Is it wise to discharge a loyalty program?

Is it wise to discharge a loyalty program?

March 21

, 2012

Is it wise to discharge a loyalty program?

By

Auke Sikke Broersma

University of Groningen

Faculty of Economics and Business

Msc Business Administration

Department of Marketing

Specialization: Marketing Management

Master Thesis

March 2012

Compagnie 2

9231BE Surhuisterveen

0612225339

as_broersma@hotmail.com

Student number: 1530259

Management Summary

This paper examines the reactions of consumers toward a loyalty program withdrawal. This is done by taking into account the level of progress the consumer has made in the discharged loyalty program, and the personality traits of the consumer. Online scenarios, which ask participants to imagine themselves in hypothetical constellations and roles, were used in this study to provide insight into consumer responses to a loyalty program withdrawal. Firstly, in the research consumers level of progress was varied (i.e., 20%, 60%, and 100%). This was done to test if consumers would react more negatively to a withdrawal when they had made more progress in the program. Secondly, the availability of a retailer solution was varied in this research (e.g., solution, no solution). This was done to test if a retailer solution would temper the negative reactions of consumers toward the withdrawal. Moreover, consumers were asked about their preferred solutions with regard to the withdrawal scenario. Finally, this study investigated the moderating effect of consumer personality characteristics (e.g.

attachment style and need for closure) on consumer’s withdrawal reaction. The results of this study showed that consumers react more negatively to a loyalty program

withdrawal once they made more progress in the discharged program. However, the

difference between the 20% progress group and 60% progress group was not significant. The results showed that consumers in the 100% progress group react far more negative to a loyalty program withdrawal. Consumers in the 100% progress group are far more likely to complain, are more likely to switch to a competing retailer, are more likely to spread negative word-of-mouth, becomes less committed to the retailer, and have less trust in the retailer. Furthermore, the results of this study revealed that the moderators attachment style and need for closure did not influence consumer’s withdrawal reaction. A small effect was found for both moderators, but this effect was not significant. Furthermore, the results of this study showed that there were no differences in reactions to a program withdrawal between consumers that were offered a solution and consumers that were offered no solution. However, this can be explained by the fact that very few customers liked the solution as depicted in this study. Moreover, the results revealed that consumers with different levels of progress prefer different solutions. Finally, the results showed that customers typically prefer tangible rather than intangible resolutions to their complaints. A low-cost reward (e.g., discount coupon,

Preface

With this research paper, I finalize my master Business Administration (Marketing Management) at the University of Groningen. The topic in this paper is loyalty program withdrawals. In this study I am interested in the reactions of consumers toward a program withdrawal. I hope this study will contribute to a better understanding of loyalty program withdrawals and their (negative) consequences.

Throughout the development of this text, several individuals were involved and made substantial contributions. First of all, I recognize the invaluable contributions of my

supervisors Laurens Sloot (professor in retail marketing) and Jacob Wiebenga (PhD student). They provided constructive comments and suggestions for my master thesis. Furthermore, I would like to thank my parents, girlfriend, family, and friends for their support and interest in my thesis.

Enjoy reading this report!

Auke Sikke Broersma

Table of contents

MANAGEMENT SUMMARY ... 3

PREFACE ... 5

1 INTRODUCTION ... 8

1.1INTRODUCTION TO THE RESEARCH OBJECTIVE ... 8

1.2PROBLEM STATEMENT ... 10

1.3RESEARCH QUESTIONS ... 10

1.4MANAGERIAL AND ACADEMIC RELEVANCE ... 11

1.5OVERVIEW OF THE PAPER ... 11

2 THEORY ... 12

2.1WHAT IS LOYALTY? ... 12

2.2LOYALTY PROGRAMS ... 13

2.3LOYALTY PROGRAM WITHDRAWALS ... 13

2.4CUSTOMER DISSATISFACTION ... 14

2.5CONSUMER RESPONSE TO A LOYALTY PROGRAM WITHDRAWAL... 15

2.5.1SWITCHING ... 15

2.5.2COMPLAINING AND BOYCOTTING ... 16

2.5.3WORD-OF-MOUTH COMMUNICATIONS ... 17

2.5.4COMMITMENT... 18

2.5.5TRUST ... 18

2.5.6INERTIA ... 19

2.5.7EQUITY AND FAIRNESS ... 19

2.5.8ATTRIBUTION THEORY ... 20

2.6RETAILER COMPLAINT HANDLING ... 20

2.7PERSONALITY CHARACTERISTICS ... 22

2.8INTRODUCTION TO ATTACHMENT THEORY ... 22

2.8.1ATTACHMENT STYLE ... 23

2.8.2HYPOTHESIS -ATTACHMENT THEORY ... 24

2.9INTRODUCTION TO NEED FOR CLOSURE ... 25

2.9.1NEED FOR CLOSURE SCALE ... 25

2.9.2NEED FOR CLOSURE -HYPOTHESIS ... 26

2.10OVERVIEW OF STUDY ... 27

3 THE PRESENT RESEARCH ... 28

3.1PARTICIPANTS AND DESIGN ... 28

3.2PROCEDURE ... 30

3.2.2DEPENDENT VARIABLES ... 34

3.3MANIPULATION AND REALISM CHECK MEASURES ... 36

4 RESULTS ... 38

4.1PRELIMINARY CHECKS AND ANALYSIS ... 38

4.2MAIN EFFECTS ... 39 4.3INTERACTION EFFECTS ... 40 4.4OTHER EFFECTS ... 43

5 GENERAL DISCUSSION ... 46

5.3LIMITATIONS AND FURTHER RESEARCH... 48

REFERENCES ... 49

APPENDICES ... 56

APPENDIX A:REPRESENTATIVENESS OF SAMPLE ... 56

APPENDIX B: SURVEY LOYALTY PROGRAM WITHDRAWAL (DUTCH) ... 57

1. Introduction

In the first part of this section some background information will be given. Next, an explanation of the purpose of the research is provided, which is followed by the problem statement and the research questions of the paper. Finally, the managerial and academic relevance of the study is outlined and the structure of the report is given.

1.1 Introduction to the research objective

Loyalty programs, both in practice and as a focus of consumer research, have become popular over the past decade, but the financial impact of these efforts rarely meets expectations

(Henderson et al, 2011). Extant empirical research provides mixed evidence of loyalty

program effectiveness. Some authors show the positive effects of loyalty programs (Taylor & Neslin, 2005; Verhoef, 2003) and other authors show that the effects of these programs are weak and transitional (Leenheer & Bijmolt, 2003; Bellizzi & Bristol, 2004; Sharp & Sharp, 1997). Everyday observation and media reports on problems that customers and companies experience with loyalty programs suggest that these programs may not function as they should be. As a consequence, loyalty programs are sometimes discharged or revised. For example, consider Starbuck’s recent decision to discontinue a poorly performing premium rewards program that required a registration fee and offered special discounts to members (Allison, 2010). The reactions of their customers on this decision were not mild. Negative reactions appeared at Facebook and at ‘My Starbucks Idea’, a site run by the coffee company to garner ideas from customers. A lot of customers were unhappy about losing another chance to earn rewards at their favorite coffee shop chain. Because of these negative reactions, it is possible that this program may even be leading to a reduction in loyalty. This will result in negative word of mouth, lower satisfaction, and a decrease in profits. Another example of a loyalty program withdrawal is Shell’s recent decision to stop with their free saving stamps in the Netherlands. But how will consumers react to this decision? It can be reasonably assumed that their reaction will be negative because customers of Shell lose the opportunity to earn free rewards.

It is however quite surprising that there is no research that investigates the effects of a loyalty program withdrawal on consumer behavior. The literature on loyalty program withdrawals is very scarce. It could be reasonably hypothesized that a negative experience with a loyalty program could turn a customer away. There is some literature that supports this idea

these studies are of limited generizability. It is therefore, the main goal of the present study to examine the effect of a loyalty program withdrawal on consumer behavior. A better

understanding of loyalty program withdrawals is essential for researchers and managers alike. Now, why is a better understanding of program withdrawals important? Firstly, loyalty consists of many dimensions (e.g., word-of-mouth, trust, commitment, customer complaints) that can be negatively affected by a loyalty program withdrawal. Take for example word-of-mouth. A loyalty program withdrawal can possibly lead to negative word-of-mouth as seen in the Starbucks example. This can have severe consequences. According to Luo (2007) higher levels of current consumer negative voice harm firms’ future stock returns. Secondly, it is difficult to stop loyalty programs when problems arise. Loyalty programs become part of the customer’s shopping experience, so customers must be informed about even the smallest changes in programs. They react negatively to any perceived ‘’take away’’ once a program is in place, even if they are not actively involved in it (Levy & Weitz, 2009). Thirdly, a thorough understanding of consumer’s reaction towards a loyalty program withdrawal opens up the possibility to handle the situation in a better way. Companies can implement solutions to temper the negative reactions of consumers and strengthen loyalty. Higher customer loyalty will result in more natural interactions, complaints being expressed earlier, more knowledge of customer wishes, positive word of mouth, more profit, and lower price sensitivity

(Reichheld, 1996; Reinartz & Kumar, 2006).

The current research contributes to a better understanding of loyalty program withdrawals in several ways. Firstly, it examines the ways in which people can respond to a loyalty program withdrawal. One research stream that can help to examine this issue is the dissatisfaction literature with regard to services. This literature can help to classify the different responses toward a loyalty program withdrawal (Lovelock & Wirtz, 2007; Singh, 1988). Secondly, this research investigates the relationship between consumers level of progress (e.g., 20%, 60% or 100%) made in the discharged program and their reaction toward the withdrawal of the program. This is important because the customer has put his or her resources into the loyalty program of the retailer. It could be reasonably argued that people who have completed the program will react more negatively to a program withdrawal than people who did not

complete the program. There is more inequity between their cost and gains. Furthermore, this research investigates the influence of personality characteristics on consumer’s response to a loyalty program withdrawal. It could be reasonably argued that people with different

characteristics that will be used in this study are attachment style and need for closure. These two concepts are proven to be useful in the marketing domain. Finally, this study examines whether an apology and/or compensation of a retailer can temper the negative reactions of consumers toward a loyalty program withdrawal. Often companies offer the consumer some sort of compensation and/or apology when they discharge a loyalty program.

To summarize, the goal of this study is to examine the effect of a loyalty program withdrawal on consumer behavior by taking into account the level of progress the consumer has made in the discharged program, and consumer’s personality characteristics. Furthermore, this research investigates whether companies can temper the negative reactions toward a

withdrawal by offering an appropriate solution. The ultimate goal of this study is to shed more light on loyalty program withdrawals and their (negative) consequences.

1.2 Problem statement

The presented research objective can be summarized into a statement which will serve as the leading thread in this study. The following statement is formulated:

‘’Getting insight into the reactions of consumers toward a retailer loyalty program withdrawal, by taking into account the level of progress the consumer has made in the discharged program, and the personality traits of the consumer’’

1.3 Research questions

The problem statement of the study can be translated into four research questions. These questions will serve as the guideline for this study. By answering these questions a

knowledgeable foundation is formed which will be used to answer the problem statement.

Research questions:

Question 1: How do consumers respond to a loyalty program withdrawal?

Question 2: In what way does the level of progress in the discontinued program (e.g., 20%, 60%, and 100%) influence the reactions of consumers toward a loyalty program withdrawal?

Question 3: Is the relationship between level of progress made in the program and withdrawal reaction significantly different for consumers with different psychological traits (e.g.,

Question 4: Will consumers react less negative to a loyalty program withdrawal when the retailer offers them a solution in the form of an apology and/or compensation?

1.4 Managerial and academic relevance

This paper will provide more insight into the reactions of consumers toward a loyalty program withdrawal. Surprisingly very limited research has been conducted with regard to loyalty program withdrawals and their effects. In view of the considerable investment required to set up and maintain a loyalty program, it appears especially relevant, both scientifically and managerially, to examine the effects of a loyalty program withdrawal. Specific insights about the effects of a withdrawal on consumer behavior can provide valuable knowledge for

retailers and marketing managers. They than know what the possible effects are of

discharging a loyalty program and can assess whether a potential withdrawal is worth the risk. Moreover, increased knowledge about loyalty program withdrawals opens up the possibility for companies to deal with withdrawals in a better way. Furthermore, this study illustrates the value of applying psychological principles to marketing. Personality characteristics of

consumers are included in this study to get more insight into consumer response styles and how this influences consumer reactions toward a withdrawal. Overall, this research paves the way for further research on loyalty program withdrawals and their (negative) consequences.

1.5 Overview of the paper

2. Theory

Firstly, in this chapter the concepts loyalty, loyalty programs, and loyalty program

withdrawals will be discussed. Secondly, this chapter will present the dissatisfaction literature with regard to services. This research stream will be used to get more insight into consumer reactions toward a loyalty program withdrawal. This literature will be used to classify the responses of consumers toward a withdrawal. Thirdly, in this chapter retailer complaint handling is discussed. The service recovery literature will be used in this paper to discuss complaint handling and how this can possibly help to temper the negative reactions of consumers toward a loyalty program withdrawal. Finally, in this section two personality characteristics of consumers will be discussed (e.g., attachment style and need for closure). The expectation in this study is that consumers with different personality traits will have significantly different responses to a loyalty program withdrawal. Important to note here is that hypotheses are also given in this chapter.

2.1 What is loyalty?

Before looking at loyalty programs per se, it is important to understand the concept loyalty. Consumer loyalty is indicated by an intention to perform a diverse set of behaviors that signal a motivation to maintain a relationship with a firm, including allocating a higher share of the category of wallet to the specific company, engaging in positive word of mouth (WOM), and repeat purchasing (Zeithaml et al, 1996). Loyalty has many different forms – for example, loyalty to a significant other, a family, an employer, or a country (patriotism) but also to a service provider, a store, or a brand (Melnyk et al, 2009). Loyalty also has many definitions. There is substantial disagreement about the exact definition or nature of the loyalty concept (Melnyk et al, 2009). Important aspects of the definition are that there is a relationship between an actor and another entity and that the actor displays behavioral or attitudinal loyalty to that entity in the presence of alternative entities (Melnyk et al, 2009). Broadly speaking, behavioral loyalty refers to the observed action customers have demonstrated towards a particular product or service (Reinartz & Kumar, 2006). Attitudinal loyalty, by contrast, refers to the perceptions of and attitudes a customer has toward a particular product or service (Reinartz & Kumar, 2006). Empirical evidence exists that loyalty programs are able to stimulate both types of loyalty (Mägi, 2003; Roehm et al, 2002).

different measures of loyalty that reflect the measures, indicators, and definitions found in the literature are used (Dick & Basu, 1994, Oliver, 1999, Palmatier et al, 2006). Examples of measures/dimensions that are used in this study are: customer referrals (word of mouth), complaints, trust, commitment, and switching. These measures/dimensions will grab the most important aspects of the loyalty concept.

2.2 Loyalty programs

A loyalty program is an integrated system of marketing actions that aims to make member customers more loyal (Leenheer et al, 2007). Through loyalty programs, firms can potentially gain more repeat business and, at the same time, obtain rich consumer data that aid future CRM efforts (Liu, 2007). The term ’loyalty program’ captures a variety of marketing

initiatives, including reward cards, gifts, tiered service levels, dedicated support contacts, and other methods that positively influence consumers’ attitudes and behavior toward the brand or firm (Henderson et al, 2011). A distinction can be made between short term loyalty programs (e.g., programs with an expiring date) and long term loyalty programs such as the frequent-flier programs (Liu, 2007; de Wulf et al, 2003). Loyalty programs are currently available in many industries, such as supermarkets, gasoline stations, clothing stores, airlines and hotels (Mimouni-Chaabane & Volle, 2010). The principal motivation for consumers to enroll in loyalty programs is to accrue benefits from rewards that can be attributed to their purchase transaction over time (Reinartz & Kumar, 2006). Thus, from a consumer perspective rewards are the key benefit.

2.3 Loyalty program withdrawals

Although loyalty programs have become widely popular, it is far from clear whether loyalty programs are effective in helping firms engender greater customer loyalty and higher profits (Reinartz & Kumar, 2006). Everyday observation and media report on problems that

could develop concerns about whether participation in a loyalty program is worthwhile. As a consequence their receptiveness to relationship marketing efforts will be hindered and customers will be less likely to participate in loyalty programs. One research stream that can help to get more insight into loyalty program withdrawals is the dissatisfaction literature with regard to services. This literature will now be discussed.

2.4 Customer dissatisfaction

When consumers have a negative evaluation of an outcome, they feel dissatisfaction. Dissatisfaction can lead to feelings of tolerance, distress, sadness, regret, agitation, and outrage (Hoyer & Macinnis, 2009). Dissatisfied customers may express their dissatisfaction behaviorally. These behavioral responses may impact the firm’s profitability (Zeelenberg & Pieters, 2004). Behavioral responses traditionally have been the focus of complaint behavior studies; constitute any or all consumer actions that convey an ‘expression of dissatisfaction’. The nonbehavioral responses are also important because research around the globe has shown that most people fail to complain (Lovelock & Wirtz, 2007). Many times in the case of failed service encounters, customers do not react at all, or they experience a lack of behavioral action in response to the failure (Zeelenberg & Pieters, 2004). For deciding whether to lodge a complaint or not, the costs involved with complaining play a major role (Singh, 1990).

Whereas remaining silent requires minimal effort, expressing the dissatisfaction to the

company necessitates more effort. Nevertheless, even a few consumer complaints can indicate marketing-related problems that need attention (Lovelock & Wirtz, 2007).

A useful concept that can be used to explore consumer complaining behavior is the customer complaint intentions and behavior concept (Singh, 1988; Singh & Wilkes, 1996). The

customer complaint intentions and behavior (CCB) concept is a set of multiple (behavioral and nonbehavioral) responses triggered by some feelings or emotions of perceived

that different consumers have different distinctive response styles due to the specific combination of responses utilized (Singh, 1990). A reason why consumers have different response styles is because of their personality traits. This is why the variables need for closure and attachment style are included in this study.

2.5 Consumer response to a loyalty program withdrawal

The three behavioral responses to dissatisfaction that dominate the customer satisfaction literature are the first responses that are discussed in this section. These responses are: switching, complaining, and word-of-mouth communications (Richins, 1987; Oliver, 1997; Zeithaml et al, 1996). Afterwards, commitment and trust are discussed. These five concepts are important dimensions/measures of the loyalty concept (Dick & Basu, 1994, Palmatier et al, 2006). Furthermore, the concept inertia is discussed in this section because many times customers experience a lack of behavioral action. Finally, at the end of this section, equity theory and attribution theory are discussed because these two concepts also have important implications for the study of customer dissatisfaction (Blodgett et al, 1993).

2.5.1 Switching

Customer switching (or churn), which is the opposite of customer loyalty, has been studied extensively in marketing literature, including various studies that discuss and investigate potential antecedents of customer switching (Wieringa & Verhoef, 2007). Switching refers to the termination of a relationship with the service provider (Zeelenberg & Pieters, 2004). Ample research has shown that dissatisfied customers are more likely to display switching intentions (Loveman, 1998; Rust & Sahorik, 1993). Across literature streams, it becomes clear that switching costs represent important determinants of customer switching behavior (Wieringa & Verhoef, 2007). Loyalty programs can serve as a useful tool to put up switching costs. In sectors with homogeneous assortments (e.g., supermarkets) loyalty programs can increase switching costs and support behavioral loyalty (Leenheer & Bijmolt, 2008).

H1a: The more progress an individual has made in the loyalty program, the more he or she is

likely to switch in case of a retailer loyalty program withdrawal

2.5.2 Complaining and boycotting

Complaining occurs when customers communicate their discontent explicitly to the firm or to a third party, such as a consumer union or a government body (Zeelenberg & Pieters, 2004). Complaining is more likely when motivation, ability, and opportunity are high. It is also more likely as the level of dissatisfaction or the severity of the problem increases. However, the severity of the dissatisfaction alone does not explain complaining behavior (Hoyer & Macinnis, 2009). In particular, consumers are less likely to act if they perceive that

complaining will take a lot of time and effort, that their chances of benefiting from doing so are low, or that the offer is insignificant (Hoyer & Macinnis, 2009; Singh, 1988). When looking at the previous discussion, the expectation in this study is that consumers who have made more progress in the discharged loyalty program will complain more. Their problem is more severe. There is much more at stake for them.

H1b: The more progress an individual has made in the loyalty program, the more he or she

will complain in case of a retailer loyalty program withdrawal

A radical form of complaining is boycotting. A boycott is an organized activity in which consumers avoid purchasing products or services from a company whose policies or practices are seen as unfair or unjust (Hoyer & Macinnis, 2009). Boycotts are an intriguing form of consumer behavior. They are unwelcome to marketers yet consistent with the marketing concept, because firms targeted by a well-supported consumer boycott have apparently failed to sustain a sufficient customer focus (Klein et al, 2004). Organized boycotts are able to gain publicity and are likely to have more impact than the same number of customers acting on their own (Hoyer & Macinnis, 2009). When looking at the previous discussion, the

expectation in this study is that consumers who have made more progress in the discharged loyalty program are more likely to boycott. Boycotting, as a radical form of complaining, is far more likely when the problem is severe and consumer dissatisfaction is large.

H1c: The more progress an individual has made in the loyalty program, the more he or she is

2.5.3 Word-of-mouth communication

Interpersonal influence has received great attention in social psychology and there is ample evidence for the relevance of interpersonal communications when individuals make choices in different contexts, including those of consumption (de Matos & Rossi, 2008). Empirical studies show that customers are even more likely to rely on these, known as ‘’word of

mouth’’ (WOM), in the service context because of the intangibility and experiential nature of services (de Matos & Rossi, 2008). WOM has recently attracted a great deal of attention among marketing practitioners (Trusov et al, 2009). WOM communication covers all communications of customers with the members of their social and professional network (Zeelenberg & Pieters, 2004). WOM can contain powerful messages about a company and its products and they are not controlled by the marketer. Although WOM may be the result of both satisfaction (positive WOM) and dissatisfaction (negative WOM), the focus in this paper will be on the latter form. When consumers are unhappy with a product or service, they are often motivated to tell others in order to relieve their frustration and to convince others not to purchase the product or to do business with the company. Negative WOM is more likely to occur when the problem is severe, when consumers are unhappy with the company’s responsiveness, and when consumers perceive that the company is at fault (Hoyer & Macinnis, 2009). It also may influence other consumers to stop (or never begin) doing business with the company. Online forums, blogs, websites, and e-mail can potentially

magnify the effect of WOM because consumers can tell others about good or bad experiences with just the click of a mouse. So marketers must act to prevent negative word of mouth and to rectify once it occurs. Rather than ignoring complaints, firms that empathize with

consumers’ complaints and respond with free goods or in another meaningful way will be more successful at reducing negative WOM (Blodgett et al, 1993; Hoyer & Macinnis, 2009). Negative WOM has a strong behavioral component compared to positive WOM which is more evaluative and attitudinal-based (de Matos & Rossi, 2008). When looking at the

previous discussion, the expectation in this study is that consumers will spread more negative word about the withdrawal when they have made more progress in the program. Negative WOM is more likely when the problem is severe and when consumers are very unhappy about a situation.

H1d: The more progress an individual has made in the loyalty program, the more he or she is

2.5.4 Commitment

Commitment is a central concept in the relationship marketing paradigm (Bansal et al, 2004). Drawing on the organizational behavior literature, marketing scholars have variously defined commitment as a desire to maintain a relationship, a pledge of continuity between parties, the sacrifice, or potential for sacrifice if a relationship ends, and the absence of competitive offerings (Gustafsson et al, 2005). These various sources create a ‘’stickiness’’ that keeps customers loyal to a brand or company even when satisfaction may be low (Gustafsson et al, 2005). There are two types of commitment: affective and calculative commitment (Bolton et al, 2004). Affective commitment is the desire to maintain a relationship and is based on feelings of loyalty and affiliation. Calculative commitment is based more on rational motives, focusing on termination or switching costs. Research in marketing has shown that

commitment positively influences purchase intentions or behavioral loyalty (Morgan & Hunt, 1994; de Wulf, 2001, Bansal et al, 2004). Moreover, committed customers are less likely to patronize other companies (Bansal et al, 2004). This will lead to higher levels of customer retention. These results underscore the importance of including commitment when examining customer loyalty and loyalty programs. The expectation in this study is consumers who are confronted with a withdrawal become less committed because a part of their relationship with the company has ended. Furthermore, this study expects that this effect will be stronger for consumers that have made more progress in the loyalty program because they are more likely to be disappointed in the retailer.

H1e: The more progress an individual has made in the loyalty program, the less he or she is

committed in case of a retailer loyalty program withdrawal

2.5.5 Trust

establishing and maintaining loyalty. In this study the expectation is consumers who are confronted with a withdrawal will lose trust in the retailer because the relation exchange between the customer and the company is hindered. Furthermore, this study expects that this effect will be stronger for consumers that have made more progress in the loyalty program because there are more likely to be disappointed in the retailer.

H1f: The more progress an individual has made in the loyalty program, the more he or she is

likely to lose trust in case of a retailer loyalty program withdrawal

2.5.6 Inertia

Inertia seems to be a relevant behavioral response for disappointment (Zeelenberg & Pieters, 2004). Many times customers do not react at all, or they experience a lack of behavioral action. In this paper inertia is defined as a lack of behavioral action of a consumer in response to a loyalty program withdrawal. It can be reasonably assumed that consumers who have made more progress in the loyalty program are more likely to react to a loyalty program withdrawal because their problem is far more severe. There is much more at stake. The

preceding discussion leads to the following hypothesis, which is tested in the empirical study:

H1g: The more progress an individual has made in the loyalty program, the more he or she is likely to react to a retailer loyalty program withdrawal

2.5.7 Equity and fairness

reasonably hypothesized that consumers who have made more progress in the program and are confronted with a withdrawal will perceive less fairness because there is more inequity between their costs and gains. A retailer loyalty program withdrawal will violate ’’the norm of reciprocity’’. The idea giving benefits to people who give us benefits has often been called a universal principle.

H1h: The more progress an individual has made in the loyalty program, the less he or she will

perceive fairness in case of a retailer loyalty program withdrawal

2.5.8 Attribution theory

Attribution theory is a theory of how individuals find explanations for events (Hoyer & Macinnis, 2009). Customers will ask themselves whether a problem could have been prevented. Customers, who feel that the problem was controllable, and could have been prevented, are likely to be somewhat angry at the retailer. According to Blodgett et al (1995), customers who believe that the problem was controllable are more likely to engage in

negative word-of-mouth behavior and are less likely to repatronize the seller. Attribution theory can provide marketers with guidance in how to deal with potential or existing perceptions of consumer dissatisfaction. Customers are more likely to be dissatisfied if the cause is perceived to be permanent, marketer related, and not under the customer´s control (Weiner, 2001). Furthermore, attributional search is more likely following a failure or negative outcome (Morales, 2005). In the case of a loyalty program withdrawal it could be reasonably argued that a consumer will blame the retailer. The problem is not under the customer’s control and the customer receives a negative outcome. Furthermore, the

expectation is that consumers that have made more progress in the program are more likely to blame the retailer because they receive a far more negative outcome.

H1i: The more progress an individual has made in the loyalty program, the more he or she

will blame the retailer in case of a loyalty program withdrawal

2.6 Retailer complaint handling

2007).Consumers who perceive that they have been treated appropriately in the complaint process will be more satisfied, more likely to buy again, and more likely to spread positive word of mouth (Maxham & Netemeyer, 2002). However, studies show that many customers feel that they have neither been treated appropriately nor received adequate compensation (Tax et al, 1998). When this happens, their reactions tend to be immediate, emotional, and enduring (Lovelock & Wirtz, 2007). Thus, it is important to compensate adequately for the harm done or to recognize the total costs incurred by customers (Tax et al, 1998).

In the service literature researchers have turned to theories of organizational justice to explain customer reactions to a service recovery (Morrisson & Huppertz, 2010). Organizational justice involves the following three dimensions: distributive justice, procedural justice, and interactional justice (see Blodgett et al, 1997). The recovery actions of retailers influence each of these dimensions of justice (Morrisson & Huppertz, 2010). In this paper, the focus is on the distributive dimension of organizational justice. This dimension is about resource allocation and perceived outcome of exchanges (e.g., equity theory). Compensation and apologies influence distributive justice. When a loyalty program withdrawal occurs consumers can desire financial compensation (e.g., coupon, free gift) and apologies from the firm. Financial compensation represents monetary value for the recipient, and apologies increase liking and forgiveness (Darby & Schlenker, 1982). An advantage of an apology is that it is much cheaper for a company than offering financial forms of compensation (Lovelock & Wirtz, 2007). Most studies of post-complaint satisfaction show that distributive justice in the form of

compensation has the greatest impact on customer satisfaction with recovery, repurchase intentions, and loyalty (Morrisson & Huppertz, 2010). ). Important to note here is that compensation can be perceived as a form of confession. Offering compensation implies an admission of guilt on part of the company, thus increasing the perception that the company had control over the service failure (Wirtz & Mattila, 2004). However, consumers seem to derive satisfaction from an effective recovery regardless of their attributions for service failure (Wirtz & Mattila, 2004). Therefore, the expectation in this study is that an apology and compensation from the retailer (e.g., offering a solution) can temper the negative reactions of consumers toward a loyalty program withdrawal.

H2: The relationship between level of progress made in the program and withdrawal reaction

Now that consumer withdrawal reactions and consumer complaint handling are discussed, it is time to discuss two personality characteristics that are relevant for this study. As said before, this study investigates whether the variables attachment style and need for closure influence consumer’s response to a loyalty program withdrawal.

2.7 Personality characteristics

Consumers vary in terms of their personality or the way in which they respond to a particular situation. Personality consists of the distinctive patterns of behavior, tendencies, qualities, or personal dispositions that make one individual different from another and lead to a consistent response to environmental stimuli (Hoyer & Macinnis, 2009). These patterns are internal characteristics that we are born with or that result from the way we have been raised. The expectation in this study is that the relationship between level of progress made in the program and withdrawal reaction will be significantly different for people with different psychological traits. In this chapter personality concepts will be reviewed that have high relevance with loyalty programs. As said before in the previous section, the focus will be on attachment style (Thomson & Johnson, 2006; Paulssen, 2009) and need for closure (Neuberg & Newsom, 1993; Vermeir & Kenhove, 2005) because they have to do with relationships and level of progress (e.g. complete something). In the case of a loyalty program withdrawal it is not possible to complete the program and a relationship is damaged.

2.8 Introduction to attachment theory

Attachment theory was launched by John Bowlby in the late 1950s. Perhaps the most central proposition of attachment theory is the prototype hypothesis. This hypothesis holds that experiences in early close relationships create internal working models that then influence cognition, affect, and behavior in relationships that involve later attachment figures (Paulssen, 2009; Simpson & Rholes, 2010). There is strong support for the prototype hypothesis in the psychological literature (Grosmann et al, 2005; Simpson et al, 2007). Research in psychology has explored features of relationships in purely interpersonal situations and between people and nonhuman objects (Hazan & Shaver, 1994; Zasloff & Kidd, 1994). It has also been proposed that consumers form relationships with service providers, brands, and other types of consumption objects (Price & Arnould, 1999). Recently, researchers in the field of marketing investigated the role of attachment style in predicting satisfaction, commitment, and

2.8.1 Attachment style

Attachment theory describes attachment style as capturing an individual’s emotional and behavioral tendencies in personal relationships, developed over time based on the individual’s prior experience (Thomson & Johnson, 2006). Attachment style influences consumers’

evaluations of their relationships with brands and with service providers (Thomson & Johnson, 2006). In a recent marketing study Paulssen (2009) fruitfully extended attachment theory to business-to-business relationships. In this study two dimensions of business

attachment were empirically derived. These two business attachment can be used to segment customers with respect their intrinsic retainability and their preference for personal bonding with their business partners. In short, Paulssen (2009) showed that attachment style is valuable in predicting the success of B2B relationships.

Attachment style is based on two dimensions labeled avoidance and anxiety (Simpson & Rholes, 1997; Thomson & Johnson, 2006). The avoidance dimension captures the extent to which people are comfortable in close relationships and the extent to which they can depend on and trust relationship partners. The anxiety dimension reflects the extent to which people experience fear of abandonment in relationships. The origins and implications of people’s scores on these two dimensions are what all attachment researchers deal with, whether knowingly or not (Simpson & Rholes, 1997).

Figure 1: The two-dimensional model of individual differences in adult attachment (Source: Bartholomew & Horowitz, 1991)

The four groups of attachment can be used to form the basis of a segmentation scheme for relationship marketing activities (Paulssen, 2009). With this segmentation scheme it is possible to identify customers with high levels of intrinsic retainability, on whom companies can focus their relationship-building efforts (Reinartz & Kumar, 2000; Paulssen, 2009). Attachment theory enhances marketers’ understanding of customer heterogeneity in marketing relationships (Paulssen, 2009). Firms should not form strong relationships with some customers because the relationship might overly sensitize them to negative experiences (e.g., loyalty program withdrawal) resulting in extreme emotionally based responses to any perceived injustice (Henderson et al, 2011). Thus, although building a relationship with a consumer might not always be beneficial, research shows that, in general, strong consumer relationships are advantageous (Palmatier et al, 2006).

2.8.2 Hypothesis – Attachment theory

Increasing avoidance and anxiety have been linked to a diminished experience of positive emotions in personal relationships (Simpson, 1990). High scores on the avoidance dimension indicate a relative lack of trust in relationship partners and a lack of comfort with intimacy (Thomson & Johnson, 2006). A person with high scores on the anxiety dimension has a poor view of self and, therefore, worries about rejection by relationship partners (Thomson & Johnson, 2006). An anxious person, fearing abandonment in the relationship and thus unable to feel comfortable in it, will be less likely to reciprocate a partner’s words or actions

(Meleshko & Alden, 1993). Anxious people do not perceive their relationships to be

interaction (Anderson, 1994). So it could be reasonably argued that a loyalty program withdrawal will produce more (negative) reactions for people that are high on anxiety and avoidance because they worry more about rejection of relationship partners. The preceding discussion leads to the following hypothesis, which is tested in the empirical study:

H3: The relationship between level of progress made in the program and withdrawal reaction

will be significantly different for people with a low versus high score on attachment style

2.9 Introduction to Need for closure

Kruglanski (1989) introduced the concept of need for closure (NFC) as a dimension of individual differences, which is related to a person’s motivation concerning decision making and judgment. NFC reflects the desire for clear, definite, or unambiguous knowledge that will guide perception and action, as opposed to the undesirable alternative of ambiguity and confusion (Kruglanski & Webster, 1996). Two highly entwined tendencies are assumed to underlie the NFC (Roets & Hiel, 2011). These are urgency and permanence. The urgency tendency represents an individual’s inclination to attain closure as soon as possible, while the permanence tendency represents an individual’s inclination to maintain closure for as long as possible. Both tendencies serve to avoid the aversive lack of closure, the first by terminating this state quickly, and the second by keeping it from recurring. This desire for cognitive closure varies along a continuum with a strong need to attain closure at one end and a high need to avoid closure at the other end (Roets & Hiel, 2011). NFC may be increased by a situationally inducible/evocable state (e.g., time pressure, environmental noise, or dullness of the task may cause a temporal heightening of an individual’s need for closure) but it is also a stable dimension of individual differences as well (Kruglanski & Webster, 1996). The main idea behind NFC theory is that a negative feeling is induced when closure is threatened or undermined, and a positive feeling is evoked when closure is attained or facilitated. The motivation to avoid these negative feelings prompts activities aimed at the acquisition of closure, and consequently biases the individual’s choices and preferences toward closure-bound pursuits (Vermeir & Van Kenhove, 2005).

2.9.1 Need for closure scale

given limited attention to the need for closure concept (Houghton & Kardes, 1998; Houghton & Grewal, 2000). Recently, Vermeir & Kenhove (2005) introduced NFC as a variable of individual difference that shows promise to help the understanding of a consumers’ effort to search for price and promotional information in the context of retail grocery shopping. Their research showed that high NFC consumers are more prone to search for promotional and price information. As a consequence, different marketing or retailing strategies should be developed for consumers with different levels of NFC. Thus, individual differences in NFC can be interesting for marketers as they are useful to identify market segments (Vermeir & Kenhove, 2005).

2.9.2 Need for closure - Hypothesis

Individuals high in dispositional NFC prefer order and structure in their lives, dislike chaos and disorder. They also prefer predictability, desiring secure and stable knowledge, which is reliable across circumstances and unchallenged by exceptions. High NFC individuals also experience an urgent desire to reach swift decisions, reflected in their need for decisiveness. Furthermore, they feel discomfort with ambiguity, experiencing situations as aversive when they are devoid of closure. Finally they are closed-minded, being unwilling to have their knowledge challenged by alternative opinions or inconsistent evidence (Kruglanski &

Webster 1996; Roets & Hiel, 2011). It can be reasonably hypothesized that an individual who possesses high NFC would react more negatively to a loyalty program withdrawal because closure is threatened or undermined. The motivation to avoid these negative feelings prompt activities aimed at the acquisition of closure and consequently biases the individual’s and preferences toward closure-bound pursuits (Kruglanski & Webster, 1996). However, in the case of a withdrawal it is impossible to complete the program. The loyalty program has ended and points cannot be collected anymore. The preceding discussion leads to the following hypothesis, which is tested in the empirical study:

H4: The relationship between level of progress made in the program and withdrawal reaction

2.10 Overview of study

The figure below (figure 2) gives a broad overview of the variables and the relationships that are being investigated in this study. This framework is the guideline in this study for testing the empirical data and to evaluate the theory in this paper.

Figure 2: Framework of study

3. The present research

The hypotheses that were presented earlier in this paper are tested in a web-based experiment. Online scenarios, which ask participants to imagine themselves in hypothetical constellations and roles, will be used in this research to provide insight into consumer responses to a loyalty program withdrawal. This form of quantitative research is used to test the formulated

hypotheses as it is able to provide cause-and-effect relationships (Malhotra, 2008).

In this study it is tested whether consumers would respond differently to a loyalty program withdrawal when their level of progress in the discharged program (i.e., 20%, 60% or 100%) is varied. The expectation is that consumers will react more negatively when they have made more progress in the program (hypothesis 1a-i). Furthermore, it is tested whether the

availability of a retailer solution (i.e., apology and compensation, no apology and

compensation) will moderate the relationship between the independent variable (e.g., progress condition) and the dependent variable (e.g., withdrawal reaction). The expectation is that consumers will react less negatively when the retailer offers an apology and compensation (hypothesis 2). Moreover, this study expects that attachment style and need for closure will moderate the relationship between level of progress and consumers withdrawal reaction. It is expected that respondents with high scores on attachment style and need for closure will react significantly different to a loyalty program withdrawal compared to people with low scores on attachment style and need for closure (hypothesis 3 and 4). Finally, this empirical study presented eight possible solutions to the withdrawal problem as described in the scenarios. In this research it was measured which solutions consumers preferred in all the withdrawal situations. The overall objective of the empirical study is to get more insight into consumer reactions toward a retailer loyalty program withdrawal.

3.1 Participants and design

One hundred and thirty-six consumers served as research participants in this empirical study. People that participated in the study were incentivized by being entered into a free drawing to win one of the three VVV vouchers that represented a value of 20 euro. The design of this study was a between-subjects design with three progress conditions (20%, 60%, and 100%) and two complaint handling conditions (solution, no solution). As a result, this study consisted of six different conditions. Furthermore, in order to test hypotheses 3 and 4, a

hundred and eighteen people successfully completed the online experiment, having been randomly assigned to one of the six conditions. Specifically, 45 participants completed the 20% progress condition, 39 recipients completed the 60% progress condition, and 34 participants completed the 100% progress condition. Furthermore, 61 participants (51.7%) were offered a solution compared to 57 participants (48.3%) that were offered no solution by the retailer in the online study.

Respondents in this study were reached through my own network (e.g., family, friends, and fellow students). Furthermore, snowball sampling was used for this study (Malhotra, 2008). Recipients were asked to spread the survey in their own network. Most of the respondents in this study were contacted by email and social media. Furthermore, flyers were distributed at the University of Groningen to reach potential recipients. The web-based survey was set out online using the online program Qualtrics (www.qualtrics.com). This program enables users to create their own web-based surveys and spread questionnaires (e.g., email, social media). Furthermore, this program offered the function to see the time that respondents needed to complete the questionnaire. If questions were simply being rushed through by respondents, the researcher removed them from the dataset. The program Qualtrics was used in this study to import the completed responses of the participants into the statistical program SPSS.

Looking at the profile of the sample used in this study (see table 1) it can be seen that the sample consists of 48 men (40.7%) and 70 women (59.3%). That women represent the largest group can be explained by the fact that most women do the grocery shopping in the Netherlands. To check if gender and other demographics are representative for the whole population of Dutch grocery shoppers, a research

report of the EFMI business school was used (www.efmi.nl). Each year the EFMI publishes a document called the ‘ConsumentenTrends’ which contains a lot of insights about the grocery industry in the Netherlands. In this report information can be found about samples that accurately reflect the characteristics of the whole grocery shopping population in the Netherlands (see appendix A). When looking at this information it can be seen that women also represent the largest group in the EFMI-sample (71%). However, this is far more than the 59.3% in this study. To check if women react differently to a loyalty

program withdrawal than men a t-test was conducted. No significant differences were found

Table 1: Profile of the sample

for the nine withdrawal reactions (e.g., complaining, switching, and so forth) as depicted in this study (p-values > 0.05). As a consequence, under-representation of women in this study is not a serious threat for this study. When looking at age in this study it can be seen that the 18-34 group consists of 85 participants (72%), the 35-54 group consists of 28 participants

(23.7%), and the 55 years and older group consists of 5 participants (4.3%). Hence, the 18-34 age group is the biggest group in the sample. However, when looking at the variable age for the EMFI-sample it can be seen that the 18-34 age group represents the smallest group (23%). In the EMFI report this is explained by the fact that most of these young people still live at their parents’ home and as a consequence they are not the main grocery shopper. To check if this over-representation of young consumers (e.g., students) in this study will be a potential threat, an ANOVA test was conducted. No significant differences were found between the three age groups with regard to the nine withdrawal reactions (e.g., boycotting, word-of-mouth, and so forth) as depicted in this study (p-values > 0.05). This means that the over-representation of the 18-34 age group is not a serious threat for this study. Finally, when looking at education it can be seen that most participants received a higher education (67.8%). This can be explained by the fact that a large part of the surveys were distributed through the mail of the University of Groningen. To check if this over-representation of highly educated participants will influence the results, an ANOVA test was conducted. Significant differences were found for the variables word-of-mouth (p-value = 0.032) and inertia (p-value = 0.035). It seems that highly educated people are more inclined to take action and to talk to other people

about the withdrawal.As a consequence, caution is required when interpreting the results of

these two variables. Overall, some caution is required in this study for interpreting the results

because there are some representativeness issues.

3.2 Procedure

be collected. When enough points were collected (i.e., 45), consumers were given the

opportunity to hand in their points for free rewards (e.g., barbecue or cooking pots). However, participants learned that they were not able to redeem their points at Grocery XL. Due to popularity, and thus high demand, the barbecues and cooking pots were out-of-stock and therefore the retailer ‘Grocery XL’ stopped the loyalty program. Yet, in the scenario

participants read that they knew in advance, before they entered the loyalty program, that the barbecues and cooking pots were available as long as supplies lasted. How would people react to this situation? At the end of the first section, participants were asked how they would react to the situation as described in the scenario (i.e., withdrawal reaction). These questions are related to the dependent variables in this study. More specific details about these questions are given later on in this chapter. In the second part of this study recipients were asked to fill-in two questionnaires that were related to the variables need for closure and attachment style, the two personality traits used in this study. For all participants a total score was computed for both variables. More details about the two variables and the scoring procedure can be found later on in this chapter. Finally, in the last part of the online experiment respondents were asked to provide some demographics. Also, some manipulation and realism checks were

put into the last part of the experiment.Important to note is that the whole survey was

translated into the Dutch language. This was done to get a more representative sample of

Dutch grocery shoppers. The whole survey can be found in appendix B.

3.2.1 Independent variables

Figure 3: The level of progress – 20%, 60% or 100%

Availability of solution: All participants in this study were randomly assigned to one of the two complaint handling conditions (e.g., no solution offered, solution offered). In the first case respondents were offered no solution from the retailer. In this case a consumer was told that the barbecues and cooking pots were out-of-stock due to popular demand. As a result, the consumer received no apology and no compensation (see figure 4, sheet 1). In the second case respondents were offered a solution from the retailer. They received a general apology from the retailer and they were given the opportunity to trade their full savings card for extra stamps in the next loyalty program of the retailer (see figure 4, sheet 2). The solution variable is used in this study to see if people react significantly different to a program withdrawal when they are offered a solution compared to no solution (hypothesis 2).

Figure 4: The availability of a solution – Solution or no solution (Dutch language)

Attachment style: The variable attachment style captures an individual’s emotional and

behavioral tendencies in personal relationships, developed over time based on the individual’s prior experience (Thomson & Johnson, 2006). To measure attachment style in this study the ECR-R scale was used (36 items). The Experiences in Close Relationships-Revised (ECR-R) questionnaire is a revised version of Brennan, Clark, and Shaver's (1998) Experiences in

avoidance across several distinct relationships, including relationships with parents, partners, and friends. For all 36 items, the participants in this study indicated the extent to which they agree or disagree with the statements. The items were measured on a seven-point scale with endpoints labeled completely disagree [1] and completely agree [7]. Two examples of statements that are used for the attachment scale are: ‘I worry a lot about my relationships’ and ‘I feel comfortable sharing my private thoughts and feelings with others’. An index of

attachment style (α = 0,931) was created by calculating the total average score of every

respondent to the 36 questions. This resulted in a 1 to 7 attachment score for every participant (M =…; SD =…) with higher scores indicating that people scored higher on the avoidance and anxiety dimensions of the attachment scale. A high score on the avoidance dimension indicates a relative lack of trust in relationship partners and a lack of comfort with intimacy (M = 3.56; SD = 0.91). A person with a high score on the anxiety dimension has a poor view of self and, therefore, worries about rejection by relationship partners (M = 2.93; SD = 1.16). The average score for all the respondents on the attachment scale is 3.24 with a standard a standard deviation of 0.90. Important to note here is that the attachment scale is translated into Dutch for this study (appendix B).

Need for closure: The variable need for closure (NFC) reflects a person’s desire for clear, definite, or unambiguous knowledge that will guide perception and action, as opposed to the undesirable alternative of ambiguity and confusion (Kruglanski & Webster, 1996). In this study a multi-item scale is used to measure NFC. Since its introduction, the NFC scale has been frequently changed. Given the scale’s substantial length (42 items), researchers often use adapted versions by selecting a number of items on seemingly arbitrary grounds without reporting or validating their item selection (Roets & Hiel, 2011). In this study the 15-items scale made by Roets and Van Hiel (2011) is used to measure personal NFC. Their 15-item selection can be used as a valid alternative measure of the NFC construct. The advantage is that the experiment in this study does not become too lengthy. For all 15 items, the

participants indicated the extent to which they agree or disagree with the item. The items were measured on a six-point scale with endpoints labeled completely disagree [1] and completely

agree [6]. Two examples of statements that are used for the attachment scale are: ‘I don’t like situations that are uncertain’ and ‘I dislike questions which could be answered in many

different ways’. An index of NFC (α = 0,830) was created by calculating the total average

NFC scale. Individuals that score high on NFC prefer order and structure in their lives, dislike chaos and dislike disorder. The average score for all the respondents on the NFC scale is 3.66 with a standard a standard deviation of 0.72. Important to note is that the NFC scale is

translated into Dutch for this study (appendix B).

3.2.2 Dependent variables

This section discusses the dependent variables that were used in this study. Multi-item scales were used to measure each dependent variable in this study. The responses of participants on the items were measured on a seven-point scale with endpoints labeled completely disagree [1] and completely agree [7]. In this study an index score was created for all the different dependent variables. This was done by calculating the total average score for each participant with respect to a dependent variable. As a result, a respondent could score 1 to 7 on each dependent variable (M =…; SD =…). An overview of the dependent variables is given in the table below (see table 2). This table shows the items that are representing every dependent variable, the literature source of the items, the Cronbach’s alpha of the multi-item scale, the meaning of the index score (e.g., what does a high or low score mean?), and the average of the scores of all respondents on each dependent variable (M =… ; SD = …).

Table 2: Overview dependent variables

Dependent variable Items + source* Cronbach’s

Alpha (α)**

Meaning index score

Switching: the termination of a relationship with a retailer

1 I will visit the retail store less than before

2 I will switch to a competing retailer Zeelenberg and Pieters (2004)

0,806 A higher score indicates that a consumer is more likely to switch to a competing retailer (M = 3.30 ; SD = 1.57)

Complaining: customers who communicate their discontent explicitly to the firm or to a third party, such as a consumer union or a government body

1 I will complain to external agencies, such as the consumer union, about the retailer

2 I will file written complaint 3 I will complain directly to other customers about the retailer 4 I will complain to employees of the retailer

Zeelenberg and Pieters (2004)

0,727 A higher score indicates that a consumer is more likely to complain about the loyalty program withdrawal (M = 3.05 ; SD = 1.29)

Boycotting: consumers who avoid purchasing products or services from a company whose policies or practices are seen as unfair or unjust

1 I will still do my grocery shopping at the store (reverse scored)

2 I am tempted to boycott this store Klein et al (2004)

Word-of-mouth: covers all communications of customers with the members of their social and professional network

1 I will talk with my partner and/or relatives about this incident 2 I will discourage others to go to this store

3 I will talk with friends and acquaintances about this experience Zeelenberg and Pieters (2004)

0,782 A higher score indicates that a consumer is more likely to spread negative word-of-mouth about the retailer (M = 4.59 ; SD = 1.33)

Commitment: reflects a person’s desire to maintain a relationship with a retailer

1 I feel loyal towards this store (reverse scored)

2 I am willing ‘to go the extra mile’ to remain a customer of this store (reverse scored)

3 Even if this store would be more difficult to reach, I would still keep buying there (reverse scored) De Wulf et al (2001)

0,682 A higher score indicates that a consumer is less likely to be committed to the retailer (M = 4.94 ; SD = 1.17)

Trust: consumer’s confidence in a retailer’s reliability and integrity

1 This retailer gives me a feeling of trust (reverse scored)

2 I have trust in this retailer (reverse scored)

3 This retailer gives me a trustworthy impression (reverse scored) De Wulf et al (2001)

0,707 A higher score indicates that a consumer is more likely to lose trust in the retailer (M = 4.66 ; SD = 1.23)

Inertia: a lack of behavioral action of a consumer in response to a loyalty program withdrawal.

1 I will not take action (reverse scored) 2 I will remain passive (reverse scored) Zeelenberg and Pieters (2004)

0,728 A higher score indicates that a consumer is more likely to take action against the retailer (M = 3.19 ; SD = 1.49)

Perceived fairness: the extent to which a consumer perceives a retailer loyalty program withdrawal as fair.

1 I did not get what I deserved 2 Given the circumstances, the retailer offered me an adequate response (reverse scored)

3 Taking everything in consideration, it was fair that I did not receive a gift (reverse scored)

Blodgett et al (1997)

0,678 A higher score indicates that a consumer perceives less fairness with regard to the withdrawal situation ( M = 4.64 ; SD = 1.33)

Attribution: the extent to which a consumer blames the retailer in case of a loyalty program withdrawal.

1 The unpleasant experience is the outcome of my own deed (reverse scored)

2 I did contribute to my unpleasant experience (reverse scored)

3 Little could be done by this retailer to stop what happened in the incident (reverse scored)

Russell (1982) Wirtz & Mattila (2004)

0,813*** A higher score indicates that it is more likely that a consumer blames the retailer with regard to the loyalty program

Willingness: the willingness of the consumer to participate in the next loyalty program of the retailer.

1 Unlikely……Likely 2 Improbable……..Probable 3 No chance………Certain De Wulf et al (2001)

0.959 A higher score indicates that a consumer is more likely to participate in the next loyalty program of the retailer (M = 3.48; SD = 1.65)

* The questions are (slightly) altered to fit the context of this study ** A Cronbach’s Alpha of 0.6 or higher is acceptable

*** The question ‘The incident was controllable by this retailer’ was removed from the attribution scale. This lead to a great improvement in Cronbach’s Alpha (from 0,573 to 0,813)

3.3 Manipulation and realism check measures

Firstly, to check whether the withdrawal scenario affected participants’ emotion, the

participants answered the Positive and Negative Affect Schedule (PANAS). The PANAS is a well-validated and widely used self-report index of positive and negative affect mood

dimensions (Watson et al, 1988). On the PANAS respondents indicated, on a scale ranging from 1 (very slightly or not at all) to 5 (very much), the degree to which they experience ten different feelings related to PA (e.g., inspired, alert, attentive, enthusiastic, and determined) and NA (e.g., fraid, upset, nervous, scared, and shaked). PANAS showed high reliability in the whole sample (α = 0.78 for PA and α = 0.84 for NA). In this study the expectation is that participants that have made more progress in the discharged program will experience an increase in negative affect because they are more likely to be dissatisfied. Furthermore, no such effects are expected for positive affect. Because scores on NA and PA were significantly correlated in this study, an affective Balance Score (ABS) was calculated by subtracting the NA score from the PA score. A positive score on the ABS indicates positive affect and a negative score on the ABS indicates negative affect. Furthermore, it was checked in this study whether participants read the scenario in enough detail. Participants were asked how much points they collected in the withdrawal scenario. They could answer this question on a scale from 1 to 45 (e.g. points). Thirdly, participants were asked if they deserved the free rewards that were offered in the program. On a scale from 1 (very slightly or not at all) to 5 (very

much), respondents indicated their answer. The expectation in this study is that the score will

Table 3: Reward expectation – Group means

4. Results

In this section the results of the empirical study are presented. All hypotheses were analyzed with a statistical test. All tests are conducted using a confidence interval of 95% (0.05).

4.1 Preliminary checks and analyses

Firstly, to check whether the withdrawal scenarios affected participants’ emotion, an analysis of variance (ANOVA) was performed on the affective balance score (ABS). This analysis revealed that there were no significant differences between the experimental groups with regard to affect (p-value = 0.204)1. Secondly, to check whether the participants read the scenario in enough detail, a descriptive analysis was performed on the question ‘How many points did you collect in the withdrawal scenario described earlier?’. This analysis revealed that the progress manipulation worked. In the 20% progress group (M = 8.96; SD = 0.56) 3 of the 45 participants answered the question incorrectly. In the 60% progress group (M = 27.15; SD = 0.81) 4 of the 39 participants answered the question incorrectly. In the 100% progress group (M = 45; SD = 0) none of the 34 participants answered the question incorrectly. Thirdly, from the ANOVA analysis of the question ‘Do you think you deserve the BBQ or cooking pots?’ it becomes clear that there are significant differences between the three groups (p-value = 0.000) with regard to reward expectation. As expected, people thought they

deserved the rewards more when they made more progress in the program (see table 3).

Furthermore, from the analysis of the question ‘Did you ever participate in a similar program as presented in this survey?’ it became clear that 61 respondents (51,7%) participated in a similar loyalty program compared to 57 respondents (48.3%) that did not participate in a similar loyalty program. It was also checked whether the withdrawal reactions would be significant different for

these two groups. From the t-test analysis it became clear that there were no significant differences between the two groups with regard to the nine withdrawal reactions (e.g., trust, commitment, and so forth) as depicted in this study (p-values > 0.05). Finally, to check whether the study was realistic enough, a descriptive analysis was performed on the realism index. The responses to the realism check reflected a sufficient level of realism of the

1

Affect did not differ significantly between respondents in the different experimental conditions. As a result, affect will not be used as a control variable in this study.