_________________________________________
The creation and effect of negative eWOM
perception on the purchase intention
_________________________________________
Master thesisAuthor: Elodie de Jong
University of Amsterdam, Faculty of Economics and Business Student number: 5827477
Date: January 2015
Supervisor: Prof. Dhr. A.C.J. Meulemans Second corrector: Prof. Dhr. J.H.J.P. Tettero
Statement of Originality
This document is written by Elodie de Jong who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the test and its references have been used creating it.
The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Preface
This research was conducted in order to complete a Master of Business Administration at the University of Amsterdam. The experiment undertaken in this study was combined with the research for another masters thesis: Webcare on
Facebook (de Jong, 2015), which was undertaken in order to finish a Master of
Corporate Communication at the University of Amsterdam. The last-mentioned study aimed to investigate which organisational response strategy is most successful.
Due to the overlapping topics and variables, it was decided to combine both experiments into one study. As can be seen in Figure 1 (below), each study had different independent variables and only one dependent variable was similar. Therefore, it was decided that two different experiments could be conducted in one survey. The procedure for both experiments was as follows. The survey started with a general section, which consisted of an introduction to the research and some general questions about the respondents’ demographics. Following this, respondents were randomly assigned to a manipulated condition for the current study (The creation and
effect of negative eWOM perception on purchase intention) and asked some questions
related to the treatments. After completing the questions, respondents were again exposed to a manipulated condition, this time related to the study Webcare on
Facebook (de Jong, 2015). However, the focus in that research was on an
organisation’s response and not on the content of the eWOM message, which was the aim of this study. Respondents were also asked to complete several questions related to the experiment in the communication section.
In conclusion, each study had a different topic, different literature reviews and different data. However, because the two experiments were combined in one online survey, the questions in the attachment from this file are similar to the attachment in the file Webcare on Facebook (de Jong, 2015).
The introduction of Web 2.0 has ensured that consumers can easily exchange a very broad range of product experiences and product knowledge with other consumers. Consumers are increasingly dependent on product-related electronic word-of-mouth (eWOM) in their online purchase decision process. This could become a problem if a consumer spreads a negative message to other consumers. In this research it is assumed that features of a negative eWOM message influence a consumer’s purchase intention. Therefore, the predictive factors of a consumers’ perception regarding the negative eWOM message on social media platforms are examined. In addition, the consequences a negative eWOM message has on a consumer’s purchase intention were considered. In this explanatory study, the results show that negative eWOM messages on organizational Facebook pages negatively influence the consumer’s purchase intention. Moreover, it was found that two predictive factors (source credibility and consumer involvement) affect the perceived eWOM credibility of a consumer. The theoretical background, methodology and implications of the results in this research are discussed in this study.
Preface ... 3
Abstract ... 5
1. Introduction ... 7
Relevance ... 10
2. Literature review and conceptual model ... 11
Social media and its opportunities ... 11
Challenges of social media: negative eWOM... 13
Purchase intention and affecting factors ... 14
Perceived eWOM credibility and eWOM adoption ... 14
eWOM involvement ... 16 Source credibility ... 17 Consumer expertise ... 17 Quality... 18 Consumer involvement ... 19 Quantity ... 20
Experience goods versus search goods ... 21
3. Methodology ... 23
Participants and research design ... 23
Context ... 23
Sample ... 35
4. Results ... 37
5. Discussion and conclusion ... 44
6. Conclusion... 47
Appendix – Questionnaire ... 56
1.
Introduction
Word of Mouth (WOM) is a phenomenon that is recognized by marketers and sociologists as naturally occurring. For almost half a century scholars have assumed that WOM affects the buying behavior of the majority of consumers (Kozinets, De Valck, Wojnicki, & Wilner, 2010). In the early marketing world, which was untouched by the Internet, consumers had a very reactive attitude and only talked about products or brands to each other. Thus, the scope and transparency of shared information had limited coverage. Following the introduction of Web 2.0 and the appearance of social network sites, consumers are no longer considered to be a sedentary audience. Nowadays, they have a proactive attitude within the communication, interaction and discussion platforms on the World Wide Web (Hennig-Thurau, Malthouse, Friege, Gensler, Lobschat, Rangawamy & Skiera, 2004). For companies, this means that consumers share their knowledge and experiences about brands and products with a multitude of other consumers via online platforms (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2010). They can share their opinions to a wider public than ever before because of the accessibility, reach and transparency factors of the Internet. The development of information technology has caused a transformation in traditional Word of Mouth (WOM) to electronic Word of Mouth (eWOM) (Hennig-Thurau et al., 2004).
In literature, electronic Word of Mouth (eWOM) is defined as: “any positive or negative statement made by potential, actual, or former consumers about a product or company, which is made available to a multitude of people and institutions via the Internet” (Hennig-Thurau et al., 2004, p.39). There are some clear fundamental differences between WOM and eWOM. Tong and Xuecheng (2010) highlighted the main, fundamentally-based differences between these two phenomena: face-to-face interaction vs. indirect interaction, spoken vs. written, slow vs. fast interactions, narrow vs. broad reach and identification vs. anonymity. Another difference between traditional WOM and eWOM is based on eWOM’s trust characteristic: people do not only rely upon opinions from people within their direct social network, they also trust opinions from people outside their direct social network (Jansen, Zhang, Sobel, & Chowdury, 2009). Furthermore, previous research shows that eWOM could be a more effective way of influencing consumers and changing consumer behavior compared to traditional WOM activities (Trusov, Bucklin, & Pauwels, 2008; Muntinga, Moorman
& Smit, 2011). Marketers should be aware of this fact and focus could be switched to online interaction between consumers and companies instead of traditional communication tools.
Social media is by far the largest area that provides a new source or platform for eWOM for consumers and companies. In academic literature these digital platforms are better known as Online Brand Communities (OBC) (Brogi, Calabrese, Campisi, Capece, Costa & Di Pillo, 2013). OBCs are defined “as a group of people who interact in computer-mediated space with computer-mediated communication support” (CMC) (Hur, Ahn, & Kim, 2011). OBCs could create many benefits and advantages for companies. According to Kim, Choi, Qualls & Han (2008), online brand communities can successfully contribute to a strong brand loyalty, a rise in revenue and the creation of positive eWOM. Moreover, it is possible to communicate with consumers and stakeholders and to gather customer information, which could serve as necessary and rich feedback. Companies can also use OBCs to inform their customers about such things as product features and product prices.
The benefits of OBCs do not differ substantially from traditional marketing. This creates the impression that there is not a big difference between traditional communication channels and new modernized channels. The difference lies within the symmetrical character of the new OBCs: they provide customers with the opportunity to interact with other customers about products or brands and also offer customers the opportunity to talk to the company. This two-way communication tool differs from traditional marketing activities. The feedback characteristic is not only useful for the company: shared facts or thoughts about products or brands are also expanding rapidly to other consumers (Hennig-Thurau et al., 2004).
OBCs exist in various types and forms. Companies have the opportunity to build a corporate website, which acts like an OBC. However, it is also possible to build an OBC on social media platforms such as Facebook, Twitter, or LinkedIn. It appears that Facebook is the most popular way to build a brand community. Facebook provides the opportunity for companies to improve communication with their stakeholders and to increase their customer knowledge (Brogi et al., 2014).
However, Facebook could be a disaster if an angry customer spreads a negative message to other consumers (Champoux, Durgee, & McGlynn, 2012). Companies try to manage all the eWOM content on the Internet to overcome the possible consequences of negative eWOM. Troubling effects on brand attitudes such
as brand evaluations (Hennig-Thurau et al., 2010) or a customer’s purchase intention, could occur as a result of negative eWOM (Lee & Song, 2010),. The voluminous flow of information about company’s brand has become interconnected and difficult to predict (Fournier & Avery, 2011; Hennig-Thurau et al., 2010). Brand marketers search through the web to address positive or negative eWOM. They try to successfully handle customer complaints and engage in online interactions with customers to decrease the possible consequences of negative eWOM. Deighton and Kornfeld (2009) stated that companies should participate in the conversation with the consumer to decrease the harmful side effects of negative eWOM. A way to participate in these conversations is by deploying webcare strategies (Harrison-Walker, 2001), which are defined as “the act of engaging in online interactions with (complaining) consumers, by actively searching the web to address consumer feedback” (Van Noort & Willemsen, 2012, p. 131).
According to the authors above, it can be assumed that negative eWOM and company webcare deserves serious attention from marketing researchers because it could have influence on customers’ brand attitudes and actual purchase intention (Hennig-Thurau et al., 2010; Lee & Song, 2010). However, the current literature includes ambiguous evidence regarding the effect of negative eWOM reviews on customers’ purchase intention. Berger, Sorensen and Rasmussen’s (2010) research illustrates that, actually, “any publicity is good publicity”. They found empirical evidence of an increasing likelihood of purchase intention after exposure to a negative eWOM review. However, other studies argue that negative publicity encourages the negative side effects of a product or brand, which could result in a decrease of product sales being related with an increase in negative eWOM (Cheung & Thadani, 2012; Davis & Khazanchi, 2008; Duan, Gu, & Whinston, 2008).
Thus, there is no consensus about the consequences of negative eWOM on a consumer’s purchase intention. A lot of opportunities would arise if customers and companies knew what the consequences of negative eWOM were and how to handle negative eWOM complaints (Hennig-Thurau et al., 2004; Trusov et al., 2008). According to the literature, more research on this topic is definitely required (Laroche, Habibi, & Richard, 2012).
In addition to more research into the consequences of negative eWOM, Laroche et al. (2012) investigated the possible affecting or predictive factors that create eWOM perception. They concluded that an example of a predictive factor of
consumer perception is the number of reviews: the receiver’s perception is more influenced by an eWOM comment if the receiver is exposed to multiple negative eWOM expressions instead of a single exposure. This could negatively affect the attitude towards the brand and also the likelihood of purchase. In addition, content related factors such as discussion or information quality, could influence the receiver’s perception about the usefulness of the information in the message (Willemsen, 2013). Park and Lee (2009) also conclude that the receiver’s perception is probably influenced by, for example, the number of exposures or the product type. They suggest that new research should focus on possible factors that affect customer perception.
According to these authors, it is necessary to conduct more research into the predictive influencers and the consequences of negative eWOM reviews. Therefore, the purpose of this research is twofold. First, an investigation to discover what predictive factors could contribute to the perceived credibility of an eWOM message after exposure. Second, an investigation into whether this perception has any influence on the customer’s purchase intention. Hence, the following research question is central in this thesis:
What are the predictive influencers for negative perceived eWOM credibility and to what extent could this perception influence the purchase intention of the receiver?
Relevance
This research, which adds to the current literature, is of managerial and theoretical importance for several reasons. First, if negative feedback from customers in an OBC has the potential to change the behavior or attitude of other customers towards a brand, it can potentially have an effect on the company’s revenue. It is important for company management to know what effect negative feedback has on the company. If a marketing strategy manager knew more regarding the formation and the consequences of negative eWOM perception, then they could actively react and adapt their strategy. In addition, previous research indicates that companies want to join Facebook and other social media channels but they don’t know how to work with it in the most efficient way (Waters, Burnett, Lamm, & Lucas, 2009). They do not know how to take any advantage from social media channels such as Facebook. This research could help companies to manage Facebook usage.
The second reason for conducting this research is because the results may contribute to the field of customer complaint management. There is still a knowledge gap in current literature about the consequences and the creation of negative eWOM perception (Davis & Khazanchi, 2008; Duan et al., 2008; Laroche et al., 2012; Park & Lee, 2009). Despite growing attention to eWOM, there is still limited understanding of the predictive influencers for perceived eWOM credibility and its effect on the purchase or repurchase intention. Research in this scientific field is fragmented and the findings are ambiguous. Cheung and Thadani (2012) stated that most studies into eWOM used different research models and did not build further on each other’s findings. This research aims to provide consistency in the findings and to discover the predictive factors that are important in the creation of a perception of the eWOM message.
For these reasons, the consequences of negative eWOM and the predictive influencers of eWOM perception are investigated in this study. In the second part of the study a review of the literature relating to the most important research objectives (social media, negative eWOM, perceived eWOM credibility, brand attitude and
purchase intention) will be undertaken. Following this, the conceptual model,
including the main research objectives and propositions will be developed. Chapter 3 provides a clear explanation of the research methodology and the results of the experiment. Finally, a discussion of the findings and managerial implications and the limitations of this research are presented.
2. Literature review and conceptual model
Social media and its opportunities
Social media refers to “a group of Internet based applications that builds on the ideological and technological foundations of Web 2.0, and that allows the creation and exchange of user-generated content” (Kaplan & Haenlein, 2010, p. 61). Over the years, social media has reached the top of the agenda for many businesses and marketers. Statics from the biggest social media channel – Facebook – show that this concern is not unjustified. In October 2013, there were 500 million registered Facebook users and half of these logged into their account every day (Wolfe, 2013). This makes Facebook the third largest ‘country’ in the world and the biggest and most
visited website on earth. In the four years since February 2009, the number of registered Facebook users almost doubled twice (Champoux et al., 2012).
With an average of 250 million users each day, organizations cannot ignore Facebook anymore (Wolfe, 2013). When Facebook opened registration to companies in April 2006, more than 400 companies joined the digital network within less than two weeks (Waters et al., 2009). The use and influence of social media is enormous for companies and gives many businesses the opportunity to manage customer purchase behavior or customer communications. It is a way for organizations to improve their management, interact with customers and other stakeholders and to create new and improve existing relationships with the public. Therefore, Facebook is an important tool for cultivating relationships and also an upcoming strategic communication tool for public relations (Waters et al., 2009). In addition, Facebook gives companies the opportunity to reach more customers for a cost that is minimal compared to traditional marketing and communication costs (Kaplan & Haenlein, 2010).
Thus, companies can reach critical and attractive audiences via social media in a more efficient and cost-effective way than ever before (Champoux et al., 2012). Social media forecasters and some Facebook employees predict that many businesses will abandon their normal corporate websites and exchange them for an organizational Facebook page. Organizational Facebook pages are less expensive than corporate websites and revenues increase due to high clicks. Starbucks announced that their Facebook page is more valuable than their corporate website. The Starbucks Facebook page has 21 million likes, while their website has only 1.8 million views per month (Champoux et al., 2012).
However, not every company seems to be comfortable with the new world of social media where information is public and transparent and where the company’s control of information is reduced. Traditionally, businesses had the control over their published information, with planned press announcements, spin-doctors and good public relations strategies (Kaplan & Haenlein, 2010). Today, the power of published information has been transferred to the users of social media; companies have been relegated to the sidelines and don’t have any power over published information. With the introduction of all this available information on social media, it seems that marketing looks more like public relations and brand building more like brand protection (Fournier & Avery, 2011; Waters et al., 2009) because a company’s
information is so transparent and users can easily find out everything about a company. This could probably result in an increase in negative eWOM.
Challenges of social media: negative eWOM
The Internet plays an important role as the fomenter of consumer complaints. Consumer behavior researchers have undertaken a lot of research into the impact of negative Internet messages on consumers’ attitudes and behavior. Bailey (2004) investigated the potential damage that ‘corporate complaint websites’ could inflict on a brand or a company. Corporate complaint websites are defined as “websites where consumers can go to voice their concerns about different corporations, marketing actions and brands” (Bailey, 2004, p. 170). The specific aim of these sites is the fomentation of negative articulations about companies, products or brands. This type of website could cause an enhancement of negative brand attitudes, a decrease in customer purchases, a customer insurrection against the company or a decrease in loyal and returning customers. Corporate complaint websites give aggressive customers the opportunity to circulate their concerns via a quick medium and to reach people who had paid no attention to that problem in first instance (Bailey, 2004).
The effects and consequences of WOM have been widely investigated in the communication and marketing fields. Early research investigated whether WOM is more effective in influencing the purchase intention of customers than personal selling or marketing activities (Muntinga et al., 2011). Compared with traditional marketing tools, WOM is more trustworthy and reduces customer anxiety because customers can rely on the experiences of others (Bickart & Schindler, 2001). Later, specific research was undertaken to investigate what motivated customers to participate in WOM. With the introduction of Web 2.0, research into the motivations and effects of eWOM evolved. Hennig-Thurau et al. (2004) carried out an interesting study about eWOM. They compared the motivations of customers participating in WOM with the motivations of participating eWOM customers. Surprisingly, they found that participants in traditional WOM had the same motivations for participating as participants in eWOM. This suggests that WOM is comparable with eWOM and that it probably has the same effects and consequences on the consumer behavioral process (Gruen, Osmonbekov, & Czaplewski, 2006).
Purchase intention and affecting factors
A lot of research has been undertaken regarding the factors that affect customers’ online purchase intention. An example of such a research is that of Siu and Cheng (2001). They concluded that most web-shoppers are frequent Internet users. The more frequently they use the Internet, the more intention they have to buy products online. There are many other online purchase process influencers such as online sales promotions, presence of brand names, willingness to purchase on the Internet, trust in the company and financial benefits (Degeratu, Rangaswamy, & Wu, 2000; So, Wong, & Sculli, 2005). Other possible influencers of purchase intention are more specifically related to the web environment of the specific website, such as website layout, amount of information on the website or ease of use of the website (Huang, 2000).
A relatively new influence on online purchase intention is negative eWOM. As stated earlier in this paper, exposure to a negative eWOM comment could influence the attitude or online behavior of a customer (Hennig-Thurau et al., 2010; Lee & Song, 2010)and thus also the purchase or repurchase intention of the customer. To understand to what extent a negative eWOM comment could influence the behavioral process of a customer, it is necessary to know to what degree customers estimate the credibility of the negative eWOM. Important and essential concepts for the explanation of this relationship are perceived eWOM credibility and eWOM
adoption.
Perceived eWOM credibility and eWOM adoption
Perceived eWOM credibility is defined as “the extent to which one perceives the recommendation as believable, true or factual” (Cheung, Luo, Sia, & Chen, 2007, p. 71). It refers to the usefulness of the information in an eWOM message. It says that people evaluate information and, on the basis of that evaluation, will make a decision. Perceived eWOM credibility is an important factor for a consumer to judge the online information and acts as a key factor in a customer’s decision process.
Wathen and Burkell (2002) stated that if a consumer values the information as highly credible, they will adopt it, which could change their attitude towards the brand. In (digital) communication literature, the adoption of information is better known as eWOM adoption. eWOM adoption can be seen as the acceptance and the use of an eWOM expression for making a decision regarding purchase intention (Cheung &
Thadani, 2012). It is a measurement of the acceptance of customer opinions.
The ‘information adoption model’ is a widely-used model that explains the phenomenon of eWOM adoption. It is a variation of the dual process theory, the
Elaboration Likelihood Model (ELM) (Cheung & Thadani, 2012). The information
adoption model explains the process in which people decide to use information or not. In case of eWOM, it predicts whether people will build their final purchase decision on information from eWOM experiences. People who are willing and able to process a message are using the central route. The central route involves a careful examination of the message before making a decision whether the message will be accepted and used for the formation of an attitude or not. In contrast, the peripheral route requires less examination and is triggered by environmental cues from the message. Information accepted via the peripheral route is used less for attitude formation. Thus, people who are motivated and able to accept the message are using the central route and take the information into account for their attitude formation (Petty, Cacioppo, & Schumann, 1983). In case of eWOM, people who evaluate the review as highly credible will treat this information as believable and trustworthy. These customers use the central route and will therefore adopt the eWOM information for their attitude formation (Cheung, Luo, Sia & Chen, 2009).
In line with the information adoption model, empirical evidence has been provided which suggests that if customers perceive an eWOM message as highly credible, it is more likely that they will adopt the information of the message (Fan & Miao, 2012; Cheung et al., 2009; Wathen & Burkell, 2002). According to these authors, a high level of perceived eWOM credibility will positively influence eWOM adoption. Therefore, the following hypothesis was formulated:
H1a: High level of perceived eWOM credibility will positively influence the consumer’s eWOM adoption
It is generally accepted that information from external channels can influence some parts of the behavioural process, and also purchase intention. eWOM is an example of an external channel that could influence purchase intention. As stated earlier, the evidence of the effect of eWOM on the purchase intention is ambiguous, particularly for negative eWOM. Some researchers found evidence of a negative relationship between negative eWOM and purchase intention, while other researchers found the
opposite, with increased sales numbers after negative publicity (Berger et al., 2010; Davis & Khazanchi, 2008; Duan et al., 2008).
To create a sound assumption about the ambiguous relationship between negative eWOM and purchase intention, it is important to recognize the distinction between positive and negative framed reviews. Cheung and Thadani (2012) argued about this distinction, stating that “positively framed eWOM highlights the strengths of a product/service and encourages people to accept a product/service, while negatively framed eWOM emphasizes the weaknesses/problems of a product/service and thus discourages people to accept the product/service” (Cheung & Thadani, 2012, p. 464). In other words: if people accept eWOM information as highly credible and adopt negative framed information, it is less likely that they will purchase the product/service. According to this theory, we can assume that adopted eWOM information will probably influence purchase intention. Based on these previous assumptions, the following hypothesis was formulated:
H1b: A high level of negative eWOM adoption will negatively affect the purchase intention of the consumer
eWOM involvement
As stated earlier, The ELM model explains the process in which people decide to use information or not. This model is driven and moderated by the central role of motivation and involvement. The extent to which a customer is motivated to engineer information about a product depends on the involvement of the customer. Involvement is defined as “a personally perceived relevance of an object based on the own needs, values, and interest of a person” (Lis, 2013, p. 3). In general, involvement refers to the engagement of a customer with a product or an object. If a customer is highly involved with a product or an object, they will also widely search for information about that product or object (Lis, 2013; Petty et al., 1983).
The same will apply to involvement with eWOM. If a customer is an advanced user of eWOM and highly involved with eWOM, it is more likely that this person will take the information from the eWOM message into account and will perceive the information as highly credible. The information in the message will provide extra value for the customer’s perception and make it more likely that he/she will adopt the information in the message (Lis, 2013). Based on this information,
involvement will strengthen the relationship between perceived eWOM credibility and eWOM adoption. Therefore, the following hypothesis was formulated:
H1c: eWOM involvement will strengthen the relationship between perceived eWOM credibility and eWOM adoption
Source credibility
WOM and eWOM information comes from a communicator. WOM literature suggests that whether somebody adopts the information depends on the reliability of the source – the ‘source credibility’. Cheung, Lee and Rabjohn (2008) defined source credibility as “the extent to which an information source is perceived to be believable, competent and trustworthy by information recipients” (Cheung et al., 2008, p. 232). If somebody perceives a source to be highly credible, they generally perceive the information from that source to be highly credible (Cheung et al., 2007). This could lead to a negative or positive effect on customer decisions. For example, imagine that a customer hears a negative experience from a friend about a product. The customer rates the friend as a very good friend and his/her source is therefore highly credible. The perceived information from his/her friend is therefore also highly credible. The customer decides to buy a different product based on the friend’s negative experiences.
Recent studies show that these findings are the same for eWOM. Whether somebody perceives the information from the eWOM message as highly credible depends on the source credibility (Cheung et al., 2008; Watts & Zhang, 2008). Based on these previous findings, the following hypotheses was formulated:
H2: High source credibility will positively influence the perceived eWOM credibility of a customer
Consumer expertise
The expertise or prior experience of the subject of the eWOM message is particularly important. If new information via eWOM is consistent with a person’s prior experience, it is more likely that this person will believe the information because they have the confidence to rely on this information. On the other hand, if the information contained in an eWOM message disagrees with prior beliefs or experiences, it is more
likely that this person will reject the information because the person doesn’t have the confidence to believe it (Bansal & Voyer, 2000).
This is in line with findings from many other studies (Bansal & Voyer, 2000; Park & Kim, 2009). They demonstrated that people with a high level of expertise about a topic do not rely on others’ opinions because they have the self-confidence to make a decision based on their own knowledge and experiences. The ELM model can explain this. As already stated, the ELM model explains why people adopt information and why other people reject information. If people are motivated and able to process the information from a message, they will use the ELM model’s central route. This group with high expertise are able to evaluate the information and they have the confidence to decide whether they adopt the information or not. In contrast, people who have less motivation and are less able to process information use the peripheral route. The information process via this route requires little conscious thinking and is mostly used by people with low expertise. This group will count on information from other people because they lack the ability and confidence to process the information on their own (Petty et al., 1983).
If this information is brought into practice, it can be concluded that high expertise could decrease the value of perceived eWOM credibility. For example, imagine that somebody with high expertise about telephones read something about a poorly functioning iPhone. This person is able to process the information and has also the self-confidence to make a decision based on their own knowledge and experiences. Her/she may conclude that the information is not reliable. Thus, he or she estimates the information as low credibility and therefore rejects the information via the central route. This example shows us that high expertise will decrease perceived eWOM credibility. Therefore, the following hypothesis was formulated:
H3: High consumer expertise will negatively influence a consumer’s perception of eWOM credibility
Quality
The quality of the eWOM message could also be of great influence on perceived eWOM credibility. The quality of eWOM refers to “the strength or plausibility of persuasive argumentation” (Cheung et al., 2008, p. 465). Previous studies have shown that the quality of information is of great influence on the attitude of the customer
towards a product/brand (Petty et al., 1983). This could also be the case in an online environment. If the eWOM is perceived as a high quality message with strong arguments and useful information, receivers will perceive that message as highly credible. As explained earlier, this could lead to benefits in the case of repurchase or trust in an organization. In contrast, if the receivers perceived the eWOM as a message with low quality arguments, they will reject the information because they will decide that the information is not useful. They will perceive the eWOM to have low credibility. As a result, they will change their attitude in a negative way (Cheung et al., 2007; Park, Lee, & Han, 2007). Thus, it is important to know how a consumer judges the quality of the information in an eWOM message because it could say something about their potential buying behaviour and changing attitude. Therefore, the following hypothesis was formulated regarding the influence of information quality on the perceived credibility of the customer:
H4: High quality eWOM messages will positively influence the perceived eWOM credibility of a customer
Consumer involvement
Consumer involvement is defined as “personal relevance or importance of a product/service” (Cheung & Thadani, 2012, p. 6). Within the topic of eWOM, several authors discussed the role of consumer involvement (Cheung et al., 2009; Park & Lee, 2009; Cheung & Thadani, 2012). Theories and relationships that have been created regarding consumer involvement are mostly based on the Elaboration Likelihood Model (Fan, Miao, Fang, & Lin, 2013). As previously stated, motivation is a very important factor within the attitude creation process. In addition to the motivation factor, Petty et al. (1983) distinguish another factor that is important in the attitude creation process. Consumers who process attitudes via the central route are not only highly motivated, but are also very involved with the product/service. If consumers are highly involved with the product or the service, they are more engaged regarding the object and they therefore experience a high need for gathering information (Lis, 2013). Attitude formation exists after comprehensive and detailed analysis of the product/service (cognitive process). In contrast, less motivated consumers use the peripheral route and attitude formation is triggered by environmental cues from the message (affective process). These consumers are less involved and experience a
significantly low need for gathering information about the product (Petty et al., 1983). If this theory is used to shed light on the formation of perceived eWOM credibility, it can be argued that the role of consumer involvement could play a really important part in this formation. Consumers who are more involved with a product/service have more motivation to search for information about that product/service. It is therefore more likely that these consumers will read and use other opinions for their cognitive attitude formation (Fan et al., 2013). Therefore, it is possible that consumer involvement will act as a predicting factor for perceived eWOM credibility. The following hypothesis was formulated for this relationship:
H5: Higher consumer involvement will positively influence the perceived eWOM credibility of a customer
Quantity
Online products are not tangible and the number of eWOM reviews is therefore of great importance for the purchase decisions from customers. The number and the amount of eWOM reviews make the product more tangible and observable for the consumer (Cheung et al., 2009). It is possible that a high number of positive eWOM reviews creates a perception of reliability among the customers: the more consumers experience the advantages of a product, the more reliable the information about the benefits will be. This reduces customer anxiety about the product and raises the possibility that a customer will purchase the product (Cheung et al., 2007, Cheung and Thadani, 2012). However, a high number of negative eWOM reviews can actually do the opposite: the more consumers experience the same problem, the more reliable the information about the problem will be. This will increase customer anxiety and it will decrease the likelihood that they will purchase the product. Some empirical studies (Cheung et al., 2007; Park et al., 2007) agree with this assumption and found evidence that the quantity of reviews indeed influences perceived eWOM credibility. Based on these findings, the following hypothesis was formulated:
H6: A higher number of eWOM reviews will positively influence the perceived eWOM credibility of a customer
Experience goods versus search goods
Experience goods are products or services where product characteristics and features are hard or difficult to observe before purchase. Experience goods are intangible goods and it is therefore only possible to judge the attributes and product characteristics of an experience good after purchase. In contrast, a search good consists of attributes and characteristics that are easy to observe before purchase. A search good is a tangible product and attributes and product characteristics are easy to observe (Fan et al., 2013).
To obtain some knowledge about an experience good, customers search the Internet to gather information from product images, product descriptions and experiences. They search for product reviews, shaped in an eWOM message written by other customers. Park and Lee (2009) stated that eWOM therefore has more impact on an experience good than it has on a search good. In line with Park and Lee’s (2009) conclusion, the following hypothesis was formulated:
H7a: Negative eWOM has more impact on perceived eWOM credibility for an experience good than for a search good
3. Methodology
Participants and research design
To answer the research question in the best way possible, this explanatory research has been conducted using mixed methods. A 2 (experience good versus search good) x 2 (high quality versus low quality) x 2 (high quantity versus low quantity) factorial design has been utilized (see also Figure 2 on the next page). The quality, quantity and product type variables were manipulated and tested via an experiment. The other independent variables – ‘source credibility’, ‘consumer expertise’ and ‘consumer involvement’ – were measured using questions in an online survey. The dependent variables – ‘perceived eWOM credibility’, ‘eWOM adoption’ and ‘purchase intention’ – were also measured using questions in an online survey.
Context
The social network site Facebook was chosen as the research context. Facebook was chosen for several reasons. Over the years, the number of Facebook users has increased significantly. In October 2013, only nine years after its introduction, Facebook had 500 million users (Wolfe, 2013). Many users have integrated the use of Facebook into their daily life (Waters et al., 2009). They use Facebook not only to stay in touch with their friends, but to talk with organisations and to search for specific product or service information as well. Therefore, Facebook is an important tool for organisations to engage with their customers and to cultivate relationships (Waters et al., 2009). Organisations are using Facebook to investigate consumer behaviour and to communicate with customers. Customers can complain about their experience with the organisation via the Facebook page and organisations can respond to these complaints via a public message or a private message (Dekay, 2012).
Thus, Facebook is an important new communication tool for organisations and customers. It offers the opportunity to communicate with each other in a more efficient way than ever before. Moreover, organisations and customers can use Facebook to spread a message about a product or a service. The wide range that Facebook has makes it the perfect context for this study.
Control condition experience good
Control condition search good
As predicted in H7, this research makes a distinction between the effect of a negative eWOM review on the purchase intention for an experience good and for a search good. Therefore, the research was conducted in a product-related organization and a service-related organization, which are examples of a search and an experience good. The Dutch air company KLM was chosen as an experience good and the Dutch beer brewery Heineken was chosen as a search good. Both brands are very popular in the Netherlands and therefore it is less common for a respondent not recognize these brands. Moreover, both brands have a very strong and comprehensive Facebook page for their customers. Customers can complain or give positive feedback to the company via the Facebook pages and it is easy for other customers to find information about the brand on these Facebook pages.
Stimulus material and manipulation
This experiment worked with ten different conditions, eight manipulated conditions and two control conditions. Manipulation took place on the Heineken and KLM Facebook pages. Five conditions were assigned to KLM, the experience-related good company. The other five conditions were assigned to Heineken, the search-related good company. Respondents were randomly assigned to one of the 10 conditions. All manipulations are shown in the Appendix.
In all eight manipulated conditions, a fictional person posted a complaint about a Heineken product or a KLM service. The name of this fictional person was Henk de Bruin. The quality of the complaint was manipulated in two different ways: the number of reactions to the complaint from Henk de Bruin (eWOM quantity) or the quality of the message from Henk de Bruin (eWOM quality).
For the manipulation of the quality of the review, Park, Lee and Han’s (2007) criteria were used. According to their research, these criteria were reliable and therefore useable for this research. Park, Lee and Han (2007) stated that high quality eWOM reviews “are product-relevant, understandable, and persuasive, with sufficient reasons based on facts about the product” (p.132). In contrast, low quality eWOM reviews are difficult to understand and consist of information that is not relevant to the product. Generally, low quality eWOM reviews are emotional and subjective,
there is less product/service information, and contain many subjective expressions (Park et al., 2007). An example of a high quality and low quality review is shown below.
High quality review (Park et al., 2007, p. 132) – Objective and understandable
“The picture on the 3.5’ LCD monitor is absolutely amazing. I am really impressed with the colors and the contrast between darks on such a small screen. Plays songs at top-quality sounds with 5.1 channels. Almost every format is supported”
Low quality review (Park et al., 2007, p. 132) – Emotional and subjective
“Wooooooow! I searched for days and compared every PMP and finally bought one. I’m really enjoying it and it is tough to put it down. All my friends envy my PMP. Right now I’m writing a review, but I can’t wait to play my PMP”
Based on previous research, the same characteristics for low and high quality reviews were used. A high quality message consists of relevant information about a product/service and is understandable for the reader, whereas low quality messages consist of a high number of emotional and subjective expressions. The manipulations for the quality of the messages are visible in the appendix (Park et al., 2007).
For the manipulation of eWOM quantity, Park, Wang, Yao and Kang’s (2011) criteria were used. In accordance with their suggestion for criteria for eWOM quantity, high quantity eWOM reviews were defined as four reviews and low quantity eWOM reviews were defined as two eWOM reviews. Therefore, respondents who were exposed to a low quantity of eWOM messages saw two messages from other respondents as a reaction to the complaint from Henk de Bruin. A high quantity manipulation consists of four eWOM message reactions and respondents were thus exposed to a complaint from Henk de Bruin with four reactions from other customers.
To measure the differences in the effect of the perceived eWOM credibility on the customer’s purchase intention, two control conditions were integrated into the research. The first control condition was assigned to Heineken and the second control condition was assigned to KLM. In both conditions, the complaint from Henk de
Bruin was not shown; respondents only saw a screenshot of the Facebook page of one of the companies. This made it possible to measure if there was indeed an effect.
Pre-test
To ensure that the messages consisted of the right manipulation for the quality of the negative eWOM message, a test among 17 respondents was conducted. The pre-test was conducted to ensure two things. First, to pre-test if the respondents recognized the brands used. This was important because in this research the effect on the customer’s purchase intention was measured. If a customer was not familiar with the brand in first instance, then measure the differences in purchase intention after exposure to a negative eWOM message could not be measured. Therefore, it was important that respondents were familiar with the brands used. The pre-test showed that all the respondents were familiar with the brands Heineken (M = 1.0, SD = .0) and KLM (M = 1.0, SD = .0). All the participants recognized the brands and pre-test results suggested that the two chosen brands could be used successfully.
Second, the pre-test showed whether the manipulation was conducted correctly. Four questions were used to measure eWOM quality, based on Park, Lee and Han’s (2007) scale. Respondents were asked to evaluate the quality of the negative eWOM message on a 7-point Likert scale (1 = totally disagree, 7 = totally agree). Respondents evaluated the messages from Heineken (M = 2.16, SD = 0.38) and from KLM (M = 3.00, SD = 0.84) as low quality messages. However, this difference was not very strong and therefore it was decided to adjust the low quality message slightly. In the final low quality messages, more emotional and subjective statements were used. Information that was not related to the product at all was also added. The fictional person explicitly stated that he was very dissatisfied with the product or the service and also explicitly stated that he would like to receive compensation.
The high quality messages for both Heineken (M = 5.83, SD = 0.76) and KLM (M = 5.67, SD = 0.80) were indeed evaluated as messages with high quality information. It was not necessary to modify the final materials for the high quality messages.
Measurement scales
An initial structured questionnaire was developed, based on reliable scales from recent researches. Some scales were combined and modified because several questions were not relevant to this research. The items used per scale are shown in Figure 3 and were all measured on a 7-point Likert scale (1 = totally disagree, 7 = totally agree). The crossed items in the figure illustrate the items who are removed in this from the original scale.
Source credibility
For the measurement of the independent variable ‘source credibility’, this study used a combination of two reliable scales. Sussman and Siegal (2003) used a reliable 7-point Likert scale for the measurement of source credibility (α = .869) in their research. However, these items did not cover the entire hypothesis for source credibility in this study. Therefore, it was decided to modify the Sussman and Siegal (2003) scale by deleting some items and adding more items to this construct from Ohanian’s (1991) reliable scale (α = .901). Items were translated to an understandable Dutch statement. Doing this, the best possible scale to measure source credibility was created.
According to a reliability analysis, the five items used proved that they formed a reliable scale for source credibility (α = .870). Factor analysis with varimax rotation showed that these items were loading on one factor (EV = 3.307, R2 = 66.14). The
items were merged into a new constructed variable called source credibility (M = 3.56, SD = 1.17).
Consumer expertise
Consumer expertise was measured using Ohanian’s (1991) reliable 7-point Likert scale (α = .856). The questions used measured the extent of the reader’s expertise about the product or the service. Items were modified from questions to statements and translated to a logical Dutch statement.
According to a reliability analysis, the four items used proved that together they formed a reliable scale for consumer expertise (α = .920). Factor analysis with varimax rotation showed that these items were loading on one factor (EV = 3.24, R2 =
80.95). The items were merged into a new constructed variable named consumer
expertise (M = 3.56, SD = 1.17).
Consumer involvement
Premazzi et al.’s (2010) scale was adopted to measure consumer involvement. This reliable scale (α = 0.847) consists of seven statements measured on a 7-point Likert scale. Respondents were asked to indicate their feelings towards the product or the service. One item was removed from the scale because, after translation to Dutch, it had exactly the same meaning as another item. The scale included items such as “this brand is important to me” and “this brand is relevant for me”.
As in Premazzi et al.’s (2010) research, this scale proved to be reliable (α = .905) and the items loaded on one factor (EV = 4.14, R2 = 69.05). The seven items
used were merged into a new constructed variable named consumer involvement (M = 4.82, SD = 1.25).
Perceived eWOM credibility
For measurement of perceived eWOM credibility, Cheung et al.’s (2009) scale was combined with Park et al.’s (2011) scale. Both scales were very simplified and limited so it was decided to combine them in order to create a comprehensive scale with six items to measure perceived eWOM credibility. Cheung et al.’s (2009) scale had a Cronbach’s alfa reliability of 0.904 and the Park et al. (2011) scale had a Cronbach’s alfa reliability of 0.810. Both were measured on a 7-point Likert scale.
According to a reliability analysis, this 6-item scale proved to be reliable (α= 0.924). All six items loaded on one factor in a factor analysis with varimax rotation (EV = 4.36, R2 = 72.77). The six items were combined into a new variable named perceived eWOM credibility (M = 4.37, SD = 1.33).
eWOM adoption
Cheung et al.’s (2009) scale was used to measure eWOM adoption. This reliable scale (α = .898) consists of five items and was measured using a 7-point Likert scale. According to a reliability analysis, it was concluded that this scale was indeed reliable
(α = .908). The five items used were loaded on one factor after a factor analysis with varimax rotation (EV = 3.73, R2 = 74.53). The five items used were merged into a
new constructed variable named eWOM adoption (M = 3.38, SD = 1.41).
eWOM involvement
To measure how involved a respondent is with eWOM, Lis’s (2013) four-item scale was used. One item was removed because it was not related to the research question. Therefore three items were used in this research to measure the moderating effect for eWOM involvement.
According to a reliability analysis, the three items used formed a reliable scale to measure eWOM involvement (α = .742). They loaded on one factor in a factor analysis with varimax rotation (EV = 1.98, R2 = 66.03). The three items were
combined to create a new variable named eWOM involvement (M = 3.51, SD = 1.36).
Purchase intention
The aim of this research was to investigate the change in purchase intention after exposure to a negative eWOM message. This change was measured by looking at the difference in the purchase intention before (Purchase IntentionBefore) exposure and the purchase intention after (Purchase IntentionAfter) exposure to the manipulated eWOM treatment. Respondents were asked to answer the two questions related to purchase intention twice. By comparing the numbers of the combined scales of Purchase IntentionBefore and Purchase IntentionAfter, an indication of whether a negative eWOM message is of influence to the purchase intention from the consumer can be illustrated. Park and Lee’s (2009) 2-item scale was used to measure purchase intention. Respondents were asked to give an indication of how likely it was that they would buy a product or a service from the specific company. Items were measured on a 7-point Likert scale.
A reliability analysis proved that the two items used to measure the purchase intention before exposure formed a reliable scale (α = .885) for the measurement of purchase intention before treatment. The two items loaded on one factor in a factor analysis with varimax rotation (EV = 1.79, R2 = 89.72) and were therefore combined
into one new variable named Purchase IntentionBefore (Mbefore= 5.43, SD =1.57).
According to the reliability analysis with the two items that were used to measure the purchase intention after exposure to the treatment, the two items together formed a reliable scale (α = .933) to measure the purchase intention after exposure. The two items were loaded on one factor in a factor analysis with varimax rotation (EV = 1.87, R2 = 93.54) and were therefore combined into one new variable named Purchase IntentionAfter (Mafter== 4.59, SD = 1.58).
Manipulation check
To check if the correct number of eWOM reactions was used in the manipulation for eWOM quantity, a manipulation check for the variable eWOM quantity was conducted. The questionnaire included one item that was designed to measure the respondents’ perceptions about the number of eWOM reactions to the negative eWOM message. This item made it possible to conduct a manipulation check for the variable eWOM quantity. The item was based Park, Lee and Han’s (2007) scale and asked respondents the following question: “I think there were a lot of comments on the message from Henk de Bruin”. Answering was possible on a 7-point Likertscale (1=totally disagree, 7=totally agree).
To measure the manipulation for the variable eWOM quantity, an independent t-test was used. With this independent t-test the conditions with a high number of reactions were compared with the conditions with a low number of reactions. The dependent variable in this analysis was the item used for the manipulation check. For the grouping variable the ‘low quantity’ and the ‘high quantity’ conditions were used.
The respondents in the conditions with a low number of reactions to the message from Henk de Bruin also perceived the number of reactions to be low (M = 2.47, SD = 1.45). The respondents who were exposed to a condition with a high number of reactions perceived that number of reactions as higher than in the ‘low quantity’ condition (M = 3.89, SD = 1.81). This difference is significant, t (129) = 4.954, p < 0,001. Thus, respondents indeed perceived the number of reactions as higher in the ‘high quantity’ conditions. However, this difference is not really large or strong.
eWOM quality (pretest)
Park, Lee and Ham (2007)
- The review has sufficient reasons supporting the opinions
The review is objective The review is understandable The review is credible The review is clear
In general, the quality of the review is high
eWOM quantity Park, Lee and Ham (2007)
The number of reviews is large
The quantity of review information is large
Source credibility Sussman and Siegel (2003)
0.869 How knowledgeable is the person who wrote this message, on the topic of the message?
To what extent is the person who wrote this message an expert on the message topic How trustworthy is the person who wrote this message an expert, on the message topic
How reliable is the person who wrote this message, on the topic of the message
Ohanian (1991) 0.901 The reviewer is undependable
The reviewer is honest The reviewer is reliable The reviewer is sincere
The reviewer is trustworthy
Consumer expertise Ohanian (1991) 0.856 The reviewer is an expert The reviewer is experienced The reviewer is knowledgeable The reviewer is qualified The reviewer is skilled
Consumer involvement
Premazzi 0.847 Please indicate your feelings about mobile phone service:
Important to me 1 – 7 Not important to me Of no concern to me 1 – 7 Of concern to me
Irrelevant 1 – 7 Relevant
Very meaningful to me 1 – 7 Means nothing to me
Matters to me 1 – 7 Doesn’t matter to me
Interesting 1 – 7 Not interesting
Significant 1 – 7 Insignificant
Boring 1 – 7 Exciting
Perceived eWOM credibility
Cheung et al. (2009) 0.904 I think this review is factual
I think this review is accurate I think this review is credible
Park et al. (2011) 0.810 I believe the online review which been read a lot
I believe the online review which is believed by others I believe online review is important and credible information I believe online review is written with responsibility
eWOM Adoption Cheung et al. (2009) 0.898 To what extent do you agree with the review
Information from the review contributed to my knowledge of the product discussed The review mad it easier for me to make my purchase decision
The review had enhanced my effectiveness making a purchase decision The review motivated me to take purchasing action
eWOM involvement Lis (2013) 0.911 I am interested in online recommendations
I always wanted to know more about online recommendations, so I appreciate if friends give me some explanations
Online recommendations are a hobby of mine Online recommendations are important to me
Online purchase intention
Park and Lee (2009) How likely is it that you will buy this product
How likely is it that you will recommend this product to your friend
The questionnaire was presented to respondents who were recruited through email and social media. Respondents were able to complete the questionnaire via a digital link. The questionnaire consisted of four parts: general, pretesting, marketing and communication.
The survey started with a general section, which consisted of an introduction to the research and some general questions about the demographics of the respondents. After completing the general questions, respondents were directly referred to the second section of the survey, pretesting. In this section, respondents were randomly assigned to one of the two embedded conditions, KLM or Heineken. Respondents were exposed to a brand logo from Heineken or KLM followed by questions about the purchase intention and the respondents’ involvement with the product or service from this brand.
After successfully completing the pretesting section of the survey, respondents were exposed to the third section: marketing. Respondents were randomly assigned to one of the eight manipulated conditions or the two control conditions for Heineken or KLM. Respondents’ exposure to a treatment from KLM or Heineken was dependent on the condition in the pre-test section of this research. After extensive examination of the treatment, respondents were asked to answer some questions about the complaint from Henk de Bruin on the Facebook page. Questions were based on the previously explained measurement scales.
As stated earlier, this research was conducted in combination with an experiment for another master’s thesis. The experiment for that research was a continuation of the experiment for this research. However, the focus in the other research was on the organisation’s reaction and not on the content of the eWOM message. Therefore, after answering the question related to the marketing section in the survey, respondents were assigned to one of the ten experimental conditions contained in the communication section. In this section, respondents were asked to answer several questions related to the experiment.
Because one questionnaire was used for both experiments, the questionnaire was also added to the document for the master’s thesis Webare on Facebook (de Jong, 2015). Therefore, it is possible that both documents have comparable questions in the
attachments. However, the manipulations for both experiments are different because two independent experiments were conducted.
Sample
An online questionnaire was provided to the respondents between 28th November 2014 and 10th December 2014. The questionnaire was available online and was completed by 363 respondents. After rejecting the missing values, 342 usable respondents remained (n = 342). This meant that, on average, there was a group of 33 respondents per condition. From the sample, 169 respondents completed the questionnaire for Heineken (n = 169) and 173 respondents (n =173) completed the questionnaire for KLM (see Figure 5).
The sample group consisted largely of women: 73.9% (n = 258) of the sample group were women. The remaining 26.1% (n = 91) of the sample group were men. Respondents were asked to indicate how active they were on Facebook in their daily life. Of the sample group, 92.8% (n = 324) of the respondents indicated that they had a Facebook account and 79.4% (n = 277) of this group indicated that they were active users. The respondents were also asked to give an indication of the time they spent on Facebook each day. From the sample, 36.1% (n = 126) of respondents indicated that they spent 10 to 30 minutes on Facebook each day. Despite these times, only 13.8% of the sample group had lodged a complaint about a product or a service on an organisation’s Facebook page (n = 48). The sample group consisted largely of well-educated people: 71.1% of the sample group had obtained an academic degree (n = 248).
Control condition experience good (n=35) Control condition search good (n=35)
4. Results
Relationship between perceived eWOM and eWOM adoption
It was proposed that a high level of perceived eWOM credibility as a reaction to a negative eWOM message engendered an adoption of the eWOM information from the message (H1). To test this hypothesis, a regression analysis with perceived eWOM credibility as the independent variable and eWOM adoption as the dependent variable was conducted. According to the regression analysis, a positive significant effect of perceived eWOM credibility on eWOM adoption (F (1,252) = 145.1, p < .01) was found. The results showed that, by means of the regression model, the eWOM adoption of respondents could be predicted. This prediction is very useful: 60% of the differences within the eWOM adoption can be explained by the perceived eWOM credibility (R2 =.604, t = 12.046, p < .01). The effect of the perceived eWOM credibility on the eWOM adoption is positive; eWOM adoption will increase by .637 if the perceived eWOM credibility value increases with 1 point.
If a one-way variance analysis is performed where the means for the eWOM adoption for the two different companies (Heineken and KLM) is compared, the results show that there is not a significant difference for the effect of perceived eWOM credibility on eWOM adoption for Heineken (M = 3.52, SD = 1.43) or KLM (M = 3.25, SD = 1.37), F (1,254) = .908, p = .342.
Confirming expectations, these results indicate that a consumer will adopt the information from a negative eWOM message if he/she perceives the eWOM message to be highly credible. These results support H1a.
Effect of eWOM adoption on the customer’s purchase intention
As stated earlier, the changes in a customer’s purchase intention after exposure to a negative eWOM message was examined. This change was measured by comparing the purchase intention before exposure and the purchase intention after exposure. A new variable named Purchase IntentionDifferences (ΔPIDifference) was created. To create this variable, the value of the customer’s purchase intention after exposure was subtracted from the value of the purchase intention before exposure (ΔPIDifference = PIAfter – PIBefore). A regression analysis with ΔPIDifference as the dependent variable and
eWOM adoption as the independent variable showed that there is a significant negative relationship between eWOM adoption and ΔPIDifference. (F (1, 251) = 39.42, p < .01)). Moreover, the regression model showed that it is possible to predict the effect of eWOM adoption on a customer’s purchase intention after exposure to a negative eWOM message. The regression model shows that 36.9% of the difference within customers’ purchase intention can be explained by the eWOM adoption (R2 = .369, t = - 6.279, p < .01). This effect is negative: the customers’ purchase intention will decrease by -.359 if the eWOM adoption value increases by 1 point. These results indicate that a consumer’s adoption of information from an eWOM message influences the purchase intention of that customer. If a one-way variance analysis is performed where the means of ΔPIDifference for the two companies (Heineken and KLM) are compared, the results show that a significant difference does not exist between the effect of eWOM adoption on respondents’ purchase intention in the Heineken condition (M = -1.11, SD = 1.52) or KLM (M = -.56, SD = 1.15), F (1, 252) = .014, p = .905.
Confirming expectations, these results indicate that purchase intention will be negatively influenced by a consumer’s eWOM adoption. The more information the consumer adopts, the stronger the influence of eWOM adoption is on the consumer’s purchase intention. These results provided enough support for H1b and therefore the hypothesis is accepted.
Enhanced effect eWOM involvement
Hypothesis H1c proposed that the more involved the consumer is with eWOM, the stronger the effect of perceived eWOM credibility on eWOM adoption will be. To test this hypothesis, a two-way variances analysis to test whether eWOM involvement has influence on the relationship between perceived eWOM credibility and eWOM adoption, was performed. The dependent variable in this analysis was eWOM adoption; the independent variables were perceived eWOM credibility and eWOM involvement. No significant main effect of eWOM involvement on the dependent variable eWOM adoption (F (1, 253) = 4.130, p = .848) was found. However, an interaction effect between perceived eWOM credibility, eWOM involvement and eWOM adoption (F (1,249) = 4.130, p < .05) was found. This means that the more
