Faculty of Economics and Business, University of Groningen, the Netherlands Newcastle University Business School, the United Kingdom
Master Thesis
EFFECT OF NEGATIVE ELECTRONIC WORD-OF- MOUTH ON BRAND LOYALTY:
THE PERSPECTIVE OF ONLINE COMMUNITY INVOLVEMENT
Xinchun Fan
MSc International Business and Management (University of Groningen)
MSc Advanced International Business Management and Marketing (Newcastle University)
Dissertation Supervisor: Drs. Henk Ritsema (Groningen) Dr. Markus Blut (Newcastle)
Student number: S2340577 (Groningen) 140173644 (Newcastle) Groningen, 8th of December 2014
Final version
2 | P a g e
ABSTRACT
This thesis aims to investigate the effect of negative electronic word-of-mouth (eWOM) on brand loyalty from the perspective of online community involvement. A theoretical model and assumptions contain the online community involvement, trust tendency, the intensity of negative eWOM and customer loyalty are made according to the literature review. In order to study the relationship of variables which mentioned above, this thesis carry on empirical analysis by using the data from questionnaire survey. We conclude that the level of online communities dependence has positive impact on the quantity of eWOM. Besides, trust tendency is positively associated with quantity and reliability of negative eWOM. What’s more, both quantity and reliability of negative eWOM has significant positive influences on brand switching behaviour.
Key words
Electronic word-of-mouth, Customer loyalty, Brand switching behaviour, Online communities involvement, Trust tendency
3 | P a g e ACKNOWLEDGEMENTS
The dissertation is the final step in my master education. Moreover, it is the most important step in the course that I could use all my knowledge acquired in my master study to write it.
In the last five months, I try to investigate the effect of negative electronic word-of-mouth (eWOM) on brand loyalty, which is an interesting and meaningful topic.
Hereby, I would like to thank everyone who helped and encouraged me to finishing this thesis and my master degree. Without your help and support, I could not accomplish my dissertation.
I would like to thank my Groningen supervisor Drs. Henk Ritsema and Newcastle supervisor Dr. Markus Blut, who gave me substantial guidance and continual help for my thesis.
In addition, I also like to thank my parents and friends who encouraged and supported me for the last two years of my education.
Groningen, December 2014 Xinchun Fan
4 | P a g e
TABLE OF CONTENT
ABSTRACT ... 2
Key words ... 2
ACKNOWLEDGEMENTS ... 3
TABLE OF CONTENT ... 4
1. INTRODUCTION ... 6
1.1 Background ... 6
1.2 Research gap ... 8
1.3 Main research question ... 10
1.4 Structure of the thesis ... 10
2. LITERATURE REVIEW ... 12
2.1 Online communities ... 12
2.2 Trust tendency ... 14
2.3 Word-of-mouth ... 15
2.4 Electronic word-of-mouth ... 18
2.5 Customer loyalty ... 19
3. CONCEPTUAL MODEL AND HYPOTHESES ... 22
3.1 Hypotheses development ... 22
3.1.1 The online community involvement ... 22
3.1.2 Trust tendency ... 24
3.1.3 The intensity of negative eWOM ... 25
3.2 Conceptual model ... 27
4. METHODOLGY ... 28
5 | P a g e
4.1 Research design strategies ... 28
4.2 Research approaches ... 29
4.3 Data collection ... 29
4.4 Research methods ... 30
5 DATA ANALYSIS ... 32
5.1 Questionnaire design ... 32
5.2 Research target ... 34
5.3 Reliability analysis ... 36
5.4 Correlation analysis ... 38
5.5 Regression analysis ... 38
5.5.1 The regression results of online communities involvement, trust tendency and the quantity of eWOM ... 39
5.5.2 The regression results of online communities involvement, trust tendency and the reliability of eWOM ... 40
5.5.3 The regression results of the intensity of negative eWOM and brand switching behaviour ... 40
5.6 Findings ... 41
6. DISCUSSION & CONCLUSION ... 43
6.1 Discussion ... 43
6.2 Limitation and recommendations for future study ... 45
6.3 Conclusion & Implication ... 45
APPENDICES ... 47
Reference list ... 47
Questionnaire ... 57
6 | P a g e 1.
INTRODUCTION
1.1 Background
Word-of-mouth, which widely exists in our daily life, act as one of most directly consumer feedbacks for products and services (Stern, 1994). There are several studies showing that word-of-mouth with its incomparable advantages and influence is widely used in the consumption field (Richins, 1983; Hennig-Thurau et al., 2004; Chevalier and Mayzlin, 2006;
Brown et al., 2005). With the increasing use of the Internet as communication platform, electronic word-of-mouth (eWOM), also known as online word-of-mouth or Internet word- of-mouth, has become more powerful and useful for consumers and markets (Phelps et al., 2004). At the same time, eWOM has the characteristic and advantages of wide audience, strong anonymity and fast spreading speed. Hence, more and more companies start to use eWOM as its marketing tools and also eWOM gradually become a hot topic to research in the marketing field (Maxham III, 2001).
However, eWOM is a typical double-edged sword. consumers tend to share their unpleasant consuming experiences to others compared with pleasant consuming experiences. To be more specific, Mangold, Miller and Brockway (1999) prove that satisfied consumers tend to share their pleasant experiences to about three people, while the dissatisfied consumers tend to spread their unhappy consumption experience to about ten people. And those who receive this information may spread it to more people.
On the other hand, the special characteristic of online environment makes eWOM spread uncontrollable (Hartman, Hunt and Childers, 2013). Furthermore, the network opinions are often spread anonymous, which makes consumers give opinions without scruple. This situation may cause negative and even unrealistic and irresponsible false information. These negative news and rumors would be amplified without confirmed, which may affect the
7 | P a g e customer loyalty and lead to the crisis of company at last. Therefore, how to contain the spread of negative eWOM and create a healthy online environment is a one of major topics faced by companies.
As one of the core elements of virtual market, the online community is the online group which people with common interests or needs aggregate to form via the Internet (Hill et al., 1995). It provides the space for people to communicate via the Internet and to create value through the network.
Besides, online community has various formations, which including trading community, social relationship community, community for same interests or topic and etc. (Hagel, 1999) An online community can act as an information system where members can post, comment on discussions, give advice or collaborate (Ho and Chignell, 2000). What’s more, the existence of online community brings about a new social pattern, which may affect the individual online consuming behaviour greatly via eWOM. Nowadays, More and more people will ask for help in online communities when they want to buy something through the Internet (for example, electronic products). At the same time, online communities members may give some suggestion or comments based on their experience. Hence, online communities are becoming one of important elements of the Internet and also a very popular platform to spread eWOM, which may affect the intensity of eWOM.
At the same time, since eWOM are spread via electronic media, people know each other through the Internet instead of face-to-face contact. It is manifest that those who trust people easily may affect more by the eWOM and also easily spread the eWOM. Hence, the trust tendency could be another important factor influencing the intensity of eWOM.
8 | P a g e With the development of E-business, the decrease of consumer brand loyalty has become a problem for the companies which use online as their marketing platform. And this phenomenon will become more seriously because of the influence of the negative eWOM.
Nowadays, Customers will consider several brands to choose a relatively best choice when they want to buy something. And consumers' brand switching behaviour affect the profitability and competitiveness of the corporations (Knox and Walker, 2001). According to the statistics for U.S. companies, Brand switching behaviour may lead to corporate profits decline by 30% - 40%. Other research (Rosenberg and Czepiel, 1984) has shown that the workload for attracting a new customer is 5-9 times compared to the workload for keeping an existing customer. And the cost for a new customer is 3-5 times of the existing customer.
It is clear that profits brought by existing customers are more than new consumers. What’s more, increasingly numbers of people are willing to give their comments on online communities or follow some useful advices from online communities. From the perspective of companies and managers, it is important to figure out the effect of negative eWOM on brand loyalty since the brand switching behaviour caused by negative eWOM may bring huge losses for corporations. If the companies could understand how this effect happen, measures can be taken to keep customer’ loyalty and reduce the losses.
1.2 Research gap
Firstly, majority of research focus on the positive word-of-mouth and how to use it for extend their market (Brown et al., 2005; Maxham III, 2001, File and Prince, 1992)).
Although more and more scholars believe that negative word-of-mouth has greater influence on customer behaviour, the literatures of negative word-of-mouth are limited compared to positive word-of-mouth. With the rapid development of Internet, the potential threat of
9 | P a g e negative eWOM has strengthened from the perspective of the companies. Hence, the study on negative eWOM has attracted more concern than before.
Secondly, when we review the literature related to brand loyalty especially the study of brand switching behaviour, the scholars have carried out extensive research. Such as brand switching behaviour caused by customer’ dissatisfaction (Richins, 1983), the relationship among customer complaints, negative word-of-mouth and brand switching behaviour (Singh, 1990) and why brand switching behaviour occurs (Keaveney, 1995). However, the scholars mainly focus on the impact of consumer's personal subjective feelings (such as price, service failures and attraction of competitors, etc.) on brand switching behaviour, but relatively few literatures investigate whether the brand switching behaviour will cause by the impact of others (such as receiving word-of-mouth spread by their families, friends or even strangers).
Besides, most companies pay much attention to customer relationship management and investigate how companies could improve the quality of services and provide innovation and convenience for customers to strengthen customer loyalty (Verhoef, 2003; Reinartz, Krafft and Hoyer, 2004). This thesis tries to discuss the factor (negative word-of-mouth) which keep customers away from the brand and how to reduce its negative impacts. It can help companies have a deeper understanding of how to strengthen customer loyalty from another perspective. The development of the Internet makes the negative word-of-mouth become more dangerous compare to traditional environment. Therefore, figuring out the influence mechanism of negative eWOM on brand loyalty on can help the company reducing the loss of negative eWOM. And companies could take appropriate measures to avoid the deterioration of negative eWOM which lead to brand switching behaviour.
Last but not least, according to the previous articles, Yin (2011), Park and Lee (2009), Fan and Miao (2012) some articles investigate the influence of negative eWOM on consumers
10 | P a g e purchase intention while several scholars (Cho et al., 2002; Park, Lee and Han, 2007;
Sheng-Hsien, 2009) concentrate on online complaint management or the effect of online reviews in online shopping website or virtual opinion platforms on purchase intention or customer loyalty. Fewer articles focus on the influence of eWOM on brand loyalty from the perspective of online communities.
Based on what has mentioned above, it is necessary and important to focus on the effect of negative eWOM on brand loyalty from the perspective of online community involvement, which has significance for the development of corporations.
1.3 Main research question
The main research question of this thesis can be formulated as follow:
What is the effect of negative eWOM on brand loyalty from the perspective of online community involvement?
In order to investigate how this effect happen, the main research question could be divided into several sub research questions.
1. What is the effect of online community involvement on intensity of negative eWOM?
2. What is the effect of trust tendency on intensity of negative eWOM?
3. What is the effect of intensity of negative eWOM on brand loyalty?
1.4 Structure of the thesis
This thesis continues with the following chapters. Chapter 2 will review the literature of concepts and topics that are relevant in this thesis, which including online communities, trust tendency, word-of-mouth, eWOM and brand loyalty and. Then, the conceptual model and hypotheses will be built based on the literature review and also the relationship among
11 | P a g e the concepts will be presented in Chapter 3. Chapter 4 describes the main research method.
And Chapter 5 presents the development of the questionnaire, the analysis of the data and empirical results. The thesis ends with Chapter 6, which includes discussion, limitations of this study, conclusions and implication.
12 | P a g e
2. LITERATURE REVIEW
2.1 Online communities
The terms of online communities, also known as virtual community, was firstly proposed by Howard Rheingold in 1993.
Table 2.1 Definitions of online community, Source: adapted from several scholars
13 | P a g e Different scientific areas have studied for the online community or virtual community, which include management, marketing, sociology, psychology, computer science and etc.
These researchers tend to present the definition of online community according to their own needs. Main definitions of online community have shown in table 2.1. It can be seen from table 2.1 that the definition of online community has the characteristics of interdisciplinary.
Scholars choose the angels of sociology, e-commerce, knowledge sharing, and technical to define online communities. We can say that online community is a comprehensive concept which can be studied from different perspective.
Based on the definition of online community which mentioned above, we can figure out the conclusion that online community has three common point:
1. The online community is based on computer or Internet communication medium, which can across the boundaries of space.
2. More than two community members who have common interests and other factors, such as have same background or target, aggregate together to become a community.
3. The community members may interact each other and build new value and relationship via the online community.
14 | P a g e Hence, we can define the online community as: More than two individuals or groups, who has common interest, background, identity, use computer and Internet as media to interact and aggregate together to build new social relationship and values.
2.2 Trust tendency
The early study of trust is in the field of psychology, which focus on the interaction between trust and interpersonal relation. Until later, trust has been studied in sociology, organizational behaviour and economics and other subjects.
Rober (1967) suggests that trust is the expectation that people judge whether others’
commitment is reliable. Boon and Holmes (1991) think that trust is reliability or positive expectation when others may have possibility to break their words.
In the field of marketing, most research on trust is about the issue of trust exist in the transactions (Chiles and McMackin, 1996; Tan and Thoen, 2001). The starting point of this study was to examine the trust from perspective of sociology and psychology, which focus on not only individual characteristics, but also interaction between the two parties during the transaction. Although the concerns about the study of trust from different area are different, many scholars are unanimously recognized that trust has a very important position in the study of human behaviour. Besides, it is believed that the more you trust someone, the more information will transfer between you two (Abrams et al., 2003).
Besides, McKnight, Choudhury and Kacmar (2002) investigate the trust under virtual environment and presented a web trust model which widely cited in various articles (see figure 2.1). This model contains three factors, which are disposition to trust (the overall trust tendency of a person), institution-based trust (perceptions of the Internet environment) and trusting beliefs (perceptions of specific web vendor attributes).
15 | P a g e Figure 2.1 Web trust model, Source: adapted from McKnight, Choudhury and Kacmar (2002)
2.3 Word-of-mouth
Word-of-mouth has been intensive studied and discussed by scholars for more than half century. Word-of-mouth (WOM) is the way that utilizes oral correspondence to transmit information between individuals. It has been regarded as one of the most powerful assets of information transmission since the start of human world (Godes and Mayzlin, 2004; Duan et al., 2008). However, the impact and spread speed of traditional WOM communication are relatively slow since it is enslaved to social contact limits (Bhatnagar and Ghose, 2004;
Ellison and Fudenberg, 1995). From the perspective of consumers, word-of-mouth is one of the most important referenced factors before the purchasing behaviour occurs. In marketing area, word-of-mouth involves the information exchange between a non-commercial communicator and a receiver concerning a brand, a product, or a service. Besides, word-of- mouth can serves as “a powerful marketing intensifier” which can be enhanced by social media. Table 2.2 summarize the definitions of WOM from previous articles.
Table 2.2 Definitions of WOM, Source: adapted from several scholars
16 | P a g e In addition, it has been recognised by the scholars that word-of-mouth has the characteristics of positive and negative (Richins, 1983; East, Hammond and Lomax, 2008; Cheema and Kaikati, 2010). The study about positive and negative word-of-mouth can be traced back to when the scholars start to study word-of-mouth. At that time, scholars mainly focus on the differences of spread effect between positive word-of-mouth and negative word-of-mouth.
Besides, research has showed that positive and negative word-of-mouth may have difference in affecting the consumer psychology. East et al. (2008) point out that positive word-of- mouth boosts buying behaviour, while negative word-of-mouth discourages purchases.
Chevalier and Mayzlin (2006) suggest that positive word-of-mouth can reduce marketing costs and increase corporate profits (successfully guide the new customers to buy their
17 | P a g e products). However, if negative word-of-mouth appears, it may reduce the reliability of product advertising. Thus, word of mouth just like a double-edged sword, positive word-of- mouth can help corporations setting up good reputation, and even increases profits. But negative word-of-mouth will damage the good image of the corporations and bring even more harm.
For consumers, negative word-of-mouth often attract more attention than positive word-of- mouth. Sweeney, Soutar and Mazzarol (2005) hold the view that the negative word-of- mouth often affect customers more than positive word-of-mouth because of the risk aversion psychology. Several studies (Richins, 1983; Blodgett, Granbois and Walters, 1994) indicate that the negative word of mouth is that consumers share their unsatisfied experience of products or services with others or the intercommunication between the consumers about their discontent with the corporations or products. Richins (1983) also reveal that one of the most important explanatory variables of negative word-of-mouth is dissatisfaction. Richins (1983) pointed out complain (feedback to dissatisfaction) has at least three different activities which are switching brand behaviour (that is, exit), making a complaint to the seller (that is, voice) and sharing their unsatisfied experience to other (that is, negative word-of-mouth). This idea is summarized by Singh (1990) and refer Hirschman’s (1974) Exit-Voice-Loyalty model to “predict and explain variation in voice, exit, and negative word-of-mouth”.
Based on what has discussed above, we can see that most scholars define the negative word of mouth emphasized the word " dissatisfaction" and use emotional elements to explain.
Hence, we define negative word-of-mouth as unpleasant word-of-mouth, which aim to discouraging other consumers to buy the specific product.
18 | P a g e And this thesis will mainly focus on the negative WOM, which consumers discourage certain brands, products and services. Their feedbacks normally contain complaints and criticism.
2.4 Electronic word-of-mouth
With the evolution of the information technology and the development of online social website, the way of information transmission has changed a lot. To be more specific, when word-of-mouth spread through electronic medium, it can be regarded as electronic word-of- mouth (eWOM) which refers to any opinions consumers express via the Internet (e.g., web sites, social networks, instant messages or web feed) about a product, service, brand, or company (Hennig‐Thurau et al., 2004). Table 2.3 conclude the definitions of eWOM from previous articles.
Table 2.3 Definitions of eWOM, Source: adapted from several scholars
The eWOM often spread with the corresponding text or image. Besides, it is more difficult to remove and can be browsing at any time. Therefore, after the consumer spreads eWOM
19 | P a g e via the Internet, the receiver will be not limited to to only the friends around them. Instead, anyone who surf the Internet can access to the eWOM. Doh and Hwang (2009) pointed out that with the increase of communication between customers based on network. It has brought new business opportunities for the corporation and also can bring more threat. In conclusion, compared with the traditional word-of-mouth, this new form of word-of-mouth increases strength and range of influence.
The main distinction between word-of-mout and eWOM is that eWOM often spread through strangers and hence weak ties exist between them (Gruen, Osmonbekov and Czaplewski, 2006). On the other hand, transforming information in traditional word-of-mouth mostly happens between small groups of individuals whereas by eWOM often includes many individuals. In addition, eWOM communications are more persistence, accessible and more measurable than the traditional word-of-mouth. More and more companies are aware of the effect of its strong dissemination and diffusion, and use eWOM as marketing tools.
Since this thesis mainly focuses on negative word-of-mouth, we define the concept of negative eWOM as: consumers express their discontent about product or service based on unpleasant consumption experience via a variety of network communication platform.
2.5 Customer loyalty
The term customer loyalty is used to describe the behaviour of repeat customers, as well as those who offer good ratings, reviews, or testimonials (Dick and Basu, 1994). Some customers may show special preference to a particular company or service by express positive word-of-mouth to their friends and family, and advising them becoming loyal customers. On the other hand, low customer loyalty will lend brand switching behaviour.
20 | P a g e Brand switching behaviour is the performance of customers’ consumption psychology and decision in brand choosing. Zeithaml, Rust and Lemon (2001) declared that the decisive factor of company's success is not find customers but keep them. In other words, brand switching behaviour should be avoided. Keaveney (1995) believed that the brand switching behaviour should be defined as clients intend to stop consuming present brand, service or products but switch to an alternative choice. This behaviour means that the companies lose of their customers and also benefits.
In addition, some scholars thought brand switching behaviour should be defined as a completely stop consumption of some products or services (Jensen and Hansen, 2006). But others held a different opinion. They believed that this behaviour is a gradually and dynamic withdraw process (Clemes, Gan and Zhang, 2010). Although academic researchers define different definitions of brand switching behaviour, they share a same viewpoint that brand switching behaviour is that customers switching from original brand to new brand or changing their consumptive decisions, caused by either internal or external causes.
At the very beginning, research on brand switching behaviour did not stand in the spotlight of consumers’ behaviour study. However, more and more scholars realize that brand switching behaviour might lead to a huge interest loss to manufactories (Knox and Walker, 2001; Van Heerde, Gupta and Wittink, 2003). Hence, this research had drawn more and more attention to both researchers and corporations.
Reynolds and Beatty (1999) indicated that it might cause a negative influence upon market share and profits of companies when relevant customers change their choice. Rust and Zahorik (1993) also pointed out that switching behaviour would cut beneficial rate and market share of companies. If customers kept their brand loyalty, these companies would save cost in advertisement. Or else, enterprises will lose previous investment in client
21 | P a g e attraction. Furthermore, they have to spend extra money to find new customers (Jones, Mothersbaugh and Beatty 2002).
Keaveney’s research (1995) revealed that consumers’ purchasing intention would be affected by prevailing of negative word-of-mouth. If a customer switches brand, other customers might be influenced by his or her negative information about this brand. Potential clients may incline to looking for new service providers. The same behaviours create new negative appraisal and more customers would be affected. Thus, a vicious circle is established in the market. Corporations may loss a lot for lacking of clients.
22 | P a g e
3. CONCEPTUAL MODEL AND HYPOTHESES
3.1 Hypotheses development
Essentially speaking, that consumers receive and react negative eWOM and the occurrence of brand switching behaviour is a procedure of information processing.
To be specifically, various factors will affect the information receivers of perception of negative eWOM. These factors contain not only the characteristics of information itself but also the source of the negative eWOM, the relationship between the receiver and disseminator and also receivers’ personality characteristic, which is the root of why eWOM can cause effects.
Specific to this thesis, in the online community, online communities dependence, virtual relationships and trust tendency will affect perception of negative eWOM and then affect brand loyalty. In this way, brand switching behaviour may happen.
3.1.1 The online community involvement
In the previous research, the factor that affect eWOM mainly focused on characteristic of disseminator (professional degree), the content of word-of-mouth (interestingness, value) and etc. (Hovland, 1953; Rieh, 2002) With the development of the research, scholars begin to pay attention to the role of environmental elements play on reliability and influence of eWOM.
In traditional word-of-mouth, the relationships between disseminators and receivers will affect receivers’ judgment towards to information and word-of-mouth spreading (Dichter, 1966; Anderson, 1998). Bristor (1990) point out that one is willing to tell others his or her true feelings if the tie between them is close enough. To be more specific, the receiver
23 | P a g e would acquire and accept more information from familiar relatives or friends (Bansal and Voyer, 2000, Writz and Chew, 2002).
Extended to the Internet and online community, social relationships and network involvement is still the key point that scholars mainly focus on. Doh and Hwang (2009) confirmed that the consumer involvement of network would affect their identification of network information and also the spread of eWOM. Besides, researchers worked a lot upon whether online relationship still have strong impact on word-of-mouth in the virtual network environment. Gilly et al. (1998) study verified if the information disseminators and receiver thought have same characteristics or share same interests, the information can be more persuasive. Vilpponen, Winter and Sundqvist (2006) present that information receiver in the online community can still feel similarity and intimacy with information disseminators, although maybe they never met or knew each other.
From the perspective of online communities dependence, existing research indicates that individuals will increase their trust on the medium with the increasing number of using the medium. In other words, people tend to believe their preferred medium. The usage and dependence of online communities will affect the initiative searching consciousness, which may increase the contact to intensity of negative eWOM.
Besides, In terms of virtual relationships, individuals tend to acquire information in a relations-close environment and have a relatively high recognition degree of the environment (Metzger, Flanagin and Zwarun, 2000). Hence, the relationship between the word-of-mouth recipient and online community could affect the quantity and reliability of eWOM.
Bearing these considerations in mind, we proposed the following hypotheses:
24 | P a g e H1A: The level of online communities dependence is positively associated with the intensity of negative eWOM.
H1a: The level of online communities dependence is positively associated with the quantity of negative eWOM.
H1b: The level of online communities dependence is positively associated with the reliability of negative eWOM.
H1B: The intensity of virtual relationships is positively associated with the intensity of negative eWOM.
H1c: The intensity of virtual relationships is positively associated with the quantity of negative eWOM.
H1d: The intensity of virtual relationships is positively associated with the reliability of negative eWOM.
3.1.2 Trust tendency
In the field of communication and persuade, trust means creditability, which will affect how much the information the receiver will accept (McKnight and Chervany, 2002). Gefen (2000) discusses the importance of trust in online communities on exchanging information, which point out when high trust tendency exist between the community members, community members tend to share their personal privacy and are willing to respond quickly. This behaviour will positively affect trust between members and enhance the intention to share their idea and information, which may increase the quantity reliability of eWOM.
We could say that the trust between members of online community is the most valuable asset of the community. Blanchard and Markus (2004) point out that the existence of trust in
25 | P a g e online community enhances community members’ sense of belonging and dependency.
When this positive sense of community reaches certain level, the community members may behave similar like in reality communities.
Nowadays, before the consumers do online shopping, they tend to acquire useful information also word-of-mouth of the target product from the Internet especially in the online communities. Hence, online communities members may give some suggestion or comments based on their experience (eWOM), which affect the purchasing intention.
Specific to Internet environment, the network communication is based on computer-based information technology without face-to-face contact. Hence, trust tendency will greatly affect the reliability of word-of-mouth via the Internet. In other words, those who trust people easily may receive more information from others and affect more by the eWOM.
Therefore, we raise the following hypotheses:
H2: The level of trust tendency is positively associated with the intensity of negative eWOM.
H2a: The level of trust tendency is positively associated with the quantity of negative eWOM.
H2b: The level of trust tendency is positively associated with the reliability of negative eWOM.
3.1.3 The intensity of negative eWOM
Arndt (1967) found out that decrease rate of food selling caused by negative word-of-mouth is more than twice higher than the rate increased by positive word-of-mouth. Several article (Kotler, 1994, Mangold et al., 1999) mentioned that consumers tend to share their unpleasant consuming experiences to others compared with pleasant consuming experiences.
Lau and Ng (1998) pointed out people would pay more attention to negative information when a new product or service launch to the market.
26 | P a g e For a specific person, people will have stronger reaction to negative information compared positive information even if the degree of two kinds of information is same (Fiske, 1980;
Skowronski and Carlston, 1987). This is so called negative influence, which affects consumption behaviour a lot. And the Internet environment would be an amplifier to enhance this phenomenon, which cause huge losses for the companies.
In previous study of brand switching behaviour, scholars mainly focused on the consumers’
subjective feeling, such as the perception of prices, service failure or more attractive products or services provided by competitors. Except for these factors, several authors stated that the word-of-mouth, as the objective environmental factor, would also have a big impact on customer behaviour. And compared with the positive word-of-mouth, it is generally believed that negative word-of-mouth have greater impact (Casaló, Flavián and Guinalíu, 2008; Gruen, Osmonbekov and Czaplewski, 2006). Negative word-of-mouth is so powerful that consumers’ behaviours and choices could be affected. It not only results in companies lost their benefits, but also decreases customers’ loyalty. (Rust and Zahorik, 1993; Money, 2004) With the development of Internet technology, the online community play increasingly important role in consumption field. Under this background, the effect of negative eWOM on switching-brand behaviour will become even bigger. Hennig‐Thurau et al., (2004) pointed out that negative eWOM will influence cognitive trust and emotional trust of information receiver, thereby affecting the purchase decision and causing brand switching behaviour. To be more specific, the quantity of eWOM will determine the frequency that information recipient receive the negative eWOM (Liu, 2006). Besides, the reliability of eWOM is the decisive factor that affects the spread speed. Both of two factors will impact the degree of perception that the information recipients receive from eWOM, which may result in brand switching behaviour. If consumers receive the negative word-of-mouth of what they want to buy, they could abandon the brand and choose other brands.
27 | P a g e Therefore, the following hypotheses are formulated:
H3: The intensity of negative eWOM is positively related with the brand switching behaviour.
H3a: The quantity of negative eWOM is positively related with the brand switching behaviour.
H3b: The reliability of negative eWOM is positively related with the brand switching behaviour.
3.2 Conceptual model
Based on the literature review and hypothesis development part, we proposed the conceptual model (see figure 3.1) to investigate the relationship among the variables mentioned below.
We assume that the online community involvement (contain online communities dependence and virtual relationships) and trust tendency will affect the intensity of negative eWOM (contain the quantity of eWOM and the reliability of eWOM). And the intensity of negative eWOM will impact the customer loyalty. Besides, we use brand switching behaviour to measure the customer loyalty. Hence, brand switching behaviour has been considered as the final output result of this research.
Figure 3.1 Conceptual model
28 | P a g e
4. METHODOLGY
“A research design is intended to convince the reader that the proposed work is significant, relevant, and interesting; that the design of the study is sound; and that the researcher is capable of successfully conducting the study.” (Blaikie 2000: 12) Hence, it is very important to formulate a good research design structure and select appropriate research methods.
4.1 Research design strategies
Based on the Yin (2003), there are three different research design strategies named descriptive, explanatory and exploratory respectively. A combination of exploratory strategy will be used in this thesis (see figure 4.1). In general, explanatory research will be adopted when the main factors are defined and want to understand the relationships exist between them (Thomas, 2004). These explanatory studies often verify the hypotheses to figure out the explanations of the essence of certain relationships. For this study, three relationships should be verify through the hypothesis testing approach, which are the relationship between the online community involvement and the intensity of negative eWOM, the relationship between the trust tendency and the intensity of negative eWOM and the relationship between the intensity of negative eWOM and customer loyalty.
Figure 4.1 Research design strategies used in the thesis
Research design strategies
Explanatory Descriptive Exploratory
29 | P a g e 4.2 Research approaches
There are two main research approaches which are deductive and inductive for research project (Patton, 2001). According to Jonker and Pennink (2010), deductive approach process generally starts with theory, then narrow into more specific hypotheses, next use empirical research to test it and finally get confirmation. On the other hand, inductive approach process begins with specific observations, and then detects patterns, next formulate some hypotheses to explore, and finally develop new theory. To be general, deductive approach is for testing a theory and inductive approach is for building theory. Deductive approaches will be used in this thesis (Figure 4.2).
Figure 4.2 Research approaches used in the thesis
4.3 Data collection
There are two basic types of research data, which are primary research data and secondary research data. Just as its name implies, primary data refer to the original data collected by the researcher. And secondary data refer to the data collected from existing research.
Figure 4.3 Data collection methods used in the thesis
Research approaches
Deductive Inductive
Data collection
Primary research data Secondary research data
30 | P a g e For this study, most of data is primary research data which can be acquired from questionnaire according to the hypotheses (Figure 4.3).
4.4 Research methods
Research methods can be often divided into two main types which are qualitative research and quantitative research. Qualitative research is research in which “the researcher makes an attempt to understand a specific organizational reality and occurring phenomena from the perspective of those involved” (Jonker and Pennink, 2010:78) And researchers should comprehend the phenomenon rather than apply previous existing theories to build a new theory or “mini-theory” which is used for one specific condition. Moreover, quantitative research refers using sampling techniques which often include numerical data to explain the phenomena (Creswell, Clark and Gutmann, 2003). Quantitative research will be used in the thesis (See Figure 4.4).
Figure 4.4 Research methods used in the thesis
The main research methods will be divided into four steps:
1. Questionnaire in accordance with hypotheses will be designed. All the questions are based on the research question and literature review. How this questionnaire is built will be presented in next chapter (see chapter 5.1). The questionnaire is anonymous. And all participants are voluntary to finish the questionnaire.
Research methods
Qualitative research Quantitative research
31 | P a g e 2. The questionnaires will be launched at open online survey platform (qualtrics.com). And the link of the online survey will be shared in social network site like Facebook, Twitter, Weibo and so on. Then, using snowball sampling to spread this questionnaire. Hence, anyone who have interested in this study can complete this questionnaire voluntarily.
3.The questionnaires will be collected and codified. All the data is based on the information collected by the questionnaires. The potential personal information will be removed from questionnaires and the researcher will only record and identify the questionnaires by numerical code.
4. For data analysis, the first phase will calculate the value of Cronbach’s Alpha, which test the reliability of the questionnaire. The second phase: If the value of Cronbach’s Alpha is high enough (normally above 0.7), we can calculate the mean across all related question according to the variables. And the average value of the question will be used in following analysis. After that, correlation analysis (Pearson product-moment correlation coefficient method) will be used to test whether there is relationship between the variables. The final phase is to run regression analyses to test the hypothesis. All data analysis will proceed in statistic software SPSS 20.
32 | P a g e
5 DATA ANALYSIS
5.1 Questionnaire design
Conceptual model (see figure 3.1) contains the concept of online community involvement, trust tendency, intensity of negative eWOM and customer loyalty. Questionnaire is designed based on the conceptual model and previous study (see table 5.1).
Table 5.1 Questionnaire design, Source: adapted from several scholars
Variables Source Number of question
Online communities dependence Kim et al., (2006) 4
Virtual relationships Frenzen and Davis (1990) 4 Trust tendency Gefen, Karahanna and Straub
(2003) 3
The quantity of negative eWOM Schubert and Selz (1999) 2 The reliability of negative
eWOM
McKnight, Choudhury and
Kacmar, (2002) 3
Brand switching behaviour Bansal and Voyer, (2000) 3
And the variables and correlative questionnaires are presented below (see table 5.2). And all the questions use seven levels Likert scales to measure, which are “Strongly disagree”,
“Disagree”, “Somewhat Disagree”, “Neither Agree nor Disagree”, “Somewhat Agree”,
“Agree” and “Strongly agree”.
Table 5.2 Variables and correlative questionnaires
Concept Variable Questions
Online community involvement
Online communities
dependence
A1: I often use online community.
A2: I tend to acquire useful information or suggestions from this online community.
A3: I enjoy the happiness brought by this online community.
A4: I am often absorbed in this online community.
33 | P a g e Virtual
relationships
B1: When I post a question in online community, I often receive replies from other net friends.
B2: I often participate in the discussion of topic in online community.
B3: I am willing to communicate with net friends in online community and build good relationships with them.
B4: I enjoy communicating with others via the online community.
Trust tendency
C1: I tend to believe that human nature can be trusted.
C2: I usually choose trusting others.
C3: I tend to trust others unless they have untrustworthy reason.
The intensity of
negative eWOM
The quantity of eWOM
D1: I have ever read such kind of posts or comments (negative eWOM) for many times in online community.
D2: There are a number of community friends to share or publish such kind of posts or comments (negative eWOM).
The reliability of eWOM
E1: These kinds of posts or comments (negative eWOM) in online community are often trustworthy.
E2: These kinds of posts or comments (negative eWOM) in online community are often clear and easy to understand.
E3: These kinds of posts or comments (negative eWOM) in online community often have a certain reference value.
Brand loyalty
Brand switching behaviour
F1: After browsing the posts or comments (negative eWOM) of one brand, I tend to use other brand's products.
F2: After browsing the posts or comments (negative eWOM) of one brand, I will not continue to use the products of this brand.
F3: After browsing the posts or comments (negative eWOM) of one brand, I wouldn't recommend my friends using the products of this brand.
34 | P a g e 5.2 Research target
The main research participants are young people since the majority of active group in social network site is young people. Besides, they are familiar with the online community and they can easily understand the concept of online word-of-mouth, which can enhance the accuracy of the study. 223 questionnaires were filled in website qualtrics.com and 153 questionnaires were valid. The valid questionnaires occupy about 69% of the entire questionnaires (See figure 5.1). And descriptive statistics can be found as follow.
Figure 5.1 Valid rate of the questionnaires
Figure 5.2 The Nationality of the respondents
It can be seen from the pie chart (see figure 5.2) that the respondents come from 15 different countries. And nearly half of respondents (49%) are Chinese while 13%, 10% and 9% of
35 | P a g e respondents are from Netherlands, Germany and the Untied Kingdom. Except this, the number of respondents from other countries is relatively small.
Table 5.3 Descriptive statistics, Gender
Gender Frequency Percent Valid Percent Cumulative Percent Valid
Male 81 52.9 52.9 52.9
Female 72 47.1 47.1 100.0
Total 153 100.0 100.0
Table 5.4 Descriptive statistics, Age
Age Frequency Percent Valid Percent Cumulative Percent
Valid
18 1 .7 .7 .7
19 3 2.0 2.0 2.6
20 4 2.6 2.6 5.2
21 6 3.9 3.9 9.2
22 22 14.4 14.4 23.5
23 24 15.7 15.7 39.2
24 27 17.6 17.6 56.9
25 25 16.3 16.3 73.2
26 8 5.2 5.2 78.4
27 6 3.9 3.9 82.4
28 3 2.0 2.0 84.3
29 2 1.3 1.3 85.6
30 2 1.3 1.3 86.9
31 1 .7 .7 87.6
32 1 .7 .7 88.2
34 2 1.3 1.3 89.5
38 1 .7 .7 90.2
41 3 2.0 2.0 92.2
42 4 2.6 2.6 94.8
43 3 2.0 2.0 96.7
44 2 1.3 1.3 98.0
45 2 1.3 1.3 99.3
46 1 .7 .7 100.0
Total 153 100.0 100.0
36 | P a g e Table 5.5 Descriptive statistics, Age & education level
What is your age? What is the highest level of education you have completed?
N Valid 153 153
Missing 0 0
Mean 25.98 2.47
Minimum 18 1
Maximum 46 4
Figure 5.3 Histogram of education level
The tables and figure above show the descriptive statistics. The sample consists of 81 (52.9%) men and 72 (47.1%) women, with an average age of 26 (Mean of age=25.98). The majority of respondents are 18 to 25 years old, which occupy 73.2% of whole sample.
Furthermore, education level ranges from high school to doctoral degree, and most of respondents have acquired bachelor degree (see table 5.5).
5.3 Reliability analysis
This part use SPSS to do the reliability analysis for all variables. Reliability refers to dependability, consistency, stability and precision of the methods. For the questions that describe same dimension of concept, if the answers of these questions are equal or similar,
37 | P a g e we can admit the methods are dependable. In this thesis, Cronbach’s Alpha is used to analyse the reliability for all variables. In general, if the Cronbach’s Alpha is above 0.6, it means the reliability of the questionnaire is accepted. If the Cronbach’s Alpha is above 0.7, it means that the questionnaire has relatively high reliability. The results are presented as follow (see table 5.6):
Table 5.6 Reliability analyses of variables
Variables Cronbach’s Alpha N of Items
Online communities dependence 0.821 4
Virtual relationships 0.757 4
Trust tendency 0.838 3
The quantity of eWoM 0.735 2
The reliability of eWOM 0.792 3
Brand switching behaviour 0.766 3
All variables 0.885 19
According to the table 5.6, we can find that all Cronbach’s Alpha of variables is above 0.7.
The questionnaire has certain reliability.
Since all the Cronbach’s Alpha of variables is relatively high, we could use average value method to present the variables (see table 5.7).
Table 5.7 Expression of variables
Variables Expression
Online communities dependence A=(A1+A2+A3+A4)/4
Virtual relationships B=(B1+B2+B3+B4)/4
Trust tendency C=(C1+C2+C3)/3
The quantity of eWOM D=(D1+D2)/2
The Reliably of eWOM E=(E1+E2+E3)/3
Brand switching behaviour F=(F1+F2+F3)/3
38 | P a g e 5.4 Correlation analysis
SPSS is used to conduct the correlation analysis for all variables. The correlation analysis mainly used for test whether there is relation between two variables and how close of this relationship. Correlation coefficient is the statistic that describes the degree of linear relation and the direction of this relation. In this thesis, we use Pearson product-moment correlation coefficient method.
Figure 5.4 Correlations matrix
Figure 5.4 (correlations matrix) shows the correlation between different variables. We can see from this correlation matrix that the variables are positively correlated in 0.01 significant levels. The hypotheses of H1A, H1B, H2 and H3 have been initially verified.
5.5 Regression analysis
The correlation analysis could explain the correlation between the variables, but it couldn’t explain the causality between the variables. In order to investigate the causality between the variables and farther test the hypotheses, we have to use the regression analysis in this thesis.
Regression analysis is a statistical process for estimating the relationships among variables.
Three groups of variables will be tested in this thesis, which are group 1 (online communities dependence, virtual relationships, trust tendency and the quantity of eWOM),
39 | P a g e group 2 (online communities dependence, virtual relationships, trust tendency and the reliability of eWOM) and group 3 (the quantity of eWOM, the reliability of eWOM and brand switching behaviour)
5.5.1 The regression results of online communities involvement, trust tendency and the quantity of eWOM
We set online communities involvement (online communities dependence, virtual relationships) and trust tendency as independents variables and set the quantity of eWOM as dependent variable. The results are presented as follow (see table 5.8, table 5.9).
Table 5.8 Coefficient (Dependent Variable: the quantity of eWOM)
Model Unstandardized Coefficients Standardized
Coefficients t Sig.
B Std. Error Beta
1
(Constant) 2.121 .371 5.722 .000
Online communities
dependence .188 .085 .215 2.217 .028
Virtual relationships .097 .091 .103 1.059 .291
Trust tendency .244 .067 .288 3.657 .000
Table 5.9 Model summary (Dependent Variable: the quantity of eWOM)
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .489a .239 .223 0.87063
a. Predictors: (Constant), trust tendency, online communities dependence, virtual relationships According to the table 5.8 and table 5.9, R square equal to 0.239, which means the three independents variables (online communities dependence, virtual relationships, trust tendency) can explain 23.9% of the total variance. And hypothesis H1a and H2a are supported (Sig<0.1) while H1c are rejected (Sig>0.1).
40 | P a g e 5.5.2 The regression results of online communities involvement, trust tendency and the reliability of eWOM
We set online communities involvement (online communities dependence, virtual relationships) and trust tendency as independents variables and set the reliability of eWOM as dependent variable. The results are presented as follow (see table 5.10, table 5.11).
Table 5.10 Coefficient (Dependent Variable: the reliability of eWOM)
Model Unstandardized Coefficients Standardized
Coefficients t Sig.
B Std. Error Beta
1
(Constant) 2.454 .409 6.002 .000
Online communities
dependence .087 .094 .116 0.932 .353
Virtual relationships .116 .101 .093 1.146 .253
Trust tendency .200 .073 .233 2.730 .007
Table 5.11 Model summary (Dependent Variable: the reliability of eWOM)
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .360a .130 .112 0.96027
a. Predictors: (Constant), trust tendency, online communities dependence, virtual relationships According to the table 5.10 and table 5.11, R square equal to 0.130, which means the three independents variables (online communities dependence, virtual relationships, trust tendency) can explain 13% of the total variance. And hypothesis H2b is supported (Sig<0.1) while H1b and H1d are rejected (Sig>0.1).
5.5.3 The regression results of the intensity of negative eWOM and brand switching behaviour
We set the intensity of negative eWOM (the quantity of eWOM, the reliability of eWOM) as independents variables and set the brand switching behaviour as dependent variable. The results are presented as follow (see table 5.12, table 5.13).
41 | P a g e Table 5.12 Coefficient (Dependent Variable: brand switching behaviour)
Model Unstandardized Coefficients Standardized
Coefficients t Sig.
B Std. Error Beta
1
(Constant) 1.067 .392 2.276 .007
The quantity of eWOM .260 .076 .241 3.419 .001
The reliability of eWOM .476 .074 .455 6.465 .000
Table 5.13 Model summary (Dependent Variable: brand switching behaviour)
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .582a .339 .330 0.87098
a. Predictors: (Constant), the reliability of eWOM, the quantity of eWOM
According to the table 5.12 and table 5.13, R square equal to 0.339, which means the three independents variables (the quantity of eWOM, the reliability of eWOM) can explain 33.9%
of the total variance. And hypothesis H3a and H3b are supported (Sig<0.1).
5.6 Findings
Based on the data analysis in this chapter, we test the hypotheses which present in chapter 3, the summary of empirical results can be found in table 5.14.
Table 5.14 Summary of empirical results
Hypotheses Results
H1A: The level of online communities dependence is positively associated with the intensity of negative eWOM.
Partly support H1a: The level of online communities dependence is positively associated
with the quantity of negative eWOM. Support
H1b: The level of online communities dependence is positively associated
with the reliability of negative eWOM. Reject
H1B: The intensity of virtual relationships is positively associated with the
intensity of negative eWOM. Reject
42 | P a g e H1c: The intensity of virtual relationships is positively associated with the
quantity of negative eWOM. Reject
H1d: The intensity of virtual relationships is positively associated with the
reliability of negative eWOM. Reject
H2: The level of trust tendency is positively associated with the intensity of
negative eWOM. Support
H2a: The level of trust tendency is positively associated with the quantity of
negative eWOM. Support
H2b: The level of trust tendency is positively associated with the reliability of
negative eWOM. Support
H3: The intensity of negative eWOM is positively related with the brand
switching behaviour. Support
H3a: The quantity of negative eWOM is positively related with the brand
switching behaviour. Support
H3b: The reliability of negative eWOM is positively related with the brand
switching behaviour. Support
43 | P a g e
6. DISCUSSION & CONCLUSION
6.1 Discussion
Based on the literature and empirical results which present above, several research findings will be discussed as follow.
Firstly, the finding shows that the level of online communities dependence has positive impact on the quantity of negative eWOM, which has been also proved in previous study (Jansen et al., 2009; Park and Kim, 2009). If the community members with high level of online community dependence, they will tend to often use online community and spend much time browsing the post in the online community. Accordingly, they will acquire more information from the online community than members with relatively low level of online community dependence, which increase the chance of acquiring the negative eWOM. Hence, it is manifest that high level of online communities dependence will lead to rise of quantity of negative eWOM.
Secondly, the results indicate that the effect of virtual relationships on the intensity of negative eWOM is not significant. With the development of the Internet, the communication between network users and strangers increased dramatically. And people know each other through the Internet instead of face-to-face contact. Although communities’ members may share same interests and have similar background (Hill et al., 1995), their relationship will be relatively weaker than the relationship with their friends or families in the real life.
Besides, We can communication thousands of people via the Internet. With the increase number of virtual friends, the relationship between them may be weaker. Hence, This may decrease the spread of eWOM, which affect the quantity and reliability of eWOM.
44 | P a g e Thirdly, the results suggest that the level of trust tendency is positively associated with the intensity of negative eWOM. According to the web trust model presented by McKnight et al.
(2002), disposition to trust (trust tendency) will affect intuition-based trust (perceptions of the Internet environment). And both disposition to trust and intuition-based trust have positive impact on trusting intentions. In the virtual environment, there are less method of judging whether the information or word-of-mouth is truth compare to in the real world. The consumers have to make subjective judgement to distinguish the truth, which enhance the effect of trust tendency. In other words, those who trust people easily may receive more information from others and affect more by the eWOM.
Last but not least, the results confirm that the intensity of negative eWOM has impact on brand loyalty. To be more specific, the more intensity of negative eWOM, the more brand switching behaviour occurs. This finding is consistent with previous studies that not only the quantity of negative eWOM but also the reliability of negative eWOM are positively associated with the brand switching behaviour (Anderson, 2003; Amblee and Bui, 2007;Bowman and Narayandas, 2001).
When consumers receive negative eWOM of certain product, they will not accept it immediately. If consumers happen to use this product, they will tend to concern about whether there are other similar negative eWOM of this products. And then they will try to make their own judgement of this negative eWOM. When the consumers receive various similar negative eWOM from different information source, the effect of negative word-of- mouth will be definitively enhanced. The quantity of eWOM will affect the frequency of negative eWOM that you receive (Liu, 2006). The more negative eWOM you receive, the more possibility that your switch brand. Besides, the reliability of eWOM will affect the degree of perception that the information recipients receive from eWOM (Bone, 1995). And
45 | P a g e the viewpoint proved by Herr, Kardes and Kim (1991) is that products or services may be deleted from one’s list after a negative word-of-mouth had been accepted.
6.2 Limitation and recommendations for future study
Due to limited time and resource, there are several limitations in this thesis.
Firstly, this study mainly focuses on the angle of negative word-of-mouth receiver. Future study could try to investigate this topic also from the angle of disseminator and discuss the interaction between them.
Secondly, the sample number of this study is relatively small, which has only 153 samples.
And nearly half of respondents are Chinese. Hence, the results may have its limitation, which cannot adapt for all the conditions. In future study, it is recommended that larger samples should be used and the distribution of samples should be more dispersive, which can summarize the conclusions with more universal significance.
Thirdly, although the reliability of the questionnaire is relatively high, some questions in the questionnaire are built according to the articles which focus traditional word-of-mouth.
However, the key concept of study in this thesis is eWOM. Therefore, there may have error when measuring the variables by using this questions. It is suggested that future scholars develop the questionnaire aiming at eWOM.
6.3 Conclusion & Implication
Large amount of literatures have investigated the topic of word-of-mouth and brand loyalty.
With the rapidly development of the Internet, online community play an increasingly important role in consumption field. Based on this background, this thesis tries to discuss the effect of negative eWOM on brand loyalty from the perspective of online community involvement. In order to study the above questions, a theoretical model and assumptions
46 | P a g e contain the online community involvement, trust tendency, the intensity of negative eWOM and customer loyalty are built according to the literature review. This thesis carried on the questionnaire and 153 valid samples are used for data analysis. We find that the level of online communities dependence has positive impact on the quantity of eWOM. Besides, trust tendency is positively associated with quantity and reliability of negative eWOM.
What’s more, both quantity and reliability of negative eWOM has significant positive influences on brand switching behaviour. This study contributes to establish eWOM and brand loyalty model from the perspective of online community involvement, which make company have a more comprehensive understanding of influence mechanism of negative eWOM on brand loyalty.
Based on empirical results, some suggestions for companies’ implication are presented as follow.
(1) Improving the service recovery system to suppress the occurrence of negative eWOM
(2) Initiatively searching the sources of negative eWOM and actively respond the negative eWOM to reduce the spread of negative eWOM
(3) Building Internet communication platform to reduce the quantity of negative eWOM (4) Establishing relevant crisis management team to deal with the crisis caused by negative eWOM
47 | P a g e
APPENDICES
Reference list
Abrams, L. C., Cross, R., Lesser, E. and Levin, D. Z. (2003) Nurturing interpersonal trust in knowledge-sharing networks. The Academy of Management Executive, 17(4), pp. 64-77.
Amblee, N. and Bui, T. X. (2007) The Impact of Electronic-Word-of-Mouth on Digital Microproducts: An Empirical Investigation of Amazon Shorts. ECIS 2007 Proceedings, pp.
36-47.
Anderson, E. W. (1998) Customer satisfaction and word of mouth. Journal of service research, 1(1), pp. 5-17.
Anderson, E. W. (2003) The Formation of Market‐Level Expectations and Its Covariates.
Journal of Consumer Research, 30(1), pp. 115-124.
Amdt, J. (1967) Word of Mouth Advertising: A Review of the Literature. New York, Advertising Research Foundation.
Bansal, H. S. and Voyer, P. A. (2000) Word-of-mouth processes within a services purchase decision context. Journal of service research, 3(2), pp. 166-177.
Bhatnagar, A. and Ghose, S. (2004) Online information search termination patterns across product categories and consumer demographics. Journal of Retailing, 80(3), pp. 221-228.
Blaikie, N. (2000) Designing Social Research. 1st ed. Cambridge, Polity Press.
Blanchard, A. L. and Markus, M. L. (2004) The experienced sense of a virtual community:
Characteristics and processes. ACM SIGMIS Database, 35(1), pp. 64-79.
Blodgett, J. G., Granbois, D. H. and Walters, R. G. (1994) The effects of perceived justice on complainants' negative word-of-mouth behaviour and repatronage intentions. Journal of Retailing, 69(4), pp. 399-428.
Bone, P. F. (1995) Word-of-mouth effects on short-term and long-term product judgments.
Journal of Business Research, 32(3), pp. 213-223.
Boon, S. D. and Holmes, J. G. (1991) The dynamics of interpersonal trust: Resolving uncertainty in the face of risk. Cooperation and prosocial behavior, pp. 190-211.
