3
Management summary
4
Preface
After some stressful, but also very interesting months the day has finally come where I can hand in my master thesis. It was a time of ups and downs, however, I also learned a lot – both about my thesis topic as well as about myself. When I look back on the last one and a half years of my master, I feel really proud. I managed to work hard on the assignments, study for the exams, but I also used the time in Groningen to socialize with other students and I got to know the country and its people. When deciding to go to the Rijksuniversiteit Groningen, I knew it will be a tough and demanding study, but also a time of adventures and excitement. To take and master this challenge would not have been possible without my ambition and passion for marketing. I would like to thank my family and friends for their support and encouragement. They made the time when I was working on my thesis more pleasant and were able to help me when I got overloaded with work.9
1 Introduction
1.1 Background problem
In 2010, Nestlé Kit Kat’s palm oil scandal demonstrated how easily and sudden a brand can come under fire. After Greenpeace had posted a video on Youtube about the destruction of tropical rainforests in Indonesia, hundreds of thousands of people started to spread the campaign on their Facebook, Twitter and other social media profiles. Even profile pictures were changed to images of orang‐utans, rainforest and Greenpeace’s Kit Kat campaign logo (Greenpeace 2010). However, as if spreading the bad news is not enough, Nestlé even started to delete negative posts on its Facebook Fan page and banned people to further comment on the incident. Consequently, the company’s denial to react on the scandal led to even more critique which was then uttered via alternative social media.
Exemplarily, this scandal shows that the unified power of globally interconnected people via World Wide Web can dramatically facilitate the dissemination of negative word of mouth (WOM) (Shankar et al. 2003) which can be defined as informal person‐to‐person communication concerning a brand, product or company (Arndt 1967; Westbrook 1987). In this particular case it is called negative WOM since the information about Kit Kat has been highly negative and harmful to the brand.
Email referrals, online forums of users and newsgroups, as well as customer reviews encouraged by merchant web sites allow consumers to share their views, preferences, or experiences with others far more easily than ever before (De Bruyn and Lilien 2008; Trusov et al. 2009). Owing to the advent of these online resources and social networking sites, access to the opinions and recommendations of others has tremendously increased. Since people often form and express their attitudes in these social contexts, they are affected by the expressed uncertainty of others (Tormala et al. 2009; Karmarkar and Tormala 2009).
10
that the negative WOM is spread any further and ultimately leads to the damage of the brand and even company reputation (Keller 2008). Numerous authors analyzing negative WOM found that its impacts are manifold as it is a very credible and influential source of information (Bickart and Schindler 2001; Bone 1995; Crowley and Hoyer 1994; Nyer and Gopinath 2005; Sen and Lerman 2007). Owing to negative information, consumers will most likely switch to another brand (Andreason 1985; Bechwati and Morrin, 2003; Blodgett et al. 1993; Fornell and Wernerfelt 1987; Keaveney 1995; Richins, 1983; Singh, 1988, 1990; Smith et al. 1999; Westbrook 1980), the success of products and its sales will be influenced which will ultimately also affect a company’s performance (Berger et al. 2010; Godes and Mayzlin 2004), and in the long‐term the company’s reputation (Bowman and Narayandas 2001) will suffer as well as its profits (Goldenberg et al. 2007). Last but not least, companies with higher negative WOM will have diminished intangible assets such as reduced customer repurchase intention, increased defection rates of existing customers, and inhibited new customer acquisition efforts, thus leading to less robust expected sum of cash flows in the future (Gupta and Zeithaml 2006; Luo 2009; Rust et al. 2004).Over the past several years, different variables such as brand familiarity, commitment toward the brand, product type and prior brand attitude have been tested for their moderating role on the impact of negative WOM. Yet, the credibility of online platforms as moderator has been largely neglected despite its irrefutable importance in times of online WOM.
Compared to “offline” word of mouth which is characterized by oral person‐to‐person communication, the World Wide Web offers people the possibility of stating opinions, experiences and information in an anonymous way. This anonymity makes it difficult for consumers to determine the quality and credibility of the eWOM based on the characteristics of the communicators (Lee and Youn 2009), especially when the information is negative. Due to the given complexity, consumers tend to find other clues in order to make causal inferences about the communicator’s intention and to be able to judge the trustworthiness of the posts. One of the cues consumers use in evaluating negative eWOM is the online platform on which the information is posted (Senecal and Nantel 2004). Hence, it is expected that the credibility of the online platform which contains the negative eWOM plays a moderating role on the resulting impact of negative eWOM on consumers’ attitude formation and decision‐making.
11
These gaps in the WOM literature contrast sharply with the reality of decision‐making where potential and actual consumers rely on other customers as source of information. Therefore, the objective of the present study is to provide an understanding of how consumers react to negative product‐related information about brands they are familiar with depending on the online platform they learn about it.
Even though companies can monitor negative comments which are articulated by consumers via the internet and even conduct marketing research in online communities (Chatterjee et al. 2003; Henning‐Thurau and Walsh 2003; Kozinets 2002; Park and Fader 2004), the vast amount of possible online platforms makes it almost impossible for companies to react in time on negative WOM in order to prevent other consumers to read and respond on it or even to spread it further.
1.2 Problem statement
Albeit negative WOM about brands is widely prevalent in the online environment, there has been no systematic investigation of how consumers process negative information about the brands they are familiar with, depending on the online platform they read about it. Thus, in this research, I attempt to bridge this gap by examining if the impact of negative WOM on consumers’ decision‐making will be moderated by the credibility of the platform. Hence, the problem statement can be formulated in the following way:
“How does the credibility of the online platform where the negative eWOM is posted impact brand attitude and consequently purchase intention?”
1.3 Research questions
12
1.4
Theoretical and managerial relevance
.This study contributes to the WOM literature by examining how the online platform’s credibility moderates the impact of negative eWOM on brand evaluations, namely brand attitude and purchase intention. In other words, it highlights the consequences on consumer behavior and ultimately on firm performance when not reacting at all or replying too late to consumer’s critique, questions and problems.
Nowadays consumers do not anymore ask only family and friends for advice before purchasing a certain brand; they also search for information online and consult the opinion of other people who already gained experience with the product. Due to the visibility of consumers’ posts on the internet, other consumers can easily come across negative reviews during their search, thus their purchase decision as well as their image of the brand might be affected. Moreover, besides potential consumers reading negative posts and deciding not to buy the product, a far more serious problem is the fast and international dissemination of negative information facilitated by the World Wide Web. The ramifications of this global interconnectivity range from trivial to severe: people living in Canada learn that Europeans can chose between far more flavors and complain to the company, consumers get affected by much more voices due to the vast amount of available information, and frustrated and disappointed consumers can mobilize mass audiences to boycott a certain brand. Therefore, it is highly essential for brand managers to undertake any necessary measure and adequately react in time on negative eWOM.
The platforms in the study only contain consumer reviews and do not display any reaction on the part of the company or brand on the criticism. Hence, the consequences on brand attitude and purchase intention as a result of not responding are illustrated in an unbiased manner. In addition, by determining which online platform consumers perceive to have the highest credibility and information trustworthiness, the present research gives marketers at hand where to respond first to harmful eWOM in order to prevent its further spread. Furthermore, the study provides essential information for brand managers in regard to the content and layout of a brand community or other marketer‐generated site by investigating how the credibility of a post by an individual person and the credibility of a web site itself account for the credibility of the platform. Moreover, it facilitates the discussion about financial means for online marketing by highlighting the irrefutable power of marketer‐administered sites particularly with regard to consumers’ brand evaluations. Last but not least, the study paves the way for further research on how companies should monitor both, their own and competitors’ reputation online, and how they should ideally react on negative eWOM depending on the platform where the harmful
information is posted.
13
1.5 Structure of the thesis
This paper is organized in five chapters. After the introduction, the next chapter will present the literature review based on existing articles on the topic at hand. Moreover, besides the conceptual model which shows the relationships of the variables being investigated, chapter 2 will also contain the derived hypotheses from theory. After having established the basis for the empirical testing, the research method and plan of analysis will be described in chapter 3. In chapter 4 the results will be presented and discussed. Finally, conclusions will be drawn and possible limitations and recommendations will be presented in chapter 5.
14
2 Theoretical framework
2.1 Word of mouth
2.1.1 The body of WOM literature
Numerous studies have shown that WOM communication plays a significant role in consumers’ decision‐making as well as post‐purchase perceptions (Allsop et al. 2007; Bone 1995; Bowman and Narayandas 2001; Engel et al. 1969; Gershoff et al. 2007; Herr et al. 1991; Katz and Lazarsfeld 1955). Moreover, recent research has explored the impact of WOM on sales and firm value (Luo and Homburg 2007), on customer acquisition (Brown et al. 2005; Kumar and Krishnan 2002; Ryu and Feick 2007; Wangenheim and Bayon 2007), employee recruitment (Collins and Stevens 2002; Van Hoye and Lievens 2009), and on maintaining customers (Gremler and Gwinner 2008; Jones and Reynolds 2006; Maxham and Netemeyer 2002, 2003). Additional studies have demonstrated the power of WOM over other forms of advertising. In certain circumstances, WOM has a greater influence over consumer behavior than personal selling, print advertisements, and radio (e.g. Engel et al. 1969; Katz and Lazarsfeld 1955). The reasons for the outperformance of WOM compared to traditional communication are manifold. For instances, due to perceived source reliability and the flexibility of interpersonal communication, WOM has a unique ability to influence consumer decisions (Day 1971; Engel et al. 1969; Richins 1983; Tybout et al. 1981). Other factors which influence the impact of WOM on brand evaluations are described in the next section.2.1.2 Factors influencing WOM
Weak vs. strong ties While much of the existing WOM literature highlights the fact that WOM is effective due to the strong social relationship between sender and receiver, Brown and Reingen (1987) showed that weak ties display a bridging function, allowing information to pass from one group to another. However, they also found that strong ties are perceived as more influential than weak ties, and that they were more likely to be utilized as sources of information. Besides, compared to strong‐ tie sources which are more powerful, weak tie sources are not limited to the social group of the decision‐maker, thus they are more numerous and more diverse. Hence, the likelihood to find more information and better expertise among weak tie sources is greater (Duhan et al. 1997).
15
Unsolicited vs. solicited WOM
Bansal and Voyer (2000) studied the relative impact of solicited and unsolicited WOM. They found that solicited WOM has more impact. However, they did not take the effect of other variables, such as the strength of ties (weak vs. strong) into account. Thus, the impact of solicited WOM could have occurred if people ask for advice more often from those they are close to (East et al. 2008). Fitzsimons and Lehmann (2004) discovered that since a decision‐maker will most likely increase the attention to and weight on recommendations provided by credible sources (e.g. an expert in this area), an unsolicited advice will be viewed as more of a threat or intrusion.
WOM extremity
It seems likely that the extremity of WOM directly affects brand attitude, thus purchase intention. More specifically, an extremely negative statement is generally assumed to have a higher impact on consumer’s decision‐making. Nevertheless, this seems to be only true for low commitment consumers. High commitment consumers who process information about a brand which is inconsistent with their attitudes will engage in reactance and defense motivation (Fitzsimons and Lehmann 2004; Pomerantz et al. 1995). This view stems from the elaboration likelihood model (ELM) which explains the processes that are responsible for changing attitudes and for enhancing the strength of attitudes. The likelihood of elaboration is influenced by the individual’s motivation and ability to process information. Thus, the ELM helps to explain the reaction of consumers to product reviews by focusing on the information processing procedures that consumers follow (Petty et al. 1983).
Consumer characteristics
Researchers who investigated the relationship between expertise and WOM, and who focused on the consumers’ ability to process WOM messages, found that the effects of WOM are stronger when consumers have high expertise (Johnson and Russco 1984; Punj and Staelin 1983). On the contrary, when the scope of the study was on consumers’ motivation to process WOM information, the relationship between expertise and WOM was negative (Bloch et al. 1986; Gilly et al. 1998). Besides, Adaval (2001) discovered that information which is similar in valence with a person’s mood is weighted more heavily in product judgments. Moreover, when the brand is unfamiliar, the negative information elicits more supporting arguments and is perceived to have a higher diagnostic value and weight. On the other hand, when the brand is familiar, there are no significant differences in the impact of positive and negative information (Ahluwalia 2002). The moderating role of brand familiarity was already detected in earlier studies analyzing the impact of WOM on brand evaluations (Sundaram and Webster 1999; Wilson and Peterson 1989).
16
brand attitude tend to insulate and consumers with high prior attitude are likely to exacerbate when negative publicity matches. Furthermore, Clarkson et al. (2008) found that increasing attitude certainty strengthens univalent attitudes, i.e. the resistance to persuasion is stronger,
whereas ambivalent attitudes get weaker, and thus, their resistance to persuasion is lower.
2.1.3 Negative vs. positive WOM
As previously defined, WOM can be positive as well as negative. According to Herr et al. (1991), negative attributes strongly imply membership in one category (e.g. low quality) to the exclusion of others, whereas positive or neutral attributes are more ambiguous with respect to category membership. Besides, East et al. (2008) found that positive WOM is more common than negative WOM because there are more people producing positive WOM. However, compared to its positive counterpart, people place more weight on equally extreme negative WOM in forming overall product evaluations, which is called the negativity effect (Ahluwalia et al. 2000; Chevalier and Mayzlin 2006; Fiske 1980; Ito et al. 1998; Mittal et al. 1998; Mizerski 1982). Hence, negative information tends to be more diagnostic, attention getting and influential than positive or neutral information (Ahluwalia 2002; Herr et al. 1991; Wangenheim 2005). Nonetheless, this seems to be only accurate for low‐commitment consumers since high‐commitment consumers extensively counter‐argue the negative information while supporting the positive information (Ahluwalia et al. 2000). Laczniak et al. (2001) found that consumers became even more committed to a brand that was subject to negative comment which has also been observed by Fitzsimons and Lehmann (2004). As a result the counterarguments generated emerged as a significant mediator of attitude change for the particular group of consumers.
Furthermore, Berger et al. (2010) found that whereas a negative review in the New York Times hurt sales of books by well‐known authors, it increased sales of books that had lower prior awareness, i.e. which were unfamiliar to the consumers. Thus, negative publicity can under specific conditions even increase purchase likelihood and sales. Moreover, East et al. (2008) uncover that for familiar brands the impact of positive WOM on brand purchase probability is generally greater than negative WOM. They further find that respondents resist negative (positive) WOM on brands they are very likely (unlikely) to choose.
17
2.1.1 Traditional WOM vs. online WOM (eWOM)
Word of mouth communication in its traditional way can be defined as oral, informal and non‐ commercial person‐to‐person communication directed at other consumers about the ownership, usage or characteristics of particular brands, goods or sellers (Arndt 1967; Westbrook 1987). Additionally, WOM has been described as short‐lived because it vanishes as soon as it is uttered since it occurs in a spontaneous manner and then disappears (Stern 1994). However, due to the proliferation of the internet, word of mouth has changed in its very definition (Breazeale 2009; Dellarocas 2003). Online WOM differs from its “offline” counterpart in several significant ways: 1) the referrals are electronic by nature, i.e. there is no face‐to‐face communication, thus it often occurs between people who have no prior relationship with each other, 2) those referrals are usually unsolicited, that is, they are sent to recipients who are not looking for information, and hence are not necessarily willing to pay attention to them, 3) their exceptional scale compared to the limited number of consumers in the offline setting, achieved through the exploitation of the internet’s low‐cost, bidirectional communication capabilities, 4) the volatile and anonymous nature of online identities and the almost complete absence of contextual cues that would facilitate the interpretation of what is, essentially, subjective information (De Bruyn and Lilien 2008; Dellarocas 2003; Sen and Lerman 2007; Weiss et al. 2008). In summary, eWOM can be defined as any positive or negative statement made by potential, actual or former consumers about a product or company, which is made accessible to a multitude of people via the internet (Hennig‐Thurau et al. 2004). Owing to the effortless and unlimited detection of eWOM by means of Google and its long‐lived character compared to the fast fading of traditional WOM which is constrained by the receiver’s social circle (Duhan et al. 1997), eWOM bears a higher risk for brand managers.
The growth and evolution of the internet has highly facilitated the conduction of eWOM research, thus, more and more studies are dealing nowadays with online WOM instead of traditional WOM. Nevertheless, it was found that both types have similar impacts on consumer behavior and are influenced by the same factors, however, sometimes in a different degree or direction. Research has shown that not only WOM, but also eWOM has a significant effect on consumers’ decision‐making (Hennig‐Thurau and Walsh 2003) as well as on sales and firm value (Chevalier and Mayzlin 2006; Duan et al. 2008; Godes and Mayzlin 2004; Luo 2007; Villanueva et al. 2008; Zhu and Zhang 2010). Moreover, the power of word of mouth over advertising was also found in the online environment (Goldsmith and Horowitz 2006; Ratchford et al. 2007).
18
greater than the impact of a positive one. Nevertheless, Liu (2006) shows that both negative and positive WOM increase box office revenue. As with traditional WOM, variables such as prior awareness and familiarity with the product or brand highly influence whether positive or negative eWOM has a greater impact on sales. Last but not least, the negativity effect, i.e. the greater weighting of negative information holds also true for eWOM. However, besides the communalities between traditional and online WOM, several differences can be discovered in regard to the factors which impact them.
Whereas the traditional WOM literature emphasizes the fact that strong tie relationships between sender and receiver makes WOM so effective, Steffes and Burgee (2009) find that eWOM passed from virtual strangers (i.e. weak tie relationships) can be equally or more preferred than some strong tie information sources. Weak ties tend to be members of more groups, thus their unified strength leads to the dissemination of negative, thus harmful WOM through networks (Goldenberg et al. 2007). The role of weak versus strong tie relationships will be discussed in more detail in section 2.3.
19
attitudes as multiplicative function of 1) the salient beliefs that consumers have about the brand, i.e. the degree to which consumers perceive the brand to possess certain attributes or benefits, and 2) the evaluation regarding those beliefs. Hence, overall brand attitudes depend on the strength of association between the brand and salient attributes or benefits and the favorability of those beliefs (Keller 2008).
According to previous research, brand attitude can be measured as the sum of four items, namely, “this brand is likeable”, “this brand is high quality”, “this brand is good” and “this brand is pleasant” (Gardner 1985; Gresham et al. 1984; Mitchell and Ohlsen 1981). Alpert and Kamins (1995) distinguish between an overall attitude measurement scale and a multiple‐attribute scale comprising items such as quality, reliability, superiority, distinctiveness, sophistication and trustworthiness.
2.2.2 Purchase intention
In consumer behavior literature, numerous studies demonstrate that a significant positive relationship exists between attitude toward a brand and intention to buy that brand (Brown and Stayman 1992; Homer 1990; Laroche et al. 1996; MacKenzie et al. 1986). Since consumers' brand attitude impacts their confidence in evaluating the brand, it will consequently influence their purchase intention (Teng and Laroche 2007).
According to Laczniak et al. (2001), purchase intention can be measured with only one particular question, namely, “If you were to buy a product from a certain category, how likely are you to purchase this particular brand?” in combination with Likert scale. It was further considered to use adjusted items from Doney and Cannon (1997) and Zeithaml et al. (1996) in the later questionnaire, even though these measures were used in a B2B and service context (e.g.
my company will consider the focal company the first choice from which to buy
products/services).
2.2.3 Overview of most relevant studies
Table 1 summarizes the studies about WOM on consumers’ brand evaluations. The prerequisites for inclusion in the list have been that 1) the article steams from at least a 3 (e.g. Journal of
Consumer Psychology, Journal of Retailing) or 5 point journal (e.g. Journal of Marketing, Marketing Science, Journal of Consumer Research), and 2) it is about the impact of WOM on brand
20
Author(s)
Purpose of study
Moderator(s)
Sample
Findings
Ahluwalia et al. (2000)
Provides an understanding of how consumers react to negative product‐related information about brands they like and use.
Commitment toward the brand
‐ 2 (high vs. low commitment of the consumer toward the brand) x 2 (valence of the publicity information: positive and negative) between‐ subjects design / 2 (low vs. high commitment) x 2 (response strategy: counter‐argumentation and diagnosticity) between‐subjects design
‐ Size: 25 / 71 / 70 students
‐ Products: athletic shoes (Nike vs. unknown brand) / Mizuno athletic shoes and cameras
‐ Country: US Low (high) commitment consumers give more weight to negative (positive) than positive (negative) information, because they perceive it to be more diagnostic.
High‐commitment consumers extensively counter‐argued the negative information while supporting the positive information.
Low commitment consumers exhibit greater attitude change due to negative publicity.
Berger et al. (2010)
Examines whether negative publicity can increase purchase likelihood and sales by increasing product awareness.
Brand awareness
‐ Econometric analysis and experimental methods
‐ Weekly national sales for 244 hardcover fiction titles that were released from 2001 to 2003 and reviewed by the New York Times Whereas a negative review hurt sales of books by well‐known authors, it increased sales of books that had lower prior awareness.
21
Author(s)
Purpose of study
Moderator(s)
Sample
Findings
East et al. (2008)
Analyze the impact of positive and negative WOM on purchase probability.
Prior probability of purchase
‐ Questionnaires were
distributed in several ways (drop‐off, distribution via friends)
‐ Size: 2544 members of public
‐ Products: 15 categories
‐ Country: UK and France
The impact of positive WOM is greater than negative WOM.
Positive WOM tends to increase purchase probability more than negative WOM reduces it.
The impact of WOM is strongly related to the pre‐WOM probability of purchase, the strength of expression of the WOM, and whether the WOM is about the respondent's preferred brand.
Respondents resist negative WOM on brands they are very likely to choose, and resist positive WOM about brands they are very unlikely to choose.
Herr et al. (1991)
Analyzes the effect of WOM and product‐attribute information on persuasion.
Information vividness, type of evidence processing (anecdotal vs. detailed attribute information)
‐ 2 (WOM or printed anecdote) x 2 (positive or negative anecdote) factorial design / 3 (positive,
neutral, or negative attributes) x 2 (impression or memory set) x 2 (positive or negative WOM) x 2
(judgment before recall or vice versa) factorial design
‐ Size: 48 / 120 students
‐ Products: PC / car
‐ Country: US WOM communications have a greater impact on product judgments than less vivid information.
The judgmental effects of accessible WOM information are reduced when anecdotal information is available since consumers trust their own opinions more than they trust others.
22
Author(s)
Purpose of study
Moderator(s)
Sample
Findings
23
Author(s)
Purpose of study
Moderator(s)
Sample
Findings
Senecal & Nantel (2004)
Investigates consumers’ usage of online recommendation sources and their influence on online product choices.
Nature of the product, the web site on which the recommendation is proposed, and the type of recommendation source (recommender system, human experts, and other consumers)
‐ 3 (web sites)× 4 (recommendation sources) × 2 (products)
‐ Size: 487
‐ Products: Computer mice, calculator, red wine
‐ Country: US
Subjects who consulted product recommendations selected recommended products twice as often as subjects who did not consult recommendations.
The recommender system is the most influential recommendation source even if it was perceived as possessing less expertise than human experts and as being less trustworthy than other consumers.
Recommendations for the experience product were significantly more influential than for the search product. The type of web site did not affect their perceived trustworthiness and did not influence consumers’ propensity to follow the product recommendation.
Zhu & Zhang (2010)
Examines how product and consumer characteristics moderate the influence of online consumer reviews on product sales.
Product and consumer characteristics
‐ Analyzing data from a leading market research firm tracking console and game sales
‐ Size: 220 game titles
‐ Product: Video games
‐ Country: US
Online reviews are more influential for less popular games and games whose players have greater internet experience.
Moreover, online reviews are more influential when consumers have relatively greater internet experience.
Table 1: Summary of WOM studies
24
2.3 Source credibility
Due to the fact that the word of friends and close acquaintances enjoys a higher credibility compared to complete strangers (Brown and Reingen 1987; Duhan et al. 1997), traditional WOM is characterized as being credible and highly trustworthy. On the contrary, eWOM lacks this essential characteristic, i.e. consumers do not anymore necessarily know the people on whose posts they might base their decisions to purchase a product or not. Owing to the anonymity of information the question arises whether the information is really unsolicited or whether companies or competitors are posting in disguise as a consumer (Dellarocas 2006; Mayzlin 2006; Shimp et al. 2007). Furthermore, the anonymous nature of eWOM makes it difficult for consumers to determine the quality and credibility of online posts (Chatterjee 2001). Hence, the concern of source credibility and information trustworthiness occurs. Source credibility refers to the extent to which consumers perceive the source to 1) possess expertise relevant to the topic of communication and 2) to give an objective opinion on the topic (Ohanian 1990). Source credibility has been widely examined in advertising and marketing studies to delineate consumers’ purchase decision‐making processes (Pornpitakpan 2004). Moreover, marketers started to influence eWOM by compensating consumers to review products and even posting their own reviews about their products (Biyalogorsky et al. 2001; Chatterjee 2001; Godes and Mayzlin 2009). Due to the mixture of unbiased as well as promotional activity, the consumer is not able to distinguish the advertising from the consumer‐generated content (Mayzlin 2006). Thus, to overcome this conflict they tend to use other cues to make causal inference about the poster’s intention, one of them being the platform on which the eWOM is posted (Senecal and Nantel 2004).
25
under certain conditions. In particular, when people generate primarily positive thoughts in response to a message (e.g., because it contains strong arguments) and then learn of the source, high source credibility leads to more favorable attitudes than does low source credibility. However, when people have primarily negative thoughts in response to a message (e.g., because it contains weak arguments), high source credibility leads to less favorable attitudes than does low source credibility.
Studies which examined why certain personal sources of information exert more influence than others have identified factors such as source expertise (Bansal and Voyer 2000; Gilly et al. 1998), tie strength (Brown and Reingen 1987; Frenzen and Nakamoto 1993), demographic similarity (Brown and Reingen 1987), and perceptual affinity (Gilly et al. 1998) as important antecedents of WOM influence. Moreover, Bart et al. (2005) compared the determinants of online trust across different web site categories and consumers. They found that navigation and brand strength are strongest for information‐intensive sites such as brand communities (e.g. foodtv.com). Besides, Fogg et al. (2001) investigated web site characteristics that constitute online credibility based on a large‐scale survey. Their conclusion was that real‐world feel, ease of use, and expertise are among the most influential web site elements in boosting the credibility of a site. Furthermore, Schlosser et al. (2006) found that consumers trust the information contained on web sites that look like they required a high degree of investment to create. In summary, the type, content and source of a platform as well as its characteristics impact its overall credibility. Thus, the impact strength of negative eWOM on brand evaluations will depend on the credibility of the platform where it is posted. Based on the described features which make a web site credible, a measurement scale will be developed comprising also parts of Bart et al.’s (2005) trust items (e.g. “my overall believability of the information on this site is high”).
The next section will place the previous described variables into a conceptual model from which
26
2.4 Conceptual model
A conceptual research framework incorporating the independent, dependent and moderating variables described above is shown in figure 1 and subsequently described in more detail.
Figure 1: Conceptual model Independent and dependent variable Even though both positive and negative WOM impact brand evaluations such as brand attitude and purchase intention, the conceptual model only comprises negative eWOM, since the focus of this research lies on analyzing the impact of harmful and critical information about the brand on consumer decision‐making (Laczniak et al. 2001; Richins 1983). The underlying reasons for the consideration of only the negative valence of eWOM are that 1) negative information is more diagnostic and informative and has a stronger influence on consumers’ decision‐making than positive information (Arndt 1967; Herr et al. 1991; Maheswaran and Meyers‐Levy 1990), 2) marketers will most likely not be at the bottom of harming their brands; thus the assumption that the negative WOM is generated by consumers can be made, and 3) brand managers urge to react on or even prevent negative posts about their brands is of enormous interest.
Moderators
27
independent review web sites, aspire to help consumer to make informed buying decisions. In other words, the source of the platform plays a significant role in the perception of its credibility. The second component is content which refers to the information on the platform being rather negative, neutral or positive and the communicators being experts in the field or just average consumers. Thus, the manipulation of platform credibility will be accomplished by using three platforms which differ in regard to their origin and their content.
According to Ahluwalia et al. (2000), commitment of the consumer toward the brand is a moderator of negative information effects. As previously mentioned, some studies found that negative WOM about a brand to which consumers are highly committed leads them in a state of
reactance and defense (Ahluwalia et al. 2000; Fitzsimons and Lehmann 2004; Pomerantz et al.
1995). Nevertheless, the analysis of brand commitment as a moderator shall not be the focus of the present study, but rather platform credibility as illustrated above (see figure 1). Yet, since it is assumed that low commitment consumers will exhibit a greater brand attitude change in response to negative eWOM compared to high commitment consumers, few questions will be included in the questionnaire to control for this effect in the later analysis.
2.5 Hypotheses
Several authors have investigated the impact of negative information on brand evaluations (see table 1). Whereas some studies concentrated more on the impact on brand attitude, product judgment and choices (Bone 1995; Monga and John 2008; Senecal and Nantel 2004), others focused on the effect on purchase intention (Berger et al. 2010; East et al. 2008) or both (Pullig et al. 2006). Even though they were all analyzing a comparable relationship, due to the different moderators they used, a more or less significant impact of WOM on consumer behavior could be drawn. Thus, the purpose of the present study is as well to test the negative eWOM – brand evaluations relationship. Moreover, numerous studies demonstrated that attitude toward a brand positively impacts consumers’ confidence in evaluating it, therefore, increases the likelihood of purchasing the brand (Brown and Stayman 1992; Homer 1990; Laroche et al. 1996; MacKenzie et al. 1986; Teng and Laroche 2007). Thus,
H1: Compared to neutral eWOM, negative eWOM will have a more negative impact on brand attitude and purchase intention. Lee and Youn (2009) investigated whether and how different online platforms on which eWOM communication is posted influence consumers’ judgments of reviewed products. Their results
show that
participants exposed to the review posted on a personal blog were more likely to
28
29
positive attitude changes toward the position advocated (Sternthal et al. 1978), i.e. compared to a lowly credible source, negative WOM from a highly credible source will lead to a stronger brand attitude change. Moreover, Griffin et al. (1991) discovered that negative publicity has a diminishing effect on brand attitude, when the source credibility is high, however, they could not detect a significant effect on purchase intention. In summary, in order to evaluate the credibility of a platform it is essential to consider both, information content and source, i.e. the actual individual who posts a review and the web site itself, since both contribute to the overall value, thus credibility of that information (Brown et al. 2007).
Under normal conditions with brand familiarity and commitment being equal, negative eWOM will influence brand attitude, thus purchase intention more when consumers perceive high credibility of the posts compared to when consumers perceive low credibility. Moreover, negative posts are supposed to enhance a platform’s credibility since it is impossible that everything is always positive and perfect (Lee and Youn 2009). A brand community as such, i.e. without the consideration of the posts on it, is assumed to have low credibility since a marketer has interest to show a positive image and probably will remove negative posts or prevent posting them on the site. Thus, when negative and credible posts appear on a marketer‐ generated platform, this will lead to a reverse effect resulting in the information on the site to become more credible. Hence,
H3: Compared to web sites with high credibility, web sites with low credibility will increase the credibility of negative eWOM. Consequently, the effects of negative eWOM on brand attitude and purchase intention will be stronger.
The next chapter will describe the research design and the creation as well as conduction of the survey in more detail. Moreover, the plan of analysis for the hypotheses testing will be presented.
30
3 Research design
3.1 Choice of research
In order to analyze the problem statement “how does the credibility of the online platform where the negative WOM is posted impact brand attitude and consequently purchase intention?” a 2 (neutral and negative eWOM) x 3 (type of platform: social media, third party provider and marketer‐provided platform) between‐subjects factorial design was implemented. Due to issues of complexity and possible dropout rates, a brief questionnaire was developed to guarantee that the response rate is satisfactory. Instead of giving the participants to evaluate each of the three platforms, they only had to judge one valence scenario (neutral or negative) on one of the three platform types. Owing to its simplicity and feasibility, an online survey was chosen as the research method and except for the feedback question at the end all questions were 5‐point Likert scales which allowed for maximal one answer. Besides reducing the complexity for the respondents, the advantages of this survey method are that data is uncomplicated to code and results can be easily interpreted.
3.2 Choice of sample
In a similar study, Lee and Youn (2009), analyzed how different online platforms (personal blog, brand’s web site, independent review web site) on which eWOM communication is posted influence consumers’ judgments of an apartment. Their study showed a first attempt to examine how different online platforms impact consumer’s perception of a product. However, their research was limited in several points: 1) only one category was analyzed, 2) a student sample was used and 3) the survey was conducted in a single country.
The present study compensated these liabilities by sending the questionnaire not only to international students, but also to the working population of at least two countries, namely Germany and the Netherlands. The broader sample is also assumed to increase the generalizability of the results.
3.3 Choice of platforms
31
media to third party provided. Thus, an online brand community, a Facebook page of the brand and an independent review web site were chosen as platforms to be investigated. The following section provides information about each of the three different platforms that were used to manipulate source credibility in more detail. Online brand communityAccording to Muniz and O’Guinn (2001, p. 412), a brand community can be defined as “a specialized, non‐geographically bound community, based on a structured set of social relations among admirers of a brand”. Moreover, in their study they find that brand communities have three things in common: shared consciousness, rituals and traditions, and a sense of moral responsibility. Compared to Muniz and O'Guinn (2001) who envision a brand community as a customer‐customer‐brand triad, McAlexander et al. (2002) hold the opinion that a brand community is customer‐centric. In particular, they describe a brand community as interplay of several relationships: customer‐brand, customer‐firm, customer‐product and customer‐ customer. Consequently, besides the exchange of information between consumers, the company respectively the brand managers can also take advantage of the brand community to enhance the relationship with their customers. In summary, the most important characteristic compared to other platforms is that an online brand community is provided and hosted by the marketers of the brand. Thus, they can check posts before placing them online and even delete comments. The company uses the online community to get feedback and generate conversations about certain topics. This generated information is then used by the brand managers to base their decisions on and it also helps them to better understand consumers’ needs. Facebook pages
Facebook provides so called pages where companies, artists and celebrities can connect with their audience, share their story and participate in real‐time conversations quickly and easily. Therefore, Facebook provides an assortment of applications and features such as discussion boards, events and a like button. Registered Facebook members can post videos, photos and of course participate in discussions, state their opinions, utter critique and exchange their experiences with other consumers on the Facebook pages which are provided by Facebook, however hosted and administered by the company. Thus, similar to the brand community, the company has control over the information on the site and can use the Facebook page for selling intents.
32
Independent review web sites
Independent review web sites are privately owned sites provided by third parties. Thus, they are supposed to be independent of marketers’ selling intents (Lee and Youn 2009). Independent review web sites are communities of people who share honest and impartial reviews of their product and service experiences with the goal of helping consumer to make informed buying decisions. They are not limited to a certain category or specialized in only a few brands or even in one particular brand. On the contrary, the reviews combine information about any product or service ranging from mobile phones over food to holidays. Consequently, independent review web sites are considered to be more trustworthy and credible than marketer‐generated or administered pages like the two platforms described above.
In summary, it is expected that the brand community will have the lowest, whereas the independent review web site will have the highest source credibility. Facebook is assumed to fall
in between these two platforms.
3.4 Choice of brand
The purpose of this study was not to test the purchase intention of a new product for its innovation diffusion, (e.g. Mahajan et al. 1984) but rather consumers’ decision‐making after having been exposed to negative WOM about an existing product. Thus, the brand had to be already established in the market for several years. Moreover, since the impact of WOM on brand evaluations (purchase intentions and brand attitudes) is moderated by brand familiarity (Laroche et al. 1996; Sundaram and Webster 1999), the brand had at least to be known by the majority of respondents. Due to the international scope of the research, the brand had to fulfill three main conditions:
1. The brand had to be internationally known, i.e. every respondent had to be at least familiar with the brand.
2. The brand had to be emotional and valuable so that respondents could actually be influenced by negative eWOM.
3. The brand had to have an own brand community, must be represented on Facebook and give at least some reasons to be criticized so that negative reviews are available on independent review web sites.
33
in this field. According to Interbrand’s ranking of “Best Global Brands 2010”, Coca‐Cola (#1), Pepsi (#23), and Nescafé (#27) are the three most valuable beverages brands in the world, thus it can be assumed that they are also highly known on an international basis. Since both Coca‐
Cola and Pepsi only provide trivial reasons1 why people should stop buying them, I decided to take Nescafé. However, classical instant coffee is not emotionally enough, thus not involving the consumers sufficiently to consider the opinion of others before buying it. Therefore, I chose the subbrand Dolce Gusto which is a combination of a high‐involvement Krups coffee machine and the coffee pads which are manufactured by Nestlé (see appendix 1). Consequently, the three different platforms to be evaluated by the respondents were the Nescafé Dolce Gusto online community, Nescafé Dolce Gusto Facebook page, and reviewcentre.com. The later is a privately owned independent review web site and known for its unbiased consumer reviews. Compared to other independent review web sites like epinions.com, dooyoo.com and
ciao.com, reviewcentre.com provides posters no financial incentive to write a review. Thus, they
have no other motive to write a review than to help consumer to make informed buying decisions.
3.5 The two pretests
Before the actual survey was conducted, two pre‐tests were run to examine whether respondents perceive the chosen negative posts significantly more negative than the neutral posts at equal credibility. This procedure was done to ensure that the manipulation of eWOM valence was evident. Moreover, participants were asked to rate the credibility of the three different platforms as such and with the assumption that they contain negative posts. Last but not least, participants’ familiarity with the brand was inquired simultaneously. The next section describes every component respectively variable of the pre‐tests in more detail and states how the post’s valence and platform credibility has been manipulated.
3.5.1 Negative and neutral eWOM
The manipulation of eWOM valence was accomplished by collecting both negative as well as neutral posts about Nescafé Dolce Gusto. The neutral posts were required to determine the relative impact of negative eWOM on brand evaluations. Due to the fact that the word of friends and close acquaintances enjoys a higher credibility compared to complete strangers (Brown and Reingen 1987; Duhan et al. 1997), the posts were all from unknown and non‐expert persons
1
34
(weak‐tie relationship). Furthermore, since it might be difficult for respondents to judge the specific platform’s credibility when the messages on it is exceedingly strong respectively weak (Bohner et al. 2002), a mixture of strong and weak arguments were used for both the negative as well as the neutral scenario. These two premises also ensured that only the credibility of the platform rather than the posts itself would have an impact. In order to be as close as possible to reality, existing complaints as well as rather neutral reviews were searched for on the three different platforms. Besides, suitable profile pictures and names had to be chosen and combined to make the real person’s identity untraceable. This was done to protect the poster’s privacy (see appendix 2 and 3 for the choice of picture and name for each post). Moreover, to avoid an interactivity effect of neutral and negative posts which ultimately impact the credibility of the platform, thus brand evaluations, it was decided to take only one valence per scenario.
3.5.2 Source credibility
Credibility of platform
According to Bart et al. (2005), the credibility of a platform can be measured with four items, namely, “this site appears to be more trustworthy than other sites”, “my overall trust in this site is high”, “my overall credibility of the information on this site is high”, and “I would highly recommend this site to others”. Since the first and second item was comparable, it was decided to take the last three questions in order for participants to rate the credibility of the platforms. As for the previous questions, a 5‐point Likert scale was chosen with 1 being “strongly disagree” and 5 being “strongly agree”. Credibility of messages In order to control for possible effects of the perception of the posts on the platform’s credibility, control questions concerning the credibility of the posts were included in the questionnaire. Since the negative respectively neutral posts were equal in respect to content, poster’s name, and profile picture, a deviation between the three scenarios concerning the credibility of the messages, would ultimately reflect a different perception of the platform where the identical posts are placed.
35
post to be highly credible and 2) I perceive the post to be very expert (Tormala and Clarkson 2007).
3.5.3 Design of pretest 1
36
Concerning the credibility of the twelve posts, it was discovered that the negative posts received generally higher ratings. Yet, no particular reason could be found for this incident, not even participants could tell why they did not perceive the neutral posts to be as credible as the negative posts. However, as a consequence, the most credible negative and the lowest credible neutral posts were removed so that the credibility of both valences was equal in regard to their level of credibility. The preceding decision to ask respondents about both credibility as well as expertise of the posts was based on Tormala and Clarkson (2007). However, since posters were all average consumers and not experts, participants had problems in rating the posts on expertise. Thus, the question “I perceive the post to be very expert” in combination with Likert scale was excluded in pre‐test 2.
Moreover, some neutral posts were not perceived neutral, but instead rather positive. This might be due to the fact that the according questions were formulated in the following way “how negative (unfavorable) or positive (favorable) is the previous post toward Nescafé Dolce Gusto?” Thus, respondents might have been misled to choose one of the anchors of the scale (either negative/unfavorable or positive/favorable) and did not have the heart to check the middle (neutral). In addition, the answers for the two questions significantly correlated, i.e. participants perceived negative (positive) posts at the same time also unfavorable (favorable) toward Nescafé Dolce Gusto. That was the reason why the question about favorability was removed and the other question was adjusted to “how negative, neutral or positive is the previous post toward Nescafé Dolce Gusto?” giving subjects the option to check neutral when they perceived the post to be neutral. In addition, the posts’ content was altered in accordance to the feedback of the respondents. Especially the neutral posts had to be changed to become more neutral (e.g. the words like “favorite” and “opportunity” were excluded). Besides, certain phrases were reformulated due to their lack of clarity (e.g. “is this a con?”) or completely left out. In summary, six posts were filtered out and names and pictures were adjusted so that each person was stating both one negative and one neutral post (see appendix 5 and 6).
3.5.5 Design of pretest 2
After having adjusted the questions and posts in pre‐test 1, two more questions were included in pre‐test 2. The first one was about the respondents’ familiarity with Nescafé before asking them how familiar they are with Nescafé Dolce Gusto. This was done to ensure that a significant familiarity with the brand as such is existent. The second one was about participants’ attribution of the negative information to the brand or to the poster. The underlying theory for the inclusion of this question is described in more detail in the following.
37
Attribution theory
Attribution theory is particularly helpful in understanding a receiver's interpretation of a sender's motives for communicating negative information about a brand (Folkes 1988; Mizerski et al. 1979). In other words, attribution theory examines whether the negative eWOM about a brand is attributed to the brand itself or to its communicator.
According to Lee and Youn (2009), respondents will attribute negative eWOM that is posted on an independent review web site more likely to the poster’s true feeling about the brand’s actual performance (brand attribution), whereas they will attribute negative eWOM posted on the online brand community more likely toward a particular circumstance (communicator attribution). To control for possible attribution effects, the question “to what extent do you attribute the described experience to Nescafé Dolce Gusto and to what extent do you attribute the described experience to the individual person” was included in pre‐test 2 after each of the three negative posts. This procedure made it possible to draw conclusions concerning the degree of negativity and credibility since it can be assumed that post which contain information that is rather attributed to the poster, will be perceived less credible and less negative toward the brand. Moreover, respondents were asked about their credibility appraisal when the platforms contain negative information. The complete pre‐test 2 design can be found in appendix 5 to 7.
3.5.6 Results of pretest 2
The 15 participants who took part in pre‐test 2 were slightly more familiar with Nescafé (M = 2.20, SD = .414) than with its subbrand Nescafé Dolce Gusto (M = 1.93, SD = .458). It was further discovered that the neutral posts were now indeed perceived to be neutral (mean between 3.1 and 3.4) owing to the adjusted question as well as the content of the posts. Moreover, through the adaption of the content and the reformulation of certain phrases, the credibility of the neutral and negative posts could be established between a mean of 3.2 and 3.7. With the exception of Helen Dillon’s negative post where respondent attributed the negative experience to both Nescafé and her (M = 2.8, SD = 1.01), respondents attributed the described negative experience in the other two posts rather to Nescafé (M = 1.8, SD = .802 respectively M = 2.0, SD = .730). Furthermore, it was assumed that the perceived credibility of the posts might differ depending on the platform where they were placed. Hence, to analyze the effect of platform credibility on post credibility, respondents were asked to rate both, the credibility of the particular platform and the message credibility of each of the three posts in the final questionnaire.