The influence of quality of eWOM and consumer brand identification
on review attitude and defending intention:
How do quality of language and quality of arguments from a negative online review and consumer brand identification effect review attitude and defense intentions?
Final Master Thesis Charlotte van de Sande 10297677
First supervisor: Dr A. Zerres MSc. Business Studies – Marketing University of Amsterdam
2 Abstract
The last decades the use of electronic word of mouth (eWOM) by consumers prior to a purchase decision has increased extremely. Quality of these online reviews is not always high, something which can have an influence on their power to pursued readers. The purpose of this experimental study is to investigate in which way the quality of negative reviews has an influence on consumers’ purchase decisions. Moreover, it is going to explore what motivates and causes consumers to defend a brand after reading a negative online review by writing a counter-review. The results of the structural equation model show that reviews including high quality of language (no phonetic mistakes) and high quality of arguments have a small direct negative effect on review attitude contradicting to the positive effect that was expected to be found in the study. These suppressor effects are contrary to the indirect mediated effects. The indirect mediation effects from reviews including high quality of language and high quality of arguments via credibility perception on review attitude are positive and much stronger than the negative direct effects. Furthermore, there is a direct negative effect from credibility perception to brand attitude and a direct positive effect from brand attitude to purchase intention. Both consumers that strongly identify with the brand and consumers that perceive the review as hostile, are more motivated to defend the brand after reading a negative online review. Consumer brand identification does not moderate the relationships between quality of language and hostility perception or between credibility perception and hostility perception. These findings may have implications: (1) for future research, to elaborate further on the proposed model and to test it for different products and in different settings (2) for managers to make more strategically decisions on when to respond to negative eWOM based on how consumer react on negative eWOM.
[Key words: electronic word of mouth, argumentation, phonetic mistakes, ELM, credibility, CBI]
3 Content 1. Introduction ... 5 2. Literature review... 10 2.1 eWOM ... 10 2.2 Negative eWOM ... 12 2.3 Credibility ... 15 2.4 Quality of argument ... 19 2.5 Quality of language ... 26 2.6 Hostility ... 29
2.7 Consumer Brand identification ... 30
2.8 Defending intentions ... 34
3 Methodology ... 38
3.1 Survey design ... 39
3.2 Sampling and method ... 41
3.3 Measurements ... 42
4 Data analyse ... 48
4.1 Descriptives ... 48
4.1.1 Normality ... 49
4.2 Factor analysis ... 50
4.2.1 Explanatory Factor Analysis ... 50
4.2.2 Reliability ... 55
4.3 Confirmatory Factor Analysis ... 56
4.3.1 Continuous to categorical ... 56
4.3.2 Model fit ... 57
4.3.3 Invariance Tests ... 58
4.3.4 Common method bias ... 59
4.3.5 Create composites from factor scores ... 59
4.3.6 Validity and reliability ... 60
4.4 Multivariate assumptions ... 61
4.4.1 Linearity ... 61
4.4.2 Multicollineartiy ... 62
4.5 Structural model ... 62
4.6 Hypothesis testing ... 63
4
4.6.2 Multi-group Moderation ... 66
4.6.3 Interaction ... 67
5 Discussion and Conclusion ... 71
5.1 Key findings ... 71
5.1.1 Research questions ... 71
5.1.2 Relationships review attitude, brand attitude and purchase attitude ... 74
5.1.3 The effects of credibility perception on hostility perception and defending intention ... 75
5.1.4 CBI as a moderator ... 76
5.2 Managerial Implications ... 77
5.3 Limitations and future research ... 77
6. Conclusion ... 80
7. Appendix ... 82
5 1. Introduction
Living in a city like Amsterdam, one can see on a daily base new restaurants and cafes coming up and going down. The turnaround of these establishments seems to be shorter than ever before. Survival of the fittest, the fittest being the place which is talked about and recommended the most. Consumer review website like Yelp.com, TripAdvisor.com or the Dutch Couverts.nl seem to keep to secret ingredient for a business success: positive online consumer reviews. With ratings and comments about service, food and ambience, reviews control which restaurant is going to make it and which one will suffer from a quick and silent death. This review power is not only the case for restaurants, bars or cafes, but similar review websites are also available and successful for electronic devices, holidays and insurances. How can comments of consumers you have never seen before be your guide to purchases and indirectly be a way to success or failure for many products? Why have some reviews and ratings no influence at all? This Thesis will shed a light on how to determine which negative online reviews will persuade consumers and which ones will be powerless.
The last decennia social media began to play an important role in many people’s lives (Vineran, Cetina, Dumitrescu, & Tichindelean, 2013); many people are on the internet, three-fourths of these people using the internet also engage in some social media such as Facebook, Twitter, Instagram and many more, sharing information with each other (Hausmann, 2012). Consumers can use social media for numerous reasons and goals; one of those reasons and goals is to search for information and to interact with information and knowledge transfer. Social media and the Internet can be used to provide consumers with enormous amounts of information on always-available devices without physical borders or distance (Peterson, Balasubramanian, & Bronnenberg, 1997). This information source is very inexpensive and has an interactive nature; participants can exchange information and react immediately on each other.
6
Finding information about products or services prior to a purchase decision is a growing activity consumers engage into. In the last decennium less faith is put into traditional media such as television, advertising and print media as information source before a purchase decision. Opinions from friends & family and next to this online review is with 70% according to AYTM Market Research seen as the most used source of information (Nielsen, 2013). Almost 20% of consumers never buys a product without reading online reviews in advance (AYTM Market Research as cited in Garcia, Drolet and Kucinskas, 2013). Informal sharing of experiences, opinions and knowledge about products or services while influencing people’s purchase intention and attitude is called word of mouth (WOM) (Westbrook, 1987; Arndt, 1967). Previous research has proven that WOM has a positive effect on sales and purchase intention (Gilly, Graham, Wolfinbarger, & Yale, 1998; Park, Lee, & Han, 2007). Companies saw the possibilities of WOM and tried to increase positive WOM about their product or services via buzz creation and other marketing tools. With the growth of the internet, consumers start to engage in online word of mouth, sharing and search for information via the internet. This form of online interaction and exchange of knowledge is called electronic word of mouth (eWOM). The exact definition for eWOM according to Hennig-Thurau, Gwinner, Walsh and Gremler (2004) is “any positive or negative statement made by potential, actual or former consumers about a product or company, which is made available to a multitude of people and institutions via the internet” (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004, p. 1). Lots of brands in business-to-consumer environments saw possibilities in this eWOM and were eager to engage in social media as well (Fournier & Avery, 2011). Unfortunately in the beginning they overlooked the possible threats resulting from the use of Social Media and eWOM. Marketeers do not have full control over social media and over eWOM. Consumers are able to talk freely and with no restrictions via the internet and social media about all the positive aspects of products/brands. However, they can also talk about negative aspects, express their
7
dissatisfaction, complain, criticise and gossip about brands. Consumers are able to attack marketers in a way that is totally new and that was not foreseen by marketeers. They could damage the brand’s reputation, resulting in reduced sales. Bowman and Wills explain the potential damage a brand or product can incur via this negative eWOM (as cited in Sandes & Urdan, 2013).
The size of this damage depends on how the audience, the consumers, react to this negative eWOM. How consumers react on negative eWOM depends on multiple factors like credibility of source and message (Samson, 2006), the consumers personal attitude towards the product category (low versus high commitment), brand loyalty (Sandes & Urdan, 2013; Dichter, 1966) and the uncertainties and risk around a purchase (Engel, Blackwell, & Miniard, 1993). A possibility is that when consumers receive negative eWOM, their attitude towards the brand and the review might be moderated by their perception of credibility or hostility from the review. Perceived credibility is seen by companies and brands as the key to why certain reviews are persuasive and which are not (Sandes & Urdan, 2013) and therefore it is seen as a very important aspect (Cheung, Sia, & Kuan, 2012). This thesis will investigate if credibility perception or hostility perception are that influential. Especially the question how quality of a review can affect this credibility perception, will be a central construct in this study. Prior research already indicates that quality of language and quality of arguments can have an influence on how consumers perceive review credibility (Cheung, Sia, & Kuan, 2012). According to a survey conducted by Battery Ventures and Kelton Global (2012), 45% of US consumers find that reviews written with poor grammar and spelling decreases their trust in online reviews (Battery Ventures & Kelton Global, 2012). This thesis will dig deeper, it will argue if these statements are valid and what the direct or indirect relationships between quality of reviews, credibility and review attitude are. Furthermore, attributions readers form about the reviewers’ hostility against the brand can have a big role in consumers attitude against the
8
review (Cheung, Sia, & Kuan, 2012; Sen & Lerman, 2007). It could be that quality of arguments or credibility perception have an effect on this hostility perception.
Next to the possible effects of quality of a review, credibility perception and hostility perception on review attitude, this study will also analyze a possible different reaction of the consumer, the defending intention. Sometimes, it occurs that consumers start to response to negative reviews, they defend the brand or company. When consumers with strong consumer brand identification (CBI) perceive reviews are hostile against ‘their’ brand, they might feel the message is attacking them personally as well. As a way to defend their self-concept, they might react on the review by a defensive counter review (Colliander & Wien, 2013). The relationships and possible influences between these variables have not been prior investigated. This opens up an opportunity to gain new knowledge and get a clearer view on how CBI and hostility can have an impact on review attitude, brand attitude and defense mechanisms from consumer brand identifiers. The objective of this research paper is to develop a conceptual framework about variables influencing defence by receivers and brand attitude by receivers. As such there are some fundamental research questions to be answered:
1. Do quality of language of negative eWOM and quality of arguments of negative eWOM have an effect on credibility perception, hostility perception and attitude to the review? 2. Do hostility perception and consumer brand identification have an effect on consumer
defending intention?
When companies have knowledge about how and when consumers are influenced by negative eWOM, this can be the start of the management of eWOM and the creation of reaction strategies. Companies have the opportunity to better allocate their resources in order to decrease reputational damage and perhaps, when their reaction is perceived as excellent by consumers, even increase potential future sales. By knowing what the effects of consumer brand identification is on defense by loyal consumers and identifying when consumers perceive
9
eWOM as hostile, companies are better able to allocate their sources into the areas where it is most needed (Colliander & Wien, 2013). With the results and knowledge from this research in mind, companies might decide to invest more in creating brand lovers; hoping consumers will start defending their brand. In this way their reputation is protected more efficiently and spending can be decreased.
This thesis will start with a literature review in which various fundamental concepts will be discussed. These are eWOM, negative eWOM, credibility perception, quality of arguments, quality of language, hostility perception, consumer brand identification and defending intention. This will be followed by the research methodology and the results. Thereafter, the results from the experiment will be explained, discussed and a conceptual framework will be presented. This all will be finalized by a discussion which includes managerial implications and a direction for future research and in the end a conclusion.
10 2. Literature review
2.1 eWOM
The internet has given rise to many channels where consumers can communicate with each other and deliberate their opinion and experiences about products and services. Word of mouth is a phenomenon alive for decades, proven to play a major role in consumer behaviour and purchase intention (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004). The easy, fast and always accessible web 2.0 has been the origin to eWOM. Consumers are no longer passive receivers of product- information given to them by the powerful companies who were able to control almost all brand-shaping messages (Hennig-Thurau, et al., 2010). In today’s new media-channel world, there is a constant changing source of information in which consumers can dig into, react to and interact with, which is just partly created by companies and brands. Companies try to make successful use of the new media-channels, by attracting consumers to hop into the conversation and support their brand; however companies are not able to control consumers. They hope to create brand lovers, brand fans which will spread the holy word about the brand and infect other consumers as well (McCLung, Eveland, Sweeney, & James, 2012; Reichheld, 2003; Pronschinske, Groza, & Walker, 2012). Henning-Thurau et al. (2010) compare this as throwing a pinball into a framework (Hennig-Thurau, et al., 2010). The ball gets partly controlled by the companies who “serve” the ball, however, once the pinball is in the playing-field, it hits “new media-bumpers”, consumers who might change the balls direction with eWOM (Hennig-Thurau, et al., 2010). What the eventual message is which ends up at the consumers and the impact of this message, the response, is the most important thing in the game, the eventual score. One part of this framework can be seen as the message reviewers sent out into the online world. According to Cheung and Thadani (2010) communication from eWOM has four elements, as the figure in Appendix 1 will show. First of all there is the communicator, the reviewer who sends out an online message. The communicator can also be
11
seen as the source and other people can assign credibility to the source. Secondly is the stimulus, also called the message and in this study the review that is sent out to the receiver. A receiver can respond to the message via assigning message credibility. Third there is the receiver, the recipient and consumer who reads the eWOM and will respond to the message via consumer behaviour. The receiver will determine via a response the credibility of the source and the credibility of the message to find out if the message can influence the consumer behaviour. Lastly, is the response, which is the consumer behaviour, attitude, perceived credibility and belief of the receiver (Cheung & Thadani, 2010). However, response can also be a eWOM reaction to the message or source in the form of a rating given to the message or source or in the form of a written response.
Many studies are seeking to find out if the message from online reviews does have an influence in the form of an impact on purchase intention. Product and brand attitude can be influenced by eWOM (Charlett, Garland, & Marr, 1995). According to Chen, Wu and Yoon (2004) and Duan, Gu and Whinston (2008) as cited by Zhu and Zhang (2010) eWOM should be seen as predictor of future sales however it is not correlated to current sales (Chen, Wu, & Yoon, 2004; Duan, Gu, & Whinston, 2008; Zhu & Zhang, 2010). Contradicting are the results from Forman, Ghose and Wiesendfeld (2008), Chevalier and Mayzlin (2006), Ghose and Ipeirotis (2011) who all found a direct influence of eWOM on sales and purchase intention. Prior research has proven there exists an association between review rating and sales of the product (Chevalier & Mayzlin, 2006; Ghose & Ipeirotis, 2011). Review rating can be translated as the rate reviews received from receivers according to their helpfulness, valance and value. Sandes and Urdan (2013) researched effects on sales by dividing eWOM into negative eWOM and positive eWOM. They found that negative eWOM has a bigger influence on sales than positive eWOM (Sandes & Urdan, 2013). Reason for this was the difference in consumer response to negative versus positive eWOM (Samson, 2006). This is based on the classic prospective theory from
12
Kahneman and Tversky (1979) which claims that losses are identified as relatively more influential than gains. Negative eWOM can be seen as a loss, and therefore has more influence than positive eWOM which can be seen as a gain.
2.2 Negative eWOM
According to multiple researches, the effect from negative WOM on consumer behaviour, particularly on the attitude and purchase intention, is much bigger than the effect of positive WOM(Chevalier & Mayzlin, 2006; Kahneman & Tversky, 1979; Marsden, Samson, & Upton, 2005). This is called the negative bias; “a psychological tendency for people to give greater diagnostic weight to negative than positive information in making evaluations” ( Herr, Kardes, & Kim, 1991; Ito, Larsen, Smith, & Cacioppo, 1998; as cited in Samson, 2006, p. 650). For many businesses this is an important issue (Sen & Lerman, 2007). When negative WOM is weighted much heavier than positive WOM, this can have an even bigger negative impact on company revenue than when it is weighted equal or less heavy (Marsden, Samson, & Upton, 2005). Furthermore negative WOM has a higher “stickiness” factor, meaning that is more easy to remember and it is remembered for a longer period (Tinkham & Weaver Lariscy, 2004). According to their research, the readers puts more energy into comparing the negative information, think about it and asses the truthfulness. This higher level of energy and attention makes the message stick better and longer into people memory (Tinkham & Weaver Lariscy, 2004). More recent research has shown that negative eWOM can also have a negative effect on consumer’s evaluations of the company (Lee & Song, 2010). Contradicting were the findings of Lui (2006) and Ghose and Ipeirotis (2011) who found that negative eWOM compared to positive eWOM can have no or even a positive effect on sales (Lui, 2006) especially when the review was “informative and detailed” (Ghose & Ipeirotis, 2011, p. 1505). According to the research of Lui (2006), volume of eWOM has a bigger influence on revenues than the valance of the eWOM, meaning that the higher the volume of eWOM is, the more publicity there is and
13
the more likely a person will hear about the product. Commonly used proverbs like “the only thing worse than being talked about is not being talked about” by Oscar Wilde and “there is no such thing as bad publicity” by Phineas T. Barnum demonstrate that bad WOM is also publicity which can create awareness for a product. Sandes and Urdan (2013) partly endorse these findings with unconfirmed hypotheses about purchase intentions. However, they explain these unconfirmed hypotheses with a possible limitation, their experiment is not taken in a perfect world, that could explain this lack of satisfying results. According to Sandes and Urdan (2013), exposure to negative eWOM does not only have an impact on purchase decisions but also can have an effect on attitude towards the overall brand. Negative eWOM provides information that can change receivers’ attitude against the company and the product. This can influence their purchase decisions even more since it is proven that attitude to the company has an influence on consumer behaviour (Armitage & Christian, 2003). In the end, negative eWOM can decrease (future) sales of the attacked company and create major reputational damage.
Besides informing receivers of negative features, which might decrease brand attitude and purchase intention, there is another aspect that might change consumer behaviour. According to Sweeny et al. (2008) consumers receiving a negative message can develop sympathy with the WOM provider, feel anger towards and can develop a poorer image of this brand (Sweeney, Soutar, & Mazzarol, 2008). Furthermore, there is a pressure to confirm to opinions of a group (Lee, Park, & Han, 2008). When consumers received negative reviews, the consumer has the tendency to confirm to the pressure of the reviewer. The reviewer influences the consumer in this way. Therefore when the attitude towards the review is positive, and therefore the consumer sees the review as helpful, credible enough and of high enough quality, this can have an influence on the attitude to the brand. This study therefore hypothesizes that:
H1. An increase in positive attitude towards a negative review will have a negative effect on brand attitude.
14
Negative eWOM provides information that can change receiver’s attitude against the brand. Brand attitude can influence the purchase intention since it is proven that attitude to the brand has an influence on consumer behaviour (Armitage & Christian, 2003). According to the theory of planned behaviour explaining the link between beliefs and behaviour, the more positive the attitude is, which is a form of product evaluation, the more a consumer prefers one brand over another brand and the higher the intention to purchase the brand is (Azjen, 1988). When purchase intention is higher, the eventual actual behaviour is also higher leading to higher sales. In Appendix 2 displays the theory of planned behaviour used by Armitage and Christian and created by Azjen (1988). In the end, negative eWOM can decrease (future) sales of the reviewed brand. This study therefore hypothesizes that:
H2. An increase in brand attitude has a positive effect on purchase intention.
H3. An increase in positive review attitude has a negative indirect effect on purchase intention through brand attitude
There are many different motives to engage into negative eWOM. Important reasons are according to Sundaram et al. (1998): 1. “Altruism, to prevent others from experiencing the problems they had encountered.” 2. “Anxiety reduction, easing anger, anxiety and frustration.” 3. “Vengeance, to retaliate against the company associated with a negative consumption experience” (Sundaram, Mitra, & Webster, 1998). Especially these last two motives can be perceived as ways to damage the brand and to get rid of frustration and anger. This is highly related to hostility since hostility can be perceived as a way to get rid of frustration. White (2010) found that emotions are important predictors of customer satisfaction and WOM. He found a negative relationship between other attributed negative emotions (anger, anxiety, disappointed) and consumer satisfaction. The reviewer might have had a negative experience, followed by negative emotions and is very dissatisfied. Therefore he is motivated to warn other consumers for this negative experience. However in case the reviewer is extremely dissatisfied,
15
this motivation has the possibility to turn into desire to harm the company and damage it via negative eWOM.
The receiver’s perception of the review has a big influence on the degree of information adoption and influence on purchase intention. According to the elaboration likelihood model (ELM) from Patty and Cacioppo (1986) the impact of eWOM depends on three variables; the message, the source, and the receiver (Patty & Cacioppo, 1986). Other literature has explained that perception on the source of the review and the message in the review; more specifically credibility of the source and message is one of the major factors influencing effect from negative eWOM (Cheung, Sia, & Kuan, 2012; Racherla, Mandviwalla, & Connolly, 2012).
2.3 Credibility
Credibility is defined as the trustworthiness or the characteristic that makes consumers trust a message or source, and mitigates the amount of uncertainty in a review (Racherla, Mandviwalla, & Connolly, 2012; Wathen & Burkell, 2002). Charron, Favier and Li (2006) find that consumer’s trust in communication from traditional media, e.g. newspapers, printed magazines and television, has declined in a great amount (as cited in Sandes & Urdan, 2013). Contrary, consumer’s trust in online communication and more specifically online recommendations has increased in recent years (Nielsen, 2013). Trust in WOM recommendations from friends and family rose up to 84 percent and together with “consumer opinions posted online” which has a trust percentage of 70 percent, they are the most influential sources (Nielsen, 2013, p. 5). The four most important differences between WOM and eWOM according to Tong and Xuencheng (2010) are 1. Direct communication versus indirect communication 2.Verbal word versus digital word 3. Narrow reach versus broad reaches 4. Full visible identity versus anonymously. Overall uncertainty during e-WOM is assumed to remain bigger than in WOM in a non-electronic environment (Zhu & Zhang, 2010; Granovetter, 1973; Bansal & Voyer, 2000; Brown & Reinigen, 1987). Since participants of eWOM miss information about the sender of
16
information, there is information asymmetry (Einwiller S. , 2003; Racherla, Mandviwalla, & Connolly, 2012). This is because of the lack of visibility between the sender and receiver and the weaker tie between them (Chatterjee, 2001; Mesch & Talmud, 2006). The three variables from the ELM can be related to the different uncertainties which consumers engaging in eWOM come across: the message; the source; and the receiver (Patty & Cacioppo, 1986). The first uncertainty is the features of the product or services itself, which is higher the more intangible the product is (Racherla, Mandviwalla, & Connolly, 2012). This part of uncertainty is not the topic of discussion in this thesis but is part of the variable message since the features written about in the review should be determined via the use of the product itself. Another uncertainty, which is part of the foundation of this thesis, is the credibility or trustworthiness of the source (Pornpitakpan, 2004). This depends on how good and positive the intentions and integrity of the reviewer are according to the receiver (Racherla, Mandviwalla, & Connolly, 2012; Gilly, Graham, Wolfinbarger, & Yale, 1998; Pornpitakpan, 2004). This corresponds to the variable source from the ELM with a direct peripheral cue as persuasion effect (Chaiken & Maheswaran, 1994). Lastly there is the credibility of the message content (Hass, 1981). This depends on the source product expertise (Gilly, Graham, Wolfinbarger, & Yale, 1998), the knowledge, qualifications and expertise the writer of the review has (Pornpitakpan, 2004). The more a source is seen as an expert, the more knowledge and awareness they will have on the subject and the more the information will be seen as valuable and credible (Gilly, Graham, Wolfinbarger, & Yale, 1998). However message credibility does not only depend on source credibility but also for a big part on argument credibility (Slater & Rouner, 1996). The stronger argument quality is, the more the overall message is seen as credible which has an effect on total credibility. This corresponds to both message and source variables from the ELM because it entails a direct and indirect effect on message credibility and can have a direct effect on source credibility. This view is supported by the ELM according to the statement of Chaiken and
17
Maheswaran (1994): “the elaboration likelihood model also assumes that source credibility may sometimes affect persuasion directly by serving as a peripheral cue but other times may affect persuasion indirectly by biasing the valence of argument processing” (Chaiken & Maheswaran, 1994, p. 470). Cues for source credibility can come from given background information about the writer, writing style and perceived social similarity or identity (Ludwig, et al., 2013). The last variable of the ELM model, the recipient himself, can be seen as the mediator “expertise and involvement” influencing the source and message relationship with credibility (Cheung, Sia, & Kuan, 2012). This can be seen in the research model from Cheung, Sia and Kuan (2012) in Appendix 3. Involvement can be seen as one of the outcomes of brand commitment (Beatty, Homer, & Kahle, 1988), which on its turn is seen as the outcome of consumer brand identification (Bergami & Bagozzi, 2000; Brown, Barry, Dacin, & Gunst, 2005), the variable which this study will test as moderator on hostility perception and credibility perception. Consumer brand identification is therefore the higher order construct over brand commitment and brand involvement, “it is a more primary representation of a consumer’s relationship with a company or brand than commitment and … it is the more proximal cause of motivated reasoning” (Einwiller, Fedorikhin, Johnson, & Kamins, 2006, p. 186).
Pornpitakpan (2004) found that for online reviews, consumers see a high credible source as superior and more valuable over a low credible one. Pornpitakpan (2004) describes how previous research on the topic source credibility found that feedback from high credible sources were more positively evaluated than feedback from low credible sources according to a research from Albright and Levy (1995). The more credible one finds a reviewer, the more trust is put in the words of the reviewer, and the more persuasive effect the review has which leads to a higher “behavioural compliance” (Pornpitakpan, 2004, p. 245). Hovland and Mandell (1952) findings do not comply fully with the previous statement. Their study shows that sources that have a low trustworthiness do not necessarily decreases persuasion compared to sources high
18
in trustworthiness (Hovland & Mandell, 1952). This means that sources which motives are doubtful and not trustable can still be very persuasive. Pornpitakpan (2004) find as a possible underlying reason for this, that differential weights for source trustworthiness, expertise and objectivity might have a varying effect on persuasiveness and do not always relate directly to the persuasiveness of the message. the Sources that are openly very hostile or give the impression they might want to hurt the brand can therefore still be persuasive in the reviews direction. The elaboration likelihood model explains that persuasion via messages can be processed by consumers via two routes. During first one, the central route, the receiver will process the information given via the central way, which means the receiver puts effort in judging the arguments to persuade and considering the message credibility. The central route requires high cognitive involvement, ability and motivation to see if the arguments given are valid and if the message is credible (Eagly & Chaiken, 1993). In the other route, the peripheral route, the receiver is less motivated or less able to process the information and therefore uses “heuristic cues or informational indicators, such as source credibility, to assess the believability of a message.” (Cheung, Sia, & Kuan, 2012, p. 620). The peripheral route does not demand high involvement or motivation to process message credibility. Attractiveness of the source, source expertise and therefore source credibility or the quantity of arguments can be cues which lead to persuasion. Concluding from this, both message credibility and source credibility can have an effect on persuasion, and persuasion has an influence on brand- or company-attitude according to the elaboration likelihood model (Patty & Cacioppo, 1986). This study therefore proposes:
H4. An increase in perceived credibility has a direct positive effect on attitude to the review. Since decisions on attitude towards the review and the brand are based on asymmetric information, it is the consumer’s goal to acquire enough arguments to know if the review and the source are credible or not. According to Chaiken and Maheswaran (1994) many experts
19
found that in case of high motivation and ability to process information, heuristic cues, via the peripheral route do not have a significant influence on persuasion. Chaiken et al. (1989) however do not agree with these findings, stating that both routes of processing information can have an influence on persuasion and judgement being either independent or interdependent. Stoltenberg and Davis (1988) studied the effects of source trusworthiness (source credibility in the peripheral route) on quality of arguments (message credibility in the central route) and their results agree with the interdependent effects from Chaiken et al. (1989) since they found “a greater impact on attitudes and behaviours when participants were faced with a highly credible source, compared to a lower one” (as cited in Pornpitakpan, 2004, p. 250). Most consumers base credibility both on the aspects from the central route as well as aspects from the peripheral route (Sussman & Siegal, 2003). The degree to which central route is used over peripheral route depends on receiver’s motivation and ability, motivation being an important outcome of strong consumer brand identification. Two variables that might have an effect on perceived credibility are quality of argumentation in the review (assuming the indirect effect from source credibility via message credibility on persuasion following the central route) and quality of language in the review (assuming the direct effect from source credibility on persuasion following the peripheral route).
2.4 Quality of argument
According to the research Berger and Calabrese (1975) did on uncertainty reduction theory (URT), during the entry phase of interaction, credibility and trust are actively deviated from argument quality of (verbal) communication (Berger & Calabrese, 1975, p. 100). One can see eWOM as asynchronous consumer to consumer communication, which makes it more comparable to the verbal communication investigated by Berger and Calabrese (1975) (Racherla, Mandviwalla, & Connolly, 2012).
20
Argument quality is a primary factor affecting review credibility, effectiveness of WOM and has a positive effect on attitude (Patty & Cacioppo, 1986; Cheung, Sia, & Kuan, 2012; Lee, Park, & Han, 2008; Chevalier & Mayzlin, 2006). Patty and Cacioppp (1986) describe argument quality as: “the audience’s subjective perception of the arguments in the persuasive messages as strong and congent on the one hand versus weak and specious on the other” (Patty & Cacioppo, 1986, pp. 264-5). Racherla et al. agree with this definition and further add the values “objective, including data which is backed with an explanation, relevant and verifiable” to the definitions of strong arguments (Racherla, Mandviwalla, & Connolly, 2012, p. 66). Kim and Benbasat (2006) and Racherla et al. (2012) adapted an argumentation model from Toulmin (1958) to online argumentation to test quality of argumentation. There are however contradicting definitions for quality of arguments available in the literature. According to Lee et al. (2008) the definition is the following “quality of online consumer review content in terms of relevance, reliability, understandability and sufficiency” (Lee, Park, & Han, 2008). This thesis will go by the definition from Racherla et al. and Patty and Cacioppo since this is based on the ELM model, the central route in which cognitive cues, such as strong argumentation are important for persuasion to happen. The definition of Lee et al. (2008) seems to only take into account the view of the peripheral route which has more to do with source credibility and the variable quality of language. Here the quality of argument and message credibility part is missing. The arguments used in this experiment are all understandable and differ only in objectiveness, relevance and verifiability (including facts and data). Argument quality, as a central cue establishes the review attitude through considering the truth and value of the arguments (Cheung, Sia, & Kuan, 2012). Argument quality has also previously proven to influence on how receivers of online information determine credibility of the information (message) and attitude towards the information (Wathen & Burkell, 2002; Cheung, Sia, Luo, & Chen, 2009). Racherla et al. (2012) and Lee et al. (2008) find that argument quality in a review
21
has a significant effect on trust, on both source and message credibility (Slater & Rouner, 1996). Besides the effect on the source-credibility from source expertise judgements when a message contains verifiable facts (Slater & Rouner, 1996; Sweeney, Soutar, & Mazzarol, 2008), data and theory there is also the possibility of decreased credibility when arguments contain fallacies. When arguments in a review contain fallacies; they are subjective, not well developed or cannot even be seen as arguments, this can increase uncertainty since the review might seem to contain invalid aspects. When arguments contain fallacies, they contain “violations of the rules for critical discussion” (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001, p. 109). Van Eemeren, Grassen and Meuffels (2012) claim that “when people are confronted with clear cases of violations of rules for critical discussion they consistently judge these discussion moves as unreasonable” (van Eemeren, Garssen, & Meuffels, 2012, p. 345). Furthermore Copi and Burgress-Jackson (1966) conclude that fallacies should be seen as arguments which should not be able to change consumers’ attitudes (as cited in Hoeken, Timmers, & Schellens, 2012, p. 396). However they state that depending on the context these fallacies may persuad nevertheless which is also the case in the research of Hahn and Oaksford (2007) (Copi & Burgress-Jackson, 1996). This thesis will test if in an online context with negative eWOM fallacies decrease persuasion or if they won’t decrease persuasion. There are different sorts of fallacies which are often used in argumentation and which will be used in this experiment’s reviews as well:
1. Fallacy of evading the burden of proof; making a standpoint without defending it. In this thesis’ reviews the fallacy of evading the burden of proof is used in a formulation that “is immune to criticism because it cannot be tested or evaluated” (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001, p. 116):
The typical Android system has a very stupid interface and there is no proper user manual available.
22
What can be seen in this statement is that the words “typical” in combination with “stupid interface” suggest that all the Android systems have stupid interface. Furthermore the “unavailability of a proper user manual” is used as statement to strengthen the first statement. However since the “stupid interface” has no direct relationship with “unavailability of a proper user manual” this is a fallacy as well. Namely the fallacy of using irrelevant argumentation, which is argumentation fallacy number two.
2. Fallacy of using irrelevant argumentation/ ignoratio elenchi; According to the rules of critical discussion, “a party may defend his or her standpoint only by advancing argumentation related to that standpoint” (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001, p. 119). In the previous statement there is a violation of this rule since the argument “unavailability of a proper user manual” does defend a standpoint that is not the standpoint, which is stated before. It would be better to say: ‘the typical android system has a very unfriendly user interface and there is no proper user manual available –which would make the user interface better understandable and therefore less unfriendly- ’
Another example used in the experiment’s review is:
“The 8 megapixel auto focus camera with flash & zero shutter makes ugly pictures. Even though it has a flashlight and high front camera resolution, pictures are vague.”
This sentence states that the camera makes ugly pictures (statement). Then it states that it does have a flashlight and high front camera resolution but pictures are vague. This defense in the second part is not an argument for the statement that the pictures are ugly. A good argumentation here would be: “the 8 megapixel auto focus camera with flash & zero shutter lag has a very slow shutter which often leads to blurred and out of focus pictures. Even though it has a flashlight and high front camera resolution, when the light is not optimal pictures are
23
often too vague, overexposed or too dark”. In this example all the statements are backed up with correlating arguments and relevant theory.
3. Another form of fallacy is the use of non-argumentation/pathetic fallacy; according to van Eemeren, Grootendorst and Snoeck Henkemans (2001) this is mostly used with the purpose of “winning over a third party – in eWOM case the receiver – ” instead of “convincing the other party – the producer of the review product/service –” (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001, p. 120). The source does not make use of argumentation to support a statement, but tries to persuade the receiver with the use of emotions or biases. An example of this would be:
“The battery life does not live up to my expectations. It is way too quick empty. Having paid for a premium smartphone which claims to have long battery life, this is a disappointment.” This sentence states that the expectations were not met since the battery was too quick empty. Good argumentation would now come up with theoretical numbers to explain why the battery was too quick empty, which was not the fault of the user itself. However this source does not make use of a real argument but uses a pathetic fallacy to back up the statement. There is an emotional reasoning given attacking the producer: the producer claims something that is not true namely that the phone has a long battery life and the source is very disappointed by this. The source tries to manipulate the reader with an emotional and accusing sentence presenting it as if it was an argument (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001) while in reality the statement is not backed at all with arguments.
4. Another argument in the previous example is the fallacy of violation of freedom/argumentum ad hominem; “an attempt is made to undermine the other party’s credibility by pointing out a contradiction in that party’s words or deeds” (van Eemeren, Grootendorst, & Snoeck Henkemans, 2001, p. 112). There is a “violation of the pragmadialectical Freedom Rule”
24
meaning that the source is hostile or violent towards the other party, the producer (van Eemeren, Garssen, & Meuffels, 2012, p. 345). The Freedom Rule entails that “-a- discussant may not prevent each other from advancing standpoints or from calling standpoints into question” (van Eemeren & Grootendorst, 2004, p. 190). Consumers react negative to these sorts of arguments because they see the argument as unreasonable according to the research of van Eemeren, Garssen and Meuffels (2009), not because of its impoliteness but because they consider the arguments as not persuasive because of the lack of reasoning in them (van Eemeren, Garssen, & Meuffels, 2007). There are three variants of this fallacy argument: 1. The abusive variant, in which the writer is personally attacking the reader or brand by calling him names or saying they are stupid or dumb. 2. A circumstantial variant, where the writer is attacking the reader or brand by writing the mistrust on the motives of the reader or brand. 3. The tu quoque variant in which the reader or brand is attacked by explaining the difference between what they say and what they do (van Eemeren, Garssen, & Meuffels, 2012). In this thesis experiment there is made use of the first and the third variant: “crappy phone”, “stupid interface” both are type 1 fallacies. And then there is the type 3 fallacy: “–the battery – is way too quick empty. Having paid for a premium smartphone which claims to have long battery life, this is a disappointment. “1 . Here the comparison is made between what the producer said and what was really experienced. Since the experienced situation is not backed up with valid argumentation and theory, the statement contains two fallacies.
5. The last fallacy of argumentation is the absence of argumentation at all; in this case a statement is made without any argumentation to back it up. An example of this would be the following statement: “The Android operating system however was disappointing.” As can be seen, there is made a statement without any related argumentation. An example of a stronger argumentation would be: “The Android operating system however was disappointing, I often
1 This last statement is also used in the review with good argumentation however than it is backed up with valid evidence and theory argumentation and therefore can be seen as less hostile.
25
had difficulties with letting e-mail working, which is kind of a hard thing to do when you do not have a g-mail account.” Here the sentence “Android operating system was disappointing” is backed up with argumentation on how there are difficulties with the systems e-mail account. When reviews lack argumentation or contain weak, fallacy arguments, the message might lack credibility on both trustworthiness dimension and expert dimension. Slater and Rouner (1996) state that there is a direct from the variable message quality on the variables credibility and belief change and a mediation effect from message quality on the relationship between credibility and belief change. The receiver might start wondering what the motives to write the negative review are. It becomes more plausible to the receiver that motives like anxiety reduction and vengeance are the reason for the source to send the negative eWOM (Sundaram, Mitra, & Webster, 1998). The source might seem hostile, frustrated or aggressive against the product or brand. Furthermore because of the lack of argumentation or the argumentation containing fallacies and lacking theoretical foundation the receiver might start doubting the expertise of the source. Both trustworthiness and lack of expertise can decrease credibility. Especially when consumers are following the central route of the ELM this might reduce persuasion effect from the negative review. In line with previous researches (Pornpitakpan; Stoltenberg & Davis, 1988; Slater & Rouner, 1996; Kempf & Palan, 2005) the following hypotheses are proposed:
H5a. An increase in quality of arguments has a positive effect on perceived credibility. H5b. An increase in quality of arguments has a direct positive effect on attitude to the review H5c. An increase in quality of arguments has an indirect positive effect on attitude to the review through perceived credibility.
26 2.5 Quality of language
Besides the quality of argumentation, credibility perception might be also affected by the quality of language in the review. If a statement is written in poor language style, containing many spelling errors, the receiver might intuitively, via the peripheral route of persuasion, attribute “amateurism” to the sender and reject the comment in turn as not credible (Wathen & Burkell, 2002). Next to this, spelling errors can cause distractions (Molich & Nielsen, 1990). Quality of language can have both effects on source credibility and on message credibility. Reviews which include less spelling errors, no slang and no grammatical mistakes might be perceived as easier to read and therefore as more valuable and helpful (Ghose & Ipeirotis, 2011) increasing the message credibility, review attitude and persuasion effect via the peripheral route. Ghose and Ipeirotis (2011) even found evidence for direct links between higher readability of a review and an increase in sales. They also found, for products which quality could not be judged before purchase, a decrease in sales when there were many spelling errors in the review (Ghose & Ipeirotis, 2011).
The other effect that low quality of language can have via the peripheral route is on source credibility; spelling errors can be seen as cue’s from which source credibility is determined. Information containing spelling errors can provide cues to signal an incompetent or an unintelligent source (Hancock & Dunham, 2001; Lea & Spears, 1992; Lynch & Horton, 2002; as cited by Everard & Galletta, 2006). Prior research shows that there is a relationship between lower intelligence and making more spelling mistakes (Finucci, Isaacs, Whitehouse, & Childs, 1983; Kreiner, Schnakenberg, Green, Costello, & McClin, 2002; Lastine-Sobecks, Jackson, & Paulo, 1998). This is only the case when spelling mistakes are phonetic, when the mistakes are dysphonetic there is no relationship with intelligence, only with lexical problems like dyslexia (Finucci, Isaacs, Whitehouse, & Childs, 1983). A phonetic mistake is a mistake which when pronouncing the word cannot be heard. The mistake is caused by the replacement of a correct
27
syllable or vocal by an incorrect syllable with the same phonetic sound, or by the exclusion of the correct not audible syllable. A disphonetic mistake is a mistake which is audible after pronouncing. In a disphonetic mistake syllables are interchanged, left out or included extra, this can be noted immediately when pronouncing the word (Finucci, Isaacs, Whitehouse, & Childs, 1983). Examples from the phonetic errors used in this study can be found in table 1.
Table 1. Examples from phonetic errors used in this study. The underlined part is wrong.
According to Kreiner et al. (2002) evidence shows that “perception of another person’s intelligence can be affected by only a small amount of information about that person” (Kreiner, Schnakenberg, Green, Costello, & McClin, 2002, p. 5). Making phonetic mistakes can have an effect on source credibility since source intelligence can be called in question. According to Kreiner et al. (2002), who did an experiment to test how readers perceive a source when a text contains spelling errors, college students attribute spelling errors to both writing ability and to general cognitive abilities such as intelligence (Kreiner, Schnakenberg, Green, Costello, & McClin, 2002). They do, however, state that in their experiment, college students attribute spelling errors more to a lack in writing ability than to intelligence.
Everard and Galletta (2005) tested if language errors had an influence on how website visitors valued reliability of the information given on the website and found that spelling errors result in a negative perceived competition and a lack of “attention to details” (Everard & Galletta, 2006, p. 61). They claim that credibility is affected by site characteristics that signal (cue) information about a source and the message. One of these signals (cues) is language errors and
correct spelling phonetic mistake disphonetic mistake
satisfied satisvied satisdief
life live lfe (missing the i syllable)
resolution resolucion resoultion
typical tipical tpyical
happy happie happt
fulfills fufills (missing the l fulillfs
28
since consumers search for signals to determine the amount of credibility to put into a message and a source and to determine if they are persuade or not, language errors can have an effect on credibility and persuasion. The results support their hypothesis claiming that perceived site quality decreases when there are language errors on the website (Everard & Galletta, 2006). Other studies agree with the findings of a relationship between spelling errors and credibility perception of an online source (Fogg, et al., 2011; Briggs, Burford, De Angeli, & Lynch, 2002; Liu, 2004; Maier, 2005; Metzger, Flanagin, & Medders, 2010). Battery Ventures and Kelton Global (2012) found that poor grammar or spelling errors is one of the most important reasons why consumer do not trust an online review of a product or service (Battery Ventures & Kelton Global, 2012). As can be seen in Appendix 4 concluding from their research 45% of the respondents agree with this statement.
Despite all these findings, lots of the spelling mistakes made in reviews are dysphonetic, or typographic, and possibly originate from typing or writing too fast, from bad coordination or from being careless (Kreiner, Schnakenberg, Green, Costello, & McClin, 2002). Therefore a receiver might be more used to seeing these sort of mistakes and might not form any perceptions or beliefs about the writer as having a lower intelligence (Finucci, Isaacs, Whitehouse, & Childs, 1983). Especially when we consider that eWOM is not only typed on computers, but also on phones and tablets it is not strange that typographic mistakes are easily made. To type a message as fast as possible people might make use of slang, abbreviations or type without capital letters. Consumers are more used to make and see textual mistakes, therefore it also has been argued that the effect of language mistakes, more specifically phonetic mistakes, has been reduced in the last decade (Chesney & Su, 2010).
To test if consumers after reading a narrative with phonetic spelling errors, perceive the writer as having a lower intelligence or ability to write and therefore find the narrative less credible or
29
if this effect is reduced because of being more accustomed to seeing spelling mistakes, the following hypotheses are proposed:
H6a. A decrease in phonetic mistakes has a positive effect on perceived credibility
H6b. A decrease in phonetic mistakes has a direct positive effect on attitude to the review H6c. A decrease in phonetic mistakes has an indirect positive effect on review attitude through credibility perception.
2.6 Hostility
Hostility can be seen as an aggressive state of mind when thinking about a brand (Blais & Renshaw, 2012). In computer-mediated communication, which is overall more hostile than face-to-face communication, being hostile is also called “flaming” (Moor, Heuvelman, & Verleur, 2010; O'Sullivan & Flanagin, 2003). Flaming can be seen as “hostile verbal behaviour…, insulting…, social aggression…, -and- emotional outburst” (O'Sullivan & Flanagin, 2003, p. 71). According to Moore, Heuvelman and Verleur (2010) there are multiple ways to express hostility in an electronic review: (1) “Displaying hostility by insulting, swearing or using otherwise offensive language” ; (2) “All kinds of emotional expression or even the use of superlatives” (Moor, Heuvelman, & Verleur, 2010, p. 1536). Of course there are many different motives for sources to engage in flaming: simple enjoying insulting other consumer or producers (Lange, 2007); expressing frustration; hate; aggression and anger (Sundaram, Mitra, & Webster, 1998; Romani, Grappi, & Dalli, 2012), to damage or harm the producer or brand and hope to persuade other consumer to use another producer or brand. When consumers perceive a review as written with such an offensive motive they might feel offended themselves and perceive the review as hostile. When a review is perceived as hostile, flaming and extremely negative, especially compared to other reviews, consumers get suspicious (Petty, Fabrigar, & Wegener, 2003). This leads them to be more aware for these sort of reviews and to
30
decrease “the influence of these affective cues” (Ludwig, et al., 2013, p. 91). Consumers even can start doubting the authenticity of the review and see it as fake (Hardey, 2011). Therefore perceived hostility can have an effect on attitude to the review since consumer will be less influenced by a hostile review. Trust in the value of such a review can decrease extremely. Moor, Heuvelman and Verleur (2010) showed that many consumers perceive flaming as a problem and 27% of them felt offended by flaming. A not-intended-hostile review can also be perceived as hostile in a form of miscommunication; a receiver could perceive a message as offensive even though this was not meant to be hostile by the source (Postmes & Lea, 2000; O'Sullivan & Flanagin, 2003). When a hostile message contains good quality of arguments, it is more likely that the message is nuanced and therefore is less seen as aggressive and hostile. This is because when the messages contain facts, is reasonable and does not contain fallacies, credibility might increase and when credibility is high, this can have an effect on hostility perception. The reader is less uncertain about the motive for the source to write the review. It is more plausible that the source is true and the source is less judged as aggressive and therefore less hostile. The following hypotheses are therefore proposed:
H7. An increase in perceived hostility has a direct negative effect on attitude to the review. H8. An increase in perceived credibility has a negative effect on perceived hostility
H9a. An increase in quality of arguments has a negatively affect perceived hostility
H9b. An increase in quality of arguments has an indirect positive effect on review attitude through hostility perception.
2.7 Consumer Brand identification
Consumer brand identification (CBI) is the degree to which consumers feel they are connected and related to a company (Mael and Ashforth, 1992). The interaction and identification with social groups, institutes or individuals is a key part of the person’s social personality (Einwiller,
31
Fedorikhin, Johnson, & Kamins, 2006). According to the extended self-concept theory from James (1890) people have a dynamic view about the self which contains multiple values, traits and roles a person can have but also social relationships with other individuals, groups, and material goods (James, 1890 as cited by Lisjak, Lee, & Gardner, 2012). Consumers that have a strong CBI get a part of their identity from a brand or company and feel their values are similar to the company’s values, their self-definition is comparable to the company’s identity (Einwiller, Fedorikhin, Johnson, & Kamins, 2006; Lisjak, Lee, & Gardner, 2012). Often they are part of a strong brand community entailing consumers “whose interactions are mainly driven by shared passion for specific brand” and who “tend to engage in pro-brand activities such as active brand community participation” (Yeh & Choi, 2011, p. 146). Prior studies have confirmed a relationship between strong brand identification, loyalty and positive eWOM (Yeh & Choi, 2011; Fournier, 1998; Algesheimer, Dholakia, & Herrmann, 2005; Chung & Darke, 2006; Kim, Han, & Park, 2001). This is especially true for the “brand evangelists” which are deliberately spreading positive word about the brand and highly recommend it to others and spreading (Muniz & O'Ginn, 2001 as cited by Yeh & Choi, 2011, p.146). CBI is influenced by five drivers according to Stokburger-Sauer et al. (2012): brand-self similarity; brand warmth, brand social benefits; brand distinctiveness and unforgettable brand experiences (Stockburger-Sauer, Ratneshwar, & Sen, 2012). These five have an even higher influence when the consumer has a strong involvement with the brand’s product category. Strong CBI has not only an effect on the spread of WOM; it can also have an effect on how consumers with a strong CBI perceive negative reviews. According to the extended self-concept when there is a threat against one of the self-related persons or objects, which are part of the self, this should cause the same reaction as to a threat against a person itself or his properties (James, 1890 as cited by Lisjak, Lee, & Gardner, 2012). According to Einwiller et al (2006) negative information about an organization can decrease organization attitude and image, however CBI can moderate this decreasing
32
organization or brand attitude. This moderation effect however vanished when the negative publicity was too extreme. Ahluwalia et al. (2000) had similar results when they found that negative information can be filtered in a biased way when consumers have strong brand commitment. When consumers with strong CBI read a negative review on ‘their’ brand, they feel the reviewer is attacking ‘their’ firm, but also that he is attacking them personally as well. Therefore they are more motivated in engaging in defense tactics like writing a reaction to the review or writing a review that is highly positive (Einwiller, Fedorikhin, Johnson, & Kamins, 2006). Consumers that strongly identify themselves with the brand, often engaging in brand communities, are highly involved into the product and therefore have high brand/product knowledge (Lee, Park, & Han, 2008; Patty & Cacioppo, 1986). They form attitudes that are harder to change and accept fewer alternative opinions or thoughts (Racherla, Mandviwalla, & Connolly, 2012). When this consumer reads a negative review; he will because of his high involvement take the central route to evaluate the review. Quality of arguments will therefore have just partly an effect on him. Quality of arguments, as stated before, contains aspects of both peripheral route and central route. Not only is this consumer better able to see if arguments are weak and therefore the review is not persuading him, he will also feel his personal beliefs and values are questioned and attacked. Cervellon (2012) agrees with her findings of consumers being highly involved with fashion having a lower attitude change towards a fashion brand after being exposed to negative messages about this fashion brand compared to low involved consumers. It is very likely he will try to defend “and preserve those beliefs” and values and therefore he will see this negative review as not true, containing lies and are therefore being hostile and invalid (Einwiller, Fedorikhin, Johnson, & Kamins, 2006, p. 187). This agrees with the motivated reasoning theory saying that people can process information with the goal to come to an truthful conclusion or with the goal to find evidence for the support of their predetermined conclusion (Kunda, 1990). In this last case the person will evaluate the
33
information in a highly biased way to make sure it fits his predetermined conclusion or otherwise to find cue’s for why the information is not correct. This can be the case when consumers have strong brand identification (Eagly & Chaiken, 1993; Crocker & Luhtanen, 1990). Since this review is attacking the consumer’s brand, identity and beliefs, it sees it as being hostile no matter if it contains strong or weak arguments. Lisjak, Lee and Gardner (2012) confirm this reasoning and state: “Participants who identified with a brand, a threat to the brand elicited the same responses as a threat to the self” (Lisjak, Lee, & Gardner, 2012, p. 1120). Consumers who have weak consumer brand identification are more motivated to process the information in a way to come to truthful conclusions. They are not biased by any personal protection motivation (Kunda, 1990; Ahluwalia, 2002). The more consumers can identify themselves with the brand the more positive the relationship is between quality of arguments, for both bad and good arguments, and perceived hostility. Based on this rational one can propose:
H10a. Consumer brand identification moderates the relationship between quality of argument and hostility perception, such that for high consumer brand identification, this relationship is less negative or even positive than for low consumer brand identification.
When credibility is high, trust in the review is high and motives for the writer are seen by the readers as more altruistic, concerning for others and product involved (Sundaram, Mitra, & Webster, 1998). There is less reason to doubt the motives from the writer. When the brand is being attacked, by a negative review, high credibility might decrease hostility perception as proposed before. However when consumers feel strong CBI, one can expect the effect from credibility on hostility perception is weakened by the feeling of being attacked themselves. Based on this rational one can propose:
34
H10b. Consumer brand identification moderates the relationship between credibility perception and hostility perception, such that for high consumer brand identification, this relationship is less negative or even positive than for low consumer brand identification.
2.8 Defending intentions
For marketing, CBI is a very important phenomenon; the more people have a strong CBI, the more people are loyal and start spreading the word like. The consumers become brand evangelists. Not only consumers can start recommending a brand via eWOM, they can also start defending a brand after a negative message about the brand (Stockburger-Sauer, Ratneshwar, & Sen, 2012). Lisjak, Lee and Gardner (2012) confirm this with their findings that consumers with strong CBI defend the brand when it is threatened since they see the brand as self-aspects, which are threatened. According to Berger and Milkman (2012) emotionally suggestive information is more often shared than information that does not touch the writer (Berger & Milkman, 2012). When a review is attacking a certain brand, the attack leads to anger and anxiety, emotions for the consumer with strong CBI. These are emotions which are often accompanied by a state of activation; the reader wants to do something (Barrett & Russell, 1998). Therefore a negative review increases the motivation for consumers to defend the brand and react on the review. The stronger the tie between the consumer and the brand is, the stronger the CBI is, the higher chance this person will engage into positive, defense eWOM (de Matos & Rossi, 2008; Lisjak, Lee, & Gardner, 2012). Based on this rational one can propose:
H11. An increase in consumer brand identification has a positive effect on defending intention. Colliander and Wien (2013) also found prove of defending behaviour from consumers after reading negative reviews by reacting on these reviews. In their research Colliander and Wien (2013) describe six different defense styles that consumers use. The different motives for these
35
defenses are per style described, as one can see in Appendix 5. Among these motives are self-experience, attribution of guilt and perceived justice.
Self-experience can be explained as the contradicting personal experiences a consumer has had which are contradicting enough from the review that he or she is willing to defend the product. According to Ahluwalia et al. (2000) these sorts of defenses are based on strong brand commitment. This motivation can be most often found in the defense style vouching.
Attribution of guilt is all about whom the defender sees as guilty one for the negative expression. This can be the attacked company or brand, the external cause, but it can also be that the defender sees the writer as the one to blame, an internal cause (Colliander & Wien, 2013). In this last case, the defender is sceptical about the complaint of the reviewer and more importantly about the motives and truthfulness of the reviewer. This defense motivation is most often expressed with the defense style doubt (Colliander & Wien, 2013).
Perceived justice depends on if the defender sees the negative review as fair and if he sees if the writer of the review is treated fair by the company (Colliander & Wien, 2013). When the review is seen as attacking the company with no justified reason, being hostile against the company, which he sees as not being justified, the defender will be triggered to react on this review. This can be seen in both defense style advocation and vouching.
The defense style advocation, which Colliander and Wien (2013) describe as “arguing in favour of a company using reason and/or logic” is used most often after reading an “irrational or based upon unreasonable demands” review (Colliander & Wien, 2013, p. 1739). This “irrational or based upon unreasonable demands” sort of review can easily be seen as a low credible review, or even a hostile one. Since this sort of defense is not based on personal experience, Colliander and Wien (2013) see this defense style not as caused by high customer loyalty or even consumer
