Branded content and cause-related marketing: success or failure for eWOM?

(1)

1

Branded content and cause-related marketing:

success or failure for eWOM?

A study on the effect of cause-focused branded content on eWOM moderated by self-enhancement and mediated through brand credibility and ad credibility.

(2)

2

Branded content and cause-related marketing:

success or failure for eWOM?

A study on the effect of cause-focused branded content on eWOM moderated by self-enhancement and mediated through brand credibility and ad credibility.

Master Thesis Marketing Management Erika Masone

e.masone@student.rug.nl s4217160

First supervisor: dr. Janny C. Hoekstra Second supervisor: dr. Judith de Groot

Date of submission: 10.01.2021

University of Groningen Faculty of Economics and Business

Department of Marketing PO Box 800 9700 AV Groningen

(3)

3

Abstract

Branded content makes blending advertising with editorial or entertainment content its peculiar characteristic, emerging as the marketing tool capable of breaking the increasing advertising clutter. Indeed, branded content aims to offer more than plain advertising. Similarly, cause-related marketing (CRM) goes beyond the simple advertising goal, linking the brand to social, environmental or humanitarian causes. Therefore, it is interesting for practitioners to understand whether and how branded content and CRM can be combined to elicit even stronger customer responses. In particular, the metric used in this study to understand the effectiveness of cause-focused branded content was eWOM. The current research adopted a between-subjects

experimental design with 70 participants in order to investigate whether cause-focused branded content resulted in greater eWOM effects (sharing, liking, commenting) than regular branded content, taking into account intermediate steps. Specifically, the moderating effect of self-enhancement and the mediating effects of brand credibility and ad credibility on the relationship between cause-focused branded content and eWOM are considered.

(4)

4

Preface

Writing this Master Thesis was the last step in finalizing my Master in Marketing Management. The research and writing process was interesting and exciting thanks to the broad topic of “branded content”. I managed to choose and explore the field of cause-related marketing in relation to branded content.

I would like to thank my supervisor dr. Janny Hoekstra for her guidance and help while following my work through every step with useful feedback. I would also like to thank dr. Judith de Groot for being my second supervisor. Furthermore, I would like to thank all the respondents who took part in my survey, whose participation was very important for the purpose of this thesis.

I will remember the Master at the University of Groningen as a challenging and fruitful experience and I’m looking forward to see what the next step will be.

(5)

5

2.4 Brand credibility ... 15 2.5 Ad credibility ... 16 2.6 Self-enhancement... 17 3 Methodology ... 19 3.1 Research Design ... 19 3.2 Data Collection ... 22 3.3 Measures... 23 3.4 Factor analysis ... 25 3.5 Reliability analysis ... 26 3.6 Manipulation check ... 26 3.7 Data analysis ... 27 4 Results ... 30 4.1 Control Variables ... 30 4.2 Hypotheses Testing ... 30

4.2.1 Hypothesis 1: CRM and eWOM ... 30

4.2.2 Hypothesis 2: CRM and brand credibility ... 31

4.2.3 Hypothesis 3: CRM and ad credibility ... 32

4.2.4 Hypothesis 4: brand credibility and ad credibility ... 33

4.2.5 Hypothesis 5: brand credibility and eWOM ... 34

4.2.6 Hypothesis 6: ad credibility and eWOM ... 35

4.2.7 Hypothesis 7: the moderating effect of self-enhancement on the relationship between CRM and eWOM ... 36

4.2.8 Hypothesis 8: self-enhancement and eWOM ... 38

4.3 The mediating effects of brand credibility and ad credibility ... 39

4.3.1 Baron and Kenny... 39

4.3.2 Hayes PROCESS model 6 ... 41

4.4 Multivariate analyses ... 43

4.5 Testing the conceptual model ... 45

5 Conclusion ... 49

(6)

6

(7)

7

1. Introduction

Branded content refers to a marketing practice that combines advertising with editorial or entertainment content and it is included in the category of online digital advertising (Harms, Bijmolt, & Hoekstra, 2019). While traditional marketing instruments saw a decline in their effectiveness (Risselada, Verhoef, & Bijmolt, 2014), online advertising has emerged as the most relevant advertising medium in the last decade and the percentage of total advertising

expenditures devoted to this tool is expected to grow from 37.6% in 2017 to 44.6% in 2020 (Harms, Bijmolt, & Hoekstra, 2019). In particular, branded content has received significant attention and recognition because it creates opportunities both for marketers and online media platforms. It represents a marketing medium capable of breaking the increasing advertising clutter and the resulting ad avoidance that most consumers manifest (Harms, Bijmolt, &

Hoekstra, 2017). Moreover, online media platforms, particularly news media, can generate new income streams and compensate for the decline of traditional advertising by presenting

(8)

8

One of the most interesting channels for branded content is social network sites. Nowadays, the majority of brands consider social media marketing as an integrated component of their

marketing communication strategy (Ashley & Tuten, 2015). Social media represent an effective medium to create and share content with a large set of customers. They enable brands to

communicate easily and directly with their customers, promoting active online engagement and consolidating the brand-customer relationship (Ashley & Tuten, 2015). Indeed, through branded content, customers derive thoughts, emotions, perceptions, images and stories that create

associations with the specific brand in their minds (Aribarg & Schwartz, 2020). As long as brands are able to deliver valuable content, they can generate electronic word of mouth

(eWOM) with shared branded content that, as Risselada et al. (2014) report, has a higher reach and impact than traditional WOM. Furthermore, results from previous studies provide evidence that exposure to branded content positively influences consumers’ brand attitudes, brand loyalty and purchase intentions, while playing an important role in brand building (Lou, Xie, Feng, & Kim, 2019). Social engagement on social media occurs in the form of sharing and commenting content. Observing these interactions may give marketers valuable insights on customer

behaviour, which should be useful in developing content that is appreciated by the specific targets of the brand. Social transmission becomes important for brands considering the most important findings from prior studies. As Berger and Milkman (2012) report, interpersonal communication affects attitudes and decision making among consumers, while eWOM has a causal impact on product adoption and sales. Such considerations become even more relevant considering that 59% of people frequently share online content with others (Berger & Milkman, 2012).

In this study I analyse the relationship between branded content and eWOM effects in the form of customers’ willingness to share, like or comment such content. Specifically, this paper aims to investigate the motives behind customers’ eWOM intentions, with a focus on message content with a cause-related marketing component. This research is also interested in understanding the factors that might contribute to the relationship between cause-focused

branded content and eWOM, namely self-enhancement, ad credibility and brand credibility. The main research question focuses on whether branded content with a CRM component positively stimulates eWOM, taking into account the moderating effect of self-enhancement and

(9)

9

CRM on eWOM by explaining the mechanism that underlies an observed relationship between cause-focused branded content and eWOM.

Self-enhancement reflects one of the most dominant human social motivations and describes individuals’ tendency to bolster their self-concept with respect to themselves and others (Wojnicki & Godes, 2017). The results presented by Fan et al (2020) , that is a positive influence of CRM on recommendation intentions, suggest a possible link between CRM

branded content and eWOM, moderated by self-enhancement. In the context of social networks, individuals engage in strategic image management which aims at presenting the most favourable version of themselves to others (Hogan, 2010). Social media platforms represent a public place where identities are displayed and multiple interactions take place. Self-enhancement thus emerges as the most important mechanism that governs these social interactions because

individuals prefer to disclose an enhanced version of the self in order to make a good impression (Sicilia, Delgado-Ballester, & Palazon, 2016).

Ad credibility is defined as the extent to which customers perceive the advertisement as credible and believable (Cotte, Coulter, & Moore, 2005). Customers’ perceptions of the ad are important in determining their behaviour towards the brand, in terms of purchase intentions and eWOM (Cotte, Coulter, & Moore, 2005). Moreover, ad credibility may also result crucial in determining the success of a marketing campaign, with lower ad credibility negatively influencing

customers’ eWOM effects (Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). This effect of ad credibility on eWOM is especially important in the context of CRM because of the initial scepticism that cause-focused brand messages evoke (Bergkvist & Zhou, 2019).

Finally, brand credibility is defined as the extent to which customers perceive that the brand has the necessary level of expertise and trustworthiness (Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). Brand credibility is a key factor in determining consumer behaviours, such as brand loyalty and purchase intentions (Rifon, Choi, Trimble, & Li, 2004). Brand credibility is also an important determinant of marketing success because the brand symbolises the

company’s position on the market and thus, it becomes a means for customers to develop judgements and preferences (Bigne-Alcaniz, Caceres, & Perez, 2010). With respect to CRM, brand credibility is used by customers as an indicator of the company’s genuine intentions behind cause-focused advertising messages (Varadarajan & Menon, 1988).

(10)

10

(11)

11

2. Theoretical framework

2.1 Conceptual model

Figure 1 presents the conceptual model for this study. This research investigates the effect of branded content with a CRM component on eWOM compared to branded content without a CRM component. The aim is to understand the process of CRM influencing eWOM taking into account intermediate ladders operating in this relationship. Specifically, self-enhancement as a potential moderator can strengthen the influence of cause-focused branded content on eWOM, while brand credibility and ad credibility as potential mediators can explain the mechanism

underlying an observed relation between CRM and eWOM.

Customers’ eWOM refers to their willingness to share, like or comment branded content on social media platforms (Facebook, Instagram, Twitter). I expect branded content with a CRM component to elicit stronger consumers’ eWOM effects (Bergkvist & Zhou, 2019).

(12)

Curras-12

Perez, & Sanchez-Garcia, 2009). Thus, I expect brand credibility to increase ad credibility. The mediation of brand credibility is assumed to clarify the relationship between cause-focused branded content and ad credibility by exploring the underlying mechanism by which CRM influences ad credibility through brand credibility. A mediating relationship exists when brand credibility plays a partial or full role in governing the relationship between the two variables (Baron & Kenny, 1986). Customers see brand credibility as a symbol guiding their perceptions of the company and its products, positively influencing all the elements surrounding the brand, including brand messages (Elving, 2013). Customers’ positive perceptions of brand messages are considered as an important indicator of eWOM intentions (Van den Putte, 2009). Therefore, I expect brand credibility to positively influence eWOM. The mediation of brand credibility is expected to explain the relationship between cause-focused branded content and eWOM by exploring the underlying mechanism by which CRM influences eWOM through brand credibility.

Figure 1: Conceptual model

(13)

13 2.2 eWOM effects

Word-of-mouth is one of the most powerful communication channels for a brand (Wojnicki & Godes, 2017). The relevance of WOM has motivated researchers to investigate what drives customers to share brand-related content. This research contributes to this field by specifically focusing on CRM as a potential driver of WOM on social media (thus, eWOM), linking the intention to share cause-focused branded content to the desire to self-enhance. In particular, eWOM effects refers to customers’ intentions to share, like or comment branded content on social networks.

Social media platforms are often used by customers as a channel to share brand-related content or messages, giving rise to eWOM (Kulkarni, Kalro, & Sharma, 2019). Social media and internet communication have enhanced the effect of WOM, leading brands to consider eWOM as a strategic marketing tool (Taylor, Strutton, & Thompson, 2012). A survey reported by Wang and Chen (2020) suggests that brand and consumer-generated contents have a significant impact on consumers’ brand evaluation. Relevant academic literature on the issue of eWOM has

documented the effectiveness of electronic word-of-mouth on purchase intentions, customer loyalty and engagement due to its speed and global reach (Taylor, Strutton, & Thompson, 2012).

2.3 Message content: cause-related marketing

Effective communication requires choosing a successful advertising strategy (Van De Putte & Dhondt, 2005). An advertising strategy consists of different elements and this research

addresses the issue of message content. The content of the advertisement can be a determinant of effective communication (Van De Putte & Dhondt, 2005). Indeed, if the advertisement fails to deliver an effective message, the frequency of advertising, the budgets or the customer segmentation will not make a difference in determining advertising success (Van den Putte, 2009). The focus of analysis is the choice of what image, values and personality of the brand are modelled into branded content and the consequences in terms of consumers’ sharing intentions of online branded content (Wang & Chen, 2020).

(14)

14

enhancing consumer perceptions of the long-term image of the brand and building emotional bonds with customers (Fan, Deng, Qian, & Dong, 2020). Cause-related marketing has received many definitions in the marketing literature. Kim and Johnson (2013) define cause-related marketing as “marketing activities that offer consumers opportunities to make purchase decisions for reasons other than personal beneﬁt” such as social or moral beliefs. More specifically, Bergkvist and Zhou (2019) define CRM “as a form of leveraged marketing communications (LMC), that is, marketing communications that aim for the brand to beneﬁt from consumers’ positive associations to another object”, in this case a social cause.

This research aims to study and evaluate whether a brand benefits more from branded content with a CRM component or from branded content without a CRM component. The impact of branded content on the brand is measured in terms of eWOM effects because the consumer plays a key role in determining the success of cause-focused messages, since branded content incorporating CRM is considered effective when customers believe and appreciate the message and the cause (Fan, Deng, Qian, & Dong, 2020). Relevant findings from the literature on CRM found that consumers exhibit favourable attitudes and perceptions towards brands engaging in CRM supporting social causes (Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). Nan and Heo (2007) found that respondents elaborated more positive attitudes towards a company associated with an ad with a CRM message compared with an ad without a CRM component. Finally, Fan et al (2020) report that consumers’ recommendation intentions are positively influenced by CRM strategies. Thus, this research expects that the positive perceptions and beliefs toward the brand, resulting from branded content with a CRM initiative, translate into consumer behaviour in the form of willingness to share.

H1: Cause-focused branded content exerts a stronger influence on customers’ eWOM on social media platforms compared to branded content without a CRM component.

However, as previous literature stressed, consumers are often sceptical about CRM initiatives due to the intuitive belief that this type of strategy is motivated by profit-oriented and egoistic interests rather than a genuine commitment (Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). Rima and Kimb (2016) claim that this scepticism is the result of the paradox between the definition of a for-profit company, namely generating revenues, and corporate social

(15)

15

(Cotte, Coulter, & Moore, 2005). Scepticism will then threaten customers’ perceptions of message believability. Therefore, I expect expects the following:

H3: Cause-focused branded content will result in lower ad credibility compared to branded content without a CRM component.

2.4 Brand credibility

Brand credibility refers to a company’s expertise, trustworthiness and likeability (Hoeffler & Keller, 2002). Brand credibility is measured as the extent to which a company proves to be expert, being competent on the market, trustworthy, showing commitment to its goals and customers’ needs, and likeable, being interesting and desirable (Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). With respect to marketing, credibility is often associated with source credibility. In the context of CRM, the company itself is perceived as the communication source and thus the effectiveness of the message strategy is determined by its own credibility (Bigne-Alcaniz, Caceres, & Perez, 2010).

(16)

16

H2: Cause-focused branded content positively influences brand credibility.

Brand credibility helps customers overcome the initial scepticism that companies’ CRM initiatives generate because they perceive credibility as an indicator of reliability and trustworthiness (Hoeffler & Keller, 2002). Indeed, previous studies have shown the positive influence of brand credibility on the persuasive and influential impact of CRM messages

(Bigne-Alcaniz, Curras-Perez, & Sanchez-Garcia, 2009). Brand credibility enhances customers’ responsiveness towards the company’s cause-focused messages because people use the brand’s credible image as a criterion to judge the company’s actions (Bigne-Alcaniz, Caceres, & Perez, 2010). Therefore, I expect the following:

H4: Brand credibility positively influences ad credibility.

H5: Brand credibility positively influences eWOM on social media platforms.

2.5 Ad credibility

Existing literature on CRM suggests that generally consumers are more favourable towards companies associated with CRM initiatives than those that are not, since they notice the

commitment and values and evaluate them positively (Kim & Lee, 2009). However, advertising credibility, in the form of scepticism towards CRM content, is a major challenge for marketers (Elving, 2013). Scepticism is the individuals’ reaction when they tend to question or distrust the company’s motives behind CRM and suspect that the brand is engaging in CRM only to

improve its image (Elving, 2013). Rima and Kimb (2016) claim that scepticism towards CRM initiatives arises from the doubt of the truthfulness of socially responsible messages when they are delivered through advertising. Indeed, sceptic consumers might suspect that the company’s motives are entirely linked to profit generation rather than seeking the benefit of the society. In this context, when scepticism prevails, consumers attribute extrinsic motives to the brand promoting CRM (Rima & Kimb, 2016). Extrinsic motives generate when the company uses CRM in an opportunistic way, in order to improve brand image and reputation (Elving, 2013). Lower ad credibility with respect to CRM originates from the paradox between the definition of for-profit organizations, whose goals are profit-oriented, and prosocial behaviour, which is motivated by a voluntary commitment to increasing social welfare (Kim & Lee, 2009). Kim and Lee (2009) report that, when authenticity and commitment are doubted, consumers are less likely to respond to CRM content. Therefore, I propose the following hypothesis:

(17)

17 2.6 Self-enhancement

People often share content on social media platforms to communicate identity or to enhance one’s self-concept (Berger & Milkman, 2012). This includes sharing informative content due to the desire to appear knowledgeable. Social transmission on social media involves tweeting, blogging or posting on your Facebook wall or Instagram. People active on social networks manage their profiles and disclose information about their identify, preferences or opinions through posts and conversations shared on online platforms (Luo & Hancock, 2020). Publicly displaying one’s own preferences or impressions by posting or sharing an online advertisement means that consumers employ the value and image of the brand to communicate their self-concept (Taylor, Strutton, & Thompson, 2012).

(18)

18

settings are the result of the interaction between social media messages and consumer identity (Ashley & Tuten, 2015). Thus, self-enhancement plays a key role in marketing efforts since it represents an instinct which is present in how individuals portray their identity online (Bazarova & Choi, 2014). Nan and Heo (2007) report that CRM can enhance the customer image because those consumers using brands associated with CRM are perceived as generous and altruistic. Since the use of social media implies demonstrating and signaling qualities or competence, self-enhancement becomes a potential motivator for individuals to respond to and share CRM branded content rather than regular branded content in order to improve and convey their positive self-concept (Mathur, Chun, & Maheswaran, 2016). Branded content with a CRM component may lead to greater eWOM effects through the mechanism of self-enhancement because those individuals who are concerned about signaling their positive qualities and competence to the self and others may be more willing to be proactive towards prosocial behaviour (the social cause presented in the branded content) to demonstrate that they are good people (Mathur, Chun, & Maheswaran, 2016). Indeed, previous research found that engaging in eWOM activities can boost an individual’s self-image (Wojnicki & Godes, 2017). Thus, CRM can act as the object that enables consumers to enhance their self-concept through eWOM, using branded content with a CRM component as cues signaling their knowledge, competence and moral values (Mathur, Chun, & Maheswaran, 2016). Therefore, I propose the following hypotheses:

H7: Self-enhancement strengthens the effect of cause-focused branded content on consumers’ eWOM on social media platforms.

(19)

19

3. Methodology

3.1 Research design

The information relevant for this research was collected by means of a between-subjects experimental design with two conditions. Respondents received a link directing to the online survey, which they could visit whenever they preferred during the time when the survey was conducted in order to ensure a natural setting. The survey was distributed adopting a

convenience sampling approach and participants were also asked to distribute it within their own networks as well, creating a snowball effect (Malhotra, 2009). The survey was based on presenting respondents with a stimulus which consisted of one of two different branded contents. Both of them involved a combination of advertising and editorial content and they were both sponsored by the same company: Netflix. The key difference between the stimuli was the presence of a social message in only one of them.

The first ad was an article commissioned by Netflix and published on The New York Times website, with the title “Women Inmates: Why the Male Model Doesn’t Work” (Figure 3.1a).It was clearly identified as a paid post and also well optimized for smartphones. The advertising component was Netflix intention to promote one of its original series, Orange is the new black. The ad consisted of video, charts and audio to supplement text about female incarceration in the U.S. Orange is the New Black is a show about a modern woman's experiences in prison. Thus, an important component of the article revolved around exploring the conditions of female prisoners in the U.S., stressing how the prison system fails to meet the needs of female inmates. The social message consists in sensitising the audience on the difficulties of female prisoners and on the unjustified difference between the conditions of female and male incarceration. The second ad was an article commissioned by Netflix which appeared on the Wired website with the title “TV got better” (Figure 3.1b). It was clearly identified as sponsored content and optimized for mobiles. The article explores changing content consumption patterns and the future of television, including text, images and videos. Moreover, it suggests how new

(20)

20

Figure 3.1a: “Women Inmates: Why the Male Model Doesn’t Work”

Description of the content of the article:

This article is a sponsored content which appeared on The New York Times website, paid by Netflix to promote Orange is the new black, a Netflix original series. Orange is the new black is a show about a modern woman's experiences in prison and, although it is never mentioned in the article, it helps introducing the discussion about female incarceration in the US. The text reports personal stories and statistics of how the prison system fails to meet the needs of female

(21)

21 Figure 3.1b: “TV got better”

Description of the content of the article:

This article is a sponsored content which appeared on the Wired website, paid by Netflix to promote its streaming service. The text explores the evolution of television history, proposing a timeline that points to the milestones in television experience. The discussion presents the changing consumption patterns and the future of television. It focuses on how the role of TV as a medium is evolving and how Netflix is at the root of it. Indeed, it suggests how new

(22)

22 3.2 Data collection

A survey was createdwith the online survey software Qualtrics and it was distributed online. The survey began with an introduction stating that the purpose of the research was to collect data for a Master Thesis project, that it would take around five minutes to complete the survey questions and that it was fully anonymous. On the next page, questions aimed at gathering demographic data on the sample were introduced. After the demographic section, respondents were asked to answer questions aimed at understanding their social media use. Respondents were then randomly assigned to one of the two experimental conditions. They were presented with the “Women Inmates: Why the Male Model Doesn’t Work” article (cause-focused branded content) or with the “TV got better” article (regular branded content). Specifically, they saw a screenshot (respectively Figure 3.1a and 3.1b) presenting the focus and design of the article and they read a short description of the content of the article. Finally, they were asked to answer questions measuring their perceptions of brand credibility, ad credibility, desire for self-enhancement and eWOM based on Likert scales. Attitudes towards the ad and the brand were also measured in order to control for their effects on the dependent variable, eWOM. As soon as the survey was finished, respondents were thanked for their participation.

The survey collected 85 responses. A data analysis and cleaning procedure followed. A total of 15 responses were deleted because the respondents failed to complete the questionnaire. Thus, the complete dataset consisted of 70 valid observations. The sample exposed to the cause-focused branded content consisted of 37 respondents, while the group exposed to the regular branded content included 33 participants.The distribution of respondents across the two conditions can be seen in Table 3.2. Following the research conducted by Cohen (1988), 30 participants per experimental condition are enough to detect differences between groups, in terms of medium or large effect size. Thus, the minimum number of participants for this study should be 60 and it was exceeded with a total of 70 responses. In addition, the number of respondents per condition is sufficient in order to obtain significant results.

The age of participants ranged from 15 to 47. The mean age was 24 (SD=5.05). The sample included 32.86% (n=23) of men and 62.86% of women (n=44). With respect to social media platforms, the most frequently used apps were Instagram (25%), YouTube (21.08%) and

(23)

23 Table 3.2: Distribution of the experimental conditions

CONDITION RESPONDENTS

Cause-focused branded content 37 Regular branded content 33

3.3 Measures

The measurement scales for the constructs were adapted from existing academic literature (see Table 3.3). Brand credibility was measured adopting a six items 7-point Likert scale proposed by Spry et al. (2011). Ad credibility was measured on a three items 7-point Likert scale based on Cotte et al. (2005). Following Taylor et al. (2012) self-enhancement was measured on a six items 7-point Likert scale. Finally, the dependent variable, eWOM, was measured in terms of willingness to share, like and comment using a three items 7-point Likert scale adapted from Kulkarni et al. (2019). Previous studies in the literature on CRM have found a positive

(24)

24 Table 3.3: Constructs measurement

MEASURE SCALE FACTOR

LOADINGS CRONBACH ALPHA eWOM (Kulkarni, Kalro, & Sharma, 2019)

Suppose this is the ﬁrst time you have seen this article. How likely would you be to share/like/comment it with others on social media platforms (Facebook, Twitter, etc.)?

Press the like button. Share the article. Comment the article.

0.83 0.90 0.85 0.816 Brand credibility (Spry, Pappu, & Cornwell, 2011)

This brand reminds me of someone who’s competent and knows what he/she is doing.

This brand has the ability to deliver what it promises. This brand delivers what it promises.

Over time, my experiences with this brand have led me to expect it to keep its promises, no more and no less. This brand has a name you can trust.

This brand doesn’t pretend to be something it isn’t.

0.69 0.89 0.83 0.87 0.87 0.78 0.898 Ad credibility (Cotte, Coulter, & Moore, 2005)

I think the ad is believable. I think the ad is truthful. I think the ad is realistic.

0.92 0.90 0.91 0.894 Self-enhancement (Taylor, Strutton, & Thompson, 2012)

This message reflects who I consider myself to be. This message reflects who I am.

Passing along this message would communicate who I am to other people.

This message is consistent with how I want to present myself to others.

I can identify with this message.

My reaction to this message would tell others something important about me.

0.86 0.86 0.81 0.88

(25)

25

Ad attitude

(Lichtlé, 2007)

I find this ad convincing. I find this ad intelligent.

I find this ad pleasant to look at. I find this ad informative.

0.86 0.81 0.81 0.81 0.842 Brand attitude (Speed & Thompson, 2000)

Thinking about Netflix, please evaluate this company by selecting the point on each scale that best represents your attitude to the company.

Good/Bad Like/Dislike Pleasant/Unpleasant Favorable/Unfavorable 0.94 0.97 0.91 0.94 _0.955 3.4 Factor analysis

Factor analysis and reliability analysis were performed in order to compute reliable and valid constructs. The aim of the factor analysis is to reduce the number of variables by testing if all the items regarding one variable are related to each other and thus, putting together interrelated variables into a factor. Before conducting the factor analysis, the Kaiser-Meyer-Olkin (KMO) statistic and Bartlett’s Test of Sphericity were consulted to see whether the factor analyses could be properly executed. The KMO measure ranges from 0 to 1 and indicates that factor analysis is feasible for KMO values above 0.5 (Malhotra, 2009). The Bartlett’s Test of Sphericity tests the null hypothesis that the correlation matrix has an identity matrix, which means that the items are unrelated and unsuitable for a factor analysis (Bartlett, 1950). Thus, the Bartlett’s Test of

Sphericity should be significant (p<0.05) in order to conduct a factor analysis (Malhotra, 2009). Finally, communalities should have values higher than 0.4 and factor loadings should exceed the value of 0.5 (Malhotra, 2009).

(26)

26

value which is higher than the minimum threshold of 60% (Malhotra, 2009). All factor loadings were above 0.5. The respective factor loadings are reported in Table 3.3.

A separate factor analysis was conducted for the dependent variable, eWOM. The KMO and Bartlett’s Test of Sphericity were performed to ensure the appropriateness of a factor analysis. The KMO measure had a value of 0.688 and the Bartlett’s Test of Sphericity was significant (p=0.000). In addition, all communalities were above 0.4. A principal component analysis (PCA) was conducted. The scree plot and the table of total variance explained indicated a one-factor solution, namely eWOM. This one one-factor explains 74.21% of the total variance, which is above the critical value of 60%, and had an eigenvalue of 2.23 (Malhotra, 2009).

3.5 Reliability analysis

A reliability analysis was conducted to test the reliability and internal consistency of the factors found in factor analysis, using Cronbach’s alpha. The value of the Cronbach’s alpha should be above the critical value of 0.60 (Malhotra, 2009). Cronbach’s alpha was computed for all six factors and, as reported in Table 3.3, all values of Cronbach’s alpha were higher than the minimum threshold, indicatingthat these components are internally consistent enough to measure the latent constructs of the factors. Therefore, the new factors were created by adding up the scores of all the items in each of the factors and dividing the total by the number of items included in one particular factor, which created mean scores for all the respondents for each individual factor.

3.6 Manipulation check

(27)

27 Table 3.6: Manipulation check

Measure Scale Factor loadings Cronbach’s

alpha

Manipulation check Please indicate to what extent do you agree with the following statements The article aims to support a social message

The article is not involved with a social cause

0.91

0.778

A factor analysis was performed to test whether the two items of the manipulation check were related to each other. First, the scores of the second item were reversed. Then, a Kaiser-Meyer-Olkin measure of sampling adequacy and Bartlett’s Test of Sphericity were used to determine if a factor analysis is feasible. The KMO measure was higher than 0.5 (KMO=0.6) and the

Bartlett’s Test of Sphericity was significant (p=0.001). Thus, the items are related and suitable for a factor analysis with principal component analysis (Malhotra, 2009). Moreover,

communalities were higher than 0.4. The factor analysis resulted in a one-factor solution with factor loadings of 0.91. Subsequently, a reliability analysis was conducted. Cronbach’s Alpha showed ⍺=0.778, which was higher than the minimum for computing a sum variable (⍺=.60) (Malhotra, 2009).Therefore, a new variable to test the extent to which the branded content was perceived as cause-focused was computed by summing the items and dividing it by two. Finally, an independent sample t-test was conducted to check whether the experimental

manipulation was successful. The difference in means and standard deviations implied that the two groups (social cause and no social cause) significantly differed from each other. The participants exposed to the cause-focused branded content perceived that the advertisement focused on a social message (M=5.52, SD=1.00)compared to the group exposed to a regular branded content (M=2.86, SD=1.42). The t-test confirmed that these differences were

significant, t(81)=2.16, p=0.042). The manipulation of message content experimental condition was therefore successful.

3.7 Data analysis

(28)

28

the relationships between the individual variables. Thus, each hypothesis was tested separately. Hypothesis 1, hypothesis 2 and hypothesis 3 were tested with an ANCOVA. Linear regressions were performed to test hypothesis 4, hypothesis 5, hypothesis 6 and hypothesis 8 since they included continuous independent variables. The moderating effect of self-enhancement was analysed by means of a two-way ANOVA and Hayes PROCESS model 1 (Hayes, 2018). The mediating effects of brand credibility and ad credibility were tested with the Baron and Kenny method (Baron & Kenny, 1986) and with Hayes PROCESS model 6 (Hayes, 2018).

(29)

29 Table 3.7: Estimated models

Model 1 (H1) ANCOVA EW = eWOM

Dummy coded (1 is no social cause / 2 is social cause)

MC = Message content BC = Brand credibility AD = Ad credibility SE = Self-enhancement AA = Ad attitude ε = Error Term EW = β0 + β1MC + β2AA + ε Model 2 (H2) ANCOVA BC = β0 + β1MC + β2AA + ε Model 3 (H3) ANCOVA AC = β0 + β1MC + β2AA + ε

Model 4 (H4) LINEAR REGRESSION

AC = β0 + β1AA + β2BC + ε

EW = β0 + β1AA + β2BC + ε

EW = β0 + β1AC + β2AA + ε MULTIVARIATE 1

AC = β0 + β1MC + β2BC+ β3AA + ε

Model 7 (H7) MODERATION

Two-way Anova _{MULTIVARIATE 2}

EW = β0 + β1MC+ β2BC + β3AC + β4SE + β5AA + β6MC*SE + ε

EW = β0 + β1MC + β2SE + β3MC*SE + β4AA + ε

Hayes Model 1

EW = β0 + β1MC + β2SE + β3MC*SE + β4AA + ε

CONCEPTUAL MODEL – Hayes model 89

BC = β0 + β1MC + β2AA + ε

AC = β0 + β1MC+ β2BC + β3AA + ε

EW = β0 + β1MC+ β2BC + β3AC + β4SE + β5MC*SE +

β6AA + ε

EW = β0 + β1SE + β2AA + ε

MEDIATION (: Baron and Kenny

(30)

30

4. Results

4.1 Control variables

Before testing the hypotheses of the study, the effects of the control variables were analysed. The control variables were brand attitude, ad attitude and gender. An ANCOVA was performed for each control variable separately, in order to check the influence of the control variables on the dependent variable (eWOM) given the effect of the independent variable (message content). Brand attitude (p=0.466) and gender (p=0.455) were not found to be significant. Ad attitude showed a significant effect on eWOM (p=0.016). When all control variables were tested together, the same variable was significant. Therefore, ad attitude was included in subsequent analyses as a covariate.

4.2 Hypotheses testing

The collected data were tested with the Shapiro-Wilk test in order to verify the assumption of normal distribution of the data.All variables showed significant results on the Shapiro-Wilk test, which implied that a normal distribution cannot be assumed. However, the sample was large enough to be able to obtain significant results (Cohen, 1988). Indeed, each experimental condition had a sample including more than 30 respondents.

Lastly, the data were checked for multicollinearity. The VIF score was computed for all the variables. The VIF score should be below 4 in order to have no multicollinearity in the dataset (Malhotra, 2009). The VIF values for all the variables were below 4, indicating no

multicollinearity.

4.2.1 Hypothesis 1: CRM and eWOM

The first hypothesis predicts that cause-related branded content exerts a stronger influence on customers’ eWOM on social media platforms with respect to branded content without a CRM component. An ANCOVA was used to test the relationship between the independent variable, message content, and the dependent variable, eWOM, including ad attitude as a covariate. The Levene’s test of equality of variances is significant (p=0.002), and therefore equal variances are not assumed. The results of the ANCOVA can be found in Table 4.1. Message content was not found to have a significant effect on eWOM (F(1.66)=1.29, p=0.260, η2 =0.019), therefore H1 is rejected. However, the effect on eWOM can be explained by the control variable. Indeed, ad attitude showed a significant positive effect on the dependent variable (F(1.66)=6.08, p=0.016, η2 _{=0.084). This indicates that more positive attitudes towards the branded content lead to}

(31)

31

Table 4.1: Results of the ANCOVA with eWOM as dependent variable

Model 1 df S.S. M.S. F p ηp2 3 1 7.820 0.000 2.607 0.000 2.812 0.000 0.046 0.989 0.113 0.000 Corrected model Intercept Message content 1 1.197 1.197 1.291 0.260 0.019

Attitude towards the ad 1 5.636 5.636 6.080 0.016 0.084

Error Total Corrected total 66 70 69 61.180 69.000 69.000 0.927 R2_{=0.113 Adjusted R}2_=0.073

An additional ANCOVA was performed, excluding the control variable ad attitude to better understand the impact of message content. The model had a R2_{of 0,031 and adjusted R}2_of

0,016. The independent variable message content was again found to have a non-significant impact on eWOM (F(1.68)=2.15, p=0.148, η2 =0.031), confirming the rejection of H1.

4.2.2 Hypothesis 2: CRM and brand credibility

Hypothesis 2 predicts that cause-focused branded content positively influences brand credibility. An ANCOVA was conducted in order to verify the effect of message content on brand credibility, taking into account ad attitude.The Levene’s test of equality of variances is not significant (p = >0.05), and therefore equal variances are assumed. The results of the ANCOVA can be found in Table 4.2. Message content was not found to be significant with respect to brand credibility (F(1.66)=8.19, p=0.991, η2 =0.000). Thus, H2 is rejected. However, the control variable can explain the impact on brand credibility. Ad attitude had a positive significant effect on brand credibility (F(1.66)=29.71, p=0.000, η2 _{=0.310), indicating that more}

(32)

32

Table 4.2: Results of the ANCOVA with brand credibility as dependent variable

A second ANCOVA was conducted to verify the effect of message content on brand credibility when the control variable ad attitude is excluded. The model had a R2_{of 0,030 and adjusted R}2

of 0,018. The independent variable message content was not found to have a significant effect on brand credibility (F(1.68)=0.431, p=0.514, η2 =0.006), confirming the rejection of H2.

4.2.3 Hypothesis 3: CRM and ad credibility

Hypothesis 3 predicts that cause-related branded content will result in lower ad credibility compared to branded content without a CRM component. An ANCOVA was conducted to test the effect of message content on ad credibility, including ad attitude as a covariate. The

Levene’s test of equality of variances is not significant (p = >0.05), and therefore equal variances are assumed. The results of the analysis can be found in Table 4.3. Message content was not found to be significant (F(1.66)=0.894, p=0.348, η2 =0.013), therefore H3 is rejected. Again, the effect on ad credibility can be explained by the control variable. Indeed, ad attitude showed a significant positive effect on the ad credibility (F(1.66)=53.62, p=0.000, η2 _=0.448).

(33)

33

Table 4.3: Results of the ANCOVA with ad credibility as dependent variable

An additional analysis of variance (ANCOVA) was conducted, excluding the control variable ad attitude. The model had a R2_{of 0,033 and adjusted R}2_{of 0,019. The analysis confirmed that}

the effect of message content on ad credibility is not significant (F(1.68)=2.32, p=0.132, η2_{=0.033), thus supporting the rejection of H3.}

4.2.4 Hypothesis 4: brand credibility and ad credibility

Hypothesis 4 predicts that brand credibility positively influences ad credibility. A linear regression was performed to test the relationship between brand credibility and ad credibility, including ad attitude as a covariate. The model was found to be significant (R2 =0.534,

(34)

34

Table 4.4: Results of the linear regression with ad credibility as dependent variable

Model 4 B SE β t p

Attitude towards the ad 0.489 0.101 0.489 4.855 0.000

Brand credibility 0.334 0.101 0.334 3.318 0.001

R2 0.534

Adjusted R2 0.520

F-value 38.442

p-value 0.000

4.2.5 Hypothesis 5: brand credibility and eWOM

Hypothesis 5 predicts that brand credibility positively influences eWOM on social media platforms. A linear regression was used to test the effect of brand credibility on eWOM, taking into account ad attitude as a covariate. The model was found to be significant (R2 =0.102, F(2.67)=3.798, p=0.027). The results of the linear regression can be found in Table 4.5. Brand credibility was not found to be significant with respect to eWOM (B=-0.106, t=-0.756,

p=0.453). Thus, H5 is rejected. However, ad attitude was found to have a positive significant effect on eWOM (B=0.366, t=2.618, p=0.011), indicating that more positive attitudes towards the ad lead to greater eWOM effects.

Table 4.5: Results of the linear regression with eWOM as dependent variable

Brand credibility -0.106 0.140 -0.106 -0.756 0.453

R2 0.102

Adjusted R2 0.075

F-value 3.798

(35)

35

A second linear regression was performed to verify the effect of brand credibility on eWOM, excluding the control variable ad attitude. The model was found to be significant (R2 =0.100, F(1.68)=0.685, p=0.023). The independent variable brand credibility was not significant with respect to eWOM (B=0.100, t=0.828, p=0.411), confirming the rejection of H5.

4.2.6 Hypothesis 6: ad credibility and eWOM

Hypothesis 6 predicts that ad credibility positively influences eWOM on social media platforms. A linear regression was performed to test the relationship between the independent variable, ad credibility and the dependent variable, eWOM, including ad attitude as a covariate. The model was found to be significant (R2_{=0.096, F(2.67)=3.56, p=0.034). The results of the regression}

analysis can be found in Table 4.6. Ad credibility was not found to be significant (B=-0.057, t=-0.361, p=0.719), therefore H6 is rejected. However, ad attitude had a significant positive effect on eWOM (B=0.345, t=2.190, p=0.032), indicating that more positive attitudes towards the branded content generate greater eWOM effects.

Ad credibility -0.057 0.158 -0.057 -0.361 0.719

R2 _0.096

Adjusted R2 _0.069

F-value 3.555

p-value 0.034

(36)

36

4.2.7 H7: the moderating effect of self-enhancement on the relationship between CRM and eWOM

Hypothesis 7 predicts that self-enhancement strengthens the effect of CRM content on

consumers’ eWOM on social media platforms. Self-enhancement is considered as a moderator of the relationship between message content and eWOM. A two-way ANOVA was conducted to test the moderating effect of self-enhancement. In addition, ad attitude was included as a

covariate. The Levene’s test of equality of variances is significant (p=0.012), and therefore equal variances are not assumed. The results of the two-way ANOVA can be found in Table 4.7. The interaction effect between message content and self-enhancement was not significant (F(1.40)=2.235, p=0.856, η2 =0.020), therefore no moderation effect can be proved. Thus, H7 is rejected. The effects of the individual variables were then considered. The effect of message content on eWOM was not significant (F(1.66)=0.337, p=0.593, η2 _{=0.078). Ad attitude was not}

found to be significant with respect to eWOM (F(1.66)=5.805, p=0.074, η2 _{=0.592). However,}

(37)

37

Table 4.7: Results of the two-way ANOVA with eWOM as dependent variable

Model 7 df S.S. M.S. F p ηp2 65 1 68.622 0.039 1.056 0.039 11.178 0.417 0.015 0.553 0.995 0.094 Corrected model Intercept Message content 1 0.032 0.032 0.337 0.593 0.078 Self-enhancement Message content*Self-enhancement

Attitude towards the ad

63 1 1 60.918 0.042 0.548 0.967 0.042 0.548 10.238 1.002 5.805 0.017 0.856 0.074 0.994 0.045 0.592 Error Total Corrected total 4 70 69 0.378 69.000 69.000 0.094 R2_{=0.995 Adjusted R}2_=0.906

(38)

38

Table 4.8: Results of Hayes PROCESS model 1 with eWOM as dependent variable

Hayes Model 1 SE β t p Message content 0.220 0.255 1.158 0.251 Self-enhancement Message content*Self-enhancement

Attitudes towards the ad

0.401 0.230 0.125 0.445 -0.050 0.124 1.110 -0.215 0.999 0.271 0.830 0.322 R2 _0.218 F-value 4.534 p-value 0.003

4.2.8 Hypothesis 8: self-enhancement and eWOM

Hypothesis 8 predicts that self-enhancement positively influences eWOM on social media platforms. A linear regression was performed to test the relationship between self-enhancement and eWOM, including ad attitude as a covariate. The model was found to be significant

(R2=0.201, F(2.67)=8.450, p=0.001). the results of the linear regression can be found in Table 4.9. Self-enhancement showed a positive significant effect on eWOM (B=0.365, t=3.000, p=0.004). Thus, self-enhancement is indeed a motivation to engage in eWOM and H8 is supported. Moreover, ad attitude was not found to have a significant effect on eWOM (B=0.146, t=1.197, p=0.236).

Self-enhancement 0.365 0.122 0.365 3.000 0.004

R2 0.201

Adjusted R2 _0.178

F-value 8.450

(39)

39

4.3 The mediating effects of brand credibility and ad credibility

Hypothesis 1, 2, 3, 4, 5 and 6 predict a mediation effect of brand credibility and ad credibility on the relationship between message content and eWOM. The mediating effect was tested with two methods: the Baron and Kenny method (Baron & Kenny, 1986) and the Hayes PROCESS model 6 (Hayes, 2018).

4.3.1 Baron and Kenny

Baron and Kenny method consists of separate regressions (Baron & Kenny, 1986). Each regression tests a different relationship. Since this study has a model with two mediating variables, multiple mediating relationships exist. The first regression determines c, the total effect of the independent variable on the dependent variable. The second regression determines a1, the effect of the independent variable on the first mediator. The third regression determines

a2, the effect of the independent variable on the second mediator. The fourth regression

determines b1, the effect of the first mediator on the dependent variable. The fifth regression

determines b2, the effect of the second mediator on the dependent variable. Then c1 is the

remaining direct effects of the independent variable on the dependent variable. Finally, the sixth regression determines d, the effect of the first mediator on the second mediator. Complete mediation is present when the independent variable no longer influences the dependent variable after the mediators have been controlled (Baron & Kenny, 1986). Partial mediation occurs when the independent variable’s influence on the dependent variable is reduced after the mediators are controlled (Baron & Kenny, 1986). The independent variable is message content, the first

mediator is brand credibility, the second mediator is ad credibility and the dependent variable is eWOM. The results of the multiple regressions according to the Baron and Kenny method can be found in Table 4.10, with the respective figure.

Table 4.10: Results of multiple regressions

(40)

40 1. The effect of message content on eWOM (c)

Message content was not found to be significant with respect to eWOM (B=0.348, t=1.465, p=0.148). Therefore, the independent variable was not found to have a significant effect on the dependent variable and path c is not supported.

2. The effect of message content on brand credibility (a1)

Message content was not found to be significant with respect to brand credibility (B=0.158, t=0.657, p=0.514). Therefore, the independent variable was not found to have a significant effect on the first mediator and path a1 is not supported.

3. The effect of message content on ad credibility (a2)

Message content was not found to be significant with respect to ad credibility (B=0.362, t=1.524, p=0.132). Therefore, the independent variable was not found to have a significant effect on the second mediator and path a2 is not supported.

4. The effect of brand credibility on eWOM controlling for message content (b1)

Brand credibility was not found to be significant with respect to eWOM (B=0.087, t=0.720, p=0.474). Therefore, the first mediator was not found to have a significant effect on the

dependent variable and path b1 is not supported. In addition, the effect of message content was

not significant (B=0.334, t=1.398, p=0.167).

5. The effect of ad credibility on eWOM controlling for message content (b2)

Ad credibility was not found to be significant with respect to eWOM (B=0.150, t=1.240, p=0.219). Therefore, the second mediator was not found to have a significant effect on the

(41)

41

dependent variable and path b2 is not supported. In addition, the effect of message content was

not significant (B=0.294, t=1.221, p=0.226).

6. The effect of brand credibility on ad credibility (d)

Brand credibility showed a significant positive effect on ad credibility (B=0.609, t=6.327, p=0.000). Therefore, the first mediator was found to have a significant effect on the second mediator and path d is supported. This indicates that higher brand credibility leads to higher ad credibility.

The results of the multiple regressions indicate that no mediating effects can be assumed. Indeed, the effect of the independent variable message content on eWOM, the dependent variable, was not significant. If the direct effect between the independent and dependent variable is not supported, no mediation can be assumed (Baron & Kenny, 1986). However, the two mediators, brand credibility and ad credibility, do have a significant relation.

4.3.2 Hayes PROCESS model 6

The Hayes PROCESS model 6 with a number of 5000 bootstrap samples and 95% confidence intervals was used to test the mediating effects of brand credibility and ad credibility on the relationship between message content and eWOM, including ad attitude as a covariate.The models were found to be significant (p=0.000). The results of the Hayes PROCESS model 6 are presented in Figure 4.11.

Figure 4.11: Results of Hayes model 6

(42)

42

Path a1 was not significant, thus message content was not found to have a significant effect on

brand credibility (B=-0.002, t=-0.012, p=0.991). However, the effect on brand credibility can be explained by the control variable. Ad attitude has a significant positive effect on brand

credibility (B=0.562, t=5.496, p=0.000).

Path a2 was not significant, which means that message content was not found to have a

significant effect on ad credibility (B=0.173, t=1.031, p=0.306). Path d was found to be significant. Brand credibility has a significant positive effect on ad credibility (B=0.334, t=3.321, p=0.002). In addition, ad attitude showed a significant positive effect on ad credibility (B=0.477, t=4.698, p=0.000).

Path c was not found to be significant, indicating that the effect of message content on eWOM was not significant (B=0.272, t=1.155, p=0.252). Path b1 was not found to be significant. Thus,

the effect of brand credibility on eWOM was not significant (B=-0.092, t=-0.608, p=0.545). Path b2 was not found to be significant, which means that the effect of ad credibility on eWOM

(43)

43 Table 4.12: Results of Hayes model 6

Hayes model 6 B SE t p

dv = brand credibility

R2_{= 0.315, f = 15.411, p = 0.000}

Message content -0.002 0.203 -0.012 0.991 Attitude towards the ad 0.562 0.102 5.496 0.000

dv = ad credibility

R2 = 0.542, f =26.006, p =0.000

Message content 0.173 0.168 1.031 0.306

Brand credibility 0.334 0.101 3.321 0.002 Attitudes towards the ad 0.477 0.101 4.698 0.000

dv = eWOM

R2 =0.120, f =2.216, p = 0.000

Message content 0.272 0.236 1.155 0.252

Brand credibility -0.092 0.152 -0.608 0.545

Ad credibility -0.039 0.172 -0.229 0.820

Attitudes towards the ad 0.366 0.164 2.235 0.029

Indirect 1 = message content → brand credibility → eWOM Indirect 2 = message content → ad credibility → eWOM

Indirect 3 = message content → brand credibility → ad credibility → eWOM

4.4 Multivariate analysis

The first multivariate analysis was a linear regression testing the effects of message content, brand credibility and ad attitude as independent variables on ad credibility, the dependent variable. The model was found to be significant (R2 =0.542, F(3.66)=26.006, p=0.000). The

Indirect effects of X on Y

Effect BootSE BootLLCI BootULCI

(44)

44

results of the regression analysis can be found in Table 4.13. Message content was not found to be significant (B=0.173, t=1.031, p=0.306). Brand credibility was found to be significant (B=0.334, t=3.321, p=0.001), reflecting that branded content generating positive attitudes is perceived as more credible. Ad attitude had a significant positive effect on ad credibility (B=0.476, t=4.698, p=0.000), indicating that more positive attitudes towards the branded content lead to higher ad credibility.

Table 4.13: Results of the first multivariate analysis with ad credibility as dependent variable

Multivariate 1 B SE β t p Message content Brand credibility 0.173 0.334 0.167 0.101 0.087 0.334 1.031 3.321 0.306 0.001

R2 _0.542

Adjusted R2 0.521

F-value 26.006

p-value 0.000

The second multivariate analysis consisted of a linear regression testing the relationship between the independent variables message content, brand credibility, ad credibility, self-enhancement and ad attitude on the dependent variable eWOM, including the interaction term between message content and self-enhancement. The model was found to be significant

(45)

45

Table 4.14: Results of the second multivariate analysis with eWOM as dependent variable

Multivariate 2 B SE β t p Message content Brand credibility Ad credibility 0.266 -0.099 -0.066 0.224 0.144 0.166 0.134 -0.099 -0.066 1.189 -0.687 -0.400 0.239 0.495 0.691 Self-enhancement

Attitudes towards the ad

Message content*Self-enhancement 0.387 0.223 -0.024 0.197 0.166 0.235 0.387 0.223 -0.019 1.958 1.342 -0.104 0.045 0.184 0.918 R2 0.230 Adjusted R2 0.157 F-value 3.137 p-value 0.009

4.5 Testing the conceptual model

(46)

46 Table 4.15: Results of Hayes model 89

Hayes model 89 B SE t p

dv = brand credibility

R2_{= 0.315, F = 15.411, p = 0.000}

Message content -0.002 0.203 -0.012 0.991 Attitude towards the ad 0.562 0.102 5.496 0.000

dv = ad credibility

R2 = 0.542, F =26.006, p =0.000

Message content 0.173 0.168 1.031 0.306

Brand credibility 0.334 0.101 3.321 0.002 Attitudes towards the ad 0.477 0.101 4.698 0.000

dv = eWOM R2 =0.279, F =2.954, p = 0.000 Message content 0.236 0.226 1.043 0.301 Brand credibility -0.156 0.145 -1.075 0.287 Ad credibility -0.013 0.171 -0.078 0.939 Self-enhancement 0.443 0.403 1.101 0.048 Message content*Self-enhancement -0.048 0.231 -0.205 0.838 Attitudes towards the ad 0.200 0.168 1.195 0.237

(47)

47

Message content was not found to have a significant impact on brand credibility (B=-0.002, t= -0.012, p=0.990) nor on ad credibility (B=0.173, t=1.031, p=0.306). The effect of message content on eWOM was not significant (B=0.236, t=1.043, p=0.301). Univariate analyses,

specifically hypotheses 1, 2 and 3, reported that message content was not found to be significant with respect to brand credibility, ad credibility and eWOM. Multivariate analyses confirmed these findings. Therefore, hypotheses 1, 2 and 3 are rejected.

Brand credibility was found to have a significant positive effect on ad credibility (B=0.334, t=3.321, p=0.002). However, the effect of brand credibility on eWOM was not significant (B= -0.156, t=-1.075, p=0.287). Brand credibility was shown to have a significant effect on ad credibility and a non-significant effect on eWOM when conducting the univariate analyses testing hypotheses 4 and 5 respectively. These effects are then confirmed by multivariate analyses. Thus, hypothesis 4 is supported while hypothesis 5 is rejected.

Ad credibility was not found to be significant with respect to eWOM (B=-0.013, t=-0.078, p=0.939). When testing hypothesis 6, ad credibility was shown to have a non-significant effect on eWOM. The relationship between ad credibility and eWOM is then confirmed by

multivariate analyses, rejecting hypothesis 6.

Self-enhancement was found to have a significant positive effect on eWOM (B=0.442, t=1.101, p=0.048). This result is in line with univariate analysis testing hypothesis 8, which showed a significant effect of self-enhancement on eWOM. Multivariate analyses confirmed this finding, supporting hypothesis 8.

(48)

48

The effect of message content on eWOM was not significant, therefore no moderation nor mediation can be assumed given that the direct effect between the independent variable and the dependent variable is non-significant (Malhotra, 2009). In addition, the interaction term between message content and self-enhancement was not found to have a significant effect on eWOM (B=0.173, t=1.031, p=0.306), confirming the rejection of the hypothesis 7 of self-enhancement moderating the relationship between message content and eWOM. Moreover, the indirect effects of message content on eWOM through brand credibility and ad credibility were not significant, confirming the rejection of the mediation. The results of the indirect effects from the analysis of Hayes model 89 can be found in Table 4.16.

Table 4.16: Results of the indirect effects from Hayes model 89

Indirect 1 = message content → brand credibility → eWOM Indirect 2 = message content → ad credibility → eWOM

Indirect 3 = message content → brand credibility → ad credibility → eWOM

Indirect effects of X on Y

Effect BootSE BootLLCI BootULCI

(49)

49

5. Conclusion

5.1 Discussion

The aim of this study was to investigate the relationship between branded content and eWOM. In particular, the main research question was whether branded content with a CRM component generated greater eWOM effects than regular branded content on social media platforms (Facebook, Instagram, Twitter). Additional factors were considered as potential contributors to the relationship between cause-focused branded content and eWOM, namely self-enhancement, brand credibility and ad credibility. The results of the hypotheses testing are summarized in Table 5.1.

Table 5.1: Hypotheses testing

Hypotheses Results

H1: Cause-focused branded content exerts a stronger influence on customers’ eWOM on social media platforms with respect to branded content without a CRM component.

Rejected

H2: Cause-focused branded content positively influences brand credibility.

Rejected

H3: Cause-focused branded content will result in lower ad credibility compared to branded content without a CRM component.

Rejected

H4: Brand credibility positively influences ad credibility. Supported

H5: Brand credibility positively influences eWOM on social media platforms. Rejected

H6: Ad credibility positively influences eWOM on social media platforms.

Rejected

H7: Self-enhancement strengthens the effect of cause-focused branded content on consumers’ eWOM on social media platforms.

Rejected

H8: Self-enhancement positively influences eWOM on social media platforms.

Supported

(50)

50

content. Cause-related marketing is a strategic tool employed by companies to increase positive eWOM while minimizing consumer scepticism (Christofi, Vrontis, Leonidou, & Thrassou, 2020). Indeed, the involvement of the brand in a social cause goes beyond what customers expect from a firm and these delighted customers should respond with positive eWOM

(Thomas, Mullen, & Fraedrich, 2011). However, this study rejected these theories. One reason could involve how the cause-focused branded content was perceived by respondents. If

customers’ attention and expectations are not satisfied, no response is expected (Christofi, Vrontis, Leonidou, & Thrassou, 2020). Another potential crucial variable that was not considered in the current study is brand-cause fit. This factor describes the perceived

compatibility, in terms of fit and similarity, between the brand and the supported cause (Bigné-Alcañiz, Currás-Pérez, Ruiz-Mafé, & Sanz-Blas, 2012). Strategic fit between the brand and the cause can determine the ultimate success of the CRM initiative, allowing strong positive associations between the two actors and stimulating eWOM (Christofi, Vrontis, Leonidou, & Thrassou, 2020). The absence of this variable in the current study could help explain the non-significance of cause-focused branded content on eWOM.

Cause-focused branded content was not found to have a significant effect on brand credibility. The research conducted by Dean (2003) on CRM suggested long-term commitment to a cause as a crucial variable to understand the relation between cause marketing and consumer

responses. Firms that show a long-term relationship with specific social causes have been found to stimulate perceptions of brand credibility (Dean, 2003). Indeed, the long-term commitment has a positive effect on the general positioning and image of the brand, directly influencing its credibility (Dean, 2003). However, the participants of this study were exposed to a single piece of cause-focused branded content and a single exposure to a CRM initiative may have a lower impact on customers compared to a long history of the brand involvement with cause-focused marketing. Furthermore, another reason behind the non-significance of cause-focused branded content on brand credibility could be found in the complexity and strength of the brand

credibility variable (Lafferty, 2007). Corporate credibility includes judgements about the length of time in the industry, analyses of its financial situation and core business in addition to its social responsibility (Lafferty, 2007). The current study didn’t provide this type of information. Therefore, since customers tend to evaluate the credibility of a company from a broader

Branded content and cause-related marketing: success or failure for eWOM?

Branded content and cause-related marketing:

success or failure for eWOM?

Branded content and cause-related marketing:

success or failure for eWOM?

Abstract

Preface

Table of contents

1. Introduction

2. Theoretical framework

3. Methodology

4. Results

5. Conclusion