Master Thesis
MSc Marketing Management
Convinced by #positiveemotions or #strongarguments?
How self-monitoring influences the decision to engage
and to purchase products advertised on Instagram
Author: Eva Charlotte Johannsen Student number: S3769577
Abstract
Table of Contents
Abstract ... II List of figures ... IV List of tables ... V
Introduction ... 1
Theory and hypotheses ... 4
2.1 Conceptual model ... 4 2.2 Advertising appeal ... 5 2.3 Engagement ... 5 2.4 Self-monitoring ... 7 Method ... 10 3.1 Research design ... 10 3.2 Procedure ... 13 3.3 Construct measurement ... 14
3.3.1 Factor and reliability analysis ... 18
3.3.2 Additional analysis for self-monitoring ... 19
3.4 Data analysis ... 20
Results ... 23
4.1 Correlation analysis ... 23
4.2 Engagement and purchase intention ... 24
4.3 Appeal and engagement ... 24
4.4 The mediating effect of engagement ... 25
4.5 The moderating effect of self-monitoring ... 26
4.6 Moderated moderation ... 27
4.7 Moderated moderated mediation ... 28
Conclusion ... 30
5.1 Discussion ... 30
5.2 Academic implications ... 32
5.3 Managerial implications ... 33
5.4 Limitations and recommendations for further research ... 34
List of figures
List of tables
Table 3.1 Instagram posts ... 10
Table 3.2 Manipulation scales ... 12
Table 3.3 Measurement scales ... 15
Table 4.1 Correlation table ... 23
Table 4.2 Engagement and purchase intentions ... 24
Table 4.3 Appeal and engagement ... 25
Table 4.4 Mediation results ... 25
Table 4.5 Moderation of self-monitoring ... 27
Table 4.6 Moderation of number of likes and comments ... 28
Introduction
45% of the global population uses social media networks and this number is growing continuously (Kemp, 2019). Following Facebook and YouTube, Instagram ranks as the third largest social network in terms of active user accounts (Kemp, 2019). Instagram is of crucial relevance for marketers, as it is the fastest growing social network with a quarter on quarter growth of +4.4% in January 2019 compared to only +1.7% growth of Facebook. In total, there are 894.9 million people using Instagram in 2019 (Kemp, 2019). It is very valuable for businesses to have an Instagram profile, where they post content about their products and advertise them in this way, as 80% of users follow at least one business on Instagram and users are very likely to purchase a product with their phone (Clarke, 2019). In addition, Instagram introduced a new function, the “checkout feature”, on the 19th of March 2019 for business profiles (Instagram-Press, 2019). This feature enables users to purchase products directly on Instagram with the “checkout on Instagram” button and thus, makes shopping through Instagram very fast and user-friendly (Instagram-Press, 2019). Thus, increasing purchase intentions is very relevant. On Instagram, engagement is expressed by the number of likes, comments and shares in response to a brands’ post (Hoffman & Fodor, 2010; Silva, Farias, Grigg, & Barbosa, 2019). The number of likes and comments are displayed underneath a post. However, recently Instagram has begun to hide the number of likes under a picture. The test started in autumn 2019 and since then, some users can still like a post, but do not see how many other people clicked like (Paul, 2019).
In the past years, the concept of engagement has gotten a lot of attention in the academic literature, as several studies have concluded that engagement is an important antecedent for purchase intentions (Hutter, Hautz, Dennhardt, & Füller, 2013; Kilger & Romer, 2007; Valentini, Romenti, Murtarelli, & Pizzetti, 2018). Thus, understanding how to increase engagement is crucial for marketeers. Scholars, that focused on how to increase engagement found, that positive emotions influence engagement positively (Eckler and Bolls, 2011; Berger and Milkman, 2012; Nikolinakou and King, 2018). Following these previous findings, this study presumes engagement as an underlying mechanism between advertising appeal and purchase intentions.
their expressive behaviour to make it appropriate to the situation they are in. Cues and guidelines for their self-presentation are taken from the behaviour of other comparison persons in the same social situation. Low self-monitors do not have such a strong concern for appropriateness of their self-presentation. Those people express their feelings, as they feel them and do not monitor them to fit in (DeBono, 2006; DeBono & Packer, 1991; Snyder, 1974; Snyder & DeBono, 1985). In consumer psychology, emotional and rational advertising are the two most common types of advertising appeals (Fennis & Stroebe, 2010). Self-monitoring is an important personality trait in explaining whether an emotional or rational advertising appeal leads to stronger purchase intentions. High self-monitors are more likely to react positively to emotional advertising displaying a certain image. This makes them believe, that by purchasing that product they will fit in better in certain social situations. Contrary to that, low self-monitors react more positively to rational advertising focusing on quality advantages, which align with their underlying values and attitudes (Snyder and DeBono, 1985; DeBono and Packer, 1991). This study aims to understand, whether those findings also hold in the context of engagement on Instagram. Thus, it is expected that high self-monitors are more likely to engage in response to emotional advertising and low self-monitors are more likely to engage in response to rational advertising on Instagram.
In addition, as high self-monitors use cues in their environment to monitor their expressive behaviour and low self-monitors do not (Ickes, Holloway, Stinson, & Hoodenpyle, 2006; Shaffer, Smith, & Tomarelli, 1982; Snyder, 1974), it can be assumed that a high number of likes and comments under a post, strengthens the intention of people scoring high on self-monitoring to engage. Previous scholars have already shown that a high number of likes influences people to like that post as well and explained this phenomenon with conformity motivation (Chin, Lu, & Wu, 2015; Egebark & Ekström, 2012; Lascu, Bearden, & Rose, 1995). This research will enhance the findings on conformity motivation of previous scholars by investigating, whether the personality type of self-monitoring strengthens the effect of conformity motivation.
underlying mechanism of advertising appeal and purchase intention and answers the following research question:
Does engagement on a brands’ instagram post, with either a positive emotional or a rational appeal, depend on people’s degree of self-monitoring and to what extent is this effect influenced by the number of likes and comments?
This is the first study to introduce the personality trait of self-monitoring in the context of engagement on social media. Thereby, it contributes to the academic literature an answer to the underlying motives of engagement. This is done by conducting a field experiment. Beyond their theoretical implications, the findings will give important insights for marketeers into the underlying mechanisms influencing the effectiveness of Instagram advertising.
Theory and hypotheses
2.1 Conceptual model
Figure 2.1 shows the conceptual model of this study. As it has been shown, engagement is a crucial mechanism leading to purchase intentions on Instagram (Hutter et al., 2013; Kilger & Romer, 2007; Valentini et al., 2018) and positive emotions lead to engagement (Berger & Milkman, 2012; Eckler & Bolls, 2011; Harvard Business, 2015; Nikolinakou & King, 2018). Thus, this study assumes that positive emotions lead to stronger purchase intentions, but only when engagement is present.
Self-monitoring is of crucial importance in this study, as it is assumed that this personality trait is a decisive criterion of whether people engage or not. This study posits that high self-monitors are more likely to engage in response to positive emotional advertising and that low self-monitors are more likely to engage in response to rational advertising (DeBono & Packer, 1991; Hadjimarcou, 2012; Johar & Joseph Sirgy, 1991; Kotler & Keller, 2009; Snyder & DeBono, 1985; Stafford & Day, 1995). Further, it is expected that a high number of likes and comments under a brands’ Instagram post influences high self-monitors to engage as well (Ickes et al., 2006; Shaffer et al., 1982; Snyder, 1974). This is due to the phenomenon of conformity, which is an intrinsic motivation to follow the behaviour of a group of people (Egebark & Ekström, 2012; Lascu et al., 1995). As a personality trait of high self-monitors is to take the behaviour of others as a guide for their own behaviour (Ickes et al., 2006; Shaffer et al., 1982; Snyder, 1974), it can be expected that they are very likely to conform to a high number of likes and comments and engage as well. As this personality trait is not typical for low self-monitors (Ickes et al., 2006; Shaffer et al., 1982; Snyder, 1974), they are not expected to be influenced by a high number of likes and comments.
Figure 2.1 Conceptual model
The control variables age, gender, Instagram use, perceived trust, attitude toward advertising and product involvement are included.
2.2 Advertising appeal
In consumer psychology two types of advertising appeals can be differentiated (Fennis & Stroebe, 2010). The soft sell approach (DeBono & Packer, 1991) focuses on the experiential facets of consumption and is based on affect and emotions such as adventure, happiness, romance or status (Cutler & Javalgi, 1993). This appeal links the advertised product to desirable symbols and images and by doing so, depicts the experience a consumer will get, or what kind of person a consumer will be when using the brand (Hadjimarcou, 2012; Kotler & Keller, 2009). In the following, this will be referred to as an emotional appeal. Contrary to that, the hard sell approach (DeBono & Packer, 1991) is a rational, straightforward presentation of factual information about a product (Stafford & Day, 1995). The focus is on the utilitarian benefits of a product and the superior quality, performance or reliability are communicated with this advertising appeal (Johar & Joseph Sirgy, 1991). Thus, this appeal aims to convince consumers with reasons and arguments (Albers-Miller & Stafford, 1999). In the following, this will be referred to as a rational appeal.
2.3 Engagement
The concept of engagement has been studied for more than two decades in various academic fields, such as information systems (O’Brien & Toms, 2010), education (Kahu, 2013), educational psychology (Fredricks, Blumenfeld, & Paris, 2004), management (Gruman & Saks, 2011), or marketing (e.g. Kumar and Reinartz, 2016; Pansari and Kumar, 2017; Valentini et al., 2018; Mirbagheri & Najmi, 2019). The definitions of engagement differ within these contexts as the
Advertising Appeal (emotional vs. rational) Purchase Intention Self-monitoring (High vs. low) H4 H2 H1 Engagement
(Likes, comments, shares)
(H3)
H5 # of likes and comments
conceptualization of engagement depends on the engagement object and the context of the study (Mirbagheri & Najmi, 2019). The context of this study is Instagram and the engagement object of this study is a brand’s post on Instagram. Therefore, the activity of liking, commenting and sharing define engagement, as they are valuable indicators for engagement in that context (Hoffman & Fodor, 2010; Valentini et al., 2018).
Ample research has shown that engagement influences purchase decisions. For instance, Kilger and Romer (2007) showed that engagement in media channels such as televisions, magazines, and the internet increases the likelihood of purchase. Furthermore, Hutter et al. (2013) chose the social network Facebook to show that user engagement on the Facebook page of the brand MINI leads to increased purchase intentions of that car and Valentini et al. (2018) research showed, that digital visual engagement on Instagram influences purchase intentions. Therefore, it can be hypothesised:
H1: Engagement positively influences purchase intentions.
H2: A positive emotional advertising appeal leads to stronger engagement than a rational advertising appeal.
As described before, ample research has found that engagement leads to purchase intentions (Hutter et al., 2013; Kilger & Romer, 2007; Valentini et al., 2018). Furthermore, scholars found that positive emotions lead to engagement (Eckler & Bolls, 2011; Nikolinakou & King, 2018). Consequently, positive emotions lead to engagement and engagement leads to purchase intentions. Thus, it can be assumed, that positive emotional content leads to purchase intentions, but this effect is only due to engagement. This suggests that engagement is an underlying mechanism between advertising appeal and purchase intentions. Therefore, it can be hypothesised:
H3: Engagement mediates the relationship between a positive emotional advertising appeal and purchase intentions.
2.4 Self-monitoring
Self-monitoring describes how far individuals monitor their self-presentation and expressive behaviour of affective experiences (Snyder, 1974). High self-monitors are very sensitive about their environment and thus, aim to monitor their expressive behaviour to make it appropriate to the situation they are in. Cues and guidelines for their self-presentation are taken from the behaviour of other comparison persons in the same social situation. Low self-monitors do not have such a strong concern for appropriateness of their self-presentation. Those individuals express their feelings, as they feel them and do not monitor them to fit into a social situation (DeBono, 2006; DeBono & Packer, 1991; Snyder, 1974; Snyder & DeBono, 1985).
more relevant for high self-monitors (DeBono & Packer, 1991; Snyder & DeBono, 1985).
Contrary to that, low self-monitors are not concerned with the image they project in a social situation, but want to reflect their inner values and attitudes (Snyder & DeBono, 1985). This explains that low self-monitors react more favourably to advertising, that gives rational information about the quality of a product and evaluate products advertised with a focus on quality as more self-relevant and higher in quality (DeBono & Packer, 1991). Therefore, low self-monitors can be convinced with a rational approach rather than an emotional approach (DeBono & Packer, 1991; Johar & Joseph Sirgy, 1991; Snyder & DeBono, 1985; Stafford & Day, 1995).
Translating these findings (DeBono & Packer, 1991; Snyder & DeBono, 1985) to the context of Instagram advertising, it can be assumed that high self-monitors express their favourable evaluations of emotional advertising appeals with engagement. This can be explained by the pursuit of high self-monitors to always project the ideal self-image in a certain social situation. By liking, commenting or sharing the post, high self-monitors can project an ideal self-image to other people on Instagram. Contrary to that, it can be assumed that low self-monitors are more likely to engage on a post advertised rationally with a strong focus on the quality of the product due to the characteristic of low self-monitors to focus on the quality and rational information of an advertised product.
H4: High self-monitoring strengthens the relationship between a positive emotional advertising appeal and engagement, whereas low self-monitoring strengthens the relationship between a rational advertising appeal and engagement.
In order to monitor their self-presentations to fit into the social situations they encounter, high self-monitors use other peoples’ behaviour as a guide for their own behaviour (Ickes et al., 2006; Shaffer et al., 1982; Snyder, 1974). This relates to conformity, which is an intrinsic motivation to follow others. People are likely to change their mind under social influence and align it with the most popular opinion in a group (Chin et al., 2015; Lascu et al., 1995). Scholars have shown, that people display conformity behaviour online. For example, people rather download a file, that has a high number of downloads than a very similar file, with a small number of downloads (Hanson & Putler, 1996). And in the context of Facebook it has been shown, that people are more likely to “like” a post when many others have liked it as well (Egebark & Ekström, 2012).
likes and comments and reciprocate other people’s behaviour by engaging on that post as well, as they assume that this is the appropriate behaviour in that situation. Consequently, it is hypothesised, that a high number of likes and comments under a post will influence high self-monitors to engage on that post as well, but low self-monitors will not be affected by that.
Method
3.1 Research design
This study uses a 2 (emotional, rational) x2 (high number of likes and comments, low number of likes and comments) between-subject experimental design is used. It was decided on a fictitious brand to avoid confounding effects of an actual brand. Sunscreen is used as it is interesting to all genders and allows for an emotional and rational appeal. The posts, which are displayed in table 3.1, differentiate from each other in the claim, description and number of likes and comments. Post 1 and post 2 display the emotional version, with a focus on the image and emotions the product reveals, whereas post 3 and post 4 show the rational version, which focuses on the functional quality of the product.
In addition, the number of likes and comments are manipulated. Post 1 and post 3 display a small number of likes and comments, whereas post 2 and post 4 display a high number of likes and comments.
Table 3.1 Instagram posts
Instagram post
Claim & Description
Post 1• Emotional appeal • 6 likes
• 2 comments
Claim:
Enjoy your summer! Description:
Post 2
• Emotional appeal • 38.691 likes • 322 comments
Claim:
Enjoy your summer! Description:
Sunsets, sandy hair and the sound of the ocean – we all love summer! Our SUNscreen protects your skin so that you can get a beautiful tan. Post 3 • Rational appeal • 6 likes • 2 comments Claim: Innovative UV protection technology! Description:
Our SUNscreen has an innovative SolarSmart technology, that stabilizes high-level protection against the aging and burning effects of UVA and UVB rays. It is up to 4h water
Post 4 • Rational appeal • 38.691 likes • 322 comments Claim: Innovative UV protection technology! Description:
Our SUNscreen has an innovative SolarSmart technology, that stabilizes high-level protection against the aging and burning effects of UVA and UVB rays. It is up to 4h water resistant.
The posts in table 3.1 are the final version, which are used in the main study. Prior to that, a pre-test was conducted, which was significant for the number of likes and comments, but not for the advertising appeal. In the pre-test the claim “Enjoy your summer!” was used for all four posts and the scale for the manipulation of the advertising appeal was a different one. It was decided to make the difference between the emotional posts and rational posts clearer with a different claim for the rational posts (see table 3.1) and to use a 7-point scale with three items (logical/emotional; objective/subjective; and factual/nonfactual) for the advertising appeal in the main study (Andreu, Casado-Díaz, & Mattila, 2015). For the number of likes and comments manipulation, a 7-point Likert scale is used (Chin et al., 2015). Table 3.2 shows the scales.
Table 3.2 Manipulation scales
Manipulation
Items
Advertising Appeal Adapted from
Andreu, Casado-Díaz and Mattila (2015)
Do you think the post of SUNscreen is rather…? 1. Logical / emotional
Number of likes and comments Adapted from Chin, Lu and Wu (2015)
1. I think the post has many “likes” and “comments”.
2. I think very few clicked “like” or “comment” under this post.*
3. I think many people clicked “like” or wrote a “comment” on this post.
(7-point likert scale, with 1= I completely disagree; 7= I completely agree)
(*= item is negatively phrased and reversed coded) An independent samples t-test was conducted for the manipulation of the appeal in the main study. Homogeneity of variance is given as the significance of Levene’s test is >0.05. The results show, that the emotional appeal (M=4.5, s.d.=1.4, n=92) is significantly different from the rational appeal (M=3.9, s.d.=1,30, n=91) as p<.05.
The number of likes and comments were also manipulated successfully. The independent samples t-test shows that there is a significant difference between high number of likes and comments (M=4.6, s.d.= 1.6, n=95) and a small number of likes and comments (M=2.4, s.d.=1.5, n=88) with p<.05.
3.2 Procedure
An anonymous online survey is shared through Instagram. As 71% of Instagram users are aged younger than 35 years old (Kemp, 2019), and the survey is shared on the authors’ Instagram account, it is expected that most participants will be younger than 35 as the author has mostly followers similar to her age of 25. As it is crucial that many participants take the survey and complete it, an incentive is offered in the form of 5 x 10€ amazon gift cards and the winners were selected in January.
notice the attention check (Liu & Wronski, 2018). However, easy trap questions are indeed able to detect cases of severe satisficing and it has shown, that in simple decision-making processes, such questions are not threatening scale validity (Anduiza & Galais, 2017; Kung, Kwok, & Brown, 2018). As this study is asking respondents for rather easy, simple answers and is not addressing critical issues which respondents might answer socially desirable, the author decided to include only one trap question in the form of an instructed response item. Following Kung, Kwok and Brown (2018) it is phrased: please select “agree” to demonstrate your attention. It is placed within the question block for self-monitoring (see table 3.3) as this contains the most items.
The questionnaire is built in Qualtrics in the following way. Participants are shown one of the four conditions (see table 3.1) and asked to have a close look at it and remember it. In the following, all questions, which relate directly to the Instagram post are asked for in the following order. Firstly, the manipulation check questions, which are displayed in table 3.2 are shown. In the next step, the questions about intentions to engage and purchase the sunscreen are raised and further, the question about participants’ perceived trust of the Instagram post. In the next step, the self-monitoring scale is presented and after that, the general attitude towards Instagram advertising and participants’ involvement with sunscreen. Finally, participants are asked for their age, gender and Instagram usage. At the very end, participants are thanked for their participation and given the opportunity to enter their email address for the lottery. It is assured that the email addresses will not be related to the previous answers and only used to select winners for the gift cards.
3.3 Construct measurement
Table 3.3 Measurement scales
Construct
Scale
Items
FL
CA
Engagement Adapted from Valentini et al. (2018) 1 = completely disagree; 7 = completely agree
“The Instagram post makes me want to […].” 1. like the post
2. post a comment under the post
3. share the post with my friends or others in my network
.704 .790 .758 .746 Purchase Intention Adapted from Chandran and Morwitz (2005)
7-point scale 1. How likely are you to buy the sunscreen in the Instagram post? (highly unlikely / highly likely)
2. How probable is it that you will purchase the sunscreen on offer? (highly improbable / highly probable)
3. How certain is it that you will purchase the sunscreen? (highly uncertain / highly certain)
4. What chance is there that you will buy the sunscreen? (no chance at all / very good chance)
.771 .854 .846 .834 .902 Perceived trust Adapted from Chen (2008); Kim, Chung and Lee (2011); Escobar-Rodríguez, Grávalos-Gastaminza and Pérez-Calañas (2017) 1 = completely disagree; 7 = completely agree
What do you think about the Instagram post? 1. The Instagram post is trustworthy. 2. The Instagram post is reliable. 3. The Instagram post has integrity.
.896 .834 .853
Self-monitoring Taken from Kim, Seely and Jung (2017) 1= completely disagree; 7 = completely agree
1. I would probably make a good actor.
2. I guess I put on a show to impress or entertain people. 3. In a group of people I am rarely the center of attention.* 4. I find it hard to imitate the behaviour of other people.* 5. I can only argue for ideas which I already believe.*
6. I can make impromptu speeches even on topics about which I have almost no information.
7. I am not particularly good at making other people like me.* 8. I am not always the person I appear to be.
9. I have considered being an entertainer.
10. I feel a bit awkward in company and do not show up quite so well as I should.*
11. I have never been good at games like charades or improvisational acting.*
12. I have trouble changing my behaviour to suit different people and different situations.*
13. I can look anyone in the eye and tell a lie with a straight face (if for a right end).
14. I may deceive people by being friendly when I really dislike them. (*= items are negatively phrased and reversed coded)
.717 .668 .319 .626 .460 .510 .022 .142 .612 .253 .489 .232 .522 .116 .620 Attitude toward Instagram advertising Adapted from Lee, Kim and Ham (2016)
Product Involvement Taken from
Chandrashekaran (2004)
7-point scale 1. In general, I am particularly interested in sunscreen
2. Given my personal interests, sunscreen is very relevant to me 3. Overall, I am not quite involved when I am purchasing sunscreen for
personal use.*
(*= item is negatively phrased and reversed coded)
.834 .808 .710
The concept of self-monitoring, which was developed by Snyder in 1974 is of crucial importance for this study. Besides defining the concept of self-monitoring, he developed the original 25-item scale, which was reduced to 18 items with higher reliability twelve years later (Snyder & Gangestad, 1986). Based on this, Kim, Seely and Jung (2017) used a 14-item scale, which is used to measure self-monitoring in this research.
Instagram features social metrics, which capture engagement of users: Likes, comments, shares and followers (Hoffman & Fodor, 2010; Silva et al., 2019; Valentini et al., 2018). Followers indicate how many users approve the profile of a brand (Virtanen, Björk, & Sjöström, 2017). As the number of followers does not relate to a single post, but rather to the whole profile, they are not of interest for this study. But the number of likes, comments and shares in response to a brands’ Instagram post are of interest, as they are valuable indicators for engagement on a post (Hoffman & Fodor, 2010; Valentini et al., 2018). Likes refer to a form of engagement, which is categorized as consuming as it shows how many users appreciate the post. Comments and shares refer to contributing as users share their opinions and topics for discussion under a post and thus, display a deeper level of engagement (Bakhshi, Shamma, & Gilbert, 2014; Muntinga, Moorman, & Smit, 2011; Silva et al., 2019; Valentini et al., 2018). Consequently, the measurement scale for engagement in this study entails the activity of liking, commenting and sharing and is adapted from Valentini et al. (2018). Purchase intention is measured with a four item scale taken from Chandran and Morwitz (2005).
Several control variables are identified and included to control for confounding effect. Age, gender and Instagram use are measured with pre-specified options. In social media, perceived trust of the Instagram post has shown to influence purchase decisions (Escobar-Rodríguez et al., 2017). Following previous scholars (Chen, 2008; Escobar-Rodríguez et al., 2017; Kim et al., 2011), it is measured in this study with three items including trustworthiness, reliability and integrity of the Instagram post (see table 3.3). Further, attitude towards advertising has shown to influence behaviour in social media (J. Lee et al., 2016; Marchand, 2010). Thus, it is included as a control as well as product involvement as it has shown to impact purchase decisions in online shopping (Hong, 2015; W. I. Lee, Cheng, & Shih, 2017) and in specific, mobile shopping (Drossos & Fouskas, 2010) with a three item scale of Chandrashekaran (2004).
3.3.1 Factor and reliability analysis
self-monitoring scale and the third question of product involvement (see “*” in table 3.3). Next, a factor analysis with all variables is conducted. The factor analysis is appropriate, because the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy is .741, which is above the recommended .5 value (Malhotra, 2009). Further, Bartlett’s test of sphericity is significant at p=.000, so the variables are not uncorrelated. The communalities all have factor loadings above .4. VARIMAX rotation is used to identify the factors. All constructs load on one respective factor except self-monitoring, which loads on different factors and will be discussed in the following chapter. Following factor analysis, the reliability analysis is conducted for all the variables. Cronbach’s alpha (CA) is bigger than .7 for all variables, so it is above the satisfactory 0.6 value (Malhotra, 2009). The exact values are displayed in table 3.3. For every construct the items are averaged to make one score per construct and those new scores are included in further analysis.
3.3.2 Additional analysis for self-monitoring
The self-monitoring scale does not clearly load on one factor. Therefore, several steps are taken to define a good solution and get a better understanding of the data. Firstly, an individual factor analysis is conducted only for the fourteen self-monitoring items with the specification that it should result in only factor. This results in a KMO of .723, Bartlett’s of .000 and Cronbach’s Alpha of .620. Thus, it is appropriate (Malhotra, 2009) Out of all these fourteen items one variable is constructed.
In addition, the factors, which have communalities <.4 are taken out and another variable containing only items 1,2,9 and 13 is constructed. This additional variable will be included in further analysis and when results differ from the first self-monitoring variable, it will be reported.
Furthermore, the data is inspected again. It was one trap question included within the block of the self-monitoring questions. Eleven people failed to answer that question correctly. Those people are removed from the data and factor analysis is run again resulting in a KMO of .724 and Bartlett’s of .000 and a slightly higher Cronbach’s alpha of .643. Again, it is decided to compute an additional variable with these items, which will be tested additionally in further analysis. When the results differ from the first self-monitoring variable, it will be reported.
either a high or low self-monitor can be seen from the data. This finding is underlined by the mean score of self-monitoring, which is 4.10. Four is exactly the middle of the 7-point Likert scale and the standard deviation is only .653, which is small. Thus, it is not surprising that self-monitoring does not clearly load on one factor, as the mean suggests that people are neither high, nor low on self-monitoring, but score in the middle at four with a small standard deviation. Concluding it can be said from the close inspection of the data of self-monitoring, that no distinct self-monitoring scores were captured for individual respondents. To proceed with further analysis, it is decided to focus mainly on the variable that resulted from the first factor analysis and thus, includes all 14 items of the self-monitoring scale as theory states that self-self-monitoring is explained by the variable as a whole (Kim et al., 2017; Snyder & Gangestad, 1986). Four outliers are identified for this variable with a boxplot and removed from the data.
3.4 Data analysis
For the following data analysis, the advertising appeal and number of likes and comments conditions are dummy coded. Next, the control variables are tested for significance with purchase intention as the dependent variable. Thus, for gender, which is also dummy coded, a one-way analysis of variance (ANOVA) is conducted as gender is categorical. This is not significant at p=.116. For the other five control variables a linear regression analysis is conducted. This results in Instagram use and age being not significant and trust, attitude and involvement being significant with a p<.05. Consequently, trust, attitude and involvement are included in further analyses of the hypotheses of the conceptual model.
Linear Regression
Hypothesis 1 is tested with a simple linear regression: H1: PI = α0 + α1Eng + α2Att + α3Inv + α4TR+ ε
if the mediation is significant (Hayes, 2018). Therefore, the confidence interval of the bootstrapped indirect effect is displayed, which should not include 0. Hayes (2018) Model 4 is used for the mediation.
H3: (c) PI = α0 + α1App + α2Att + α3Inv + α4Tr + ε (a) Eng = α0 + α1App + α2Att + α3Inv + α4Tr + ε (b) PI = α0 + α1Eng + α2Att + α3Inv + α4Tr + ε
(c’) PI = α0 + α1App + α2Eng + α3Att + α4Inv + α5Tr + ε Where:
PI = Purchase intention Eng = Engagement App = Appeal
Att = Attitude towards advertising Inv = Involvement
Tr = Trust ε = Error term
ANCOVA
The effect of appeal on engagement described in hypothesis 2, the interaction of the moderation of hypothesis 4 and the effect of number of likes and
comments of hypothesis 5 are tested with an analysis of covariance (ANCOVA) as the independent variable, which is either appeal or the number of likes and comments is categorical. The following models are estimated:
H2: Eng= α0 + α1App + α2Att + α3Inv + α4Tr+ ε
H4: Eng= α0 + α1App + α2SM + α3App*SM + α4Att + α5Inv + α6Tr + ε H5: Eng= α0 + α1LikComm + α2SM + α3LikComm*SM + α4Att + α5Inv + α6Tr + ε
Where:
PI = Purchase intention Eng = Engagement
App = Dummy variable, equal to 0 if appeal is rational and 1 for emotional appeal
LikComm= Dummy variable, equal to 0 if number of likes and comments is low and 1 if it is high
Tr = Trust ε = Error term
Results
183 participants filled in the survey. First, some outliers are identified by looking at the boxplot of the variables and subsequently removed from the data. Next, the data is checked for normal distribution with the Kolmogorov-Smirnov test and the Shapiro-Wilk test. Except self-monitoring, all variables have a p< .05 and are thus, not normally distributed. The histograms underline this finding and thus, the assumption of normality is violated. However, the data are used for further analysis and treated as normally distributed data as the sample is large enough and therefore, significant results can still be obtained (Ghasemi & Zahediasl, 2012). Furthermore, the analyses are checked for multicollinearity with the VIF scores. In the following analyses the VIF scores are all =/< 1.2 and thus, multicollinearity is not an issue (Alin, 2010).
4.1 Correlation analysis
A Pearson correlation analysis is conducted to understand whether there are strong relationships between purchase intentions, self-monitoring and engagement (Malhotra, 2009). The categorical variables advertising appeal and number of likes and comments as well as the control variables are not included. Table 4.1 shows the correlation coefficients, means and standard deviations. In line with hypothesis 1 a correlation of .427 (p<.01) between engagement and purchase is found.
Table 4.1 Correlation table
Mean S.D. Engagement Purchase
Intention Self-Monitoring Engagement 2.087 1.063 1 Purchase Intention 2.799 1.332 .427** 1 Self-Monitoring 4.104 .653 -.073 -.077 1
4.2 Engagement and purchase intention
Hypothesis 1 predicts that engagement positively influences purchase intentions. A linear regression is conducted with purchase intention as a dependent variable and trust, involvement and attitude as control variables. Engagement has a positive influence on purchase intention as the overall model is significant with F=23.276 at p=.000. As the results in table 4.2 indicate, engagement is significant and the adjusted R2 = .331, so 33.1% of the variance in purchase intention can be explained by engagement. The coefficient is .324 so there is a positive significant effect of engagement on purchase intention. Therefore, hypothesis 1 can be confirmed.
Table 4.2 Engagement and purchase intentions
Model 1 Main Variables Engagement Control variables .000a Attitude Involvement Trust .069c .039b .000a R2 (Adjusted R2) (.331) .346 F-value 23.276a
Note: ap-value < .01; bp-value < .05; cp-value < .10
4.3 Appeal and engagement
that appeal does not influence engagement. An R square of .164 and an adjusted R squared of .145 can be reported. Concluding, hypothesis 2 is rejected.
Table 4.3 Appeal and engagement
df SS MS F p ηp2 Appeal 1 .915 .159 .164 .686 .001 Attitude 1 .015 .015 .015 .901 .000 Involvement 1 7.788 7.788 8.060 .005a .004 Trust Error 1 176 26.045 170.059 26.045 .966 26.955 .000a .133
Note: ap-value < .01; bp-value < .05; cp-value < .10
4.4 The mediating effect of engagement
Table 4.4 shows the mediation analysis with linear regression in SPSS. Model 1 has an F-value of 18.690 at p=.000. However, appeal is not significant. Model 2 is also significant with F=8.646 at p=.000, but again, appeal is not significant. Model 3 is also significant with F=23.276 at p=.000 and as it was already found for hypothesis 1, engagement has a significant effect on purchase intention. Model 4 is significant with F=19.260 and P=.000. Engagement has a significant effect on purchase intention and appeal fails to show significance.
Table 4.4 Mediation results
Model 1 Model 2 Model 3 Model 4
Main Variables
DV=PI DV=Eng DV=PI DV=PI
Engagement Appeal Control variables .167 .686 .000a .000a .120 Attitude Involvement Trust .072c .003a .000a .901 .005a .000a .069c .039b .000a .067c .029b .000a R2 (Adjusted R2) (.282) .298 (.145) .164 (.331) .346 (.337) .355 F-value 18.690a 8.646a 23.276a 19.260a
Note: a p-value<.01; b p-value < .05; c p-value < .10
Figure 4.1 Mediation model
Additionally, the analysis is run with the Hayes PROCESS macro in SPSS using model 4 for mediation (Hayes, 2018). The results are identical with those of the linear regression. And the bootstrapping method confirms that only engagement on purchase intention is significant and all other models fail to show significance. The direct effect of appeal on purchase intention is p=.1058 [effect=-.2892, LLCI:-0.6404; ULCI:0.0619]. As there is a 0 in the bootstrap interval it cannot be said that there is a significant effect. However, the upper interval is close to 0 and p is close to marginal significance at a p<0.1 value. Thus, very cautiously, it can be said that an emotional appeal leads to less purchase intentions than a rational appeal (effect=-.2892) as rational appeal is dummy coded as 0 and emotional appeal is dummy coded as 1.
4.5 The moderating effect of self-monitoring
To test hypothesis 4, which predicts that self-monitoring moderates the effect of appeal on engagement, an ANCOVA is performed. From testing of hypothesis 2 we already know, that the interactions between the IV and the covariates are not significant and thus, homogeneity of regression slopes is not violated. However, Levene’s test has a p<.05 and thus, homogeneity of variance is violated (Levene, 1960). The model is significant with F=1.999 and P=.001. However, neither the interaction of self-monitoring and appeal, nor appeal or self-monitoring individually show significance. The mean for rational appeal is 2.059 and the standard deviation is 1.022. For emotional appeal the mean is 2.117 and the standard deviation is 1.107. Therefore, hypothesis 4 is rejected. The results are
Appeal
Engagement
PI
(a)0.060
0.324
(b) a (c’)-.0255
(c)-0.235
displayed in table 4.5 and an R square of .445 and an adjusted R squared of .099 can be reported.
Table 4.5 Moderation of self-monitoring
df SS MS F p ηp2 Appeal*self-monitoring 24 17.793 .741 .728 .813 .136 Appeal Self-monitoring 1 41 .156 30.220 .156 .737 .153 .724 .697 .880 .001 .211 Involvement 1 2.743 2.743 2.694 .104 .024 Trust Attitude Error 1 1 111 12.643 1.101 113.022 12.643 1.101 1.018 12.417 1.081 .001 .301 .101 .010 Note: a p-value<.01; b p-value < .05; c p-value < .10
The analysis is run again using Hayes PROCESS model 1 (Hayes, 2018). The results are very close to the ones displayed in table 4.5. Thus, no significance is found and hypothesis 4 is rejected. Furthermore, the analysis is run without the covariates, but that also does not change the result.
To get a better understanding of self-monitoring an additional step is taken. The self-monitoring scale is transformed into a categorical scale with all values 0-4=0 for low self-monitoring and all values 4-7=1 for high self-monitoring. This median split is a well-known method to convert a continuous variable into a categorical one (Kim et al., 2017). This conversion allows for an independent samples t-test of self-monitoring and engagement, which provides information about the means. It reveals that the mean of engagement is 2.172 with a standard deviation of 1.050 for low self-monitoring and 2.000 with a standard deviation of 1.074 for high self-monitoring. There is no significant difference between the two groups as t(181)= -1.073 and p=.285 for engagement, which explains why no significant effect was found.
4.6 Moderated moderation
Table 4.6 Moderation of number of likes and comments df SS MS F p ηp2 LikComm*self-monitoring 67 14.114 .543 .509 .975 .108 LikComm Self-monitoring 1 41 .050 34.734 .050 .847 .047 .794 .829 .798 .000 .230 Involvement 1 1.455 1.455 1.363 .246 .012 Trust Attitude Error 1 1 109 12.951 .580 116.354 12.951 .580 1.067 12.132 .544 .001 .463 .100 .005 Note: a p-value<.01; b p-value < .05; c p-value < .10
4.7 Moderated moderated mediation
Finally, the previous individual analyses are tested again in one model using Hayes PROCESS model 11 (Hayes, 2018), with a number of 5000 bootstrap samples and a 95% confidence interval. Figure 4.2 shows the results.
Figure 4.2 Moderated moderated mediation
This model reaffirms the previously described findings of the separate analyses. The effect of engagement on purchase intention is significant (coeff= .3284, p= .001). Therefore, hypothesis 1 can be supported as there is a positive effect of engagement on purchase intention. Hypothesis 2 is rejected as appeal does not have a significant effect on engagement (coeff= .8337, p= .5445). Hypothesis 3 and thus, the mediating effect of engagement can also not be supported, but there is a slightly significant direct effect of appeal on purchase intention (effect: -.2545, p=.1203). Hypothesis 4 assumes a moderating effect of self-monitoring,
Appeal
Engagement
PI
0.8337
0.3284
aSM
LikComm
-0.1801
-0.312
Conclusion
5.1 Discussion
Table 5.1 Hypotheses results
H1: Engagement positively influences purchase intentions. à confirmed
H2: An emotional advertising appeal leads to stronger engagement than
a rational advertising appeal. à rejected
H3: Engagement mediates the relationship between emotional
advertising appeal and purchase intentions. à rejected
H4: Self-monitoring moderates the relationship between advertising
appeal and engagement. à rejected
H5: The number of likes and comments moderates the extent to which self-monitoring moderates the relationship of advertising appeal and
engagement. à rejected
The aim of this research was to investigate the role of self-monitoring in explaining motivations to engage on Instagram advertising and to purchase products on Instagram. In the following, the results of this research are discussed for each of the five hypotheses displayed in table 5.1.
Hypothesis 1 is confirmed, as this study reveals that engagement leads to purchase intentions. Thus, this study is line with previous scholars who also found that engagement leads to purchase intentions (Hutter et al., 2013; Kilger & Romer, 2007; Valentini et al., 2018).
message processing (Petty, Cacioppo, & Schumann, 1983). Based on this theory, people process an emotional social media post by relying on feelings and heuristics and do not feel the need to actively engage with a like or comment, but simply consume the post (Dolan, Conduit, Frethey-Bentham, Fahy, & Goodman, 2019). This relates to the system 1 and system 2 thinking theory of Kahneman (2011). System 1 is the most common way of thinking, which is automatic and fast and results in only passive engagement in a social media environment (Swani, Milne, Brown, Assaf, & Donthu, 2017). Contrary to system 1, system 2 is a more controlled and effortful way of thinking resulting in logical, conscious behaviour (Kahneman, 2011). Hypothesis 3 revealed, that very cautiously it can be said that there is a marginal significant effect of the rational advertising appeal on purchase intentions. This can be explained by system 2 thinking, which is used for rational advertising and thus, results in more conscious decision making (Dolan et al., 2019; Kahneman, 2011; Swani et al., 2017).
Another explanation, which also gives a reason why this study rejects hypothesis 3 as it does not find, that engagement is a mediator of advertising appeal and purchase intentions, might be that the emotional post used emotional wording but consumers did not feel any emotions towards the brand. Pansari and Kumar 2017) find in their study, that a satisfactory relationship between the firm and the customer is needed for engagement to result in purchase intentions. Only if a firm has a relationship with the customer based on trust and commitment, it can result in satisfaction, and subsequently, in purchase intentions. Furthermore, they explain that a close relationship and prior positive experiences with the brand lead to emotions, which lead to customer engagement in the form of referring the brand to others (Pansari & Kumar, 2017). In this study, participants did not have any connection with the fictitious brand, as they have never seen it before. Thus, another important aspect for engagement might be a strong felt connection between the consumer and the brand, as this leads to emotions and satisfaction resulting in engagement and purchase intentions.
in the context of Instagram. A reason might be, that participants did not see sunscreen as a product that can help them to project a certain image about themselves, which is the statement of ground, why emotional advertising appeals to high self-monitors (DeBono & Packer, 1991; Snyder & DeBono, 1985). Thus, for sunscreen, the quality of the product seems to be more important for high, as well as low self-monitors. Scholars suggest, that product category plays an important role for online advertising effectiveness (Liu-Thompkins, 2019).
Another finding of this study is, that the self-monitoring scale apparently did not capture the essence of the self-monitoring score for participants in this study. An inspection of the individual questions of the self-monitoring scale revealed, that participants answered the 14 questions quite differently. From a mean of four and a small standard deviation in can be concluded, that the self-monitoring scale did not produce a high or low self-monitoring score for the individual participants. This proposes an explanation why no significant effect for self-monitoring is found. Kim, Seely and Jung (2017) used the same self-monitoring scale successfully. A reason might be that they used their sample from students of a south eastern university enrolled in a communication course in exchange for research credits. An assumption is that those students might have given better answers as English is their native language. The survey of this study was posted on the author’s Instagram account and reached mostly German participants.
Finally, hypotheses 5 is rejected as well. The assumption, that a high number of likes and comments influences high self-monitors to engage is based on the character trait of high self-monitors to use other peoples’ behaviour as a guide for their own behaviour (Ickes et al., 2006; Shaffer et al., 1982; Snyder, 1974). However, this study does not show a significant effect. Given the problems with the self-monitoring scale in this study, it is not surprising that hypotheses 5 is rejected as it is based on the self-monitoring scale.
Summarized, this study found that self-monitoring does not explain, why some people engage in response to a brands’ Instagram post, whereas others do not. This study proves that engagement leads to purchase intentions.
5.2 Academic implications
was conducted among people using Instagram, and found, that people’s self-monitoring score was neither high, nor low. Consequently, this research does not confirm the finding, that people using Instagram, score high on self-monitoring. Furthermore, as people scored neither high, nor low on self-monitoring in this study, it can be recommended, that the self-monitoring scale used by Kim, Seely and Jung (2017), which was adapted from the 25-item scale of (Snyder & Gangestad, 1986), should be treated carefully in future studies, as it did not capture the essence of self-monitoring in this study. Thus, this offers opportunities for academics to revise the self-monitoring scale and study self-monitoring again in the context of engagement on social media, to understand, whether the findings of previous scholars, namely, that emotional advertising is more effective for high monitors and rational advertising is more effective for low self-monitors (DeBono & Packer, 1991; Snyder & DeBono, 1985), can be extended to engagement.
In addition, this study brought valuable insights about engagement and purchase intentions in the context of Instagram. It contributed to the academic literature a better understanding of those mechanisms in the context of a brand’s post. Previous scholars found that engagement leads to purchase intentions (Hutter et al., 2013; Kilger & Romer, 2007; Valentini et al., 2018), which was confirmed by this study. In addition, this study adds to the academic literature the finding, that a rationally advertised post of a brand on Instagram, leads to purchase intentions without engagement.
5.3 Managerial implications
This study brings valuable findings for marketeers. Firstly, as Instagram is introducing the checkout feature, which makes it possible for users to purchase products directly on Instagram, increasing purchase intentions is an important topic (Instagram-Press, 2019). This study reveals, that indeed, engagement leads to purchase intentions and thus, initiating engagement is important. In addition, this implies for marketeers, that it might be effective to target potential customers, who engaged on a post with additional advertising, as those customers are more likely to be convinced to purchase a product.
more successful, even if it does not initiate engagement. Marketeers need to analyse the product they are advertising carefully and establish a corresponding marketing strategy. In the case of this study, a new sunscreen brand, which is not known, is more successful in convincing potential customers by using a rational advertising appeal focusing on the quality of the product, than an emotional appeal.
5.4 Limitations and recommendations for further research
This study comes with some limitations and further research suggestions. Firstly, this study confirms that engagement leads to purchase intentions, but also found at a very marginal significant level that rational advertising leads to purchase intention without engagement. This should be addressed in further research with a larger sample. Sunscreen seems to be a product, which is better advertised rationally. According to these findings, further research should be done about the effect of product category, as suggested by other scholars in the context of engagement as well (Liu-Thompkins, 2019).
Secondly, this study did not find, that an emotional appeal leads to engagement and thus, further research could replicate this study by using high arousal evoking emotions as suggested by Nikolinakou and King (2018). Therefore, it should be tested firstly, whether the emotional appeal is successful in generating arousal and in a second step, whether that leads to engagement. In addition, Pansari and Kumar (2017) state that emotions are based on a relationship between the brand and the customer. This should be considered in further research as well.
Thirdly, this research was not successful in capturing a meaningful self-monitoring score of participants. Further research could revise the self-self-monitoring scale and find solutions, on how to conduct meaningful results. Maybe, this scale is not suitable for an online questionnaire, but works in a controlled experiment, where people take the time to really think about themselves and answer the questions better (Aronson, Wilson, & Brewer, 1998). Another reason might be, that this study was in English, but distributed in the authors Instagram network, which consists mostly of German speaking people. Therefore, an insufficient level of English might have been a reason, why the self-monitoring scale failed to bring significant results. Consequently, the level of English language skills should be implemented as a covariate in future studies or the questionnaire should only be distributed to native speakers.
References
Albers-Miller, N. D., & Stafford, M. R. (1999). International services advertising: An examination of variation in appeal use for experiential and utilitarian services. Journal of Services Marketing, 13(4), 390–406. https://doi.org/10.1108/08876049910282682
Alin, A. (2010). Multicollinearity. Wiley Interdisciplinary Reviews: Computational
Statistics, 2(3), 370–374. https://doi.org/10.1002/wics.84
Andreu, L., Casado-Díaz, A. B., & Mattila, A. S. (2015). Effects of message appeal and service type in CSR communication strategies. Journal of
Business Research, 68(7), 1488–1495. https://doi.org/10.1016/j.jbusres.2015.01.039
Anduiza, E., & Galais, C. (2017). Answering without reading: IMCs and strong satisficing in online surveys. International Journal of Public Opinion
Research, 29(3), 498–518. https://doi.org/10.1093/ijpor/edw007
Aronson, E., Wilson, T. D., & Brewer, M. B. (1998). Experimentation in social psychology. In The Handbook of social psychology.
Bakhshi, S., Shamma, D. A., & Gilbert, E. (2014). Faces engage us: Photos with faces attract more likes and comments on instagram. Conference on Human
Factors in Computing Systems - Proceedings, 965–974. https://doi.org/10.1145/2556288.2557403
Baron, R. M., & Kenny, D. A. (1986). The Moderator-Mediator Variable Distinction in Social Psychological Research – Conceptual, Strategic, and Statistical Considerations. Journal of Personality and Social Psychology, 51(6), 1173– 1182.
Berger, J., & Milkman, K. L. (2012). What makes online content viral? Journal of
Marketing Research, 49(2), 192–205. https://doi.org/10.1509/jmr.10.0353
Chandran, S., & Morwitz, V. G. (2005). Effects of Participative Pricing on Consumers’ Cognitions and Actions: A Goal Theoretic Perspective. Journal
of Consumer Research, 32(2), 249–259. https://doi.org/10.1086/432234
Chandrashekaran, R. (2004). The influence of redundant comparison prices and other price presentation formats on consumers’ evaluations and purchase intentions. Journal of Retailing, 80(1), 53–66. https://doi.org/10.1016/j.jretai.2004.01.004
Chen, C. (2008). Identifying Significant Factors Influencing Consumer Trust in an Online Travel Site. Information Technology & Tourism, 8(3/4), 197–214. https://doi.org/10.3727/109830506778690849
Chin, C. Y., Lu, H. P., & Wu, C. M. (2015). Facebook users’ motivation for clicking the “like” button. Social Behavior and Personality, 43(4), 579–592. https://doi.org/10.2224/sbp.2015.43.4.579
https://blog.hootsuite.com/instagram-statistics/
Cutler, B. D., & Javalgi, R. G. (1993). Analysis of print ad features: Services versus products. Journal of Advertising Research, 33(2), 62–69.
DeBono, K. G. (2006). Self-monitoring and consumer psychology. Journal of
Personality, 74(3), 715–737. https://doi.org/10.1111/j.1467-6494.2006.00390.x
DeBono, K. G., & Packer, M. (1991). The Effects of Advertising Appeal on Perceptions of Product Quality. Personality and Social Psychology Bulletin,
17(2), 194–200. https://doi.org/10.1177/014616729101700212
Dolan, R., Conduit, J., Frethey-Bentham, C., Fahy, J., & Goodman, S. (2019). Social media engagement behavior: A framework for engaging customers through social media content. European Journal of Marketing, 53(10), 2213– 2243. https://doi.org/10.1108/EJM-03-2017-0182
Drossos, D. A., & Fouskas, K. G. (2010). Mobile Advertising: Product involvement and its effect on intention to purchase. ICMB and GMR 2010 - 2010 9th
International Conference on Mobile Business/2010 9th Global Mobility Roundtable, 183–189. https://doi.org/10.1109/ICMB-GMR.2010.27
Eckler, P., & Bolls, P. (2011). Spreading the Virus: Emotional Tone of Viral Advertising and Its Effect on Forwarding Intentions and Attitudes. Journal of
Interactive Advertising, 11(2), 1–11. https://doi.org/10.1080/15252019.2011.10722180
Egebark, J., & Ekström, M. (2012). Like What You Like or Like What Others Like? Conformity and Peer Effects on Facebook. SSRN Electronic Journal, (886). https://doi.org/10.2139/ssrn.1948802
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.
Escobar-Rodríguez, T., Grávalos-Gastaminza, M. A., & Pérez-Calañas, C. (2017). Facebook and the intention of purchasing tourism products: moderating effects of gender, age and marital status. Scandinavian Journal
of Hospitality and Tourism, 17(2), 129–144. https://doi.org/10.1080/15022250.2015.1137784
Fennis, B. M., & Stroebe, W. (2010). The psychology of advertising. In The
Psychology of Advertising. https://doi.org/10.4324/9780203853238
Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of the concept, state of the evidence. Review of Educational
Research, 74(1), 59–109. https://doi.org/10.3102/00346543074001059
Ghasemi, A., & Zahediasl, S. (2012). Normality tests for statistical analysis: A guide for non-statisticians. International Journal of Endocrinology and
Metabolism, 10, 486–489. https://doi.org/10.5812/ijem.3505
https://doi.org/10.1016/j.hrmr.2010.09.004
Hadjimarcou, J. (2012). An Investigation of informational versus emotional advertising appeals during life transitions. International Journal of
Management and Marketing Research, 5(1), 55–66.
Hanson, W. A., & Putler, D. S. (1996). Hits and misses: Herd behavior and online product popularity. Marketing Letters, 7, 297–305. https://doi.org/10.1007/BF00435537
Harvard Business. (2015). Why Some Videos Go Viral. Retrieved October 24, 2019, from Harvard Business Review website: https://hbr.org/2015/09/why-some-videos-go-viral
Hayes, A. F. (2018). Introduction to Mediation, Moderation, and Conditional
Process Analysis: A Regression-Based Approach (2nd Editio; Todd D. Little,
Ed.). New York: The Guilford Press.
Hoffman, D. L., & Fodor, M. (2010). Can you measure the ROI of your social media marketing? MIT Sloan Management Review, 52(1), 41–49.
Hong, I. B. (2015). Understanding the consumer’s online merchant selection process: The roles of product involvement, perceived risk, and trust expectation. International Journal of Information Management, 35, 322–336. https://doi.org/10.1016/j.ijinfomgt.2015.01.003
Hutter, K., Hautz, J., Dennhardt, S., & Füller, J. (2013). The impact of user interactions in social media on brand awareness and purchase intention: The case of MINI on Facebook. Journal of Product and Brand Management,
22(5), 342–351. https://doi.org/10.1108/JPBM-05-2013-0299
Ickes, W., Holloway, R., Stinson, L. L., & Hoodenpyle, T. G. (2006). Seli-monitoring in social interaction: The centrality of self-affect. Journal of
Personality. https://doi.org/10.1111/j.1467-6494.2006.00388.x
Instagram-Press. (2019). Introducing Checkout on Instagram. Retrieved October 10, 2019, from https://instagram-press.com/blog/2019/03/19/instagram-checkout/
Johar, J. S., & Joseph Sirgy, M. (1991). Value-expressive versus utilitarian advertising appeals: When and why to use which appeal. Journal of
Advertising, 20(3), 23–33. https://doi.org/10.1080/00913367.1991.10673345
Johnson, T. R. (2016). Violation of the homogeneity of regression slopes assumption in ANCOVA for two-group pre-post designs: Tutorial on a modified Johnson-Neyman procedure. The Quantitative Methods for
Psychology, 12(3), 253–263. https://doi.org/10.20982/tqmp.12.3.p253
Kahneman, D. (2011). Thinking, Fast and Slow. In Book. New York: Macmillan. Kahu, E. R. (2013). Framing student engagement in higher education. Studies in
Higher Education, 38(5), 758–773. https://doi.org/10.1080/03075079.2011.598505
We are social website: https://wearesocial.com/global-digital-report-2019 Kilger, M., & Romer, E. (2007). Do measures of media engagement correlate with
product purchase likelihood? Journal of Advertising Research, 47(3), 313– 325. https://doi.org/10.2501/S002184990707033X
Kim, D. H., Seely, N. K., & Jung, J.-H. (2017). Do you prefer, Pinterest or Instagram? The role of image-sharing SNSs and self-monitoring in enhancing ad effectiveness. Computers in Human Behavior, 70, 535–543. https://doi.org/10.1016/j.chb.2017.01.022
Kim, M. J., Chung, N., & Lee, C. K. (2011). The effect of perceived trust on electronic commerce: Shopping online for tourism products and services in
South Korea. Tourism Management, 32, 256–265.
https://doi.org/10.1016/j.tourman.2010.01.011
Kotler, P., & Keller, K. L. (2009). Marketing Management (13th ed.). Upper Saddle River, New Jersey, USA: Pearson Education, Inc.
Kumar, V., & Reinartz, W. (2016). Creating enduring customer value. Journal of
Marketing, 80(6), 36–68. https://doi.org/10.1509/jm.15.0414
Kung, F. Y. H., Kwok, N., & Brown, D. J. (2018). Are Attention Check Questions a Threat to Scale Validity? Applied Psychology, 67(2), 264–283. https://doi.org/10.1111/apps.12108
Lascu, D., Bearden, W. O., & Rose, R. L. (1995). Norm Extremity and Interpersonal Influences. Journal of Business Research, 32(3), 201–212. Lee, J., Kim, S., & Ham, C. D. (2016). A Double-Edged Sword? Predicting
Consumers’ Attitudes Toward and Sharing Intention of Native Advertising on Social Media. American Behavioral Scientist, 60(12), 1425–1441. https://doi.org/10.1177/0002764216660137
Lee, W. I., Cheng, S. Y., & Shih, Y. T. (2017). Effects among product attributes, involvement, word-of-mouth, and purchase intention in online shopping. Asia
Pacific Management Review, 22, 223–229. https://doi.org/10.1016/j.apmrv.2017.07.007
Levene, H. (1960). Robust tests for equality of variances. In I. Olkin (Ed.),
Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling (1st ed.). Palo Alto: Stanford University Press.
Liu-Thompkins, Y. (2019). A Decade of Online Advertising Research: What We Learned and What We Need to Know. Journal of Advertising, 48, 1–13. https://doi.org/10.1080/00913367.2018.1556138
Liu, M., & Wronski, L. (2018). Trap questions in online surveys: Results from three web survey experiments. International Journal of Market Research, 60(1), 32– 49. https://doi.org/10.1177/1470785317744856
Malhotra, N. K. (2009). Marketing Research: An Applied Orientation (6th Editio). New Jersey: Pearson.
