The influence of Negative Word of Mouth
on loyalty intentions
“The one number you need to grow(?) “ Frederick Reichheld (2003)
Master thesis
Author: Remco Temmink
Student number: 1386484
@: r.temmink@student.rug.nl
Tel: 06-24865753
February 2011 University of Groningen Faculty of Economics and Business Msc Business Administration Marketing Management
February 2010 Master Thesis – Remco Temmink 2
The Influence of Negative Word of Mouth on Loyalty Intentions
Remco Temmink
University of Groningen, The Netherlands
Abstract
The purpose of this research was to empirically test whether people in the mobile telecommunication sector that are willing to give negative Word of Mouth (WoM) referrals to other people, also influence their own loyalty intentions. Results could not confirm the hypothesis that negative WoM referral does have a negative effect on loyalty intentions of customers, although this was suggested from earlier research. However, the moderating effect of customer commitment on this relationship is confirmed, in line with our second hypothesis. Customer commitment seems to enhance the relationship between negative WoM and loyalty intentions. Furthermore, perceived reliability of online information sources is tested, corresponding to the upcoming of the Internet and ‘new’ Social Media. Although it could not be confirmed that these online information sources do have more or less influence than ‘offline’ sources, several ideas for further research in this field of marketing are provided.
February 2010 Master Thesis – Remco Temmink 3
Preface
This is the last assignment of my study Business Administration at the University of Groningen. By finishing my Master Thesis finally the end of my career as a student in Groningen has arrived. For me this feels like saying goodbye, because in Groningen I have had one of the best times in my life. However, it is also the start of a new and exciting period as a working man!
My final study assignment started during my traineeship at H.J. Heinz in Zeist, where I became more inspired by working and the inspirational environment of the fast moving industry. Therefore, I decided to start writing my final thesis after this internship. However, during that time I got inspired by a presentation and the book of Steven van Belleghem about the ‘Conversation Manager’, who was after all the main inspiration for the first subject I wanted for my Master Thesis.
Finishing my Master Thesis was not an easy process. Unfortunately, it took much more time than I first expected. For the main part, this has to do with me, myself and I. Especially during the start of the process I found myself not very inspired and I rather did a lot of other interesting things, except starting and/or finishing my master thesis. However, in the end, with the finish line within sight I finally saw the light and worked harder! Unfortunately, some problems made the finishing time longer as well…
At this point, I would like to thank my supervisor Janny Hoekstra for her inspiration and useful feedback the last eight months. I also would like to thank dr. K.J. Alsem as my second corrector. Furthermore, I would like to thank my parents for the financial support and all my family, friends and especially my girlfriend Roos for their patience and moral support while writing this Master Thesis.
Remco Temmink
February 2010 Master Thesis – Remco Temmink 4
Table of Content
Abstract 2
Preface 3
1. Introduction and Problem Statement 6
1.1 Background 6
1.2 Research Method 10
1.3 Academic and Managerial Relevance 10
1.4 Structure of The Thesis 11
2. Literature Framework 12
2.1 Word of Mouth 12
2.2 The Conceptual Model 14
2.3 Negative Word of Mouth 16
2.3.1 Characteristics of Negative WoM 17
2.3.2 Consequences of Negative WoM 19
2.4 Customer Loyalty Intentions 20
2.4.1 The Role of Advertising 20
2.4.2 Creating Loyalty Intentions 22
2.5 Perceived Reliability of Online Information Sources 23
2.5.1. The Role of Information Sources 23
2.5.2 The Upcoming Role of Social Media 24
2.5.3 Online versus Offline Information Sources 25
2.6 The Role of Customer Commitment 26
3. Methodology and Research Design 29
3.1 Research Design Explanation 29
3.2 The Sample 30
3.3 Data Collection 31
3.4 Measurement Scales and Survey Measures 32
3.5 Method of Analysis 34
February 2010 Master Thesis – Remco Temmink 5
4.1 Descriptive Results 37
4.2 Analysis 38
4.2.1 Hypothesis Testing 38
4.2.2 Incremental F-test 41
5. Conclusions and Recommendation for Further Research 43
5.1 Conclusions 43
5.1.1 Points of Research Improvement 44
5.2 Answering the Research Question 45
5.3 Reflection and Further Research 45
5.4 Limitations 46
Epilogue 48
References 49
Appendix 1: Questionnaire 54
Appendix 2: Crosstab Bar Chart 56
February 2010 Master Thesis – Remco Temmink 6
1. Introduction and Problem Statement
1.1 Background
Worth of Mouth (WoM) has been a topic of interest in marketing ever since the post-war 1940s. WoM can be described in several ways, but as the most usual description is passing information from one person to another. Historically the term referred to oral communication (words that come from the mouth), but these days it entails all forms of human communication (Arndt, 1967).
WoM is a perfect subject for discussion, and a lot of research has been done the past decades in this area of marketing. Some 2400 years ago, Aristotle was one of the first people to think about interpersonal influences (Buttle, 1998). Nowadays, WoM is still an important topic in marketing and a lot of research is done to see the impact of WoM in daily life, for example, in the area of customer loyalty, brand attitude and decision making of consumers. Furthermore, data developed by the U.S. office of Consumer Affairs suggested that satisfied customers for consumer services are likely to tell five others about it (Heskett, Sasser and Schlesinger 1997). This fact emphasizes the importance of positive WoM for marketing managers once more. Because in the end, isn’t advertising without spending money the dream of every marketer?
February 2010 Master Thesis – Remco Temmink 7
question that arises might be: “How can this process in creating loyalty intention for customers be influenced?” Which antecedents are the most important?
When talking about WoM, it is presumed most of the time that it is positive WoM. Antecedents and effects of positive WoM have been studied by many people and are definitely important. However, especially considering today’s critical customers, negative WoM is probably even a more interesting subject. A negative and dissatisfied voice can influence much more people than a decade ago, with the help of modern media. However, the exact engagement and expected greater participation in negative WoM has long not been examined in detail (Richins, 1983; Anderson, 1998) and has been given much less attention in literature in the past decades.
Samson (2006) claims that when consumers’ experience is averse, the content of WoM is mostly negative as well, since WoM is a direct reflection of customers experience with a brand. People can therefore expect brand loyalty and loyalty intentions to be better captured by negative WoM than positive WoM. Negative WoM is for both loyalty and advocacy a good measure, but negative information also has strong effects on purchase decisions by potential future customers (Samson, 2006). Furthermore, with the upcoming of Social Media (e.g. Hyves, Facebook, Twitter) at the Internet, the immediate information sources available every second of the day (also think about Internet on your phone) and consumers becoming more critical each day, negative WoM deserves more attention. This is supported by Wangeheim and Bayón (2006) who pointed out the specific importance of customers having negative experiences.
So, because of the critical customers, important developments made with upcoming Social Media and the changing role of the Internet, the focus of this thesis will be specifically on negative WoM.
February 2010 Master Thesis – Remco Temmink 8
changing digital environment, understanding exactly how peer-to-peer communication works is therefore particularly important for customers, companies and managers. Moreover, the manner in which WoM affects decision making and in what specific direction (positive or negative) people are affected, is a key in this process. Comparative websites online can play a role in understanding this process, because of their visibility.
Electronic Worth of Mouth (e-WoM) is, because of the changing role of the digital environment, a term that continually receives more attention of companies (Gruen et al, 2006). Therefore, Social Media and the homepages of companies receive a lot of attention; huge capital investments are made by companies for numerous Internet-projects to influence their communication possibilities on the World Wide Web. This might create larger brand awareness and interest among customers, especially in these
days, where people are spending more and more time online1. However, because
information is mostly exchanged in private conversation, direct observations traditionally have been difficult (Godes and Mayzlin, 2004) and more information is crucial for managers and companies.
Despite the fact that the role of Internet and Social Media is growing, still 94% of all conversations took place offline in 2007 (Brown et al. 2007). Only 6% of conversations occurred online but obviously, this number is rising. However, one could simply consider the role of e-WoM more important than just its current share, since customers will have more time to spend on searching for exactly the type of information they want and easily become more committed to their buying behavior.
Besides, we need to consider what the importance and difference of online versus offline communication is. When talking about WoM, there are specific differences between WoM on the Internet and traditional WoM face-to-face. It is very well possible that consumers may start searching on the Internet for some specific information, after
1
February 2010 Master Thesis – Remco Temmink 9
they have heard something via traditional WoM with friends or family. Thus, the role of ‘traditional’ WoM and e-WoM online might be completely different, although they exist together in one world. Despite this fact, it might be very useful to look in more detail to the (perceived) reliability of an online information source (versus offline) for consumers as a mediator on the relation between negative WoM and loyalty intention.
Further, another important subject mentioned earlier is the involvement or commitment of people in giving and receiving negative WoM (Matos and Rossi, 2008). Reasons to consider giving negative or positive WoM to someone can be several, and based on either functional or emotional clues (Keller, 2008). This depends of the type of WoM that is given. Moreover, there are also several reasons why someone might listen to or search for an information source and gain knowledge. This all depends on the level of customers’ commitment with the product.
Looking at WoM, perceived reliability of online Social Media and companies’ (comparative) websites and also commitment of customers can be seen as important objects now and in the future for the area of loyalty intention formation and WoM (Keiningham et al. 2008).
After careful investigation of the information provided in literature, we come to the following research questions for this master thesis:
• To what extent does negative Worth of Mouth (WoM) affect the loyalty intention of consumers?
o What is the influence of customer commitment on this relationship? o What is the influence of the perceived reliability of online information
February 2010 Master Thesis – Remco Temmink 10
1.2 Research Method
After the construct of a conceptual model, the sample data is collected with the help of a questionnaire. Users of mobile telecom providers in The Netherlands are asked to answer questions in a short online survey. The research focuses mainly on the moderating role of customer commitment and the perceived reliability of online information sources for customers, when creating loyalty intentions for future behavior. Several regression analyses and t-tests are performed during the analysis to find the information needed. Finally, conclusions will be given, which give answers to the research questions.
1.3 Academic and Managerial Relevance
Using the results and findings of this thesis, this study makes several contributions to the field of WoM activity, on both academic and managerial level. Primarily, it gives a better understanding for managers about the negative effects of WoM towards creating loyalty intentions for customers, and a better understanding of the variable negative
WoM itself. However, there is other practical and academic relevance as well; the
influence of two important moderating variables on the relationship are much further investigated. For managerial implications in the nearby future in the field of negative WoM, commitment and the reliability of online information sources, this can be an important step forward. Managers can use the findings to influence loyalty intentions of consumers in the future.
February 2010 Master Thesis – Remco Temmink 11
about perceived reliability of online sources in this thesis is a first step for specific academic research in that direction. It can be a useful inspiration for other researchers and gives support for the developing of new models, in which new Media are investigated. While there are also indications about differences between man and woman in their response and intentions, this is an addition to academic literature as well. The creating of a new variable in this field can also be improved after interpretation of this article and better developed and described in future academic literature.
Lastly, this Master Thesis identifies research questions worthy of future investigations into the field of online research, negative WoM and creating loyalty intentions. Especially with the described upcoming of the Social Media and online networks available, managers and marketers from all over the world should work hard to find ways that influence the results negative WOM can have for their own brand, service or product.
1.4 Structure of The Thesis
February 2010 Master Thesis – Remco Temmink 12
2. Literature Framework
In this chapter literature is described, which is important to understand the conceptual model developed. After giving more information about Word of Mouth in general in section 2.1, the conceptual model is shown and shortly explained in section 2.2. In the following sections 2.3 till 2.6 the four variables of the model are explained in more detail, while explanation is given and the choices made are clarified. The hypotheses developed are also given in the last three sections.
2.1 Word of Mouth
Word of Mouth (WoM) is simply a reference to the passing of information from one person to another. It now includes any type of human communication, such as face to face-, telephone-, email-, and text messaging conversations. According to Arndt (1967), WoM can be described as ‘oral, person-to-person communication between a receiver and a communicator’. Brown et al. (2005) sees WoM communication in the broadest sense as the transfer of any information about a target subject from one individual to another, in person or via a communication channel.
February 2010 Master Thesis – Remco Temmink 13
manner. Moreover, WoM is just likely to be perceived as really credible, because consumers generally feel it is a trustworthy source of information (Bone, 1995).
Nowadays, especially with the Internet as a growing communication and information tool, Social Media and comparative reference sites on the Internet are becoming more and more important in the referrals from one person to another (e-WoM). According to research of Insites Consulting (2009), it is expected that Social Media will overtake search engines (e.g. Google, Yahoo) as the most important source of product specific information for consumers at the Internet in 2011. In the USA this is currently already the case(!).
Talking about WoM, products and services recommended by friends and other people are seen as more reliable than products and services recommended by advertising in magazines or on television (Bone, 1995) and will therefore have a more positive influence on the perceptions of customers. Moreover, it seems obvious that products advised against will probably be a strong discouragement for most people to buy a specific product, brand or service. Samson (2006) states in this case that negative WoM has superior predictive power compared to positive WoM. However, others as Reichheld (2003) and Reinartz and Kumar (2003) are more convinced by the predictive power of positive WoM.
February 2010 Master Thesis – Remco Temmink 14
companies therefore do even reward customers that provide a company with new customers (Reinartz and Kumar, 2003).
As mentioned before, for a lot of companies the conventional marketing instruments and conventional media do still play a vital role in the current marketing mix, although
expenditures are declining2. All these marketing instruments do influence the perceived
brand attitudes and loyalty intentions of consumers and importantly also feed the WOM of consumers (Keller and Berry, 2006). Therefore, marketing instruments and media can be a significant antecedent of WoM. WoM itself has further shown to influence a variety of brand conditions like awareness, expectations, perceptions, attitudes, behavioral intention and behavior itself (Keller and Berry, 2006).
Further, with explicit focus on behavioral loyalty intention, marketing influence can be both in a positive and negative manner; but WoM has definitely shown to be subsequently more influential on behavior than other marketed-controlled sources of managers and companies (Keller and Berry, 2006). Whether this is positive or negative WOM, is in this stadium irrelevant.
2.2 The Conceptual Model
Now that we have discussed the general importance of WoM, the conceptual model of this thesis is shown in the figure below. First, a short explanation of the model and its variables will be given, after that all variables will be further discussed in more detail.
In the conceptual model in figure 2.1 the relation between the two main variables ‘Negative WoM’ and ‘Loyalty intention’ is graphically represented. The main research question of this thesis focuses on the influence of negative WoM on loyalty intentions. The role of negative WoM in today’s fast moving environment is totally different from the role of positive WoM, as discussed in the introduction. Therefore, the added value of this thesis will specifically consider the role and influence of the negative aspects of
2
February 2010 Master Thesis – Remco Temmink 15
communication between people and the differences with the influence of positive WoM. Negative WoM includes, according to Wetzer et al. (2007), all “negatively valenced, informal communication between private parties about goods and services and the evaluation thereof”.
Figure 2.1: Conceptual Model
Several moderators can have additional influence on the relation between negative WoM and loyalty intention. We investigate two of these variables. The moderating variables that we investigated are (1) Perceived reliability of online information sources, and (2) the level of commitment of the specific customer in the process of forming loyalty intentions while experiencing negative WoM. The choice for these specific variables is made because of the growing importance of the Internet in today’s digital environment and opinion making. Both variables are closely related to this issue.
First, the perceived reliability of the information source for WoM is important for a customer, because he or she will not simply trust every source of information (Keiningham et al. 2008). A well documented source might be much more trustful than a
February 2010 Master Thesis – Remco Temmink 16
vague remark from someone not even known. Family and friends, for instance, are far more trustworthy than most other people (Bone, 1995). Nevertheless, the influence of an online source versus an offline source is of greater meaning in this thesis, because the quality and importance of online (negative) information is increasingly important (Dellarocas, 2003) and quality increases year after year. This point of discussion is also suggested by Van Hoye and Lievens (2009).
Second, customer commitment is a moderating variable, which is expected to have influence on the relation between negative WoM and loyalty intention. Gregoire et al. (2009) found that relationship strength and time are important in the evolution of customer intentions towards companies. Relationship strength definitely has a strong connection with customer commitment, mentioned by Matos and Rossi (2008). The relationship is investigated in an online public complaining context. This finding might also be important for the first moderating variable ‘Perceived reliability of online information sources’, where Gregoire et al. (2009) looked at reasons and time span for the grudge of online complainers. In past research, customer commitment has been a hot topic of discussion, since it is investigated in several ways and there are differences between affective, sacrifice and behavioral commitment of customers (Harrison-Walker, 2001).
Finally, a control variable is included in the regression analysis of the conceptual model. This variable will be called ‘satisfaction’ and is included in the final model, because it could exemplify some of the variance found in the total model.
2.3 Negative Word of Mouth
February 2010 Master Thesis – Remco Temmink 17
consumers becomes increasingly important. There are several reasons why customers can engage in and influence negative WoM, which will be discussed in this section.
2.3.1 Characteristics of Negative WoM
The influence of negative WoM has been a subject of discussion for some authors over time (Samson, 2006; Keiningham et al. 2007; Wetzer et al. 2007). However, this is mostly the case, since a lot of discussion in literature is about the exact differences between the amount of influence of positive and negative WoM on brands, products and companies. Not because negative WoM itself was considered so important. Anyhow, in the last decades marketing specialists have collected increasing insight into the antecedents of negative WoM (Wetzer et al. 2007). These factors, which influence whether or to what extent dissatisfied consumers engage in communication about their negative experience, are important in understanding the underlying reasons of negative WoM. For example, research shows that the amount of negative WoM consumers engaged depends on both perceived justice (Blodgett, Granbois & Walters, 1993), and the severity of the problem (Richins, 1983). Consumers nowadays seem to control the main message and companies can, in practice, only respond to the WoM of consumers. This WoM given might be positive, but is, as mentioned before – also very often – more negative (Samson 2006). Therefore, it deserves specific attention and investigation.
February 2010 Master Thesis – Remco Temmink 18
Facebook and Hyves, which are increasing in both numbers and size. What a client declares online can have permanence and also a large audience on the Web. This cannot be compared with a passing comment made in a backyard to the neighbors. According to Sammson (2006), “A review on Amazon.com or a strategically placed blog posting currently has the power to radiate to more viewers than a primetime TV advert and can also be perceived to have more credibility”. Therefore, especially the way companies handle negative WoM after receiving it, is becoming increasingly important. Damage control for companies seems necessary all the time.
Moreover, the study of Carl (2006) further illustrates the power of negative WoM and the retelling of stories. Indeed, according to the study, "the complaints have an even greater impact on shopper who is not directly involved as the story spreads and is embellished. Almost half those surveyed, 48%, reported they have avoided a store in the past because of someone else's negative experience." This statement shows the risks of negative stories in public and the importance of customer involvement in the process. Involved customers probably need more arguments to change their behavior when comparing them to uninvolved ones. We will discuss the importance of this in another paragraph about customer commitment.
At last, we do not have a clear overview of the specific differences between the effects of both positive and negative WoM yet, although it seems that negative WoM may have greater predictive power than positive WoM in many situations (Wetzer et al. 2007; Samson, 2006) influencing brand loyalty and buying intention. But still, not everyone agrees with this statement (Reichheld, 2003).
February 2010 Master Thesis – Remco Temmink 19
role of emotions while spreading negative WoM is important and that content and implications are related to specific emotions in consumer experiences. The fact that emotions in brand experience are connected with commitment of customers may sound logical.
2.3.2 Characteristics of Negative WoM
Research has repeatedly shown that negative WoM is equally, if not more, influential as positive WoM in affecting customer attitudes (Bone, 1995). Therefore, careful investigation of the negative, dissatisfaction WoM-non-acquisition chain is desirable. The chain shows why customers that receive negative WoM do not buy certain products or buy less of it, after receiving information. Besides, there is a big difference between the consequences of retention and acquisition of new customers (Wangenheim and Bayón, 2007). Customers already familiar with a brand or product will focus their attention on different characteristics of the company. In contrast, Customers who need to be convinced to a first purchase or experience do need other ways of attention characteristics.
Evaluations of customer experiences are affected by previously received WoM, despite one’s own experiences with a brand, product or service provider (Wangenheim and Bayón, 2006). However, this evaluation also depends on customers’ own involvement.
February 2010 Master Thesis – Remco Temmink 20
2.4 Customer Loyalty Intentions
Creating loyalty intentions has always been important for marketers. However, the ways to create intentions evolved over time. Findings about brands play an important role in this process, since they can be used to understand underlying thoughts. Traditional advertising is changing and creating attitudes is different than before. Johnson et al. (2006) described the relatively high loyalty of Germans in the phone market in the period 1996-2000, but in the evolution of loyalty intentions over time things might have been changed dramatically…
2.4.1 The Role of Advertising
It has become clear to most companies that their brands and branded products or services are the most valuable asset they posses (Keller, 1993; Keller & Lehmann, 2006). Although creating a valuable brand name takes time and a lot of effort for companies, most companies do not have much of a choice. For those companies, advertising is (and has always been) a very helpful way in creating such a strong brand. However, there should be considered if it is still the best way. Several researches (Gruen et al. 2005; Vakratsas, 1999) have shown that advertising alone is, for quite some time, not good enough anymore to create a strong brand, product or service and ultimately those loyalty intentions wanted.
February 2010 Master Thesis – Remco Temmink 21
Mitchell and Olson (1981) in earlier times described brand attitude as: ‘Individuals internal evaluations of a brand’. These evaluations will definitely be responsible for future intentions and loyalty towards a brand. Wilkie (1973) defined brand attitudes as consumers' overall valuations of a brand. Brand attitudes are important because they often form the basis for consumer behaviour (e.g. brand choice). Though different models of brand attitudes that have been proposed, one widely accepted approach is based on the multi-attribute formulation in which brand attitudes are a function of the associated attributes and benefits that are salient for the brand. Both Fishbein and Ajzen (1975) along with Ajzen and Fishbein (1980) proposed what has been probably the most influential multi attribute model to marketing, according to Bettman et al. (1986). Their model views attitudes as a function of (1) the salient beliefs a consumer has about the product or service (i.e., the extent to which consumers think the brand has certain attributes or benefits) and (2) the evaluative judgment of those beliefs (i.e. how good or bad it is that the brand has those attributes or benefits). These beliefs and judgements are essential in the process to creating long-term loyalty intentions. This will also be essential in the process of customer commitment, which will be discussed later. Attitudes can be related to beliefs about product-related attributes and the functional and experiential benefits (Keller, 1993), consistent with work on perceived quality (ZeithamI 1988). These antecedents can be taken into account when looking at the direct influence of (negative) WoM.
February 2010 Master Thesis – Remco Temmink 22
the opposite direction and leave companies and people with negative associations, negative intentions and a negative attitude.
2.4.2 Creating Loyalty Intentions
Yet, central in the theory about brands and branding is ‘what customers think and feel about a brand’ (Keller, 2008). The opinion of consumers towards a product is determined through good market research. Brand attitude and loyalty intention will tell managers and companies what people think and believe about a product or service, whether the product answers a consumer need, and just how much the product is really wanted and talked about by the consumer. Loyalty intentions are dynamic, evolve over time and are a function of perceived values earlier (Johnson et al. 2006). Knowledge about the intentions of customers is therefore vital for companies and managers when making decisions about their marketing strategy. The role of negative WoM in affecting these intentions is an important part of this process, since it does not seem to stimulate these intentions.
Currently, customer loyalty is viewed as the strength of the relationship between an individual’s relative attitude and repeat patronage. The relationship is seen as mediated by several social norms and situational factors. Negative WoM mostly makes things worse. Cognitive, affective, and cognitive antecedents of relative attitude are identified as contributing to loyalty intentions, along with motivational, perceptual, and behavioral consequences. Implications for research and for the management of loyalty are derived (Dick and Basu, 1994). Customers are supposed to have certain loyalty intentions towards brands, and of course, it is important to create positive loyalty intentions which as high as possible, by as many customers as possible. However, this will remain a big challenge.
Altogether, the above information results in the first hypotheses:
February 2010 Master Thesis – Remco Temmink 23
2.5 Perceived Reliability of Online Information Sources
People rely on several information sources, when creating opinions about brands,
services and products. In the 19th century, without radio, television and telephones you
told people stories. Today, the fast digital environment has changed everything…
2.5.1. The Role of Information Sources
It has been researched that recently acquired customers (“switchers”) differ from long-term customers (“stayers”) and that switchers acquired through WoM (customer referrals) differ from those recruited by advertising or direct mail, in their satisfaction, loyalty and WoM behavior (Wangenheim and Bayón, 2006; Godes and Mayzlin, 2004). Switchers and customers acquired through WoM are higher in active loyalty, lower levels of reactive loyalty and positive WoM giving (Trusov et al. 2009). Therefore, the highest value for companies lies in the referral which is made shortly after switching. WoM is employed as a behavioral measure for researching differences between stayers and switchers. Both attitudinal and behavioral loyalty should therefore be measured. However, in the last decades the manner of using WoM seems to be changed.
Customer satisfaction can be described as the outcome of a comparison process between perceiver product performance and previously held expectations (Wangenheim and Bayón, 2006). Therefore, switchers after hearing WoM will have lower comparison standards when entering a new relationship, because they have experienced outcomes below expectations in the past. Moreover, there might also be a difference between the information sources of consumers able to use WoM in daily life. Referrals and attitude creation can be different in an online environment, compared to ‘offline’ face-to-face and people do also act different here (Brown et al. 2007).
February 2010 Master Thesis – Remco Temmink 24
online environment, especially with the specific (new) role of Social Media, comparative websites and search engines like Google is growing. This increases the trustworthiness of online sources in the last years or even months, at least for some people. Moreover, ‘media richness’ is another important term, meaning that richer media are more persuasive (Allen, Van Scotter, & Otondo, 2004). This is determined by the capacity of
the medium for immediate feedback, the number of cues and channels utilized for
personalization, and language variety (Daft and Lengel, 1986). Face-to-face might be expected to work better than, for instance, email. Possibilities for immediate feedback are present and the personal cues are higher in existence. However, email or advice websites give you the time and opportunity to think more carefully about reactions and feedback. Furthermore, older people (65+) do differ much from the new generations (<25) in their opinion about these new media, because of their background. Therefore, it is a complicated point and age might be a factor of additional influence.
2.5.2 The Upcoming Role of Social Media
February 2010 Master Thesis – Remco Temmink 25
While the improvement of Social Media usage by companies is an ongoing process, we should be well aware of the principles of the media itself. LinkedIn for instance, is a professional site aimed at work related issues, the international site Facebook is an international site for friends and relatives, while Twitter is a medium that can be used all day long to communicate to people that are interested in your habits and life. These Media do have really different roles, when we are talking about the ‘new generation’ of WoM communication.
2.5.3 Online versus Offline Information Sources
WoM influences both short- and long term product judgments. However, the influence is greater when a customer faces a disconfirmation experience and when the WoM-communication is presented to the customer by an expert (Bone, 1995). Since more and more people and also experts are using the Internet and/or Social Media as a online place to face their expertise, this medium is also growing in importance (Brown et al. 2007). Social networking sites typically allow a user to build and maintain networks for friends and family, for both social and professional interaction (Trusov et al. 2009). No wonder that Godes and Mayzlin (2004) suggest that people often make offline decisions based on online information. Furthermore, it illustrates the growing reliability of online information sources.
February 2010 Master Thesis – Remco Temmink 26
group. It might be expected that their will be large differences between the several mediums of information, which we will research in this thesis.
We will look more closely at the moderating effect of the online information source as a moderating variable on the relation between negative WoM and loyalty intentions. Thereby the difference between an online or offline source of information will be researched, in which it is expected that the amount of time to influence the decision plays an important role, as mentioned by Wangenheim and Bayón (2006). Since there is more time to process the information online, this might be crucial in making important decisions with high involvement. However, it may be quite different when one does not know the person giving the advice personally. In this case the role of experts may play a crucial role as well in the perceived reliability of an online source, as mentioned by Bone (1995).
This results in the second hypotheses:
H2: Higher perceived reliability of an online information source strengthens the influence of negative WoM on loyalty intention.
2.6 The Role of Customer Commitment
A buying decision process creates tension and therefore, contemporarily increases customers’ involvement in decision-making (Richins and Block, 1986). This tension can be eased by WoM communication (Dichter, 1966). These findings are two reasons why involvement or commitment of customers should be present in this study. Customer commitment is defined as a customer’s enduring desire to maintain their relationship with the firm (De Wulf et al. 2001; Morgan and Hunt, 1994) or a brand.
February 2010 Master Thesis – Remco Temmink 27
some knowledge about a product you need (at least) a bit of involvement or commitment. Consequently, brand knowledge and commitment with a product probably will be an important moderating variable in the process of influencing loyalty intentions through WoM.
However, Brown et al. (2005) claim that the moderating role of consumer commitment depends on the interaction effects between satisfaction and consumer commitment. Furthermore, according to Suh and Yi (2006) loyalty is also influenced by product involvement and satisfaction with the product. The relative effect of satisfaction and therefore loyalty intentions would decrease at higher levels of commitment. Wangenheim and Bayon (2007) claim that the satisfaction-WoM link is non-linear and moderated by several customer involvement decisions. Since more commitment will lead to less positive WoM intentions and behaviour if the experience is positive, the same effect is expected when the experience is negative. So, more commitment will probably decrease the effect of negative WoM on satisfaction and loyalty intentions.
Furthermore, numbers of brand choices and brand commitment are addressed across industries. It is proposed that products (and particularly services) in high commitment/low-choice sectors are more sensitive to negative WoM, while positive WoM may be a better predictor for business growth in low-commitment/high-choice industries (Samson, 2006). Consumers are very much involved in finding the right brand in a product category with high investment, status and/or utility. Therefore, we should expect them to have a stronger opinion about these brands (Samson, 2006). It also makes sense to state that, compared ceteris paribus; consumers are forced to be more committed to subscription category products brands that are provided by only one brand at a time (Sharp et al. 2003).
February 2010 Master Thesis – Remco Temmink 28
Therefore, negative WoM is extremely bad for service companies or companies with only a few products or only a few clients which they depend on. The other way around, there are far more individual products than services, what indicates that in the total amount of WOM, products are larger in number.
Managers need to focus on the discussions consumers are having, according to Bone (1995), while consuming a product/service. Effective listening can give managers an idea about what aspects of the product/service customers feel are worthy of discussion and are evaluated positively. This indicates that both the consumer and the companies need involvement with the buying decision. Altogether, the chosen industry for this research will be of great importance.
This last section leads to the third and last hypotheses:
February 2010 Master Thesis – Remco Temmink 29
3. Methodology and Research Design
In section 3.1 the arguments for the chosen Research Design are given. Section 3.2 will explain what the sample chosen for this research consisted and section 3.3 will describe how the data is collected and why specific methods for analyses are chosen. The measurement systems that are used have been brought under attention in section 3.5. Therefore, in section 3.4 the scales that are needed for the numerous variables are also explained. Together with the methods of analysis, it concludes the total methodology and research design.
3.1 Research Design Explanation
The Research Design is about the WoM the respondent is giving himself, while the hypothesis and the whole thesis is about the WOM others give to a person and whether you are influenced by this. This is a mistake, but was just found in the final stage of the process. However, this measurement is also important, especially since several articles (Brown et al, 2005; Wetzer et al, 2007; De Matos and Rossi, 2008) claim the importance of the own commitment and feelings of customers when giving WoM. Customers are also more willing to give WoM (especially negative) when they are more involved with the company (Gregoire and Fisher, 2008). These involved people are important, because these should be convinced about the positive sides of the company and retained. Otherwise, negative information influence other consumers and decrease market share. When people themselves are influenced by WoM, this also happens. The involvement and reliability are also important in this part of the process.
February 2010 Master Thesis – Remco Temmink 30
have resulted in a better and more extensive survey. Now there are arguments given in this thesis why current results about WoM messages are expected to be similar with results in an experimental setting.
3.2 The Sample
The sample for this research consisted of more than one hundred mobile telecom users in The Netherlands who have filled in a questionnaire. A convenience sample was used here for the distribution of the questionnaires; all people were known and selected by the author itself. This group of consumers consisted only of adults, both man and
woman, aged 18+. The used questionnaire (which can be found in the Appendix) was
distributed to fellow students, family and friends at once via a direct link in an email on
July 27th 2010, which redirected everyone to the website www.thesistools.com. In total,
around 241 emails were sent, and there were 107 questionnaires filled out appropriately, which is a high response rate of around 44%. However, it could be very well possible that the questionnaire had a so-called snowball effect of spin-off referrals (Helm, 2007). This is a valuable effect caused by the respondents asking friends to fill in the questionnaire as well. This might have led to a higher response rate and higher bias than usual, so 44% is probably not the actual response rate of the emails that have been sent. However, descriptive results of the sample in section 4.1 indicate that the sample is representative in most aspects. According to t-tests men and women are equally divided and tests do not show significant differences between residents in Groningen (which is around 50%) and other cities in the country, on most variables. Residence is expected to be most effected by the snowballing effect of the convenience sample. As 107 respondents were an appropriate number of respondents for this survey, no more emails were needed.
February 2010 Master Thesis – Remco Temmink 31
age. When analyzing the difference in age, a difference is found on the customer commitment level between respondents older than 25 and of 25 and lower. Younger people show significant higher levels of customer commitment. This seems a bit weird, because this group has less experience with their current provider. However, this can be caused by the fact that older people own telephones which are provided by their employer.
3.3 Data Collection
Questions in the questionnaire are about the Dutch mobile telecom market. The questionnaire was spread in The Netherlands, where telecommunication is a rather oligopoly market with only a handful of providers. KPN (51%), T-Mobile(24%) and
Vodafone (25%) are the only three network providers3 although there are several other
(sub)providers using their network, for example Telfort, Hi, Ben and Orange. The telecom market is considered a good market for this survey, because of the relatively high commitment and usage. Therefore, it is also used in the article of Samson (2006) for research, were it has proven its value already. In this research, it was not really important to have a clear division between men and women, since no difference in response between these two groups were previously known for the telecom industry. So everyone between the age of 18 and 65 was suited to answer in the questions. Furthermore, no difference in the spreading of WoM has been shown in literature for man and woman. Therefore this was not expected to be a problem for the survey.
Mobile networks are representatives of the so-called ‘subscription’ brands, where consumers tend to use just one product over a given time period. In several other categories, one finds ‘repertoire’ products, where the consumption of multiple brands within shorter periods is more common. This is an essential point of consideration, because customers tend to be more committed to subscription brands than repertoire brands (Ehrenberg et al. 2004), since there is often no possibility to change during the
February 2010 Master Thesis – Remco Temmink 32
time of a contract. The contract time for telecom providers in The Netherlands is usually one or two years, which is expected to lead to more commitment.
Another possibility for a mobile telecom contract in The Netherlands is a so-called prepaid contract. With this contract a person purchases calling credit for a certain amount of money (10 or 20 euros). Since this is not popular anymore, especially among adults (18+), these cards are not taken into account for this survey. Furthermore, the involvement with these cards is definitely lower, because there are only little differences between the different card providers. Therefore, it is not included in this survey.
The results found from the questionnaire can be found in the next chapter. The questions in the questionnaire were measured on a 1 (totally not agree) to 7 (totally agree) point Likert-scale, because statistically this has proven a good measurement system (Malhotra, 2004).
3.4 Measurement Scales and Survey Measures
As mentioned before, questions in the survey were asked on 7-points Likert-scale. In table 3.1 an overview is given which contains all the questions that are used in the questionnaire. For three variables, existing scales from literature could be used, and where necessary the items were adapted to the specific telecom provider setting. All Cronbach’s Alpha’s found were higher then 0,6. This included the self-constructed variable ‘Perceived reliability of online information sources’. The first Cronbach’s Alpha found for all five questions was 0,457. Factor analysis was used to find the questions that belonged to the variable in the best possible way. After deleting two items (see Table 3.1) and the program using only one item for the construct, the Alpha moved up to 0,633, which is an adequate number according to Malhotra (2006). Therefore, only three (out of five) questions are combined in this constructed variable.
Table 3.1: Items and Cronbach’s Alpha
Items: Source: Cronbach’s Alpha
Negative WoM Gregoire and Fisher, 2006 & 2008 α= 0,847
February 2010 Master Thesis – Remco Temmink 33 - I bad-mouthed against the firm to my friends
- When my friends were looking to a similar product or service, I told them not the buy from this firm
Loyalty Intentions Wangenheim and Bayón, 2006 α= 0,688
- I do not intend to switch my current provider in the near future
- I would strongly recommend my current provider to friends and family
- If my current provider would raise its prices, I would still continue being a customer
- If a competing provider would offer a better rate, I would switch providers (*reverse scaled)
- As long as I live in this country, I do not foresee myself switching to a different provider.
Customer Commitment Brown et al. 2005 α= 0,894
- I am committed to my relationship with my current provider
- I really care about my ongoing relationship with my current provider
- The relationship that I have with my current telecom provider is something I am very committed to.
- The relationship that I have with my current provider deserves my maximum effort to maintain.
Perceived reliability (of online sources) Self constructed α= 0,633
- I make use of (some of) the Social Media and comparison websites on the Internet4
- People can influence my buying behavior via online information sources more than via face to face contact 4 - On the Internet I can find opinions of experts, which I cannot find somewhere else
- I trust information find online
- I have more confidence in the independent opinions of people and experts online, compared to the opinion of acquaintances
The variables in table 3.1 were all accounted by adding the total score of the questions belonging, and divided by this number of questions. The negative WoM measure describes the amount of negative WoM given by consumers to friends and family. The loyalty intentions include repurchase intentions and recommending the provider to
4
February 2010 Master Thesis – Remco Temmink 34
others. The customer commitment measure includes customers’ commitment to maintaining a relationship and personal interactions with the provider. The self-constructed variable ‘Perceived reliability’ in table 3.1 is formulated to indicate the trustworthiness of online sources compared to offline sources (face-to-face, text messaging or telephone). A high score on this variable can be explained as people have much (perceived) reliability and confidence in online information sources.
The control variable ‘Satisfaction’ was created by asking the respondent the question: ‘How satisfied are you with your current telecom provider?’, on a 10-points scale. This scale is normally used for creating NPS scores.
3.5. Method of Analysis
To test the differences in demographic information of the respondents, several t-tests were performed, to indicate if there are differences in answers within the sample group, based on these demographics. These tests could show deviations in measurement of the variable scales in age, residence and gender groups.
All further data is carefully evaluated to find necessary information with the help of the statistical program SPSS version 16.0 (SPSS inc., Chicago, USA). Factor analysis here also confirmed the uni-dimensionality of the constructs in table 3.1. These factor scores can be found in table 3.2, the numbers of the questions refer to the questions in table 3.1 as well. For all variables only one components was found, with all independent scores above 0,68. The own value of the components are given in the second column, with Loyalty Intentions having the lowest variance explained with 56%, and Negative WoM the highest. However, we used Cronbach Alpha scores, based on average scores of the questions.
Table 3.2: Factor analysis results
Factor analysis scores:
Question 1 Question 2 Question 3 Question 4
Negative WoM 0,939 0,912 0,776
February 2010 Master Thesis – Remco Temmink 35
Loyalty Intentions 0,791 0,706 0,677 0,804
Own value (% variance) 55,8 73,3 88,3 100,0
Customer Commitment 0,859 0,873 0,904 0,849
Own value (% variance) 75,9 86,2 94,0 100,0
Perceived Reliability 0,790 0,725 0,766
Own value (% variance) 57,9 80,7 100,0
To test the three hypotheses, a repeated-measures regression analysis was performed in SPSS. Loyalty Intentions was the dependent variable in the model, and negative WoM the independent variable, as can be seen in the models shown below. The first model illustrates the main relation of the conceptual model.
Model 1: Y = α + β1*X1 + δU
Y = Loyalty Intentions α = Constant
Β1 = Factor regression coefficient for Negative WoM (X1) X1 = Negative WoM variable
δ = Parameter error term U = Error term
In the second model tested, the two moderating variables are included as an independent variable. This has been done, to investigate the direct influence of these to variables on Loyalty Intentions.
Model 2: Y = α + β1*X1 + β2*X2 + β3*X3+ δU
B2 = Factor coefficient for Customer Commitment X2 = Customer Commitment (1st Moderating variable) B3 = Factor regression coefficient for Perceived Reliability X3 = Perceiver Reliability (2nd Moderating variable)
Both the third and fourth model are almost the same as model 2, but instead of a third variable, the interaction effect of the moderating variable is included. In model three the moderating variable Customer Commitment is included, together with the Interaction effect created by Customer Commitment and Negative WoM.
Model 3: Y = α + β1*X1 + β2*X2 + β4*(X1*X2)+ δU
February 2010 Master Thesis – Remco Temmink 36
In the fourth model, the variable Perceived Reliability is included instead of Customer Commitment, combined with the accompanying interaction effect between Perceived Reliability and Negative WoM.
Model 4: Y = α + β1*X1 + β3*X3 + β5*(X1*X3) + δU
Β5 = Regression coefficient Interaction Effect Perceived Reliability
In the fifth and last model, both moderating variables, together with the accompanying Interaction effects are given. This is the largest model and includes all variables en effects. Consequently, it is also expected to have the highest predictive value.
Model 5: Y = α + β1*X1 + β2*X2 + β4*(X1*X2) + β3*X3 + β5*(X1*X3) + δU
February 2010 Master Thesis – Remco Temmink 37
4. Analysis and Results
In section 4.1, the descriptive results of the data are given. In section 4.2, regression analyses are made and shown in table 4.2. Furthermore, the first sub conclusions are given, indicating results of the survey.
4.1 Descriptive Results
In Table 4.1 the mean, standard deviation and frequencies of the variables are given. These numbers give an overview of the spreading in the answers given by the sample.
Table 4.1: Variable frequencies and Mean + St. Deviation Descriptive information main variables
107 respondents
Variable: Mean: Standard Dev.: Percentages (Frequencies)
≤ 3 3 - 4 4≥ Negative WOM 2,474 1,423 74,8 (80) 8,4 (9) 16,8 (18) Loyalty Intentions 4,052 1,200 22,4 (24) 24,3 (26) 53,3 (57) Customer Commitment 2,294 1,176 77,6 (83) 12,1 (13) 10,3 (11) Perceived Reliability 4,140 1,041 19,6 (21) 22,5 (24) 57,9 (62)
Control variable: Satisfaction 7,280 1,842
A large difference can be noticed in the means of Negative WoM and Customer Commitment, compared to that of Loyalty Intentions and Perceived Reliability. However, these paired groups of two times two variables look pretty much the same.
Negative WoM has a low mean score, and seems to have little spreading as well. This is confirmed by a Cross table (see Appendix) bar chart with the control variable Satisfaction. Therefore, it might sometimes be difficult in a model to notice and generate good linear relationships.
February 2010 Master Thesis – Remco Temmink 38
a 0,01 level there is a significant difference between the ‘perceived reliability of online sources’ between men and women.
Moreover, 53 out of the 107 respondents, almost half of the population size, is currently living in Groningen. This might create a bias, because inhabitants of Groningen can, on average, have different answers than people in other cities, caused by surrounding people in the city with the same ideas developing. However, independent t-tests show that there is no significant difference between these two groups for any of the four variables. Therefore, we can conclude that there is no large difference between the residences of consumers.
4.2 Analysis
In section 4.2.1 analysis will be done to test the hypothesis. In section 4.2.2 we will perform an additional incremental F-test, to clarify the strength of the individual models.
4.2.1. Hypothesis testing
The first hypothesis H1, testing the relationship between negative WoM and loyalty intentions is tested with a bivariate Regression Analyses in Model 1.
Model 1: Y = α + β1*X1 + δU
Y = Loyalty Intentions, α = Constant, Β1 = Factor regression coefficient for Negative WoM (X1), X1 = Negative WoM variable
February 2010 Master Thesis – Remco Temmink 39
Table 4.2: Regression analysis results
Regression Results (standardized coefficients) for Loyalty Intentions testing for Moderators
Hypothesis
(effect) Model 1 Model 2 Model 3 Model 4 Model 5
Main Variables: Negative WOM 1. (-) -0,15 -0,114 0,241 0,327 0,509 Control Variable: Satisfaction 0,36* 0,302* 0,299* 0,319* 0,267** Moderators: Customer Commitment 0,394* 0,686* 0,687* Perceived Reliability -0,187** 0,039 -0,077 Interaction effects
Customer Commitment * Neg WOM 2. (+) -0,483** -0,456**
Perceived Reliability * Neg WOM 3. (+) -0,552 -0,33
R2 0,227 0,397 0,399 0,260 0,438
(Adjusted R2) (0,212) (0,374) (0,375) (0,231) (0,405)
R2 change (compared to model 1) 0,170 0,172 0,033 0,211
F-value 15,23* 16,81* 16,92* 8,97* 13,01*
Note: * p-value < 0,01, ** p-value <
0,05
The most important outcomes of the moderating effects of both ‘Customer Commitment’ and ‘Perceiver Reliability of the online information source’ on the relationship between Negative WoM and Loyalty Intentions are discussed in Hypothesis 2 and 3. Results can also be found in table 4.1. Both variables are independently tested with Regression analysis in Model 3 and 4, including both the independent variable and the Interaction effect.
Model 3: Y = α + β1*Neg. WoM + β2*Cust. Comm. + Neg. WoM * Cust. Comm.+ δU
February 2010 Master Thesis – Remco Temmink 40
WoM, which was also expected based on earlier findings of Wangenheim and Bayon (2007).
Model 4, the model with the variable Perceived Reliability, testing for Hypothesis 2, does not show a significant influence within the model. Both the moderator and interaction effect are not significant. Since ‘Perceived Reliability of online information sources’ does not show a significant effect we cannot confirm any hypothesis here. So, it cannot be concluded that Perceived Reliability does strengthen the influence of the main relation. Therefore, Hypotheses 2: ‘Higher perceived reliability in an online information sources setting strengthens the influence of negative WoM on loyalty intention’, should be rejected!
In Model 2 and 5 the two moderators are tested together, respectively with and without
their interaction effect. In the last model, Model 5, with both the moderating variables
and interaction effects included and the highest R2 of 0,438, customer commitment is also found to be significant. This confirms the findings in Model 3 and strengthens the hypothesis that customer commitment strengthens the relationship between Negative WoM and loyalty intentions. Looking at the Interaction effect, model 5 also confirms the other finding of model 3, about the weakening of the influence of negative WoM.
After all, especially model 1 and 4 do not convince us that our hypothesis is right. However, model 4 shows such a low R2 value and F-value, compared to the other models, that some additional research might be useful. An adjusted model with other variables could be found, accounting for higher explained variability.
4.2.2 Incremental F-test
February 2010 Master Thesis – Remco Temmink 41
table 4.2 is much lower than that of the three other models, which indicates a low explanatory value of this model. The independent moderating or interaction effect of ‘Perceived reliability of online information sources‘, can disturb the rest of the model. This idea is confirmed in a regression analysis with a ‘Backward’ testing method (Criteria: Probabality of F to remove > 0,1). This analysis removes the moderating variable ‘Perceived Reliability’ out of the model. Remarkably, than the rest of the variables in the model (5+) all become significant!
Table 4.3: Regression analysis after incremental F-test
Regression Analysis (after Backward method)
Model 5+ Main Variables: Negative WOM 0,642* Control variable: Satisfaction 0,259** Moderators: Customer Commitment 0,673*
Perceived Reliability Out of model!
Interaction effects
Customer Commitment * Neg WOM2 -0,438**
Perceived Reliability * Neg WOM3 -0,502*
R2 0,437
(Adjusted R2) 0,409
R2 change (compared to Model 1) 0,210
F-value 15,68*
Note: * p-value < 0,01, ** p-value < 0,05
February 2010 Master Thesis – Remco Temmink 42
effects of Customer Commitment and both the Interaction effect of the moderators are responsible for this. The total R2 of the model remains the same, so the explanatory value is the same.
February 2010 Master Thesis – Remco Temmink 43
5. Conclusions and Recommendation for Further Research
In the first section, section 5.1, final conclusions are given and in section 5.2 the answer on the research question is formulated. Furthermore, several points of improvement, reflection and implications for further research are given in section 5.3. Limitations are described in section 5.4, which will be a critical evaluation.
5.1 Conclusions
The purpose of this research was to empirically test whether people in the mobile telecommunication sector were willing to give negative Word of Mouth (WoM) referrals to other people and thereby influence the loyalty intentions of these people. Important to consider is that it first was about the loyalty intentions of the respondent himself/herself, whether it should actually be about you being influenced by the WoM of other peoples. The results that were found could either way not confirm the first hypothesis that negative WoM does have a negative effect on loyalty intentions of customers, although this was suggested in earlier research by both Matos and Rossi (2008) and Gregoire et al. (2009). The negative relationship that was expected was not found and more importantly, was not significant. In another test, questions should be more specifically based on the influence of the content of the message on the receiver of the WoM message.
There are several reasons that might be responsible for this outcome in general. First, the questions creating the variables were maybe not suitable for the mobile telecom provider environment. Second, the cross table discussed in section 4.1, showing the relation between the control variable satisfaction and negative WoM, has a small degree of spreading within the construct. Therefore, it is much harder to find a relationship. Reasons for the low variance in this variable are given in section 5.1.1.
February 2010 Master Thesis – Remco Temmink 44
strengthens the relation between negative WoM and loyalty intentions in a negative manner. This is in line with the expectations and can logically be explained because the commitment of people is supposed to strengthen the relationship (De Wulf et al. 2001; Morgan and Hunt, 1994), and therefore weakens the influence of negative WoM on loyalty intentions.
In contrast, Perceived Reliability has proved to be the most difficult construct. Unfortunately, a significant relationship for this construct was not found, and this might be due to a lack of good preparation for the questionnaire questions. Findings of Herr, Kardes & Kim (1991) about the accessibility of face-to-face communication are probably still relevant. Bone (1995) mentioned that consumers tend to trust experts, and most likely, not enough expertise is found online. Perhaps in a couple of years, the results will turn out to be quite different. Interesting to see in this light, is that men already show more trust, and are more likely to have confidence in online information sources. Moreover, they also provide less negative WoM. Male consumers therefore can be considered as earlier adaptors and less resistance to innovations than women. For marketers and mangers this is a real eye-opener and it can be used in approaching the best target group for new products, brands and services.
Additional results of the model, with and without Perceived Reliability, show that both interaction effects tend to weaken the relationship between negative WoM and Loyalty Intentions. However, as Perceived Reliability has no significant effect, only the weakening effect of Customer Commitment can be presumed as correct. Furthermore, the proclaimed change in research design might show different results.
5.1.1 Points of Research Improvement
