THE INFLUENCE OF COMMITMENT AND
WORD-OF-MOUTH-MOTIVES ON NEGATIVE
WORD-OF-MOUTH
by
Petra Lambers
University of Groningen
Faculty of Management and Organization
Msc Business Administration, Marketing Management
June, 2010
Petra Lambers 1st Supervisor: S. Gensler
Lingestraat 9A 2nd Supervisor: J. Liu
9725 GL Groningen 06-13057906
MANAGEMENT SUMMARY
Word of mouth (WOM) is a much researched area within the marketing literature. WOM is a very effective and cheap marketing tool. It can come in different forms: oral communication or through electronic WOM (eWOM), acquired actively or passively. This research focuses on passively acquired traditional WOM; usually the information comes from people the receiver knows and is therefore seen as more credible. This research will focus on only the negative WOM. As opposed to most WOM-research, this research will focus on the sender of the negative WOM instead of the receiver. Also commitment is an important part of this study. It is researched whether or not
commitment has an influence on the likeliness of engaging in negative WOM, and for what reason consumers engage in negative WOM. This lead to the following research question:
When and with what motives do consumers engage in negative word of mouth, and what are the differences between high committed and low committed consumers?
Negative WOM is triggered by dissatisfaction; the extent to which a product or service is worse than expected. Dissatisfaction in its turn is supposed to be a function of both regret and disappointment. It is expected that brand commitment (consumers focus on a relationship with a brand) has an influence on negative WOM. Two contradicting findings are seen in literature about commitment and negative WOM. One the one hand it is argued that a strong brand relationship controls negative responses, but on the other hand high brand commitment should amplify negative responses.
People will engage in negative WOM for different reasons (Sundaram, Mitra & Webster, 1998); (1) altruistic reasons; preventing others from experiencing the same problems they did, (2) anxiety reduction; easing anger, frustration, and anxiety, (3) vengeance; take revenge against the company associated with the negative experience, and (4) advice seeking; getting advice about how to solve problems.
disappointment, and dissatisfaction. The likeliness of the respondents engaging in negative WOM was studied. After that some statements about the motives were given, and at last some demographics were asked. Through t-tests, correlation analyses, and regression analyses the hypotheses were tested.
Hypotheses: Outcome: H1. The more negative the scenario, the more likely consumers will engage in
negative WOM.
Rejected H2. The more dissatisfied consumers are, the higher the change they will
engage in negative WOM.
Accepted H3. The degree of brand commitment has influence on whether or not
consumers engage in negative WOM.
Rejected H4. Altruism is a more important motive for highly committed consumers to
engage in negative WOM than it is for low-committed consumers.
Rejected (although signifant
difference for the opposite was found) H5. Anxiety reduction is a more important motive for low-committed
consumers to engage in negative WOM than it is for highly committed consumers.
Rejected
H6. Vengeance is a more important motive for low- committed consumers to engage in negative WOM than it is for highly committed consumers.
Accepted H7. Advice seeking is a more important motive for highly committed
consumers to engage in negative WOM than it is for low-committed consumers.
Rejected
Differences in expectations and outcomes may be caused by personal
characteristics. These were not included in the research, but may be very important. Not everybody may be likely to engage in negative WOM, no matter how negative their experience. They might respond in different ways, like complaining to store personnel or looking for a new brand or company to take their business.
INDEX
MANAGEMENT SUMMARY ... 3
INDEX... 5
1. INTRODUCTION... 7
1.1 Background... 7
1.2 Aim of the research & research question... 10
2. THEORETICAL FRAMEWORK ... 12
2.1 Negative word of mouth... 12
2.2 Regret, disappointment, and dissatisfaction... 14
2.3 Commitment... 15 2.4 WOM-motives... 17 2.4.1 Altruism... 19 2.4.2 Anxiety reduction... 21 2.4.3 Vengeance... 22 2.4.4 Advice seeking... 23 2.5 Conceptual model... 24 3. RESEARCH DESIGN ... 27 3.1 Research method... 27 3.2 Data collection... 28 3.3 Plan of analysis... 30 3.3.1 Data preparation... 30 3.3.2 Data analysis... 31 4. RESULTS ... 32 4.1 Respresentativeness... 32 4.2 Data description... 33 4.3 Reliability... 35 4.4 Hypothesis testing... 36 4.4.1 Analyses... 36
4.4.2 Engaging in negative WOM... 37
4.4.3 Dissatisfaction, negative WOM & commitment... 39
5. CONCLUSIONS & RECOMMENDATIONS ... 44
5.1 Conclusions... 44
5.2 Recommendations... 45
5.2.1 Research implications... 45
5.2.2 Managerial implications... 46
5.3 Limitations & further research... 47
LITERATURE ... 48
APPENDIX I ... 52
APPENDIX II... 53
1.
INTRODUCTION
People receive product- or brand information from many different places. This can be from searching their physical environment, from the mass media, or from other people (East, Hammond, Lomax & Robinson, 2005). Information that people receive from other people can come in two forms; information received from commercially interested parties, such as sales representatives, and information received from other consumers, who have experience with the product or brand. This last form of information transmission from one person to another is called word of mouth (WOM), which will be researched in depth in this thesis, especially focussing on negative word of mouth.
1.1 Background
Word of mouth is a much researched area within the marketing literature. Word of mouth can come in different forms. The most used form is oral communication, but with the rise of the Internet, people are starting blogs and writing reviews, which are also rich sources of word-of-mouth communication (Hoyer & MacInnis, 2007). This last form of word of mouth is called electronic word of mouth (eWOM). The most important
difference between traditional WOM and eWOM is that traditional WOM usually comes from a person the receiver of the WOM knows, while eWOM usually comes from people the receiver of the WOM does not know. Also people might have different motives (reasons) for engaging in traditional WOM or eWOM. eWOM can have a much larger reach than traditional WOM, so it it used for different purposes and motives. Studying the motives for both forms would make this research too extensive. Although eWOM
becomes more and more important because it is easily accessible, this research does focus on traditional WOM. Usually information that comes from people the receiver knows (as is the case with traditional WOM), is seen as more credible than information that comes from people the receiver does not know.
reviews and blogs). Passively acquired WOM communication people do not look for, but come in contact without asking. They talk to people that are so enthusiastic/mad
about/with a brand/product, that they feel the need to talk about their feelings for a certain product or brand. This can be through traditional WOM as well as through eWOM.
Word of mouth is not only a very cheap, but also a very effective marketing tool for companies. Already in 1955, Katz & Lazarsfeld discovered that word of mouth is seven times more effective than newspaper and magazine advertising, four times more effective than personal selling, and twice as effective as radio advertising in influencing consumers to switch brands (Harrison-Walker, 2001). This effectiveness can be explained by the fact that personal recommendations are usually seen as more reliable and are of more influence to consumers than advertising and other sources of marketing (Sjödin, 2008). WOM has an important impact on consumers’ judgments and opinions of products and brands, and on consumer choice (East et al., 2005; East, Hammond & Lomax, 2008; Herr, Kardes & Kim, 1991). This is because recommendations are more accessible and diagnostic (Laczniak, DeCarlo & Ramaswami, 2001). Recommendations are usually the most important reason for brand choice (East et al., 2008). According to Sjödin (2008) this is because consumers rely for their opinions on others.
But not all WOM is effective and good for a company or brand. Even though most WOM is positive WOM, also a significant part of all WOM is negative. And this
negative WOM can really damage a brand or company (Blodgett, Granbois & Walters, 1993). Just like positive WOM, negative WOM is seen as a credible source of
information. In the current literature about WOM, often both sides of (positive as well as negative WOM) are being discussed. But since they discuss both sides, this is usually not investigated very deep. When literature deals with only one side of WOM, it is usually positive WOM. Because of this, negative word of mouth is a very interesting topic.
Literature about WOM is always focussed on the receivers of the WOM-communication. Only in a few cases the researcher(s) also looked at the senders of the WOM, but this is really an area that can be researched in more depth. This will be done in this research.
and detractors (Romaniuk, 2008). The research by Romaniuk (2008) shows that
promoters are the people who are usually committed to a brand and provide 80 to 90 per cent of all positive WOM that is spread. Passives do not talk or think about the brand, they do not really have an opinion about the brand. Detractors are usually accountable for all the negative WOM that is spread about the brand and are likely to defect from the brand. Opposed to other research in this field, this study will try to find out whether also the promoters of a brand or product might spread negative WOM. When something negative happens, this might influence the opinions of promoters, which might make them talk negatively about their experiences or the product.
Most research about negative word of mouth discusses the consequences of negative WOM (for the consumers or sales of the product) or looks at the antecedents of WOM (of which commitment is the most important one (de Matos & Rossi, 2008)). However, a much less studied area of negative word of mouth is the side of the sender of negative WOM. When and for what reason someone would engage in negative WOM is a scarce researched area.
Some important research about the sender of the message is done by Sjödin (2008). He researched the likeliness of people spreading negative information about a brand extension. In his research Sjödin focused on customers with favourable, but not exceptional, views of a brand. To add a new dimension to the research about the senders of negative WOM, this study will look at the differences between high and low
committed consumers.
1.2 Aim of the research & research question
The aim of this research is to investigate the differences in when and for what reasons high- and low committed consumers engage in negative word of mouth. It will be interesting to see whether highly committed consumers talk negative about a brand after a negative experience. This research will also give insight in the different motives high- and low committed consumers have to engage in negative WOM. It is expected that highly committed consumers have different motives (or reasons) to engage in negative WOM than do low committed consumers. Highly committed consumers have different feelings towards the product or brand than low committed consumers. Therefore they may act in a different way to a negative experience, have different motives to engage in negative WOM, e.g. getting advice on how to handle the situation, while a low
committed consumer might want to take revenge on the company. It is also expected that low committed consumers more easily engage in negative WOM than highly committed consumers do. They have less bonding with the brand, and therefore it might be easier for them to say something negative about the brand and are expected to engage in negative WOM sooner than highly committed consumers. An important research in the same direction was done by Ahluwalia et al. (2000). They also investigated the influence of commitment on negative experiences, but they looked at the influence on attitude of the respondents towards the target brand. Investigating the influence of commitment on the motives of negative WOM is an area that is not dealt with in current literature. The results of this study can give even more insight in how valuable committed consumers really are to a company or brand.
To investigate this, one brand will be chosen of which there are really obvious high- and low committed consumers. The respondents will be asked how they respond to a situation in which something happens to the chosen brand. By means of three different cases (neutral situation, moderately negative situation, and negative situation) it must become clear if people engage in negative WOM and when they start talking negatively about the brand. Also insight can be obtained in why people engage in negative WOM.
When and with what motives do consumers engage in negative word of mouth, and what are the differences between high committed and low committed consumers?
2.
THEORETICAL FRAMEWORK
Companies try to make consumers committed to their company or their brands. This because it is assumed that when people are committed to a brand, or at least like a brand, they will talk positively about it. This positive word of mouth is very valuable for a company. But negative word of mouth can really damage a brand. This chapter will first focus on the concept of negative word of mouth. After that it will discuss the concept of commitment. Next product- or brand (dis)satisfaction will be explained, followed by whether the construct of commitment. After that the motives for engaging in negative WOM will be discussed, one by one. In these paragraphs also hypotheses are drawn up. At last the conceptual model will be presented.
2.1 Negative word of mouth
Word of mouth is informal advice passed between consumers (East, Hammond & Lomax, 2008). Even though word of mouth can be very valuable for a company or brand, this is only the case for positive word of mouth. Also positive WOM has a significant influence on consumer decision making (Breazeale, 2009). On the contrary, negative word of mouth can really damage a brand. And it can have even more impact on consumer decision making than positive WOM (Einwiller, Fedorikhin, Johnson & Kaming, 2006).
This can be explained by the fact that although negative word of mouth is supposed to occur less than positive word of mouth, people who are dissatisfied with a brand or company tend to tell more people, than do people who are satisfied (Breazeale, 2009; Sjödin, 2008). Blodgett et al. (1993) find that, on average, dissatisfied consumers tell about their negative experiences to nine others, which may lead for some businesses to a loss in volume of ten to fifteen percent. Considering that it costs five times as much to attract a new customer than it does to retain an old one (although many companies seem to forget this), negative word of mouth has serious consequences for companies and brands. Also negative WOM often leads to advising potential consumers against
Ahluwalia et al. (2000) talk about negative publicity, which can also be seen as a form of negative word of mouth. They state that publicity is seen as a relatively credible source of information, which makes it a more influential form of communication than other marketing-driven communications. The authors also mention that negative
information usually gets more attention from consumers than positive information. East et al. (2008) and Laczniak et al. (2001) also argue that negative information is rarer than positive information, which makes negative information more valuable and useful than positive information.
Not every consumer who has had a negative experience with a brand or product will engage in word-of-mouth. And also they will not all start talking at the same time. Whether or not a consumer will engage in negative WOM is dependent of a few things. Consumers might have reasons to avoid engaging in negative WOM (Anderson, 1998). Individuals may not want to be the senders of bad news to avoid feelings of guilt or being associated with bringing bad news. Besides that, Richins (1983) found that the more dissatisfied consumers become, the more they have a tendency to engage in negative WOM.
H1: The more negative the scenario, the more likely consumers will engage in negative WOM.
H2: The more dissatisfied consumers are, the higher the change they will engage in negative WOM.
2.2 Regret, disappointment, and dissatisfaction
Word of mouth is triggered by (dis)satisfaction (Panther & Farquhar, 2004). According to Day, Grabicke, Schaetzle & Staubach (1981) there are three key elements to the satisfaction/dissatisfaction process. The first is some a priori basis of evaluation (e.g. the expectations a consumer has about the product performance). The second is an aspect or circumstance of the purchase or consumption experience that triggers an evaluation. The last element is the judgment that the experience was noticeably better or worse than expected, leading to feelings of (dis)satisfaction. These same elements are also mentioned by De Matos & Rossi (2008), although they do not include the experience that triggers an evaluation. Day et al. (1981) do state here that exact confirmation of consumers’ expectations do not lead to feelings of satisfaction as well as dissatisfaction.
According to De Matos & Rossi (2008) the level of satisfaction is an important factor in the spreading of word of mouth by the consumer. The extent to which the product or service exceeds the expectations of the consumer motivates the consumer to talk about his or her experiences; positive word of mouth. This also works the other way. When the performance of a product or service is worse than the expectations, this might lead to negative word of mouth.
Zeelenberg & Pieters (2004) mention that recent publications show that customer dissatisfaction is a function of both disappointment (i.e., the amount of negative
disconfirmation) and regret (i.e., the performance of forgone alternatives). Zeelenberg & Pieters (2004) say that regret and disappointment are the result of bad decisions and disconfirmed expectancies. Other reasons why regret and disappointment define
dissatisfaction is that they both originate in counterfactual thinking (the obtained outcome is compared to the outcome that was expected or might have occurred. Also both
outcomes that might be obtained. The similarity of these emotions makes that literature says they are highly related, and related to dissatisfaction (Zeelenberg & Pieters, 2004).
2.3 Commitment
There are many antecedents that can influence word of mouth. De Matos & Rossi (2008) developed a framework which comprises the relationships between word of mouth and the most investigated antecedents. These antecedents are loyalty, quality, satisfaction, commitment, trust, perceived value. Other antecedents that are considered of influence on word of mouth are brand familiarity and brand love. As a result of their research de Matos & Rossi (2008) find that commitment is the most important antecedent of word of mouth. This also answers up to what Ahluwalia et al. (2000) find. Commitment is very important in determining resistance to counter attitudinal information. The constructs that look most like commitment will be explained hereafter.
Although brand commitment and brand loyalty in some literature are seen as the same construct, this research sees it as two different constructs. Both can be seen as some sort of attachment to a brand, but consumers are usually more aware of brand
commitment. Brand loyalty is often seen as repeat buying (Ahluwalia et al. 2000), which can also happen out of habit instead of a real relationship with a brand. Brand love can be defined as the level of ‘passionate emotional attachment’ a consumer has for a particular brand (Carroll & Ahuvia, 2006). Brand love therefore can be seen as a real tight
connection to a brand. But the fact that consumers love a brand does not necessarily mean that they are very involved or committed to a brand (Ahuvia, 2005). People can love things they are currently not involved with (e.g. a book they love, but have not read in years), or they can hate something they are very involved with. Brand trust is not a very usable construct since it is more seen in B-to-B relationship, where trust is usually more important. This research focuses on B-to-C relationships.
it represents the feelings of a customer about preserving a relationship with a company. Commitment therefore could be seen as the desire of consumers to be attached to a company or brand, that they are willing to build a relationship with this brand or company.
In the existing literature commitment is measured as a uni-dimensional construct as well as a multidimensional construct. In the latter construct commitment is composed of ‘affective (positive emotional attachment), continuance (perceived costs associated with leaving the brand), and normative (perceived moral obligation toward the brand) commitment’ (de Matos & Rossi, 2008). Although in most research the normative construct is left behind; Fullerton (2005) as well as Amine (1998) recognize that consumer commitment has two components; (1) an affective component, and (2) a continuance component (or called calculated commitment at the research by Amine).
The affective component comprises shared values, identification, and attachment. This relationship is built on positive affect; consumers enjoy the brand or company and trust the brand or company (for example, the relationship between a hairdresser and a client can be defined as an affective relationship). High-committed customers have a high identification with the brand as well as hold feelings of attachment toward maintaining the relationships they value. This reflects the affective part of commitment. Fournier (1998) found that consumers build relationships with brands they use. It is said that the affective component lies at the heart of these relationships. The continuance component focuses more on commitment a consumer maintains as long as the benefits attached to the brand are higher than the switching costs to another brand.
The normative component is often seen as a less important component of
commitment for several reasons. First, the impact of normative commitment has been in the same direction as the affective component, but weaker. Second, the normative component is often highly correlated with the affective component, and therefore it is questioned if it even is a distinct component. This research also leaves out the normative component and only focuses on the affective and continuance component, since the normative component does not add real value to the construct.
with negative WOM about ‘their’ brand (Ahluwalia et al., 2000). Grégoire, Tripp & Legoux (2009) call this a ‘protection effect’; this means that a strong brand relationship controls negative responses. On the other hand, other researches find that high brand commitment amplifies negative responses. Einwiller et al. (2006) even find that when information becomes extremely negative, commitment does not moderate this effect anymore. This research will investigate whether there is a limit to what highly committed consumers can bear, if they will also talk negatively about a brand if they have a really negative experience with it.
This research assumes that low committed consumers will more easily engage in negative WOM. This is supported by Richins (1984); it is expected that for people who are highly committed to a brand more negative things, or worse negative things need to happen before they start talking negatively than do people who are low committed to that same brand. A reason for highly committed consumers to engage in negative WOM later than low-committed consumers, is that telling others about a purchased product that was unsatisfactory, can be seen as a failure as a consumer (Richins, 1984). When consumers are highly committed this might be seen as worse than when consumers are low
committed to the brand or product.
This leads to the following hypothesis:
H3: The more consumers are committed to a brand, the less they will engage in negative WOM.
2.4 WOM-motives
from the motives for engaging in negative WOM (e.g. wanting to prevent others from what they experienced).
One of the first and a very extensive study about the motives for WOM is done by Dichter (1966). He found four motives people can have to engage in word of mouth; (1) product-involvement, (2) self-involvement, (3) other-involvement, and (4) message-involvement. Although the study by Dichter (1966) is a very prominent study, an
important weakness is that no detailed information is provided about how the typology is obtained (Hennig-Thurau, 2004). Engel, Blackwell & Miniard (1993) modified Dichter’s typology. They renamed the motives (involvement, self-enhancement, concern for others, and message intrigue) and added one additional motive; dissonance reduction. Both studies assume that the motives for engaging in positive WOM and negative WOM are the same.
But the most extensive study to word of mouth-motives is done by Sundaram, Mitra & Webster (1998). This study used 730 critical-incident interviews to define the motives people can have to engage in word of mouth and made a distinction between positive WOM-motives and negative WOM-motives. They found eight different
categories, in which most of the categories mentioned by Dichter (1966), and Engel et al. (1993) come back. The motives for positive WOM are: altruism, product involvement, self-enhancement, and helping the company. The motives for negative WOM are: altruism, anxiety reduction, vengeance, and advice seeking. The motives for negative WOM are described in table 1.
Negative WOM-motives Description
Altruism Preventing others from experiencing the same problems they did. Anxiety reduction Easing anger, frustration, and anxiety.
Vengeance Take revenge against the company associated with the negative experience. Advice seeking Getting advice about how to solve problems.
Table 1: Descriptions of motives by Sundaram, Mitra & Webster (1998)
For this research, the motives by Sundaram et al. (1998) will be used as a
positive WOM-motives and negative WOM-motives. Besides that, a lot of the categories correspond with the originally suggested categories by Dichter (1966) and Engel et al. (1993). These studies therefore will not totally be rejected. When looking at other studies done in this field, it can be seen that they apply almost the same categories (i.e. Cheung, Anitsal & Anitsal, 2007; Wetzer, Zeelenberg & Pieters, 2007). Also, since this study only looks at the negative WOM-motives, it is clearer to use the categories by Sundaram et al. (1998). But to add a new dimension to these categories, in this research the emphasis will be on the differences between motives for high and low committed consumers. This is an area that has not yet been researched. The different motives will be discussed in the following paragraphs. But first it will be discussed whether a consumer will engage in negative WOM at all.
2.4.1 Altruism
Altruism can be defined as the caring about each others’ welfare (Foster & Rosenzweig, 2001; Croson, 2007). People engaging in negative WOM for altruistic reasons, do this to protect others from experiencing the same problems they have encountered. Powers & Hopkins (2006) examine altruism in terms of three dimensions: (1) ethnocentrism; the tendency people have to favour their own group over other groups; (2) cognitive moral development; the development through which people (over time) obtain an increasingly precise understanding of their moral obligations; and (3) pro-social behaviour; the behaviour an individual carries out voluntarily, to benefit an other person. This research is focussed mostly on the last dimension; pro-social behaviour. For this research it is important to see what a consumer does without the influence of other people on their motives. In the case of altruism that is that a consumer who has had a certain experience (negative or positive) wants to let others know about this, so they can benefit from this information; pro-social behaviour. This dimension is also mostly discussed in other literature about altruism.
have. The consumers do this voluntarily, without the expectation of a reward (Powers & Hopkins, 2006). Knowing this, it can be expected that commitment plays a large role for altruism being the motive for negative WOM; high commitment to a brand or product might trigger altruism more than low commitment to a brand or product, since there is no reward for sharing their experiences, it is an altruistic act.
Jex, Adams, Bachrach & Sorenson (2003) studied how commitment influences altruism in the presence of organizational constraints in the workplace (with constraints they mean conditions within the organization that make it more difficult for employees to do there jobs, like a lack of office supplies, or budget costs). Their research has shown that when employees are low committed to the organization, the presence of these organizational constraints has a negative relationship with altruism. This means that the worse the organizational constraints are, the less these low-committed employees will engage in altruistic behaviour. In addition to the former finding, Jex et al. (2003) also find that for employees with high commitment to the organization, the presence of
organizational constraints has a positive relationship with altruism, meaning that the worse the organizational constraints, the more highly committed employees will engage in altruistic behaviour. Even though this research is focussed on employees within an organization, the same reactions could be expected when looking at consumers and their commitment to a brand. The employees in the research by Jex et al. (2003) are exposed to a situation that makes them feel regret and/or disappointment. As said earlier in this chapter, feelings of regret and disappointment are expected to lead to feelings of dissatisfaction. The same is expected for consumers who are exposed to negative information about a brand. It is foreseen that commitment plays the same role for the consumers as it did for the employees in the research by Jex et al. (2003). Therefore it can be expected that the same results can be found.
This leads to the following hypothesis:
2.4.2 Anxiety reduction
An other motive for engaging in negative WOM is anxiety reduction. Consumers can use negative WOM communication to vent their frustration, anxiety, or anger. Sharing their negative emotions helps in easing these negative feelings (Sundaram et al., 1998). In some literature this sharing of negative emotions is called complaining,
although complaining can be divided into two categories. The first form of complaining is talking to the company (or sales representative) about the negative feelings. The other form of complaining is negative WOM, talking to other people about the negative emotions. Only this last form of complaining is relevant for this research. The main reason of complaining to other people is to vent negative emotions (Nyer, 2000).
People associate anxiety, anger and frustration with feelings ‘as if they would explode’ and ‘being overwhelmed by their emotions’ (Bougie et al., 2003). When these negative feelings become worse, consumers become more and more likely to complain and feel the need to engage in negative WOM. Besides that, they are less likely to repurchase the product or brand.
Anxiety reduction as a motive for negative WOM can have a therapeutic effect (Chueng et al., 2007). Highly committed consumers have more need to reduce dissonance and anxiety. Consumers having anxiety reduction as a motive for engaging in negative WOM communications just want to talk about their experiences, get it off their chest or blow off steam, with as only result feeling better. Through research (i.e. Nyer, 2000; Nyer & Gopinath, 2005) it is found that complaining or venting indeed lowers the consumers’ level of dissatisfaction, although some research suggests that this works only on the short-term.
Nyer & Gopinath (2005) find that when consumers are committed to a brand and this is publicly known (people around them are aware of this commitment), these
consumers have a strong desire to appear consistent in the eyes of others. Which means they would keep their dissatisfaction to themselves and will not start complaining, or venting. This would mean that anxiety reduction is a more used motive for consumers who are low committed to a brand than for consumers that are highly committed.
H5: Anxiety reduction is a more important motive for low-committed consumers to engage in negative WOM than it is for highly committed consumers.
2.4.3 Vengeance
Another motive for consumers to engage in negative WOM Sundaram et al. (1998) found is vengeance. The dissatisfied consumers want to take revenge against the company they are dissatisfied with. Consumers want to stop others from buying a brand they perceived did not care enough about them, or did not listen to complaints, and therefore should not be allowed to do business. Consumers who seek vengeance want to get back at the organization (Bougie et al., 2003) and explicitly advice others not to do business with the company or brand they are dissatisfied with and may spread messages that malign the company or brand (Wetzer et al., 2007).
Finkel, Rusbult, Kumashire & Hannon (2002) talk about betrayal by partners. People can feel let down or betrayed. These same feelings can be perceived when negative things happen with a brand someone is highly committed to. This can also be seen as a relationship. But instead of taking revenge on the brand one is highly committed to, the consumer is more likely to forgive the brand, because he or she has more to loose (e.g. a highly valued relationship with a brand) when ending the relationship, than does a consumer who is low committed to that same brand. Finkel et al. (2002) find that
commitment is positively associated with forgiveness. They find that this likeliness to forgive comes from a will to persist, the decision to remain dependent on a partner. The more an individual wants to persist and remain dependent on the relationship, the more he or she is willing to renounce from revenge, so they can hold on to what they have. Also McCullough, Rachal, Sandage, Worthington, Brown & Hight (1998) find that people are more likely to forgive when their relationship is close, committed, and satisfactory. Also their revenge motives reduce, because of great closeness with a brand. Finkel et al. (2002) also find that commitment is associated with derogation of
alternatives, adaptive behaviour, and willingness to sacrifice.
they are to want to take revenge on the product or brand, the other researches mentioned support that highly committed consumers do not tend to take revenge. The reason Grégoire & Fisher (2008) find for highly committed consumers to do want to take
vengeance is that the high level of commitment makes the consumers take offense if they feel they are victims of a company, product, or brand. When they feel they are treated poorly by a product or brand with which they feel a strong connection, can be very disconcerting and hurtful. To get back at the product or brand these people may proceed to taking vengeance.
But Grégoire et al. (2009) find that highly committed consumers have a lower desire for revenge than low-committed consumers do. Many research shows that a desire for revenge is difficult to sustain and tends to decrease over time. What Grégoire et al. (2009) find is that this desire for revenge decreases more slowly over time for highly committed consumers that it does for low-committed consumers. The research by Grégoire & Fischer (2005) talks about the ‘love is blind’-effect; when highly committed consumers are confronted with a service failure, they do not find taking revenge the right solution. They tend to forgive and give the company the benefit of the doubt.
Since in current literature it is more suggested that low-committed consumers have a higher motive for vengeance when engaging in negative WOM, this is also assumed in the drawing up of the hypothesis:
H6: Vengeance is a more important motive for low- committed consumers to engage in negative WOM than it is for highly committed consumers.
2.4.4 Advice seeking
Consumers who have had negative experiences with a product or brand, and who are usually not aware of the possibilities to seek compensation from the brand or
The dissatisfied consumers might attempt to find out why the negative things have happened (through seeking information and trying to solve the problem) (Bougie et al., 2003). Bougie et al. (2003) give three possibilities the dissatisfied consumers might find. First, the company, product, or brand might be responsible for the negative experience. Second, the consumer himself might be to blame. Or finally, there might be some uncontrollable circumstances responsible for the negative experience (e.g. an other person, the government, or the weather might have caused the negative experience). When the company, product, or brand is to blame, consumers are likely to engage in negative WOM. The response the consumers get, are of great influence in determining whether the goal for negative WOM is achieved (Wetzer et al., 2007). On the other hand, when consumers blame themselves, they are less likely to tell others about what they have experienced.
Zand (1972) states that the effectiveness of advice seeking or solving the problems the consumer has encountered with a brand or company, is related to the trust people have in brand. Trust is also connected with commitment. The more they trust the brand that has let them down, the more effective the advice seeking can be. Also the more they are committed to a brand, the more effective the advice seeking can be. The
consumers are more eager to restore the relationship. Therefore the following hypothesis is drawn up:
H7: Advice seeking is a more important motive for highly committed consumers to engage in negative WOM than it is for low-committed consumers.
2.5 Conceptual model
very important in this model is whether the degree of commitment has an influence on engaging in negative WOM.
Figure 1: Conceptual model I
Consumers can engage in negative WOM for several reasons; altruistic reasons, anxiety reduction, vengeance, and advice seeking (see figure 2). The expectation is that these reasons for engaging in negative WOM will be influenced by the commitment of the consumers toward the brand or product. Ahluwalia, Burnkrant & Unnava (2000) mention that stronger consumer attitudes towards a brand or product are known to demonstrate greater resistance to negative information.
3.
RESEARCH DESIGN
After the theoretical framework and the drawing up of the hypotheses and the conceptual model, the next step is the research design. This chapter will discuss the research method, followed by the data collection and a plan of analysis.
3.1 Research method
A research design can be exploratory or conclusive (Malhotra, 2007). Exploratory research is done to provide insights and understanding. The information that is needed is defined only loosely, through a flexible and unstructured research process, usually using a small sample size. Conclusive research is done to test specific hypotheses and examine relationships. The information that is needed is clearly defined, the research process is formal and structured, and the sample size is large.
In this research both designs will be used. Respondents of the actual research will be served a scenario which is negative or moderately negative. Before the questionnaire can be distributed, cases have to be tested to make sure that people really perceive the cases as moderately negative and negative. This will be done through exploratory research. Only a small sample size is used, and the purpose of this research is to gain insights into what people find negative or moderately negative. Respondents are asked to give their opinion about eight different cases, whether they found the cases positive or negative (judged on a 5-point scale), and how this case would influence their own opinions (positive to negative, also judged on a 5-point scale). This led to the choice of two cases (one moderately negative and one negative). The results that led to this choice can be found in appendix I.
research will be causal research; it will be tested if there is a relationship between
different variables, like the degree of commitment and the tendency to engage in negative WOM and for what reasons.
3.2 Data collection
The experiment will be conducted among students from the Rijksuniversiteit Groningen. The design of the research will be a 2x2 design (see figure 3).
Low commitment High commitment Moderately negative scenario
Negative scenario
Figure 3: Experimental design
The level of commitment will be measured in a survey-part of the experiment. To measure commitment, the following three-item brand commitment measure (as used by Ahluwalia et al. (2000) and tested by Beatty, Kahle & Homer (1988)) is used:
1. If my preferred brand or type of … were not available at the store, it would make little difference to me if I had to choose another brand (reversed). 2. I consider myself to be highly loyal to one brand of …
3. When another brand is on sale, I will generally purchase it rather than my usual brand (reversed).
These statements are values on a 7-point Likert-scale, ranging from (1) totally agree to (7) totally disagree). After analyzing the data retrieved from these questions, respondents will be categorized into high or low commitment categories.
A respondent is only exposed to the moderately negative scenario or the negative scenario, which is put into a newspaper clipping. After reading the scenario, first some questions will be asked to measure the degree of regret, disappointment, and
dissatisfaction. These questions (which are adjusted a little bit to fit with the scenarios) are derived from Zeelenberg & Pieters (2004):
2. In retrospect, how bad do you judge your decision to buy …? Disappointment: 1. After this experience how much disappointment do you feel
about buying … after the negative experience?
2. To what extent is … worse than you expected beforehand? Dissatisfaction: 1. Overall, how dissatisfied are you with the … after the negative
experience?
2. Overall, how good or bad do you feel after the negative experience?
These questions are measured on a 7-point Likert-scale, ranging from (1) totally agree to (7) totally disagree. These questions will be used to find out whether there is a
relationship between regret, disappointment and dissatisfaction and whether this has an influence on engaging in negative WOM. Zeelenberg & Pieters (2004) find that regret and disappointment together are the most important variables that explain dissatisfaction. In this research it will be tested whether this is also true for this study.
After answering these questions, the respondents will be asked whether they would talk to somebody about what they have read (indicating the likeliness on a scale from 0 to 100), and what they would say if they would (in an open-ended question). Also they will answer some questions about the scenario to find out what the motives for engaging in negative WOM are (when they do engage in negative WOM). For this, statements from Wetzer et al. (2007) are used. They developed statements to measure different motives for engaging in negative WOM:
Altruism: 1. I wanted to help my conversation partner with making a decision about what brand to buy.
2. I wanted to prevent my conversation partner from buying this brand after what happened.
3. I wanted to warn my conversation partner not to use this brand. Anxiety reduction: 1. I wanted to pour my heart out about what I read.
2. I had to blow off steam. 3. I wanted to vent my feelings.
2. I wanted to give the brand a bad reputation. 3. I wanted the company to lose customers. Advice seeking: 1. I wanted to understand what happened.
2. I wanted to know whether I judged the situation right. 3. I wanted advice on how to handle my feelings.
Also these statements are measured on a 7-point Likert-scale, ranging from (1) totally agree to (7) totally disagree. These statements will be used to find out how strong the relationships are between the motives and negative WOM.
To have a representative research, in each group have to be at least 30 respondents, which makes a total of at least 120 respondents.
3.3 Plan of analysis
After the questionnaire is sent out and filled in by the respondents, the data can be prepared and appropriate analyses can be found.
3.3.1 Data preparation
The first step of the data preparation (which is already done while the
questionnaire is still running) is to check all questionnaires for completeness (Malhotra, 2007). All questionnaires are checked if they are filled in completely and whether the responses show enough variance (eg. a respondent who fills in only 4s on a 7-point rating scale). A total of 26 responses (of a total of 162) was deleted through these reasons. It is also checked whether there are enough highly committed and low-committed respondents to end the questionnaire.
After this check the remaining questionnaires are all coded in SPSS. The
questionnaire contained 1 open-ended question. This question is about what people would say if they would talk about what happened. The answers are rated on a 7-point scale by two independent coders, where 1 is very negative and 7 very positive. This coding is necessary to make analyses with the data.
but not every question was measured in the same direction (e.g. a rating of 7 on the commitment statements, means low commitment for one statement, but high commitment for an other statement). All questions were recoded so 1 indicated a low score for the variable and 7 a high score (also easier to interpret).
3.3.2 Data analysis
Once the data has been prepared for analysis, some basic analyses are done. Means, frequencies and standard deviations are calculated and compared.
Since for every variable (commitment, regret, disappointment, et cetera) a few questions are asked to measure this variable, it is checked whether these values can be added up to make one new variable. This can be done by calculating the Cronbach’s alpha. When the scores are over 0.6, this means they measure the same concept, and the values can be aggregated to make a new variable.
Once this is done, the real analyses to test the hypotheses will be done. Practically everything is measured on a 7-point Likert-scale, which is actually an ordinal measure, but can be used as an interval measure (Malhotra, 2007). The statistical techniques that can be used are univariate and multivariate techniques. In this research both techniques are used. Both techniques can again be classified as dependence techniques or
4. RESULTS
After the data has been collected, the data can be imported into SPSS and
analyses can be done. But first the dataset has to be checked for representativeness. Also some descriptive data will be dealt with, and several variables will be checked for reliability. After that the hypotheses will be tested through t-tests, correlation analyses and regression analyses.
4.1 Representativeness
First the dataset is checked for representativeness. No exact data is available about the sample. For that reason it is only checked whether the distributions of the scenarios, commitment level, and demographics are comparable. When looking at table 2, it can be seen that the dataset has a really good distribution of the scenarios and also of the commitment level (which is not only measured for the whole dataset, but also for the different scenarios to check whether this distribution is also good). All distributions are around 50 per cent, which makes the dataset very usable.
Table 2: Distributions scenarios & commitment level
The demographics also focus on the whole dataset, but a distinction is made between the moderately negative- and negative scenario as well. This is done because there were actually two surveys conducted and it might be important to see whether there are big differences between the scenarios regarding the demographic data. This
description can be found in table 3.
Distributions: Mod. negative scenario Negative scenario
Scenario: 50,7% mod. negative scenario 49.3% negative scenario
Commitment: 50.8% highly committed 49.2% low committed
50.0% highly committed 50.0% low committed
Table 3: Demographics
The age in the whole dataset ranges from 17 to 31. The mean age in the whole dataset is 23.95. In the different scenario’s this deviates with only 0.1 years, which is very close. The gender does differ much. In the whole dataset, the distribution men – women is very good (53.4 per cent and 46.6 per cent respectively), but between the scenarios there are big differences (64.2 per cent males in the moderately negative scenario and 42.4 per cent males in the negative scenario). But even though these distributions are different and not equally distributed, the data will be used anyway. The model does not assume gender influences, so the differences should not matter too much. To reinforce this, t-tests are conducted for all variables and can be found in appendix II. All variables are not significant (> 0.05), which means that there is no difference between men and women. The dataset consists of 100 per cent students, which was a condition in for this survey.
4.2 Data description
After the checking for representativeness, it is checked whether or not the respondents even drink cola. Like in the previous paragraph, also here a distinction has been made between the moderately negative scenario and the negative scenario to check for large differences. When looking at table 4, it can be seen that 86.8 per cent of the respondents ever buy cola. This can be in bottles or cans, at the supermarket, school or work. Hardly any difference can be found when looking at the moderately negative- and negative scenario (87.0 per cent and 86.6 per cent respectively).
Demographics: Mod. negative scenario Negative scenario
Age Mean age: 23.95 Mean age: 24.04 Mean age: 23.85
Gender 53.4% male, 46.6% female 64.2% male, 35.8% female 42.4% male, 57.6% female
Do respondents ever buy cola: Mod. negative scenario Negative scenario Yes 86.8% 87.0% 86.6%
No 13.2% 13.0% 13.4%
Table 4: Do respondents buy cola?
In table 5 can be seen how often respondents buy cola. Almost 65 per cent buys cola at least once every two weeks. Although, there is a difference in this between the two scenarios. Almost 75 per cent per percent of the respondents of the moderately negative scenario buys cola at least once every two weeks, against only 55 per cent of the respondents of the negative scenario. But this does not lead to extreme differences when looking at commitment (which was already shown in table 2), through which can be concluded that the data is valid and usable.
Table 5: How often respondents buy cola
Remarkable is that more than 80 per cent of all the respondents that ever buy cola, usually buy Coca-Cola (see table 6). Also between the two scenarios there is hardly any difference (83.3 per cent versus 79.3 per cent). Other brands are hardly ever bought; the private labels of supermarkets come second with an average of 9.3 per cent.
What brands respondents buy: Mod. negative scenario Negative scenario
Coca-Cola 81.4% 83.3% 79.3%
Pepsi-Cola 4.2% 3.3% 5.2%
Dr. Pepper 0.8% - 1.7%
River Cola 3.4% 3.3% 3.4%
Private label supermarket 9.3% 8.3% 10.3%
Other 0.8% 1.7% -
Table 6: What brands respondents buy
How often respondents buy cola: Mod. negative scenario Negative scenario
Every day 1.7% 3.3% -
3 – 6 times a week 9.3% 10.0% 8.6%
1 – 2 times a week 28.8% 30.0% 27.6%
Once every 2 weeks 23.7% 30.0% 17.2%
4.3 Reliability
To find out to what extent a scale variable produces consistent results if repeated measurements are made, reliability will be tested. Not only will this show if the variables test the same thing, it also tells whether or not the variables can be made into one new variable. But before this can be done, it has to be examined whether all variables are tested in the same direction. If this is not the case, the variables will have to be recoded. For the different variables that together test commitment this was the case. These variables were recoded, so they are all measured iin the same direction.
After this is done, Cronbach’s alpha is used to test internal consistency reliability. This is the average of all possible split-half coefficients resulting from different ways of splitting the scale items (Malhotra, 2007). The Cronbach’s alpha is a value ranging from 0 to 1, and a value of 0.6 or higher generally indicates satisfactory internal consistency reliability. A satisfactory internal consistency reliability means the variables can be made into one new variable.
Internal consistency reliability is tested for different constructs. Almost all constructs in the conceptual model are tested with several questions. The internal consistency reliability scores can be found in table 7.
Reliability: # of variables Cronbach’s alpha
Regret 2 0.822 Disappointment 2 0.715 Dissatisfaction 2 0.638 Commitment 3 0.854 Altruism 3 0.911 Anxiety reduction 3 0.834 Vengeance 3 0.912 Advice seeking 3 0.778
Table 7: Reliabilities of different variables
4.4 Hypothesis testing
In this paragraph the different hypotheses that are drawn up in chapter 2 are tested through t-tests, correlation analyses, and regression analyses. But first the analyses used are described in more detail.
4.4.1 Analyses
To get an insight in the relationship between variables, correlation analysis can being conducted (Huizingh, 2002). Correlation analysis assumes variables are metric. But it does not measure every relationship between variables, only linear relationships. This means that it could be that two variables do have a relationship, but this is not linear. This would give a low correlation coefficient. All hypotheses in this research expect a linear relationship (when the value of one variable goes up, so does the value of another one, or vice versa), so correlation analysis will be used. Correlation analysis is being conducted with use of the Pearson correlation coefficient. This coefficient shows the strictness of the linear relationship. Also it shows the direction of the relationship. The closer the Pearson correlation score comes to 1 (or -1, to indicate negative linear relationship), the more there is a linear correlation. Correlation analysis also gives a significance score. All tests will be done on a confidence level of 95 per cent. When the significance levels are below 0.05, the null-hypothesis is rejected, which means that there is a linear relationship between the variables. All values over 0.05 indicate no linear relationship.
When there is linear correlation and more than two variables are used in the correlation analysis, regression analysis can be conducted. This analysis is used to estimate a linear relationship between a dependent variable and one or more independent variables (Huizingh, 2002). The dependent variable will be explained through the
independent variables. Regression analysis tests to what extent the dependent variable is really explained through the independent variables. Regression analysis gives some important figures. First the R, which is the multiple correlation coefficient. This measures the correlation between all variables. The closer to 1 this coefficient is, the more
coefficients (B) and significance levels. This is also tested on a 95 per cent confidence level. When the significance level is below 0.05, the null-hypothesis is rejected, which means that the regression coefficient is not significantly equal to 0. All values over 0.05 indicate that the regression coefficient is equal to 0. B shows how much the dependent variable will increase or decrease when the independent variable increases with 1.
To test whether the differences in the mean values of a dependent variable are different from each other, an analysis of variance (ANOVA) or t-test can be used. ANOVA is usually used when more than one independent variable are taken into account. When there is only one independent variable, a t-test is a better option (Malhotra, 2007). In this research only the t-test will be used, since there is only one independent variable per analysis. Before the results of the t-test can be processed, it has to be decided whether or not equal variances are assumed. This is done through Levene’s test for equality of variances. This again is tested on a confidence level of 95 per cent. When the significance level is below 0.05, the null-hypothesis is rejected, which means that equal variances are not assumed. Equal variances are assumed when the significance level is over 0.05. The results of this test tell which results of the t-test to use (equal variances or unequal variances). Tested on a confidence level of 95 per cent, a significance level below 0.05 again rejects the null-hypothesis, which means that the means of both groups are not the same. All values over 0.05 indicate that the means of the two groups are the same.
4.4.2 Engaging in negative WOM
Respondents with scores between 3.5 and 4 were left behind to get a better distinction between the two groups.
Table 8 starts with giving the mean scores of the likeliness of engaging in
negative WOM in the two scenarios. This shows which group is more likely to engage in negative WOM. In the whole dataset there is little difference; people who are exposed to the negative scenario are a little more likely to engage in negative WOM. High
committed people are more likely to engage in negative WOM when exposed to the negative scenario, while the low committed respondents are more likely to engage in negative WOM when exposed to the moderately negative scenario. Through t-tests it is tested whether these differences are significant.
T-test Means Levene’s Test for
Equality of Variances
Equal variances assumed
Mod. negative scenario Negative scenario F Sig. t Sig. (2-tailed) Sig. (1-tailed) Whole dataset 41.69 41.81 1.318 0.253 -0.019 0.985 0.493 High commitment 40.34 42.93 2.046 0.158 -0.303 0.763 0.382 Low commitment 43.00 40.61 0.028 0.869 0.281 0.780 0.390
Table 8: Results of t-test for engaging in negative WOM
All three analyses (with significance levels of 0.253, 0.158, and 0.869 for respectively the whole dataset, high committed respondents, and low committed respondents) imply that equal variances are assumed. Only these results are therefore displayed in table 8. Since hypothesis 1 assumes that a more negative scenario would lead to more negative WOM, the t-test can be done 1-sided. The significance levels are 0.493, 0.382, and 0.390 for the whole dataset, high committed respondents, and low committed respondents respectively. All three significance levels are over 0.05, which means that the means of both groups (moderately negative scenario and negative
4.4.3 Dissatisfaction, negative WOM & commitment
To find out whether dissatisfaction and commitment have an effect on the
likeliness of engaging in negative WOM, first correlation analysis is being used. Not only likeliness of engaging in negative WOM, dissatisfaction, and commitment are included in this analysis, also the interaction effect of dissatisfaction and commitment is taken into account to see if this has a significant influence. The results of this analysis can be found in table 9 (the significant scores are in bold).
Correlation engaging in negative WOM, dissatisfaction, commitment &
interaction dissatisfaction-commitment
Likeliness of engaging in negative WOM
Dissatisfaction Commitment Interaction dissatisfaction-commitment Pearson correlation 1 0.522 -0.016 0.338 Likeliness of engaging in negative
WOM Sig. (2-tailed) 0.000 0.864 0.000
Pearson correlation 0.522 1 -0.002 0.664 Dissatisfaction Sig. (2-tailed) 0.000 0.986 0.000 Pearson correlation -0.016 -0.002 1 0.678 Commitment Sig. (2-tailed) 0.864 0.986 0.000 Pearson correlation 0.338 0.664 0.678 1 Interaction
dissatisfaction-commitment Sig. (2-tailed) 0.000 0.000 0.000
Table 9: Correlation between likeliness of engaging in negative WOM, dissatisfaction, commitment & interaction dissatisfaction-commitment
Between the likeliness of engaging in negative WOM and commitment as well as between dissatisfaction and commitment no linear relationship can be found (significance levels of 0.864 and 0.986 and correlation scores of -0.016 and -0.002 respectively)
But when looking at the interaction between dissatisfaction and commitment and the other variables, we find significant scores and relatively high correlation scores (0.338 for likeliness of engaging in negative WOM, 0.664 for dissatisfaction, and 0.678 for commitment). Since there are more than two variables in this correlation analysis, also regression analysis can be conducted (see table 10).
Regression engaging in negative WOM:
Dissatisfaction Commitment Interaction dissatisfaction-commitment R 0.523 R Square 0.274 B 15.247 1.929 -0.554 Sig. 0.002 0.604 0.596
Table 10: Regression engaging in negative WOM with dissatisfaction, commitment & interaction dissatisfaction-commitment
As can be seen in table 10, the R Square is 0.274, which indicates that
dissatisfaction, commitment, and the interaction between dissatisfaction and commitment together explain only 27.4 per cent of t he variance in the likeliness of engaging in negative WOM. When looking at the different variables, it can be seen that only the variable dissatisfaction is significant. The B of 15.247 indicates that when the
dissatisfaction score increases with 1, the likeliness of respondents engaging in negative WOM increases with 15.247.
From the analyses above it can be seen that commitment has no influence on the likeliness of engaging in negative WOM. But before hypothesis 3 is rejected, one other thing will be tested. The reviews the respondents wrote were judged on a 7-point scale, and through a t-test it can be seen whether or not there is a difference between high- and low committed respondents. The results can be found in table 11.Low committed
Levene’s Test for Equality of Variances shows that equal variances are assumed (sig. is 0.391). The 1-sided significance level of the t-test is 0.054, which indicates that the differences between the two groups are not significant (appendix III also shows these results, but with a distinction made between the two scenarios, which lead to no significant differences).
T-test Means Levene’s Test for
Equality of Variances
Equal variances assumed
Low commitment High commitment F Sig. t Sig. (2-tailed) Sig. (1-tailed) Judging of reviews 5.32 4.96 0.744 0.391 1.625 0.109 0.054
Table 11: Results of t-tests for judging of reviews
From all analyses above it can be concluded that hypotheses 4 can be rejected: the degree of brand commitment has no influence on whether or not consumers engage in negative WOM.
4.4.4 Negative WOM-motives
To test hypotheses 4 to 7 (about the effect of the negative WOM-motives on commitment) first a correlation analysis is conducted to find out whether there is a linear relationship between the variables. After that t-tests are done to see if there are significant differences between high- and low committed respondents.
As said before, first it is checked whether there is a linear relationship between the five variables through correlation analysis. The results of this can be found in table 12 (significant scores are in bold).
Correlation commitment & negative WOM-motives
Commitment Altruism Anxiety reduction
Vengeance Advice seeking
Pearson correlation 1 -0.361 0.076 -0.258 -0.245 Commitment Sig. (2-tailed) 0.002 0.531 0.030 0.041 Pearson correlation -0.361 1 0.176 0.577 0.351 Altruism Sig. (2-tailed) 0.002 0.142 0.000 0.003 Pearson correlation 0.076 0.176 1 0.410 0.433 Anxiety reduction Sig. (2-tailed) 0.531 0.142 0.000 0.000 Pearson correlation -0.285 0.577 0.410 1 0.329 Vengeance Sig. (2-tailed) 0.030 0.000 0.000 0.005 Pearson correlation -0.245 0.351 0.433 0.329 1 Advice seeking Sig. (2-tailed) 0.041 0.003 0.000 0.005
Table 12: Correlation commitment, altruism, anxiety reduction, vengeance & advice seeking
Through t-tests it is tested whether there is a significant difference between high and low commitment for the four negative WOM-motives. Table 13 gives the results of the t-tests. In this table it can be seen that for altruism, vengeance, and advice seeking the low-committed respondents have higher scores than highly committed respondents. For anxiety reduction the opposite is true. Only for vengeance the differences in mean are as expected in the hypotheses. For altruism, anxiety reduction, and advice seeking they are not as expected, but the other way around. Therefore hypotheses 4, 5, and 7 can already be rejected; altruism and advice seeking are more important motives for low committed respondents to engage in negative WOM than it is for highly committed respondents, and anxiety reduction is a more important motive for highly committed respondents to engage in negative WOM than it is for low committed respondents.
0.010) and vengeance (significance level 0.029) are significant, which means there are significant differences between the means of the highly committed respondents and the low-committed consumers. Anxiety reduction and advice seeking are of equal importance for highly committed consumers and low-committed consumers to engage in negative WOM, no significant difference is found.
Hypothesis 6 is accepted; vengeance is a more important motive for low committed consumers to engage in negative WOM than it is for highly committed consumers. But also about altruism something can be said; although hypothesis 4 is rejected, it is significant proven that altruism is a more important motive for low committed consumers than it is for highly committed consumers.
T-test Means Levene’s Test for
Equality of Variances
Equal variances assumed
Low commitment High commitment F Sig. t Sig. (2-tailed) Sig. (1-tailed) Altruism 3.17 2.28 0.133 0.717 2.408 0.019 0.010 Anxiety reduction 3.61 3.63 1.840 0.179 -0.047 0.963 0.482 Vengeance 2.19 1.67 1.227 0.272 1.934 0.057 0.029 Advice seeking 3.80 3.31 0.191 0.664 1.423 0.159 0.080
