpretation of results
This study researched whether facial emotion expressed on a brand’s Instagram post has an effect on consumer brand engagement (CBE) and what role the caption plays in this. With the widespread use of image sharing social media platforms like Instagram, most of which are used by brands, a key challenge in researching brands is to understand the role of the image and textual content in online consumer brand engagement. Facial emotional expressions are shown to be powerful visual tools and a caption can be used for textual content in order to transfer positive content, which can increase CBE. Therefore, the following research question was answered: To which extent do photos, made by or for an organization posted on their Instagram, with a facial emotional expression influence the online consumer brand engagement? And what effect does the caption emotional expression in the Instagram post of the brand has as a moderator? To answer the research question, an experiment was conducted by means of an online survey in which participants were randomly assigned to one of four conditions.
The first hypotheses in this study looked at the main effect of facial emotional expression on affective (H1a), cognitive (H1b) and behavioural intention (H1c) CBE. The results showed that the effect was only valid for affective engagement. Remarkably, when the covariate gender and only participants who passed the manipulation checks were included in the ad hoc analyses of H1a, H1b and H1c, also a significant effect was found for behavioural CBE and almost a significant effect for cognitive CBE (p = 0.054). The second hypotheses looked at whether there was a main effect of the caption emotional expression on affective (H2a), cognitive (H2b) and behavioural (H2c) CBE. With these hypotheses, no significant effects have been found. Also, no interaction effect has been found, in which the caption emotional expression would have a moderating effect on CBE.
These findings are an addition to the existing knowledge about CBE, which has increased the field of knowledge about this. Namely, because there is an effect of facial
emotional expression on affective and behavioral CBE which assumes it is better to use a joyful facial expression of emotion.
With regards to social relevance, it can be stated that the results can contribute to organizations, brands and managers in implementing their social media or Instagram strategy. Customer brand engagement is very important for the success of any brand. Without active consumers there will not be much to read or lurk about. Therefore, brands need to encourage customer engagement in such a way that it not solely leads to more commenting and liking, but also to purchase behavior. Giving the massive scale of social networks such as Instagram, even small effects can have large aggregated consequences.
Explaining the significant effect and mostly non-significant effects, there are multiple angles of elucidation. First of all, the effect of a joyful facial or caption emotion expressed on CBE may depend on the participant’s characteristics (de Vries, Möller, Wieringa, Eigenraam & Hamelink, 2018). This affects the way people process messages and, as a result, in how they affect the viewer (Valkenburg & Peter, 2013). Therefore, participants may have
responded differently with the joyful expression on affective, cognitive and behavioral CBE. Another explanation could be the way people make decisions or take action. People can make a decision to do something, for example take action in consumer brand engagement, through reason or emotion (Kühne, 2012). If the participants wanted to behave in consumer brand engagement based on cognitive CBE but did not see the joyful expression as a reason, it makes more sense that the participants also did not engaged on a behavioral level.
Also, faces have an emotional effect (Shiota, Campos & Keltner, 2003). Emotions can cause emotional priming, where joy creates positive thinking. In addition, positive moods least to relaxation and relaxation increases saving affective information (Han, Lerner & Keltner, 2007). This means that more emotion-specific information is being processed when
feeling happy. Maybe this could explain why the effect of facial emotional expression was only found consistently significant on affective CBE.
Explanation of the non-significant effects might have to do with the dimensional and discrete emotion perspective (Nabi, 2010). The dimensional approach to emotion refers more to how people experience their emotions than how they think about their emotions. So, when participants may use this approach, they felt affected by the joyful emotion but would not cognitively engage with it. The discrete emotion theory claims that there are a small number of core emotions, including joy. These core emotions are biologically determined emotional responses whose expression and recognition are fundamentally. Maybe participants did not respond on CBE with the specific emotion joy. Therefore, for further research, the effect on CBE could be carried out with different emotions which could indicate different effects.
Also, as been said before in this study, textual content ensures less engagement than a picture or image. This could explain the fact that the effect on affective and behavioral
consumer brand engagement has been found for joyful facial emotional expression and not for the joyful caption emotional expression.
Limitations
In addition to the explanation of the non-significant effects, there are some limitation to the research. For example, the sample in this study is not a representative reflection of Instagram users. The sample came from the network of the researcher and mainly consisted of younger female participants. Besides, there were significant differences between men and women in the conditions. Follow-up research should strive to draw a representative sample of Instagram users in order to increase external validity in this way.
Also, a situation was sketched for the participants that did not really take place in their own timeline of Instagram. Because the answers may not be directly related to their own time
and place, it will be more difficult to generalize the results. To improve this, follow-up research could run the Instagram post through an actual timeline of Instagram within a lab or within the Instagram of the participants themselves.
Even though the two manipulation checks were succeeded, different results were found after only including participants who passed the two manipulation checks. This could be, because the stimuli differ regarding the direction of the testimonials’ gaze. The look of the condition with the neutral facial emotional expression was straight into the camera, while in the condition with the joyful facial emotional expression the model looked away. This may affect the results because not only was the facial emotional expression manipulated, also the gaze or position of the eyes were different. Further research should strive to create the same testimonials’ gaze.
Finally, the Leven’s Tests of cognitive and behavioral CBE were significant, causing violated homoscedasticity. Therefore, some analysis results should be read with care. These limitations require more follow-up research before a clear and stronger conclusion can be drawn. But with the results found from this research, the field of knowledge can certainly be further expanded.
Further research
Besides the considerable suggestions on future work in the limitation section, they can also look at the testimonials’ gaze. In the condition with the neutral facial emotional
expression, the model looks straight into the camera, whereas in the condition with the joyful facial emotional expression, the model looks away. Maybe this is not the problem of the research design but has a more theoretical reason. Hartmann and Goldhoorn (2011) showed that looking at someone through a screen that looks straight in the camera can raise parasocial interaction, which indicates more engagement because of eye contact. Future work could look
at the differences between direct gaze and not direct gaze on CBE or keep the gaze the same when investigating effects on CBE.
Further research can look at effects of similar factors on other photo sharing communities such as Pinterest with biased gender demographics. Future work can look at other visual characteristics of social media platforms and study their impact on online CBE. Other signals can be gathered from people in photos, including gaze direction, as well as, body posture and movement. Besides, the findings are also limited to one brand community. Therefore, the findings have implications for further research on CBE with different brands and organization genres.
Our quantitative results illuminate what the engagement towards the image and textual expression with emotion is, but not why users behave this way or what kind of connections they make with such content. Additional work, particularly using qualitative methods, is needed to answer these questions.
Finally, future work can also investigate the relationship between emotion perception theories and other aspects of online engagement. For example, are emotions effective when it comes to spreading the content on the social network? Are photos or topics, accompanied with emotions more/less persuasive in terms of delivering the content of a brand?
In this paper, we took a first step toward uncovering an important feature of some of images and caption, the joyful emotion. In addition to speaking to the ongoing studies in online consumer brand engagement, our findings open a new thread of future work, suggesting research in visual analysis.
References
Ashley, C., & Tuten, T. (2015). Creative strategies in social media marketing: An exploratory study of branded social content and consumer engagement. Psychology & Marketing, 32(1), 15-27.
Bakhshi, S., Shamma, D. A., & Gilbert, E. (2014, April). Faces engage us: Photos with faces attract more likes and comments on instagram. In Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems (pp. 965-974). ACM. Berger, J., and Milkman, K. L. What makes online content viral? Journal of marketing
research 49, 2 (2012), 192–205.
Boeije, H., ’t Hart, H., & Hox, J. (2009). Onderzoeksmethoden. Amsterdam: Boom.
Burke, M., and Kraut, R. Mind your ps and qs: the impact of politeness and rudeness in online communities. In Proceedings of the 2008 ACM
Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and evaluative space: A critical review, with emphasis on the separability of positive and negative substrates. Psychological bulletin, 115(3), 401.
Dessart, L., Veloutsou, C., & Morgan-Thomas, A. (2015). Consumer engagement in online brand communities: a social media perspective. Journal of Product & Brand
de Vries, D. A., Möller, A. M., Wieringa, M. S., Eigenraam, A. W., & Hamelink, K. (2018).
Social comparison as the thief of joy: emotional consequences of viewing strangers’ Instagram posts. Media Psychology, 21(2), 222-245.
Dobele, A., Lindgreen, A., Beverland, M., Vanhamme, J. and van Wijk, R. (2007), “Why pass on viral messages? Because they connect emotionally”, Business Horizons, Vol. 50, pp. 291-304.
Erkan, I. (2015). Electronic word of mouth on Instagram: customers’ engagements with brands in different sectors. International Journal of Management, Accounting and Economics, 2(12), 1435-1444.
Frijda, N. H. (2016). The evolutionary emergence of what we call “emotions”. Cognition and Emotion, 30(4), 609-620.
Griskevicius, V., Shiota, M. N., & Neufeld, S. L. (2010). Influence of different positive emotions on persuasion processing: A functional evolutionary
approach. Emotion, 10(2), 190.
Gummerus, J., Liljander, V., Weman, E., & Pihlström, M. (2012). Customer engagement in a Facebook brand community. Management Research Review, 35(9), 857-877.
Han, S., Lerner, J.S., & Keltner, D. (2007). Feelings and consumer decision making: The appraisal-tendency framework. Journal of consumer psychology, 17(3), 158-168. Hartmann, T., & Goldhoorn, C. (2011). Horton and Wohl revisited: Exploring viewers'
experience of parasocial interaction. Journal of communication, 61(6), 1104-1121. Heath, R., Brandt, D., & Nairn, A. (2006). Brand relationships: Strengthened by emotion,
weakened by attention. Journal of advertising research, 46(4), 410-419. Hoffman, D. L., & Fodor, M. (2010). Can you measure the ROI of your social media
Hollebeek, L. D., Glynn, M. S., & Brodie, R. J. (2014). Consumer brand engagement in social media: Conceptualization, scale development and validation. Journal of interactive marketing, 28(2), 149-165.
Jamali, S., and Rangwala, H. Digging digg: comment mining, popularity prediction, and social network analysis. In International Conference on Web Information Systems and Mining, 2009. WISM 2009, ieee (2009), 32–38.
Kleinginna, P. R., & Kleinginna, A. M. (1981). A categorized list of emotion definitions, with suggestions for a consensual definition. Motivation and emotion, 5(4), 345-379. Kramer, A. D., Guillory, J. E., & Hancock, J. T. (2014). Experimental evidence of
massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences, 111(24), 8788-8790
Kühne, R. (2012). Media-induced affects and opinion formation: How related and unrelated affects influence political opinions. Living Reviews in Democracy, 3.
Luarn, P., Lin, Y. F., & Chiu, Y. P. (2015). Influence of Facebook brand-page posts on online engagement. Online Information Review, 39(4), 505-519.
Millen, D. R., and Patterson, J. F. Stimulating social engagement in a community network. In Proceedings of the 2002 ACM conference on Computer supported cooperative work, ACM (2002), 306–313.
Nabi, R. L. (2010). The case for emphasizing discrete emotions in communication research. Communication Monographs, 77(2), 153–159.
https://doi.org/10.1080/03637751003790444
Schere, K.R. (1982). Emotion as a process: Function, origin and regulation.
Shiota, M. N., Campos, B., & Keltner, D. (2003). The faces of positive emotions. Annals of
Sherman, L. E., Greenfield, P. M., Hernandez, L. M., & Dapretto, M. (2018). Peer influence via instagram: effects on brain and behavior in adolescence and young
adulthood. Child development, 89(1), 37-47.
Smith, L. R., & Sanderson, J. (2015). I'm going to Instagram it! An analysis of athlete self-presentation on Instagram. Journal of Broadcasting & Electronic Media, 59(2), 342-358.
Teixeira, T., Wedel, M., & Pieters, R. (2012). Emotion-induced engagement in internet video advertisements. Journal of Marketing Research, 49(2), 144-159.
Valkenburg, P. M., & Peter, J. (2013). The differential susceptibility to media effects model. Journal of Communication, 63(2), 221-243.
Yuki, T. (2015). What makes brands’ social content shareable on Facebook? Journal of Advertising Research, 55(4), 458-470.
Appendix
Appendix 1 – The experimental stimuli
Affective engagement
Item 1: I feel enthusiastic about the brand Item 2: The brand makes me enthusiastic Item 3: I am interested in this brand Item 4: I find this brand not interesting Item 5: I enjoy interacting with the brand
Item 6: When interacting with the brand, I feel happy Answer options:
Strongly disagree 1 2 3 4 5 6 7 Strongly agree
Cognitive engagement
Item 1: I pay a lot of attention to the brand
Item 2: I spend a lot of time thinking about the brand Item 3: I make time to think about the brand
Item 4: When I interact with this brand, I forget everything else around me Item 5: When I am interacting with this brand, I get carried away
Item 6: When interacting with the brand, it is difficult to detach myself Answer options:
Strongly disagree 1 2 3 4 5 6 7 Strongly agree
Item 1: I would show support to what the brand says or does (by liking, commenting or sharing the Instagram post)
Item 2: I would share content from the brand to my wider network Item 3: I would promote the brand
Item 4: I would try to get others interested in the brand Item 5: I would actively defend the brand from critics
Item 6: I would say positive things about the brand to other people Answer options:
