Online word-of-mouth, what are the triggers for certain motivations to complain online? : a study about the relationship between dissatisfied consumer experiences, and motivations why people engage in eWOM : with an ana

(1)

ONLINE WORD-OF-MOUTH,

WHAT ARE THE TRIGGERS

FOR CERTAIN MOTIVATIONS

TO COMPLAIN ONLINE?

A study about the relationship between

dissatisfied consumer experiences, and motivations

why people engage in eWOM. With an analysis to

examine a possible moderating effect for social

value orientation.

Schimmel, L.

6103634

University of Amsterdam

Master Thesis- Business Studies

Supervisor: Dhr. dr. A. Zerres.

(2)

(3)

__________________________________________________________________________

1. Abstract

The era for Internet nowadays is getting more and more important than the traditional media. Consumers can easily interact with each other and with firms via Social Media sites as, Facebook and Twitter. Complaining about services and firms happens every day all over the world. One of the biggest review sites Tripadvisor, has 260 million unique visitors each month, and more than 150 million reviews online in the database (“Tripadvisor”, 2014). Based on the growing importance and interest in the topic of Internet and WOM, this study investigates the relationship between several different failures (outcome or process), the magnitude of those failures (single or multiple), and if there is a moderating effect for social value orientation. Results in this study provided empirical evidence that a lower social value orientation results in more venting feelings, revenge, empowerment, and self-enhancement motivations in the relationship between the scenarios and motivation. The scenarios provided few significant results, ‘multiple process failure’ and ‘multiple outcome failure’ show the same result for all five motivations. ‘Multiple outcome failure’ shows higher results for helping others than the scenario ‘Single outcome failure’.

(5)

___________________________________________________________________________

2. Introduction

All around us marketers are trying to influence our thoughts, and marketing departments are spending millions to achieve that with all kinds of advertisements. But what really makes up a customer’s mind doesn’t costs millions or extensive marketing campaigns, a word-of-mouth (WOM) recommendations from a trusted source can achieve more in the consumers’ mind than a marketing campaign (Bughin, Doogan & Vetvik, 2010). WOM has a great influence on how (future) customers feel or think about a certain service, company or product. WOM is one of the most influential channels of communication in the marketplace, because WOM is seen as more credible than marketer initiated communications, the communication through WOM is initiated by ‘people like me’, it is addressed as more reliable than communication through a marketer (Allsop, Bassett & Hoskins, 2007). 20 To 50 percent of all purchasing decisions is influenced by WOM, and it’s greatest when people are buying a product for the first time or if a product is relatively expensive (Bughin et al., 2010). Interest in the phenomena WOM started in the early 60s ( Arndt, 1967; Dichter, 1966; Engel, Kegerreis & Blackwell, 1969) with studies about the effects of WOM. Westbrook (1987, p. 261) explained consumer WOM as “the transmissions consist of informal communications directed at other consumers about the ownership, usage, or characteristics of particular goods and services

and/or their sellers”.

Over the past decade WOM has become even more powerful due to the technological explosion of informal communication channels such as the Internet, email, cell phones, text messaging and blogs (Allsop et al., 2007). The Internet and other technological innovations caused for more interest in the phenomena WOM, because through the Internet more people

(6)

get involved in WOM. The eWOM platforms empower consumers to share their experiences and opinions with others, that empowerment makes understanding why they share that information more and more crucial (Breazeale, 2009). Sharing of information through an Internet channel is called electronic Word-Of-Mouth (eWOM). Hennig-Thurau, Gwinner, Walsh and Gremler (2004, p. 39) define the extension of traditional WOM, eWOM, as “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet”. In this study the focus will be on when people engage in negative eWOM (N-eWOM). N-eWOM “includes all negatively valenced, informal communication between private parties about goods and services and the evaluation thereof” (Wetzer, Zeelenberg, & Pieters, 2007 p. 661-662), which is posted through an Internet channel. It is important to know for marketers what negative eWOM means for them, it is difficult to influence people who engage in eWOM and the only way they can do that is through a delivery of great experiences. Delivery of a great experience can be done through delivering a good outcome, and/or by the experience a consumer has while the process, which is called the process. A lot of research is done by thinking about the effects of WOM on potential and actual customers, but the first step in the process lies by the person who is engaging in WOM in the first place. Therefore, it is important how people react on certain dissatisfied experiences and how the forthcoming failures influence the motivation of engaging in eWOM. Failures which can occur during the experience will be distinguished as outcome failure and process failure, and there will be examined if there occurs a difference when a consumer is faced with a one-time failure or a multiple-time failure. The relationship between the experience and the motivations of engaging in eWOM can also be influenced by certain personality traits why people engage in eWOM. Through this study we will examine this

(7)

relationship and the moderator social value orientation which can influence this relationship. Therefore the research question this study will address is:

Is the relationship between dissatisfied consumer experiences, measured by outcome failure, process failure and magnitude, and the motivations why people engage in eWOM, helping others, venting feelings, empowerment, revenge, and self-enhancement moderated by the construct social value orientation?

Contributions this study could bring will be in the field of the marketers. The main theoretical implication of this study is to fill the gap in the literature, there is a blind spot in the literature about the influencing factor of social value orientation and the influencing factor of process failure and outcome failure in combination with the magnitude of the dissatisfied experiences. The most important managerial implication which this study wants to show is how an outcome failure and a process failure differ in why people will engage in eWOM, and thereby

influence potential customers.

The next part of this study will start with a review of the existing literature on motivations why people engage in WOM and eWOM. This will be done with theoretical frameworks of inter alia; Sundaram, Mitra & Webster (1998) and Hennig-Thurau et al. (2004), and will be supported by other studies. Secondly, a review of the existing literature on the dissatisfied experience will be given, especially looking at the difference between outcome failure and process failure, and between failures that occurs once and multiple times will be examined. Lastly, the moderator social value construct will be explained, and how this construct can have a moderator effect on the relationship between the dissatisfied experience and motivations of engaging in eWOM.

(8)

___________________________________________________________________________

3. Literature review

3.1. Motivations for eWOM.

Several studies claim the motivation why people engage in WOM (i.e. Anderson, 1998; Dichter, 1966; Sundaram, Mitra, & Webster, 1998), or eWOM (i.e. Akyüz, 2013; Hennig-Thurau et al., 2004; Verhagen, Nauta & Felberg, 2013). The notion of both traditional WOM and eWOM are about the same, studies about the concept of WOM could be combined or associated with certain motives why people engage in eWOM. The difference between the two concepts is the anonymity of the individual who engaged in eWOM, the recommendations are typically from unknown individuals with which people do not have strong ties which is the case with traditional WOM (Bronner & de Hoog, 2011). An effect of this unknown part in the recommendation is that people can receive the online recommendations with less credibility. For this study, where the effect of the WOM doesn’t play a role, the findings of studies about motivation to engage in traditional WOM and eWOM can be integrated with each other. Jeong & Jang (2011), explained that WOM and eWOM differ from each other when looking at it as a physical communication tool, but their role in interpersonal influence remains the same. That’s why this study will integrate the two together, and shows them as similar functions.

Dichter (1966) studied as one of the first what people drive to engage in WOM communication, according to Dichter (1966) there are four main motivational categories of positive WOM communication: product-involvement, self-involvement, other-involvement, and message-involvement. Sundaram et al. (1998) stated that findings from consumer behavior, sociology and cognitive psychology studies suggests that consumption experiences

(9)

produce affect, which causes for some source of human motivation. The resulting motivation determines the nature of post-consumption behavior such as WOM communication (Westbrook, 1987). This relationship makes it clear that some emotions have a significant influence on motives why people engage in eWOM. In some earlier literature they already mentioned that the motive to engage in WOM is the degree of truly satisfaction and dissatisfaction with a consumer experience (Anderson, 1998), or when consumers’ consumption-related expectations are disconfirmed (i.e., Anderson, 1998; Hennig-Thurau et al., 2004). Dichter (1966), as mentioned earlier, was one of the first to mention some motives (product-involvement, self-involvement, other-involvement, and message-involvement) which influence WOM, his study only focused on the WOM which had a positive content, and was thereby the first step in this research field. Engel et al., (1969) modified Dichter’s list and introduced a new additional motive, namely dissonance reduction, which they see as a reason to engage in negative WOM. One of the first study which focused as much on negative as on positive WOM is the study by Sundaram et al. (1998), this study is a starting point for this present study. Sundaram et al. (1998) identified eight motives for consumer WOM communication, also corresponding to the original categories suggested by Dichter (1966). Four of the eight motives explain why people engage in positive WOM (i.e. altruism, product involvement, self-enhancement, and helping the company), while the other four motives give explanation why consumers involve in negative WOM (i.e. altruism, anxiety reduction, vengeance, and advice seeking). Hennig-Thurau et al. (2004) studied the motivations of eWOM and stated that the motivation for traditional WOM are also applicable to the concept eWOM. The study done for eWOM had covered most of the motivations which are used in studies for traditional WOM. The authors suggested eight motivations for eWOM, and their study showed that, concern for other customers, extraversion/positive self-enhancement, social benefits, and economic incentives were the main motivations behind engaging in

(10)

eWOM. Looking at the motivations for engaging in negative eWOM, social motivations are not as obvious as when people engage in positive eWOM (Yap, Soetarto, & Sweeney, 2013). However, Yap et al. (2013) found that people engage especially in negative eWOM to warn other consumers, and to vent their negative feelings. Jansen, Zhang, Sobel, & Chowdury (2009) discuss the motivations for engaging in negative WOM more as pro-self, namely hostility and vengeance. A study about experiences in the traveling industry by Sparks & Browning (2010), found that the motives of hotel guests to complain online can be through altruistic feelings or at the other end of the continuum, by revenge. Willemsen, Neijens, and Bronner (2013) came to almost a same conclusion as Yap et al. (2013), with venting, altruism, and empowerment as almost as important drivers for negative eWOM in the webcare.

These previous studies and their motives can be taken together to suggest five overall motives for engaging in negative eWOM, venting, helping others, revenge, self-enhancement and empowerment, which cover several sets behaviors. The most common motivation under complaint behavior is the desire to vent (i.e. Wetzer et al., 2007; Sundaram et al., 1998 ), and therefore has been suggested as a motive for negative eWOM (Willemsen et al., 2013). Venting the negative feelings which is associated with a discontent that they have experienced can help to reduce the frustration and the anger which is associated with a particular event (Sundaram et al., 1998). Accordingly, consumers share a negative consumption experience through engaging in eWOM can help in order to reduce the discontent that they have experienced (Hennig-Thurau et al., 2004).Other factors which belong to the umbrella of venting are; reduction of cognitive dissonance, dissonance

reduction, anxiety reduction.

The second motive helping others is also an important motive underlying negative eWOM (i.e. Willemsen et al., 2013; Wetzer et al., 2007; Sundaram et al., 1998). If people’s

(11)

motivation to engage in eWOM are driven by altruistic feelings, they want to warn the other consumers about their unsatisfactory experience, without anticipating any reward in return (Sundaram et al., 1998). They are concerned about the welfare of their fellow consumers, and they want to spare them the problems that they have encountered with particular products or

services (Willemsen et al., 2013).

Revenge is a motive which is in reaction of egoistic thinking’s, and can be seen as the most

aggressive motivation for engaging in (negative) WOM (Wetzer et al., 2007). The purpose of revenge is to give the person a relief from the anger. Bechwati and Morrin (2003, p. 6) define ‘desire for consumer vengeance’ (DCV) as, “the retaliatory feelings that consumers feel toward a firm, such as the desire to exert some harm on the firm, typically following an extremely negative purchase experience”. An action of revenge is thus the result of an extremely negative experience, as Bechwati and Morrin (2003) describe it. The last overall motive to engage in (negative) eWOM is empowerment (i.e. Bronner and de Hoog, 2011; Hennig-Thurau et al., 2004). “For consumers who are driven through empowerment, online complaining with negative eWOM is an instrument of power. Aware of the impact that negative eWOM can have for companies, consumers voice their complaints online in order to draw the attention of companies, and enforce redress” (Willemsen et al., 2013). People are aware of the impact their review could have, and therefore use it to achieve attention, or better their negotiating position with the firm (Hennig-Thurau et al., 2004). People who express themselves negatively will be noticed earlier by the firm and other consumers, so the negatively written eWOM can be used as an instrument to gain power.

Self-enhancement is a motive which allows a person to gain attention. Engel, Blackwell &

Miniard (1993, in Hennig-Thurau, et al., 2004) illustrate that not only getting attention from the crowd, but also claiming superiority, showing connoisseurship, and suggesting status are part of the motivations which is called self-enhancement. The creator of the review wants to

(12)

show other consumers that (s)he is an intelligent consumer. They want to build their own reputation (Wasko, & Faraj, 2005), when participating in WOM.

3.2. Dissatisfied experience.

The motive to engage in eWOM depends on the consumer experience one had. Anderson (1998) stated that the most important reason why WOM communication arises is when consumers’ consumption-related expectations are disconfirmed. Some (negative) emotions such as, disappointment, frustration, anger and regret follow these disconfirmed expectations. Motives to engage in WOM are significantly related to consumption experiences (Sundaram et al., 1998). Some early studies about causes for WOM concluded that product dissatisfaction is the root cause of WOM (Day, Grabicke, Schaetzle, & Staubach, 1981; Blodgett, Granbois, & Walters, 1994). Anderson (1998) also found an explanation why extremely satisfied people involve in WOM, he found an asymmetric U‐shaped relationship between customer satisfaction and WOM. The chance that a customer engages in WOM is higher for extremely dissatisfied and extremely satisfied customers, than for customers who are moderately satisfied. Those people are motivated to engage in WOM and tell their experiences online. Other research by East, Hammond, and Wright (2007) found that negative WOM is more common than positive WOM. To make a clear direction in this study there is chosen for some explicit negative dissatisfied experiences. The fact that consumers are dissatisfied does not tell much about what kind of specific behaviors someone will engage in (Zeelenberg & Pieters

1999).

Failures in the consumer experience can occur for different reasons. The experience can result in a dissatisfied experience when one of the components involved in the experience doesn’t meet the expectations the consumer had on forehand. These dissatisfied experiences can occur

(13)

during the interaction between a customer and a product, a company, or a part of the organization. Gentile, Spiller & Noci (2007) state that a customer experience consists of interactions between a customer and a product, a company and the organization itself. And if one of the components doesn’t meet the expectations, a customer can create negative thoughts pertaining to the whole experience. To make a distinction in this study, we need to divide the customer experience into parts. Zhu, Shivkumar, & Parasuraman (2004) classified the process of satisfied services into outcome failures and process failures. An outcome failure was described as getting the wrong book when you ordered via an online bookstore because the shipping went wrong. An example of process failure can include a night-shift hospital nurse who treating an emergency patient rudely. In this study we will use the process failure and the outcome failure to make a distinction in the component customer experience. Smith, Bolton, & Wagner (1999, p. 358) state that through mental accounting customers can classify the various types of resources lost due to failure into different categories. “The valuations of customers will differ by type of failure because outcome and process failures represent different categories of loss.”

3.2.1. Process failure.

The process failure dimension involves the delivery process, how do people receive the service (Smith, et al., 1999). When a process failure occurs it doesn’t affect the final core service, an process failure thereby typically involves a symbolic exchange (Smith et al., 1999). There is no economic loss of value incorporate in this failure, and Smith et al. (1999, p. 362) call “inattentive service” as a representative of process failure. In the study of Smith et al. (1999) they conclude that for restaurants and hotels, customers who experience process failures are more dissatisfied than customers who experience outcome failure. The results

(14)

indicate that face-to-face service encounters cause for more dissatisfied thoughts than if a customer experience an outcome failure during an experience, such as an unavailable service). Stevens, Knutson & Patton (1995) use five of the ten dimensions of service quality developed by Parasuraman, Zeithaml, & Berry (1988) to measure service quality in restaurants. The measurement Parasuraman et al. (1988) use is called SERVQUAL. SERVQUAL is an instrument to measure the gap in quality between the service that consumer think should be provided and what they think actually has been provided (Stevens, et al., 1995). The five dimensions used by Stevens et al. (1995) to measure the service quality in the restaurant industry are called DINSERV and consist of the following determinants; reliability (to perform the promised service reliably and accurately); assurance (knowledge and courtesy of the employees and their ability to convey trust and confidence); responsiveness (willingness to help customers and provide prompt service); tangibles (physical facilities, equipment, and appearance of personnel); empathy caring (individualized attention). The process outcome entails the restaurant experience and the overall service of the employees at the restaurant. Several studies declare that the process outcome in a restaurant setting has a greater impact on the satisfaction than an outcome failure (Smith, et al., 1999; Yüksel & Yüksel, 2002; Andaleeb & Conway, 2006 BRONBRO). Yüksel and Yüksel (2002) results show that service quality have the most significant effect on satisfaction at an aggregate market level. They suggest that “no matter how brilliant the marketing plan is, it all comes down to nothing if there is a breakdown at the most critical interface: the point at which the customer comes in contact with the employees” (Yüksel and Yüksel, 2002 p. 64). Andaleeb and Conway (2006) compared the food quality/reliability, physical design, price, and service responsiveness to investigate which is the most important contributor to customer satisfaction. The outcome of the study shows that service responsiveness had the greatest influence on customer satisfaction.

(15)

3.2.2. Outcome failure.

The organization does not fulfill their basic need and does not perform their core services (Smith, et al., 1999), when customers experiences an outcome failure. A customer is experiencing economic loss due to the fact the organization is not as performing as expecting to do so. The outcome failure is thus related to the utilitarian exchange (Smith et al., 1999). An explanation by Smith et al. (1999) why customers value a process failure as more dissatisfied is that an outcome failure results from a behind-the-scene event, where they couldn’t address the real reason behind the failure.

Outcome failure in the setting of a restaurant can be explained as the food quality. Peri (2006) mentioned that food quality can be describes as “the requirements necessary to satisfy the needs and expectations of customers”. The quality of the food is described as: the presentation; variety of choice; healthy options; taste; freshness; and temperature (Namkung & Jang, 2007). Jang and Namkung (2009), completed a second study in this field looking at perceived quality, emotions, and behavioral intentions. They mention that a product, in this case the served food, has to be visually attractive, healthy, and fresh, to create satisfaction in the eyes on the consumer. A third study, which looked at what kind of restaurant experiences trigger positive eWOM, declared that people get triggered when the food served was tasty, the presentation was visually attractive, and had an appropriate temperature (Jeong, & Jang, 2011). Outcome failure in this case will occur when at least one of these aspects isn’t fulfilled.

3.2.3. Magnitude of dissatisfied customer experience.

In this study there will be a distinction between a single failure, process failure and outcome failure, and a failure which occurs multiple times. In this study there will be assumed that

(16)

when a failure occurs more than once the dissatisfaction about the experience will grow. Loo, Boo & Khoo-Lattimore (2012) results indicate that customers are more likely to give companies with certain failures a second change to rectify the problems. The failures described in the study like, slow service, wrong order, and unavailable service are typical process failures as described in this current study. So, it’s credible that a single failure will receive the benefit of the doubt compared to a failure which occur multiple times. Another result from the study by Loo et al. (2012) declares that the number of complaints rise with almost ten percent when a failure which is related to product occurs multiple times. Product failure in this study refers to what customers receive from the delivered service. In the current study this can be linked with the outcome failure in our setting.

3.3. Social Value Orientation.

It is well known that individuals evaluate outcomes for self and others in different ways (de Dreu & Boles, 1998; Kelley & Thibaut, 1978). Social value orientation describe people as being pro-self and/or pro-other interest when they make decisions. Deutsch (1960) studied the different effects of three motivational orientations which relate to the aspects of the social value orientation. The three categories which Deutsch (1960) refer to are cooperative, individualistic, and competitive, who can also be defined as other for cooperator, and pro-self for the categories individualist and competitor (de Dreu & Boles, 1998). Cooperators has as main goal to maximize joint outcomes, the individualist wants to maximize their own outcome without thinking of others, and competitors wants to maximize their relative advantages over others’ outcomes (de Dreu & Boles, 1998). Relative to individualist and competitors, pro-others (cooperators) possess greater levels of fairness and caring for others (de Dreu & van Lange, 1995). These findings by de Dreu & van Lange (1995) are in the same

(17)

direction as what is stated for pro-others and pro-selfs. Pro-others tend to give people the benefit of the doubt, their judgments of others are more positive, and therefor are more concerned about the well-being of others. The pro-selfs, individualists and competitors, are more concerned about their own well-being and act in a way in which their own interest is the most important part.

(18)

___________________________________________________________________________

4. Research gap & research question.

Prior studies about these topics provide practically none guidelines for using the single failures and multiple failures in order to determine the motivations for engaging in eWOM. As Anderson (1998) stated, people engage in WOM communication when the consumers’ consumption-related expectations are disconfirmed, in what magnitude this disconfirmed expectation moves when facing an single failure and multiple failure isn’t clear in this study or any other study. The effects of WOM are studied in great detail, and it is proven that in many cases WOM has a greater impact on sales than traditional marketing has (Trusov, Bucklin, & Pauwels, 2009). The effects of negative WOM are of great need to examine, due to the availability of a large number of Social Network Sites and discussion fora it is easier for a dissatisfied consumer to reach a large audience. Understanding what consumers motivates to speak out loud (online) about a negative experience gives clarity in how firms can manage their failures. Concluding a study by Loo, et al. (2012) it’s credible to say that firms most often receive the benefit of the doubt when consumers face an single outcome failure. In the case of a process failure, people tend to have greater problems overlooking a failure (BRON?). Why people engage in WOM is also in line with the personality one has. To give a complete explanation for the motivations, it could be important to look at the Social Value Orientation one has. SVO as a moderator and motivations for WOM are not yet combined by any noteworthy study, the two-fold of the SVO is expected to have an influence on the type of

motivations people have.

Studies in the past 50 years explored the motivations why people engage in WOM (e.g. Bronner, & de Hoog, 2011; Dichter, 1966; Engel, 1969; Hennig-Thurau et al., 2004;

(19)

Sundaram, et al., 1998; and Wetzer, et al., 2007). Nevertheless, the effect of different types of failures (outcome single, outcome multiple, process single, process multiple), and the possible moderating effect of SVO are under-represented in these studies. The aim of this thesis is to get more insights into these three variables, and the relationships among them. This leads to the following research question:

Is the relationship between dissatisfied consumer experiences, measured by outcome failure, process failure and magnitude, and the motivations why people engage in eWOM, helping others, venting feelings, empowerment, revenge, and self-enhancement moderated by the construct social value orientation?

(20)

___________________________________________________________________________

5. Theoretical framework.

This section contains the theoretical framework and the hypotheses. Firstly, the independent variable, failures caused by process or outcome and magnitude of those failures, and their effect on the motivations as dependent variable is explored. The next step contains the SVO that may moderate the previously described relationship. The conceptual model in Figure 1. portrays the relationships between these three variables in this study.

(21)

In this study the relationship between process failure and outcome failure, which arise from customer experience which are not as expected and leave a dissatisfied feelings, and the motivations (helping others; venting; empowerment; revenge; and self-enhancement) why people engage in eWOM will be studied. The studied literature argue that process failures causes more dissatisfied emotions than an outcome failure (Smith et al., 1999). When consumer experience a process failure they experience more dissatisfied emotions like, anger, frustration, and anxiety than facing another kind of failure, like outcome, associate with an experience. But, Therefore, this study expect that people who experience a process failure will have more aggressive and egoistic motives to engage in WOM like, venting feelings, revenge, self-enhancement and empowerment. Loo, Boo, & Khoo-Lattimore (2012) declares that customers are more willing to give companies with failures like, slow service, wrong order, and unavailable service, a second change to rectify the problems. Those failures are linked to the process failures in this study. Concluding Loo, Boo, & Khoo-Lattimore (2012), people are probably less likely to harm the company when they face a failure once, but when a failure occurs more often firms are throwing away their chances. Differences between the two different types of failure, outcome and process, will become smaller when a failure occurs multiple times. Explanation for this tendency will be given by the next set of hypotheses.

For that reason, the first three hypotheses are:

Hypothesis 1: ‘Process Failure Single’ will score a (slightly) higher result for the motivations venting feelings, revenge, self-enhancement, and empowerment, than a ‘Outcome Failure Single’.

(22)

Hypothesis 2: ‘Process Failure Multiple’ will have nearly the same result as ‘Outcome Failure Multiple’ for the motivations venting feelings, revenge, self-enhancement, and empowerment.

Hypothesis 3: ‘Process Failure Multiple’ will result in higher motivations for, venting feelings and revenge, than a ‘Process Failure Single’.

The literature gives some explanation why people behave the way they behave through analyzing their behavior with the attribution theory. Attribution theory deals with how and why people form an opinion about an event or observation (Winkler, 2010). The theory is based at the start of human understanding, sense making, and behavior. Attribution theory claims that the opinions of people or consumers are formed through how they perceive the behavior and the reality surrounding it (Alony, 2014). Attributions play an important role in determining what kind of response people give after a product failure. The response depends on how people relate the attributes of a failure to different parties. Attribution theory therefore offers a great prediction of people’s complaining behavior (Folkes, 1984). Folkes (1984), determine that not only the failing of a product determines a consumers response, but consumers try to investigate the why and how question’s. Consumers find it important to know where the problem lies for the product failure, the company, themselves, or another external attribute which plays a role. The attributional approach claims that the cause inferred for product failure influences how the consumer will respond on the failure (Folkes, 1984). When people can blame a certain attribute this has an influence on the response, in contrast when there exist haziness it’s uncertain who of what to blame. In order to explain people’s motivations after they experience an outcome failure, this study looks at the attribution theory.

(23)

Smith et al. (1999) explain the nature of an outcome failure as a ‘behind-the-scene’ event, the consumers cannot address the real reason of the failure. The example used in the scenarios uses the food quality which was below standards for the consumers, as a outcome failure. It’s difficult for consumers to address were the problem lies, is it the kitchen, the wholesaler, or themselves and their sense of taste. Due to the vagueness of the cause, people don’t know who to blame and to who they need to address their anger or frustration to (Folkes, 1984). The motivations venting feelings and revenge will be lower compared to the scenario process failure, helping others will have a higher result for the outcome failure, because of the given reason supported by the attribution theory. Expected is that the difference between the outcome failure and process failure when the failure occurs multiple times will be smaller, because people will set a limit to their dissatisfied experience, when enough is enough. Those two groups will come together for all the five scenarios, helping others, venting feeling, empowerment, revenge, and self-enhancement.

For the above given declarations the following hypotheses are conducted:

Hypothesis 4: ‘Outcome Failure Single’ will score a (slightly) higher result for the motivations for helping others than a ‘Process Failure Single’.

Hypothesis 5: ‘Outcome Failure Multiple’ will have (nearly) the same result as ‘Process Failure Multiple’ for the motivation helping others.

Hypothesis 6: ‘Outcome Failure Single’ will result in higher result for the motivation helping others, than a ‘Outcome Failure Multiple’.

(24)

In this study we expect that people who score high on the social value orientation (SVO), pro-others, will engage in eWOM because they are motivated through the conduct of helping others. Therefore, people who score low on social value, pro-selfs, will not been motivated through altruistic feelings to engage in eWOM. Likewise, people who score low on social value orientation will engage in eWOM because of the motivations: venting feelings, revenge, self-enhancement, and empowerment. Therefore, people who score high on social value orientation will show low motivations of venting feelings, revenge, self-enhancement, and empowerment to engage in eWOM.

The theory of SVO proposes the last hypotheses.

Hypothesis 7: People who score high on social value are more motivated to help others than people who score low on social value.

Hypothesis 8: People who score low on social value are more motivated to vent their feelings, take revenge, empowerment, and self-enhancement than people who score high on social value.

(25)

___________________________________________________________________________

6. Method

In this study the focus encompasses three variables. The independent variables in this study consists of the dissatisfied experiences and the magnitude of those experiences. Five scenarios will be used to cover the five different experiences; single process failure, multiple process failure, single outcome failure, multiple outcome failure, and a control group. Each participant will be ascribed to one of those five scenarios. The relationship and the impact of these dissatisfied experiences on motivations on engaging in creating eWOM will be studied. Last, to measure the impact of social value orientation (SVO) on the relationship of dissatisfied experiences and the motivation to engage in eWOM, the social value orientation (SVO) will be used as the moderator in the previously described relationship.

6.1. Research method.

The focus of this study is to analyze why people engage in eWOM regarding different scenarios, single process failure, multiple process failure, single outcome failure, and multiple outcome failure. To measure these motivations it is important that there is contact directly after an experience, in this way the consumers still have a clear thought about the experience they had. It would be difficult to find those groups online who exactly qualify themselves to one of the groups studied in this research. People express their feelings about their experience on Social Media-sites such as, Facebook, Blogs, Twitter or on discussion fora, and the best way to get the right reactions and feelings about the experience they wrote about is directly

(26)

confront the consumer about it. Through this method you would collect some very specific individual data, nonetheless it would be difficult to find a large sample and thereby find significant results due to the fact that people need to be confronted right after they wrote a review. To eliminate these consequences, and the timespan given for this study, we make use of scenarios in this study. Using this method people will be asked to imagine they are in the described situation and visualize what their reactions will be as if they were actually in the described situation. To collect data the survey will be held on the Internet, this will give an opportunity to reach more people in a limited time span. This could also be a negative point, due to the Internet the sample would be biased due to the fact that not everyone has access to an Internet medium. Another disadvantage when conducting a survey online is that there is a self-selection bias (Wright, 2005). The tendency that some individuals respond to an invitation to participate in an online survey, while others just ignore it (Wright, 2005). This tendency can cause for a systematic bias in this study, which could have influence on possible outcomes. However to gather an appropriate sample in this time span it’s the most effective way to do it via an online medium. To avoid a language barrier the survey is translated into Dutch, which is the native language of most respondents who can be reached. This translation has been back-translated by a fellow student, to ensure the translation is done properly.

6.2. Survey.

This paragraph will describe the three items used and measured in this survey; scenarios, motivations, and the social value orientation (SVO), and will give a description on how the participants within this study were found.

(27)

6.2.1. Scenarios.

Participants will be shown one of the five scenarios and all scenarios describe a different situations and failure. They participants will be asked to imagine if they were in the described situation. The different scenarios all describe one of the failures stated previously in the hypothesis, namely; single process failure, multiple process failure, single outcome failure, and multiple outcome failure. The fifth scenario used in this study is the control group. The participants who get assigned to this group will read a neutral message instead of a negative message due to a failure. In this way there could be tested how people react on a neutral message to compare it with the failures written previously. All the five scenarios have the same design and are taking place at the same restaurant. A scenario is described by Huss (1988) as ‘narrative description of a consistent set of factors which define in a probabilistic sense alternative sets of future business conditions’ (p. 377). The situation described in the scenarios provide certain future behaviours, which we could analyse and compare for the different scenarios. The scenarios are part of a vignette study. Vignette studies have an experimental design where respondents are proposed to similar, but not identical scenarios. Vignettes are short descriptions of situations or persons shown to respondents to elicit their judgements about the scenarios (Alexander & Becker, 1987, p.94). The purpose of a vignette study is to examine the intentions and/or actions of individuals who read the scenario (Wallander, 2009). A vignette is often used because they “provide realism”, “improve construct validity”, reliability and time constraints when analysing real experiences (Wason, Polonsky, & Hyman, 2002). This study wants to investigate the differences in motivations when people face different scenarios/failures. All the scenarios have the same design the only thing which differs is an extra failure when the participant is assigned to a multiple failure group instead of an single failure, or the type of failure when people are assigned to or process

(28)

failure or outcome failure. To help visualize the described situation, respondents were asked to explain what they would write when writing an online review about the just described experience. For scenario analysis it is important to describe a situation in which participants can use their own imagination. Kahneman and Tversky (1982; in Huss, 1988) defined a good scenario ‘when the path leading from the initial to the terminal state is not immediately apparent, so that the intermediate stages actually raise the subjective probability of the target event’ (p. 207). The question is asked right after the respondents read the scenario, and is identical for all five scenarios. The participants are asked to write the review without elaborating the exact content of the review, and thoroughly consider why they would write a review in the specific situation.

6.2.2. Motivations.

The five motivations which are used in this study are conducted from several studies who all examined traditional and or electronic Word-Of-Mouth. The motivations which will be used to investigate why people engage in eWOM in this study are: helping others, venting feelings, empowerment, revenge, and self-enhancement. To ensure the items used to measure the motivations are valid and reliable, items will be used from other studies who also measured motivations. The items for the motivations helping others, and venting feelings will be used from the studies of Hennig-Thurau et al. (2004) and Yap et al. (2012) which both use the same items for these motivations. The two items for the motivation empowerment were used from the articles Bronner & de Hoog (2011) and Willemsen et al. (2013). Wetzer et al. (2007) used the motivation revenge in their study, and those items are replicated in this study. The items for the motivation of self-enhancement were solely used from the study of Yap et al. (2012). The items are measured using a five-point Likert scale, which range from 1 (clearly

(29)

does not describe my feelings) to 5 (clearly describes my feelings). The whole list of motivations can be found in the Appendix.

6.2.3. Social Value Orientation.

The third part of the survey entail the measuring of the social value orientation (SVO). In order to measure the SVO, the scale by Murphy, Ackermann, & Handgraaf (2011) was used. Respondents get a comprehensive description about the task they need to fulfil, with an explanation and an example of the task. The task consist of six items who all have the same form and intention. The respondent can chose between nine different allocations of payoffs between him/her-self and an unknown person. Each item is a resource allocation choice over a well-defined continuum of joint payoffs (Murphy, et al., 2011, p. 778). Depending on the respondents allocation choices, the scale can specify whether the person is an altruist or on the other end of the continuum is a competitor.

6.2.4. Participants.

Because of the design of the research question and the overall study it was important to collect respondents who could imagine themselves in the scenarios. The respondents were asked to imagine they had to write a review about the experience, in this way it was important to have respondents who could imagine themselves doing that. To ensure that, the experiments were done online. With an offline experiment there is a probability that people without any Internet knowledge participate, and this could make it hard to imagine themselves

(30)

The Internet users were asked to participate in this study through Social Media (Facebook), direct e-mail contact to obtain acquaintances, and via a special group under the guise “You scratch my back, i'll scratch yours”- which contains other students who need participants.

(31)

___________________________________________________________________________

7. Results.

This section describes the results from the statistical analyses, who were conducted with the data related to this study. The analyses for this study are conducted with the statistical program SPSS. The first paragraph in this section will report the descriptive data. Following section will report the factor analysis, reliability analysis, and correlations matrices between the motivations. Subsequently, the effect of the scenarios will be tested and analysed. Last, the moderating effect of SVO will be measured.

7.1. Descriptive data.

A total of 279 participants completed the survey. The largest group of participants belong to the group with the age range between 21 and 30 years old (i.e. 65.2%). The distribution among the participants is somewhat biased, with male (42.7 %) and female (57.3 %) participants. Most of the people in this survey have their Bachelor’s degree (51.6 %), followed by Master’s degree (33.3 %). Almost all participants had the Dutch nationality, with 92.4 percent this is the largest represented group.

(32)

Demographics __________________________________________________________________________________ Gender Male 119 42.7 % Female 160 57.3 % Age < 21 years 18 6.5 % 22-30 years 182 65.2 % 31-45 years 45 16.1 % 46 > years 34 12.2 % Nationality Dutch 265 92.4 % Belgian 7 3.3 % German 4 2.3 % Spanish 2 1.3 % Luxembourg 1 0.7 % ___________________________________________________________________________

Education level Elementary/ Middle School 1 0.4 % High School/ Secondary School 9 3.2 %

Associate’s Degree 32 11.5 %

Bachelor’s Degree 144 51.6 %

Master’s Degree 93 33.3 %

Other 0 0.0 %

(Table 1. Descriptive data)

By using the program Qualtrics, one of the five scenarios was assigned to one individual participant. By using the program, all scenarios have almost equal sample sizes. Scenario ‘Process Single’ had 57 respondents which matches a percentage of 20.4, ‘Process Multiple’ had 59 respondents, with 21.1 %, ‘Outcome Single’ had 52 respondents which corresponds with 18.6 %, ‘Outcome Multiple’ had 56 respondents with 20.1 %, and the last ‘Control’

(33)

scenario had 55 respondents which corresponds with 19.7 % of the total sample size.

Social Value Orientation N %

Pro-social -Altruistic 0 0.0 % -Prosocial 199 71.3 % Pro-self -Individualistic 80 28.7 % -Competitor 0 0.0 %

(Table 2. Social Value Orientation allocation)

As shown in Table 2. above, the sample size consist of only two types out of the four types used by Murphy et al. (2011). The calculations are done using the same way Murphy et al. (2011) describes it. The data sample didn’t consist of altruists or competitors, in this manner the two defined groups, only consist of prosocial tested respondents for the ‘Pro-Social’ group and the Self’ group out of individualistic tested respondents. In this manner, the ‘Pro-Social’ group counts for 199 respondents and the ‘Pro-Self’ group for 80 respondents. The calculations are done by using the SVO calculations by Murphy et al. (2011), and can be found in the Appendix.

7.2. Factor, reliability analysis, and correlations matrix.

To ensure that the questions asked in the survey relate to the constructs that this study intent to measure, a factor analysis is provided in the table below. A first check to ensure the construct which are used in this tests are not causing any problems while doing the analysis. All constructs in our data are correlated legitimately well (non above 0.9), so no singularity of

(34)

data, and all significance values, with p < 0.05. A second test provided below, with the Kaiser-Meyer-Olkin measure and Bartlett’s test, checks if a factor analysis is appropriate with these data.

Kaiser-Meyer-Olkin Measure .905

Bartlett’s Test of Sphericity Approx. Chi-Square 3709.150

Df 120

Sig. .000

(Table 3. KMO and Bartlett’s test)

A value of the KMO Measure close to 1 indicates that patterns of correlations are relatively compact, and factor analysis can be done properly (Field, 2005). The KMO Measure of .905 for adequacy is appropriate, the measure is satisfied at a level of 0.5 (Norusis, 2008). The Bartlett’s measure tests the null hypothesis, a significant test shows that the correlation matrix is not an identity matrix, which means that not all correlations are zero (Field, 2005). In this case the Bartlett’s test tested p<0.0001, therefore the data is appropriate to use for a factor analysis.

(35)

X Y Helping others (1) 0.799 Helping others (2) 0.811 Helping others (3) 0.799 Venting feelings (1) 0.774 Venting feelings (2) 0.738 Venting feelings (3) 0.900 Venting feelings (4) 0.684 Empowerment (1) 0.783 Empowerment (2) 0.592 Revenge (1) 0.849 Revenge (2) 0.905 Revenge (3) 0.885 Self-enhancement (1) 0.685 Self-enhancement (2) 0.653 Self-enhancement (3) 0.681 Self-enhancement (4) 0.690

(Table 4. Factor analysis. Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 3 iterations.)

The factor analysis gives insight in the adequacy of certain motivation groups. This study makes use of five motivation groups or components, nonetheless the Factor Analysis shows the sixteen items can be combined into two clear components. The loadings of each motivation exceed the 0.5 minimum point, concluding that those formed components were good measures to determine eWOM motivations. However, in this study we use five different motivations to measure why people engage in eWOM. This factor analysis is considered a limitation of this study, and will be further discussed in the ‘Limitation’-section. The next sections of this study will continue with the five motivations stated earlier, explanations of why this study doesn’t continue with the two resulted motivations are stated in the ‘Limitation’-section.

To test whether the items for the given motivations are internal consistent, and can be used to measure the same motivations the Cronbach’s Alpha is analyzed. All sixteen items and in

(36)

total five motivation were measured, all tested as reliable motivations with Cronbach’s Alpha above 0.67 which Cronbach (1951) stated as the limit for being measured as reliable. An overview of the given overall motivations is given in Table 5.

__________________________________________________________________________

Motivations Cronbach’s Alpha Number of items

Helping others .876 (3) Venting Feelings .888 (4) Empowerment .816 (2) Revenge .924 (3) Self-enhancement .793 (4) ___________________________________________________________________________ (Table 5. Cronbach’s Alpha)

To ensure that those reliabilities also count for the individual scenarios used in this study, each motivation in each scenario will be tested as well. The Cronbach’s Alpha of the motivations are summarized in Table 6. below.

PO PM OO OM C ___________________________________________________________________________ Helping others α 0.844 0.770 0.663 0.687 0.927 Venting feelings α 0.867 0.934 0.683 0.838 0.896 Empowerment α 0.640 0.720 0.902 0.725 0.840 Revenge α 0.914 0.943 0.904 0.869 0.915 Self-enhancement α 0.876 0.831 0.669 0.700 0.825

(37)

Almost all motivations in Table 6. exceed the limit of Cronbach (1951). Some motivations shows a slightly reliability, but due to the small amount of constructs who measure the overall motivation, these alpha’s are appropriate. George and Mallery’s (2003) scale the reliability and a .64 Cronbach’s Alpha can be named as questionable, and can be explained through a limited number of items used to measure the reliability. Due to the small amount of items used to test ‘Empowerment’ in this study (i.e. 2 items), this could be a possible explanation.

To investigate the relationships among the five motivations a correlation matrix is used, Table 7. below provide that correlation matrix. All correlations among the dependent variables were significant at a p < 0.01 level. This correlation test has been done prior to the multivariate analysis of covariance MANCOVA test to ensure all dependent variables would correlate, which is appropriate for a MANCOVA test. All correlations below show a positive correlation with each other, in the weak, moderate, and strong (i.e. 0.25-0.5; 0.5-0.75; 0.75-1) range. These correlations suggests that a MANCOVA test is appropriate with the given data.

___________________________________________________________________________ Helping Venting Empowerment Revenge Self-enhancement others feelings Helping others 1 Venting feelings .552** 1 Empowerment .551** .477** 1 Revenge .441** .875** .367** 1 Self-enhancement .600** .812** .543** .756** 1 (Table 7. Correlation matrix, motivations; **. Correlation is significant at the 0.01 level (2-tailed).)

(38)

7.3. Effect of scenarios.

To claim whether or not there would be one or more differences of the means between the type of failures (single process failure, multiple process failure, single outcome failure, multiple outcome failure) and the motivations for engaging in eWOM a MANCOVA test was conducted. To test whether the control variables (age, gender, nationality, and education level) had an influence, a MANCOVA was conducted. The MANCOVA gave a significant influence for control variable gender (Wilks’ λ = .894, F(5, 265) = 6.263, p < .001) and age (Wilks’ λ = .891, F( 5, 265) = 6.465, p < .001). These significant control variables will be added to further analysis as covariates to justify the effect.

The MANCOVA test gave a significant result about the relationship between the scenarios and the motivations why people engage in eWOM with Wilks’ λ = .548, F(20, 886) = 8.806, p < .001, ηp² = .132. The Partial Eta Squared (ηp²) was estimated at .139, this outcome suggests that 13.9% of the variance in the dependent variable motivations was accounted for by the

given scenarios.

In order to measure whether the variances are equal, a Levene’s Test for Equality of Variances is computed. A significant value was measured for all five motivations, which shows that the variances aren’t equal.

The main analysis of MANCOVA is partly shown in table 8. It shows the data for the dependent variables and scenarios including the covariates. The table shows a significance value for all dependent variables (p’s < .001), which suggests that the scenarios have an influence on the motivations. Due to the significance values of the motivations the assumption has not been met, but this is not as problematic because the group sizes are (almost equal).

(39)

The scenarios divide the sample into five groups which are almost equal, the limit for equal sample sizes lies between, the smallest group is not more than one and a half times smaller than the largest group size (Leech, Barrett, & Morgan, 2005).

___________________________________________________________________________ Type III Sum

SCENARIO Of Squares df Mean Square F Sig.

___________________________________________________________________________ Helping others 109.678 4 27.420 34.711 .000 Venting feelings 52.544 4 13.136 14.615 .000 Empowerment 49.716 4 12.429 14.621 .000 Revenge 42.115 4 10.529 11.663 .000 Self-enhancement 39.725 4 9.931 15.502 .000

(Table. 8, MANCOVA scenarios)

There was a significant effect of scenarios on the motivation helping others after controlling for the effect of SVO, F (4, 269) = 34.711, p < .001, ηp² = .343. There was a significant effect of scenarios on the motivation venting feelings after controlling for the covariates, F (4, 269) = 14.615, p < .001, ηp² = .177. There was a significant effect of scenarios on the motivation empowerment after controlling for the effect of the covariates, F (4, 269) = 14.621, p < .001, ηp² = .174. There was a significant effect of scenarios on the motivation revenge after controlling for the effect of the covariates, F (4, 269) = 11.663, p < .001, ηp² = .147. There was a significant effect of scenarios on the motivation self-enhancement after controlling for the effect of the covariates, F (4, 269) = 15.502, p < .001, ηp² = .181.

Using the Multivariate analysis of variance (MANOVA) any difference between the scenarios and motivation can be found. If any significant outcomes between scenarios are found, a Post-Hoc Tests is held to check the groups, this is done through the Games-Howell test. The Games-Howell test is chosen because the data doesn’t meet the homogeneity of variances assumption, and the group sizes used doesn’t provide equal sample sizes (Hilton, &

(40)

Armstrong, 2006).

Starting with the motivation helping, in Table 9. all differences were significant and looking at the ηp² of .353 it can be conclude that there were moderated strength differences between the scenarios. Between the scenarios ‘Single Process Failure’ and ‘Single Outcome Failure’ didn’t exist a significant difference, with p-value = .989. Due to this result we cannot support hypothesis 4. ‘Single Outcome Failure’ resulted in a higher result for the motivation helping others compared to ‘Single Process Failure’, however this difference isn’t significant.

When ‘Multiple Process Failure’ and ‘Multiple Outcome Failure’ are compared a significance of p = 0.379 can be found. Those scenarios with a higher significance level than p = 0.05, do not significantly vary. With a non-significant value between the ‘Multiple Process Failure’ and ‘Multiple Outcome Failure’, hypotheses 5 can be supported, stated that those two failure don’t differ significantly.

It can be said that, people who experience the ‘Single Process Failure’ are significantly less motivated to help others than people who experience ‘Multiple Process Failure’(p = 0.006). The same conclusion accounts for the scenarios ‘Single Outcome Failure’ and ‘Multiple Outcome Failure’ (p < .001), people who were assigned to the later scenario were significantly more motivated to help others. Hypothesis 6 stated that the inverse was true, for this case the hypothesis is rejected.

Scenario ‘Multiple Process’ has a significantly higher mean for helping others than ‘Single Outcome’ (p < .001). Respondents score almost a scale higher when assigned to ‘Multiple Outcome’ compared to ‘Single Process’, and this outcome has also significant values (p < .001).

The second motive, venting feelings, shows significance differences between the ‘Single Outcome’ and the scenarios ‘Multiple Outcome’ and ‘Multiple Process’, with the multiple

(41)

failure condition people are significant more motivated to engage in eWOM to vent their feelings about the experience (p < .001, p = 0.002). The same counts for scenarios ‘Single Process’ and ‘Multiple Outcome’, those differ significant, with ‘Outcome Multiple’ having a larger mean for venting feelings (p = 0.004). The difference between ‘Single Process’ and ‘Multiple Process’ just missed significance, with p = 0.072.

The scenario ‘Single Outcome’ differs significance from the scenarios ‘Multiple Outcome’ and ‘Multiple Process’, those two score higher on empowerment (both p’s < .001). Respondents assigned to ‘Single Process’ score lower on empowerment than respondents of ‘Multiple Outcome’ and ‘Multiple Process’ (both p’s < .001).

The motivation revenge gives online significant differences between the scenarios ‘Single Outcome’ and ‘Multiple Outcome’, and ‘Single Outcome’ and ‘Multiple Process’ (p < 0.001; p = 0.003). Respondents assigned to ‘Single Outcome’ scored the lowest on revenge, and thus are the least motivated to take revenge upon the company.

Finally concluding the Post- Hoc, ‘Outcome Single’ differs significantly from ‘Multiple Outcome’ and ‘Multiple Process’ (p = .003; p = .034), with the second relationship as nearly significant for the motivation self-enhancement. The differences between those scenarios is halve a Likert-scale point, as shown in Table 9. ‘Process Single’ resulted in the lowest score on self-enhancement, and is significant dissimilar from ‘Process Multiple’ and ‘Outcome Multiple’ (p = .004; p < .001), hereby people assigned to ‘Process Single’ are the least motivated through self-enhancement, but the difference between ‘Outcome Single’ is non-significant so it could be said that those two scenarios are at the same level.

Combining the last four motivations, the hypotheses can be tested. Hypothesis 1 is rejected, the motivations venting feelings, revenge, self-enhancement, and empowerment don’t score significantly higher when people are assigned to ‘Process Single’ compared to ‘Outcome Single’. The second hypothesis is supported. ‘Process Multiple’ and ‘Outcome Multiple’

(42)

result both in the same significantly values for the motivations, venting feelings, revenge, self-enhancement, and empowerment. Hypothesis 3 just missed significance, ‘Process Multiple’ and ‘Process Single’ don’t differ for the motivations venting feelings, and revenge. ___________________________________________________________________________ Scenario’s: PO PM OO OM C F ηp² __________________________________________________________________________________ Helping others 34.711** .340 Meanª 3.329 3.927 3.207 4.152 2.287 SD 0.96258 0.85173 0.81947 0.47125 1.21664 Venting feelings 14.615** .179 Meanª 2.358 2.925 2.106 2.978 1.840 SD 0.92432 1.30988 0.70216 0.86038 0.91510 Empowerment 14.621** .179 Meanª 2.696 3.362 2.612 3.494 2.399 SD 0.80041 0.88108 0.94254 0.83101 1.17873 Revenge 11.663** .148 Meanª 2.137 2.587 1.683 2.506 1.623 SD 0.91176 1.41829 0.70515 0.96960 0.75200 Self-enhancement 15.502** .187 Meanª 2.314 2.914 2.405 2.963 1.929 SD 0.85698 0.89202 0.73866 0.78809 0.89584 (Table 9. ** p < .00. ª= means are evaluated at the following covariates: age and gender)

The ‘Control Group’, which is included in this test to determine a baseline, has a significantly lower value for the motivation helping others compared to the other four scenarios (p’s < .001). Both venting feelings and taking revenge have a non-significant difference comparing ‘Control Group’ and ‘Single Outcome’ (p = .337; p = .940), those two scenarios groups can be seen as similar for these two motivations. Motivation empowerment shows similarities between the ‘Control group’ and ‘Single Outcome’ and ‘Single Process’ (p = .907; p = .665). ‘Single Outcome’ shows insignificant differences for venting feelings, empowerment, and

(43)

revenge. Self-enhancement only shows an insignificant difference with ‘Process Single’ (p = .915), and those two can be seen as similar.

7.4. Moderating effect Social Value Orientation.

To measure to what extent the SVO has an impact upon the relationship among the scenarios and the five motivations, a MANCOVA has been conducted. AMANCOVA can control a covariate, which is a factor who probably has an influence on the relation between the independent and dependent variable. A first step before conducting a MANCOVA is to test the Levene’s Test of Homogeneity of Variance. The following table is the Levene’s Test of Homogeneity of Variance. As seen in the Table 10., the assumption of p > .05 is not met. All five significance levels are below the < .05 level.

Dependent Variable Sig.

___________________________________________________________________________ Helping others .000 Venting feelings .015 Empowerment .001 Revenge .001 Self-enhancement .019

(Table. 10 Levene’s Test of Homogeneity of Variance.)

As explained earlier, the significant values in this study are seen as non-problematic. The analysis for the motivations will thus be done, despite of the Levene’s Test of Homogeneity of Variance.

To measure the effect of the SVO on the relationship between the above described relationship, we are now looking at the SVO as covariate of the model. For the total sample the table below is provided to illustrate the effect of the covariate SVO on the dependent

Online word-of-mouth, what are the triggers for certain motivations to complain online? : a study about the relationship between dissatisfied consumer experiences, and motivations why people engage in eWOM : with an ana