國立交通大學
管理科學系碩士班
碩士論文
失驗與購後口碑對顧客滿意度之影響
The Effects of Disconfirmation and
Post-purchase Word-of-Mouth on
Customer Satisfaction
研究生: 廖為新
指導教授: 張家齊 博士
The Effects of Disconfirmation and
Post-purchase Word-of-Mouth on Customer Satisfaction
研究生：廖為新 Student: Wei-Hsin Liao
指導教授：張家齊 博士 Advisor: Dr. Chia-Chi Chang
國立交通大學
管理科學系
碩士論文
A Thesis
Submitted to Department of Management Science
College of Management
National Chiao Tung University
In Partial Fulfillment of the Requirements
for the Degree of
Master
in
Management Science
June 2007
Hsinchu, Taiwan, Republic of China
研究生：廖為新 指導教授：張家齊 博士
國立交通大學管理科學系碩士班
中文摘要
本研究主要是檢視失驗與購後口碑對顧客滿意度的影響，失驗指的是顧客主觀的知覺 期待與產品或服務表現之間的差異，先前的研究指出失驗會正向的影響顧客滿意度。而購 後口碑在此定義為消費者在購買之後才接觸到的口碑訊息。本研究以一個民宿服務的實驗 情境，再次的評估失驗對顧客滿意度的影響，並且進一步的了解是否顧客在接觸購後的口 背後，會改變他們的滿意度評估。 研究結果指出，正面失驗會導致最高的滿意度，其次是無失驗，最後是負面的失驗。 購後口碑的確會影響顧客滿意度，且此影響會受到失驗情境的調節。在負面失驗之下，購 後口碑並不會對滿意度造成顯著的影響。正面失驗下，在購後接觸WOMCP的口碑訊息， 發覺他人體驗到的民宿表現與自己一致的受測者，比起沒有接觸口碑的受測者有較高的滿 意度;相反的，正面失驗之下，在購後有接觸WOMAP的口碑訊息，發覺其他人體驗了比 自己還要好的民宿表現的受測者，滿意度較沒有接觸口碑的受測者要來的低。在無失驗之 下，購後接觸 WOMCP或是WOMAP的受測者，滿意度均較無接觸任何口碑的受測者來的 高。最後，本研究會根據這些結果，提供一些解釋並給予行銷人員一些建議。 關鍵字：失驗、購後口碑、WOMCP、WOMAP、顧客滿意度Word-of-Mouth on Customer Satisfaction
Student：Wei-Hsin Liao Advisor：Dr. Chia-Chi Chang
Department of Management Science
National Chiao Tung University
ABSTRACT
This research examined the effects of disconfirmation and post-purchase word-of-mouth (WOM) on customer satisfaction. Disconfirmation, the customers’ subjective perceived
discrepancy between expectations and performance, has been confirmed in much previous research to have a positive influence on satisfaction. Post-purchase WOM in this research was defined as the WOM which consumers received after a purchase. This research used an
experimental situation of hostel service to evaluate the effect of disconfirmation again and then examined whether customers would change their satisfaction evaluations after being exposed to post-purchase WOM.
The results revealed that positively disconfirmed participants have the highest satisfaction followed by confirmed participants, and then negatively disconfirmed participants who have the lowest satisfaction. Post-purchase WOM did influence customer satisfaction and further,
disconfirmation moderated its effect on satisfaction. Post-purchase WOM did not significantly influence satisfaction in negative disconfirmation. When experiencing positive disconfirmation, participants who received WOMCP, which indicated that others’ perceived product performance
was consistent with theirs, after a purchase had higher satisfaction than those who had not received any WOM. On the contrary, when experiencing positive disconfirmation, participants who received WOMAP, which indicated that others’ perceived product performance was better
than theirs, after a purchase had lower satisfaction than those who had not received any WOM. Moreover, in zero disconfirmation, both participants who received WOMCP and WOMAP after a
purchase felt more satisfied than those who had not received any WOM. Finally, according to the results, the author provided some explanations and suggestions for marketers.
Keywords: Disconfirmation, Post-purchase Word-of-Mouth, WOMCP, WOMAP, Customer
這份論文之所以能夠順利的完成，必須感謝許多人，首先，感謝我的指導老師，張家 齊博士，對於我的論文細心、耐心的指導，給予我許多研究方面的建言，讓我在研究的路 途上不至於迷失方向。感謝我的父母給我經濟以及精神上的支持，讓我能無後顧之憂的完 成我的學業並取得碩士學位。 除了老師以及家人，還有許多管科 94 級的同學們的陪伴，才能讓我順利的走過碩士 生涯的這兩年。感謝我的紅粉知己，奐宇，給予我精神的支持，在我的研究遭遇一些問題 時給我最溫暖，最窩心的鼓勵。感謝張門的其他同學，慧芸、佩瑜、以江、秋君以及 109 研究室的夥伴們，大銓、杰宏，一路相隨，互相打氣，才能使我不斷向前，完成我的學 業，我的碩士論文。感謝交大管科提供我一個好的學習環境，以及許多優秀同學的良性競 爭、相互學習，才能成就如今的我。 最後，必須感謝每個協助我完成碩士論文的人，由於有許多好心的人願意幫我填寫問 卷，許多 PTT 實業坊的板主願意讓我放置問卷，許多給予研究建議的朋友，我才能順利的 完成研究，很多人雖然都不知道你們的姓名，但這份感謝的心意將一直留在我的心裡。 再次感謝所有陪伴我，支持我，給予幫助的所有人！ 廖為新 謹誌於 民國 96 年於交大管科
中文摘要... I ABSTRACT ...II ACKNOWLEDGEMENT ... III CONTENTS ... IV LIST OF TABLES...VII LIST OF FIGURES...VIII CHAPTER 1: INTRODUCTION ... 1 1.1 Background ...1 1.2 Research Motivation...1 1.3 Research Objectives...3 1.4 Thesis Structure ...3
CHAPTER 2: LITERATURE REVIEW ... 5
2.1 Customer Satisfaction...5
2.2 Disconfirmation and Customer Satisfaction ...7
2.3 Post-purchase Word-of-Mouth...10
2.4 Post-purchase WOM and Satisfaction...12
CHAPTER 3: METHODOLOGY ... 18 3.1 Research Produce...18 3.2 Stimulus ...19 3.3 Manipulations...19 3.3.1 Disconfirmation ... 19 3.3.2 Post-purchase WOM ... 20 3.4 Experimental Procedure ...20 3.5 Sampling Plan ...22
3.6.1 Expectations ... 22
3.6.2 Performance... 24
3.6.3 Disconfirmation ... 25
3.6.4 Satisfaction ... 26
3.6.5 Perceived Performance Discrepancy with Others... 27
3.7 Pretest ...27 CHAPTER 4: RESULTS ... 28 4.1 Background of Participants ...28 4.2 Reliability Analysis ...28 4.3 Manipulation Checks...28 4.4 Hypothesis Tests...32
4.4.1 The Effects of Disconfirmation and Post-purchase WOM on Satisfaction ... 33
4.4.2 The Effect of Post-purchase WOM on Satisfaction in Negative Disconfirmation ... 36
4.4.2 The Effect of Post-purchase WOM on Satisfaction in Zero Disconfirmation ... 38
4.4.3 The Effect of Post-purchase WOM on Satisfaction in Positive Disconfirmation ... 39
4.5 Results of the Tested Hypotheses ...41
CHAPTER 5: DISCUSSION AND IMPLICATIONS... 42
5.1 Discussion ...42
5.1.1 Post-purchase Effect in Negative Disconfirmation... 43
5.1.2 Post-purchase Effect in Positive Disconfirmation ... 44
5.1.3 Post-purchase Effect in Zero Disconfirmation ... 45
5.3 Limitations...49
5.4 Future Research Suggestions...49
REFERENCES ... 51
APPENDIX 1 ... 54
Expectations Manipulation ...54
(Low Expectations Scenario) ... 54
(Moderate Expectations Scenario)... 56
(High Expectations Scenario)... 58
APPDENX 2 ... 61
Performance and Post-purchase WOM Manipulations...61
Low Performance Scenario... 61
Moderate Performance Manipulation... 63
High Performance Scenario... 64
APPENDIX 3 ... 66
Internet Questionnaire ...66
APPENDIX 4 ... 73
The Hyperlinks of the Research Websites...73
APPENDIX 5 ... 74
The Results of Pretest ...74
APPENDIX 6 ... 77
WOMAP Manipulation in Zero Disconfirmation ...77
APPENDIX 7 ... 78
Table4. 1 Frequency Distribution of Sample Characteristics... 30
Table4. 2 Reliability ... 31
Table4. 3 Manipulation Check for Expectations... 31
Table4. 4 Manipulation Check for Performance ... 31
Table4. 5 Manipulation Check for Disconfirmation ... 32
Table4. 6 Manipulation Check for Post-purchase WOM ... 32
Table4. 7 Descriptive Statistics for Satisfaction... 33
Table4. 8 Two-way ANOVA ... 35
Table4. 9 Assumption Tests of Error for ANOVA ... 35
Table4. 10 LSD Test for Disconfirmation on Satisfaction... 36
Table4. 11 LSD Test for Post-purchase WOM on Satisfaction ... 36
Table4. 12 Profile Analysis in Negative Disconfirmation ... 37
Table4. 13 Profile Analysis in Zero Disconfirmation ... 38
Table4. 14 Profile Analysis in Positive Disconfirmation... 40
Figure1. 1 Consistency Triangle... 2
Figure2. 1 Disconfirmation Paradigm ...8
Figure2. 2 Research Framework... 17
Figure3. 1 Research Procedure...18
Figure4. 1 The Interaction between Disconfirmation and Post-purchase WOM ...34
Figure4. 2 Mean Plot of Satisfaction in Negative Disconfirmation ... 37
Figure4. 3 Mean Plot of Satisfaction in Zero Disconfirmation ... 39
Figure4. 4 Mean Plot of Satisfaction in Positive Disconfirmation... 40
CHAPTER 1: INTRODUCTION
1.1 Background
Through marketing communication (e.g., advertising, sales promotions, direct marketing, and public relations), companies deliver messages about their products (or services) to create consumers’ expectations prior to a purchase. Previous research has demonstrated that consumers would compare their perceived product performance with pre-purchase expectations and then generate satisfaction judgments (Churchill & Surprenant, 1982; Oliver 1980; Tse & Wilton, 1988). These messages from the company’s sources can be controlled and modified by marketers. However, there are some messages from other people that marketers cannot control directly, namely word of mouth (WOM) which refers to a kind of interpersonal communications in which none of the participants are marketing sources. Burnkrant and Cousineau (1975) demonstrated that people would use others’ product evaluations as a source of information about products. WOM has a significant effect on pre-purchase attitudes (Herr, Kardes, & Kim, 1991) and post-purchase evaluations (Bone, 1995). Previous research emphasized the effect of WOM which consumers come into contact with prior to a purchase on product/service evaluation (e.g., Arndt, 1967; Bone, 1995; Herr, Kardes, & Kim, 1991). However, consumers can receive WOM in any period of decision-making, even when a purchase has occurred. It is worth investigating whether WOM which consumers receive after a purchase would influence responses to product/service.
1.2 Research Motivation
Duncan’s textbook of integrated marketing communication mentioned a “consistency triangle” (Duncan, 2005, p. 335), as seen infigure1.1.
Figure1. 1 Consistency Triangle
z “Say” messages are MC (marketing communication) messages that set expectations.
z “Do” messages are messages delivered by the company’s product and service messages. They are conveyed by how products actually perform, what they actually cost, how convenient they are to get and use, and the brand’s supporting services.
z “Confirm” messages are messages from other people who either criticize or praise the brand or company. Personal and positive third-party communication is considerably more persuasive than most brand messages.
(Source: Duncan, 2005, p335)
All of these above messages must achieve consistency to create great brand relationships which contain customer satisfaction. The “say” messages delivered by marketing communication must be consistent with the “do” messages about how products and services perform, as well as with what others or word of mouth “confirm” about the brand. The concept of the consistency triangle is somewhat similar to the disconfirmation paradigm in customer satisfaction literature. According to the disconfirmation paradigm (Oliver, 1980), customers judge satisfaction by
Say Brand integrity Do Confirm Consistency Consistency Consistency
comparing previously held expectations with perceived product or service performance. To satisfying customers, product or service performance must be at least consistency with previous expectations. It also implies the importance of the consistency between the “say” and “do” messages. The motivation of this research is elicited by the consistency triangle and expectancy-disconfirmation paradigm. It is necessary to understand the importance which WOM information confirms product performance in satisfaction formation.
1.3 Research Objectives
The present research focuses on post-purchase WOM which consumers come into contact with after a purchase. The purpose of this study is to understand its influence on satisfaction judgments. As a result, this research will reaffirm disconfirmation model, demonstrate whether disconfirmation would be a moderator between post-purchase WOM and satisfaction, examine the influence of consistency between post-purchase WOM and consumers’ perceived
performance on satisfaction, and then provide some managerial implications.
1.4 Thesis Structure
This thesis includes five chapters, and the outline of each chapter is as follows:
Chapter One introduces the research background, research motivation, research objectives, and the research structure.
Chapter Two reviews the antecedent literatures relevant to this research. Customer satisfaction serves as the dependent variable; disconfirmation and post-purchase WOM
post-purchase WOM information and customer satisfaction. After reviewing the literatures, it will present the hypotheses of this research.
Chapter Three illustrates how the experiment was designed and the data was collected. It presents a conceptual research framework, the experimental procedure, sample selection, data collection, measurements, and pre-test.
Chapter Four examines the hypotheses and shows the statistical results of this research. It includes descriptive statistics, independent sample t test, reliability analysis, GLM, ANOVA, and multiple comparisons. With this information, some evidence to demonstrate the relationships among variables could be provided.
Chapter Five summarizes the findings, describes the limitations of this research and provides suggestions for future research.
CHAPTER 2: LITERATURE REVIEW
2.1 Customer Satisfaction
Customer satisfaction is a vital field in marketing literature. From the consumer’s perspective, satisfaction represents a pleasurable consumption experience and reaffirms the consumer’s decision-making prowess (Oliver, 1997). It can influence the consumer’s product attitude and repurchase intention (Oliver, 1980). From the firm’s perspective, satisfaction considerably contributes to the increase of a firm’s profitability. Research has supported that there is a positive relationship between customer satisfaction and financial performance (Anderson, Fornell, & Rust, 1997). Prior studies have found that higher levels of customer satisfaction can also lead to greater customer loyalty (Mooradian & Olver, 1997; Oliver, 1980) and it implies that satisfaction helps to secure future revenue. Satisfied customers are wiling to provide new referrals through positive word-of-mouth (Brown, Barry, Dacin, & Gunst, 2005; Mooradian & Olver, 1997). Moreover, the more customers feel satisfied, the more they are willing to pay (Homburg, Koschate, & Hoyer, 2005). Accordingly, researchers are concentrating their attentions on exploring and demonstrating the antecedents and consequents of satisfaction, and marketers are devoting their efforts to making their customers satisfied.
Generating a generally acceptable definition of satisfaction is not easy. Through reviews of previous literature, some proposed definitions summarized by Oliver (1997) are shown below. Satisfaction can be defined as:
z “an evaluation rendered that the consumption experience was at least as good as it was supposed to be” (Hunt, 1977, p. 459)
z “the summary psychological state resulting when the emotion surrounding disconfirmed expectations is coupled with the consumer’s prior feelings about the consumption experience” (Oliver, 1981, p. 27)
z “the consumer’s response to the evaluation of the perceived discrepancy between prior expectations (or some other norm of performance) and the actual performance of the product as perceived after its consumption”. (Tse & Wilton, 1988, p. 204).
These definitions contain the key concepts and mechanisms by which these concepts interact. They all mention that satisfaction is the end state of a psychological process about the consumption experience and is an evaluation or response toward the perception of the difference between the actual experience and the pre-experience standards. However, these definitions do not specify what the psychological state, evaluation or response is. Oliver (1997) maintained, “Satisfaction is the consumer’s fulfillment response. It is a judgment that a product or service feature, or the product or service itself, provided (or is providing) a pleasurable level of
consumption-related fulfillment, including levels of under- or overfulfillment” (p. 13). Based on two reasons, the present research adopts Oliver’s definition (1997). First, his definition specifies what the consumer’s summary judgment and explicitly reveals what can be measured in
satisfaction, a pleasant level of consumption-related fulfillment. Second, fulfillment implies there is a goal, something to be filled. Therefore, it can be judged with reference to a standard. “A fulfillment, and hence a satisfaction judgment, involves at the minimum two stimuli - an outcome and a comparison referent” (Oliver, 1997, p. 14). It corresponds to the concept that satisfaction is the perceived consistency/inconsistency between an actual experience and a pre-experience standard.
2.2 Disconfirmation and Customer Satisfaction
Consumers would compare actual product performance with prior expectations which serves as a standard or reference point to generate satisfaction evaluations and disconfirmation is assumed to occur when perceptions of product performance are different from previous
expectations (Oliver, 1980). Thus, disconfirmation is the extent to which expectations are disconfirmed by perceived product performance. Zero disconfirmation or confirmation means that a product has performed as expected; positive disconfirmation refers to the positive
discrepancy that occurs when performance is above expectations; negative disconfirmation refers to the negative discrepancy that occurs when performance is below expectations (Churchill & Surprenant, 1982; Oliver, 1980, 1997). Expectations refer to subjective, predictive expectations, or the consumers’ pretrial beliefs about the performance of the product in the future (Olson & Dover, 1979).
According to different approaches to conceptualizing disconfirmation, there two types of disconfirmation, subtractive disconfirmation and subjective disconfirmation (Tse & Wilton, 1988). Subtractive disconfirmation is the objective, numeric deviation which subtracts an expectation score from a performance score (Tse & Wilton, 1988). Subjective disconfirmation (e.g., Churchill & Spurprenant, 1982; Oliver, 1980) represents a distinct psychological construct encompassing a subjective evaluation of the difference between product performance and expectation. An important distinction between the two approaches is drawn by Oliver (1980), who suggests that subtractive disconfirmation may lead to an immediate satisfaction judgment, whereas subjective disconfirmation represents an intervening “distinct cognitive state resulting from the comparison process and preceding a satisfaction judgment” (p.460). Tse and Wilton (1988) demonstrate that subjective disconfirmation offers a better explanation of the satisfaction
formation than subtractive disconfirmation. This present research adopts the subjective
disconfirmation approach because it emphasizes the consumers’ subjective comparison process prevailing in most consumption-related situations.
The full disconfirmation model contains expectations, perceived performance,
disconfirmation, and their influences on satisfaction. Some previous research has been reviewed and integrated, and the relationships among these constructs were captured – see figure 2.1.
Figure2. 1 Disconfirmation Paradigm
z relationship A (Churchill & Surprenant, 1982; Spreng & Page, 2001)
z relationship B (Churchill & Surprenant, 1982; Spreng & Page, 2001; Tse & Wilton, 1988)
z relationship C (Churchill & Surprenant, 1982; Mooradian & Olver, 1997; Oliver, 1977, 1980, 1981, 1993;
Spreng & Page, 2001; Tse, Nicosia, & Wilton, 1990; Tse & Wilton, 1988) z relationship D (Oliver, 1981; Tse et al., 1990; Tse & Wilton, 1988)
z relationship E (Churchill & Surprenant, 1982; Spreng & Page, 2001; Tse et al., 1990)
According to the definition of disconfirmation, it is easy to comprehend that there is a negative relationship is between expectations and disconfirmation and a positive one between
Expectations Disconfirmation Performance Satisfaction C:+ D:+ E:+ B:+
A:-performance and disconfirmation. Much research (relationship C in figure 2.1) has supported that disconfirmation positively influences satisfaction. When consumers’ perceived performance is better than their previous expectations, they will feel satisfied; when consumers’ perceived performance is worse than previous expectations, they will feel dissatisfied. Furthermore, expectations may also have a direct, positive effect on satisfaction. Oliver (1981) suggests that this results from an assimilation effect whereby the expectation level establishes the baseline around which satisfaction judgments are made. Prior research (relationship B in figure2.1) also demonstrates that perceived performance is a determinant of satisfaction. Performance may directly and positively influence satisfaction regardless of the extent of disconfirmation (Tse & Wilton, 1988). For example, users of some brands who experience unfavorable disconfirmation of high expectations (generated through advertising) may still feel satisfied with the brand if it has more of the desired attributes than competing brands. In Churchill and Surprenant’s (1982) two studies, disconfirmation and perceived performance both have positive effects on satisfaction of house plants but for video-disk player products, only perceived performance can significantly influence satisfaction. Although an expectancy disconfirmation model consists of several
constructs, the focus of this present research is drawn on the relationship between
disconfirmation and satisfaction. Through the preceding discussions, it is reasonable to suppose that disconfirmation has a positive effect on satisfaction and hence hypothesis 1 is developed as follow.
H1: Satisfaction of positively disconfirmed customers is higher than confirmed customers followed by negatively disconfirmed customers.
2.3 Post-purchase Word-of-Mouth
Word-of-mouth is a kind of interpersonal communication in which none of the participants are marketing sources (Bone, 1995). One of the earliest researchers on word-of-mouth was Arndt (1967) who characterized word-of-mouth as oral, person-to-person communication between a receiver and a communicator whom the receiver perceives as non-commercial, regarding a brand, product or service. Westbrook (1987) defined word-of-mouth as “informal communication directed at other consumers about the ownership, usage, or characteristics of particular goods and services or their sellers” (p. 261). According to Harrison-Walker (2001), word-of-mouth is defined as “a type of informal, person-to-person communication between a perceived non-commercial communicator and a receiver regarding a brand, a product, an organization, or a service” (p. 63). Based on these definitions, word-of-mouth (WOM) in this research is defined as the informal and interpersonal communication directed from a perceived non-commercial
communicator to a receiver and about a brand, product, service or organizational related experiences.
Prior research has clarified the importance of interpersonal influence on brand purchase decision (Bearden & Etzel, 1982), product evaluations (Burnkrant & Cousineau, 1975) and variety-seeking behavior (Ratner & Kahn, 2002). There are two type of interpersonal influence, normative and informational influence (Deutsch & Gerard, 1955). Normative influence is defined as the tendency to conform to the expectations of others; informational influence is defined as the tendency to accept information from others as evidence about reality (Deutsch & Gerard, 1955). It is not hard for us to imagine that information from WOM sources might have the two types of interpersonal influence simultaneously.
People tend to trust interpersonal sources of information more than commercial sources of information, because the communicator is perceived to be independent from the seller. Harrison-Walker (2001) cited Katz and Lazarsfeld’ work (1955), and suggests that WOM is the most important source of influence in the purchase of household goods and food products. It 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 (Katz & Lazarsfeld, 1955). Due to the vividness of WOM (Bone, 1995; Herr et al., 1991), it is retrieved more easily from memory and its impact on consumers is relatively greater. Herr, Kardes and Kim (1991) demonstrate that vivid WOM communication has a greater impact on pre-purchase attitudes about personal computers and automobiles than less vivid printed information, even if the information is held constant. Bone (1995) finds that WOM
communications can influence both shot-term and long-term post-usage perceptions. When adopting a new product, compared with low-risk perceivers, high-risk perceivers tend to make more efforts to seek WOM information (Arndt, 1967). Moreover, positive WOM can reduce the performance risk perception for a brand purchase by consumers in online auctions (Ha, 2002). Because WOM communications transmit consumers’ own experiences vividly to other
consumers, people perceive that such positive information about products can reflect high quality. In addition, online WOM also influences consumers’ offline decisions. Godes and Mayzlin(2004) support that WOM activities online can influence TV viewership behavior. They suggest that “online conversation may be a proxy for offline conversation”(Godes & Mayzlin, 2004, p. 558) and people sometimes make offline decision based on online information.
Most of previous research emphasized the influence of WOM received before a consumer’s purchase on product evaluation (either pre-purchase attitudes or post-usage judgments). However,
there is a lack of research examining whether WOM received after a consumer’s purchase would influence previously formed product evaluations. In this research, post-purchase WOM might be defined as the WOM which customers received after a purchase or after forming post-purchase product (or service) evaluations. Furthermore, this research develops different categories in the light of the content of post-purchase WOM information. It could be argued that consumers would compare their own usage experiences of products with others’ from WOM information after a purchase. The above consumer’s perception WOM (WOMAP) denotes that it indicates that others’
perceived product (or service) performance is better than the consumer’s own perceived
performance; the consistent perception WOM (WOMCP) denotes that this kind of WOM indicates
that others’ perceived product (service) performance is consistent with the consumer’s; the below consumer’s perception WOM denotes that this kind of WOM indicates that others’ perceived product performance is worse than the consumer’s. This research focuses on WOMCP and
WOMAP because it would like to examine whether post-purchase WOM which indicates others’
perceived performance is better than the consumers would result negative damage to the companies.
2.4 Post-purchase WOM and Satisfaction
People usually keep absorbing product information even after a purchase. Consumers with high enduring involvement have ongoing concerns about products (Laurent & Kapferer, 1985; Venkatraman, 1989) and hence they might update their information about the products during post-usage. People with any level of involvement or knowledge in post-purchase might have the motivations to gather product information to understand others’ usage experiences, product qualities, or confirm their own decisions.
Burnkrant and Cousineau (1975) demonstrate that people use others’ product evaluations as a source of information about products. Thus, it is reasonable that consumers might use WOM information which reveals others usage experiences to confirm their own post-purchase
evaluations. WOM has two characteristics, accessibility and diagnosticity (Bone, 1995; Herr et al., 1991). “The influence of a particular piece of information depends on the accessibility of that information in one’s memory and the diagnostic of that information when predicting actual performance” (Bone, 1995, p. 213). Accessibility is whenever information is easy for consumers to retrieve. Herr, Kardes, and Kim (1991) provide evidence that WOM is highly accessible because such information is vivid. Diagnosticity is high whenever consumers feel that the information allows him or her to categorize the product clearly into one group (i.e., high quality or low quality). WOM is likely to be perceived as diagnostic, because consumers generally feel this kind of interpersonal information to be credible and trustworthy (Bone, 1995). Due to these two characteristics, when consumers understand other’s experience from WOM in a
post-purchase situation, they might have more reliable information about how the product (or service) performs. Because of the increase of performance information, consumers might change their satisfaction judgment formed initially by disconfirmed expectancy. One study has supported the idea that people would modify their satisfaction judgments after they interact with other group members (Bohlmann, Rosa, Bolton, & Qualls, 2006). When consumers experience disconfirmed expectancy and then form initial satisfaction, they also discuss their product evaluation with other group members, such as family members and buying group members. It was argued that based on the desire to confirm to the expectations of others, when individuals discover there is discrepancy between their satisfaction judgments and other group members’, they may modified their
However, it is not necessary for consumers to make their post-consumption evaluations conform to the expectations of others. In most time, consumers’ post-purchase evaluations might be influenced by others’ through informational influence. WOM might reveal some credible information about products or services. As a result, it is worthwhile questioning whether consumers would modify their satisfaction judgments after receiving WOM.
According to above discussion, it could be inferred that post-purchase WOM might influence satisfaction, but perhaps the effect of post-purchase WOM on satisfaction might differ at different disconfirmation levels. This study does not directly predict the influence of post-purchase WOM on satisfaction. Therefore, the following discussion involves the inference of the impact of post-purchase WOM on satisfaction at each disconfirmation level.
Marketing communications usually present the focal product or service in the positive side. Customers with confirmed or positively disconfirmed expectancy would perceive product
performance to be as good as or better than their expectations and hence feel satisfied. Such consumption experiences could serve as good experiences. Aron (2006) has demonstrated customers with good consumption experience would feel more satisfied after receiving positive communication messages about the product or service in a post-purchase situation. In his study, he found subjects with good consumption experience of a MP3 player felt more satisfied after being exposed to a positive advertising message. Positive post-purchase messages could
strengthen the positive belief about the product or service which customers have/had purchased (Aron, 2006). Thus, if confirmed or positively disconfirmed customers could receive WOMCP
and then perceive others’ evaluations of product or service performance are consistent with theirs, such positive messages might confirm their positive consumption experiences; meanwhile
strengthen their positive belief about the product or service, and hence enhance their satisfaction evaluations.
H2a: When experiencing positive disconfirmation, customers who received WOMCP after a
purchase are more satisfied than those who did not receive any WOM.
H2b: When experiencing zero disconfirmation, customers who received WOMCP after a
purchase are more satisfied than those who did not receive any WOM.
However, when consumers with zero confirmation or positive disconfirmation receive WOMAP information after a purchase, which indicates that others’ perceived performance is
better than theirs, they might generate inequitable feelings. Given that inputs (e.g., price paid and efforts) are equal, these consumers would perceive their output (perceived performance)/input (e.g., price paid) ratios to be disproportionately lower than others’ and then perceive inequity. Through an airline service in their study, Fisk and Young(1985) have demonstrated that consumers who perceived inequity after making comparisons with other buyers would feel dissatisfied. As a result, the inequitable feelings resulting from receiving WOMAP information
might have a negative influence on satisfaction.
H3a: When experiencing positive disconfirmation, customers who received WOMAP after a
H3b: When experiencing zero disconfirmation, customers who received WOMAP after a
purchase are less satisfied than those who did not receive any WOM.
Consumers with negatively disconfirmed expectancy would feel dissatisfied because perceived performance is worse than their expectations. Since they have learned that others have the same experiences as theirs from WOMCP information, it is likely that this product failure
would be attributed to firm-related responsibility (which contains locus and controllability (Tsiros, Mittal, & Ross, 2004)) and occurring frequently. Due to the diagnosticity of WOM, when consumers understand it is not only themselves who encountered this situation, they would think the product failure is firm-related, controllable for the firm and stable. Prior research supports that the product failure attributed to firm-related responsibility and a frequent event would induce angry reactions (Folkes, 1984; Folkes, Koletsky, & Graham, 1987). Oliver (1993) suggests that negative affects resulting from such attributions would reduce customer satisfaction. Thus, negatively disconfirmed consumers might reduce their initially formed satisfaction again after receiving WOMCP information.
H4a: When experiencing negative disconfirmation, customers who received WOMCP after a
purchase are more dissatisfied than those who did not receive any WOM.
If negatively disconfirmed consumers find others’ perceived product performances are better than theirs from WOMAP information, they might generate the additionally inequitable
H4b: When experiencing negative disconfirmation, customers who received WOMAP after a
purchase are more dissatisfied than those who did not receive any WOM.
Figure2. 2 Research Framework
Disconfirmation
Customer
Satisfaction
Post-purchase
WOM
Information
H1 H2-H4CHAPTER 3: METHODOLOGY
3.1 Research Produce
Figure3. 1 Research Procedure
Fabricate a hostel as the stimulus.
Decide how to manipulate disconfirmation and post-purchase WOM information.
Collect relevant scales and develop a questionnaire for this research.
Due to a 3*3 factorial design, it requires 30 subjects for each cell. Consequently, the study requires a total of 270 subjects.
Adopt a convenient student sample.
Conduct a pretest.
According to the results of the pretest, modify the questionnaire.
Assign the subjects to nine cells randomly and give each of them an appropriate
questionnaire.
Instruct the participants to finish the questionnaire.
Select a stimulus
Variables manipulations
Determine the sample size Develop a questionnaire
Select a sampling technique
Pretest
Modify the questionnaire
Execute the sampling process
3.2 Stimulus
This study was made up using a hostel in Yi-Lan as the stimulus of the experiment. Churchill and Surprenant (1982) suggested that when conducting an experiment of
disconfirmation, it is better that participants did not have any preconceived notions of what the product would deliver. Due to the hostel being fabricated by the author, the participants would not have any expectations before the experiment. This gives the advantage of facilitating the manipulation of expectations. Furthermore, the quality of hostels in Taiwan varies and even the same hostel might not maintain a consistent performance. It is acceptable that different
consumers have different performance perceptions and then deliver different WOM information.
3.3 Manipulations
3.3.1 Disconfirmation
Disconfirmation levels were manipulated indirectly through the manipulations of expectations and performance because the construct is derived from the comparison between expectations and performance. Although pragmatic limitations dictated that disconfirmation could not be manipulated independently, it was possible to obtain independent measures of disconfirmation which could then be used in the analysis (Churchill & Surprenant , 1982). In this study, expectations and performance were manipulated through two dimensions, service
encounters and suite quality. Three different website messages about the hostel’s suite and three different telephone records about reserving a suite were provided to set up three levels of
expectation about the hostels (high, moderate and low) (appendix 1). To generate different performance perceptions, the study designed three scenarios to make participants imagine the actual experiences in the hostel and then created three levels of performance (high, moderate and
low) (appendix 2). Although the full combinations are nine situations (3 expectation levels*3 performance levels), too much situations would increase the complexity of group comparisons. Finally, the study combined high expectations with low performance to form negative
disconfirmation, moderate expectations with moderate performance to form zero disconfirmation, and low expectations with high performance to form positive disconfirmation.
3.3.2 Post-purchase WOM
According to each level of manipulative performance, the study designed two articles on a BBS (Bulletin Board System) site about others’ experiences of visiting this hostel (appendix 2). To generate WOMCP information, one article was framed as the manipulated performance which
the subjects perceived, and to generate WOMAP information, the other was framed to make the
participants think others’ experiences were better than their perceived performance. Therefore, the study generated three levels of post-purchase WOM (no WOM, WOMCP, and WOMAP). The
condition with no WOM might serve as preliminary satisfaction after purchase. It could be the baseline compared with WOMCP and WOMAP conditions to test whether people would modify
their satisfaction judgments after contacting post-purchase WOM information.
3.4 Experimental Procedure
The study used a 3*3 between-participant factorial design. Three levels of disconfirmation (positive disconfirmation, zero disconfirmation and negative disconfirmation) and three levels of post-purchase WOM information (no WOM, WOMCP and WOMAP) were manipulated and
asked to role-play the situation because in an experimental setting, it is difficult to provide an actual hostel service experience.
According to the factorial design of the experiment, nine versions of internet
questionnaires were developed (appendix 3). At first, all participants would be given the url of the research website (appendix 4) to the section of the instructions and expectations manipulation. Participants were asked to imagine that they were planning to have a trip with a friend to Yi-Lan the following weekend and hence they had to seek a hostel to stay in for one night. They were provided with the website messages of the hostel’s suites and then listened to a conversation about reserving a suite between a clerk and a guest to imagine they were the guest to form expectations about the suite quality and hostel service. After the expectations manipulation, they were led to the internet questionnaire to complete the following manipulations and measurements.
First, participants were given two questions about the content to ensure they had not skipped the conversation, and then a manipulation check to assess their level of expectations. Second, they were asked to read a scenario to imagine the performance of the hostel they encountered. They were then asked to complete the performance and disconfirmation
measurements. After the disconfirmation manipulation check, some participants directly filled out the satisfaction scale (didn’t receive WOM information), others read WOMCP or WOMAP
information on a BBS site and then completed the WOM perception check and the satisfaction measurement. Finally, all participants had to answer some basic questions, such as gender, age, education and so on.
3.5 Sampling Plan
In Taiwan, there are more and more WOM information delivered on the Internet. This study attempted to take online WOM information on BBS sites as an example to understand the effect of post-purchase WOM on satisfaction. For the consideration of cost and feasibility, the study used the most popular BBS site, PTT, as the medium to deliver the internet questionnaires. 27 billboards (e.g., online shopping, shopping, customers, hotel, traveling, and outdoors billboard) which are more likely to contain WOM information about products, services, hotels and trips were picked to post the hyperlink of the research website on them. The reason to select these billboards is that their browsers might have high involvement of WOM information and that might strengthen the effect of post-purchase WOM manipulation.
The 9 versions of internet questionnaires were randomly assigned to the 27 billboards (each version was posted on three billboards). An article in which contained the introduction of the research, the url of the research website and the researcher’s name and institute would be posted on each billboard, and then it would be promised that every participant would be rewarded 200 PTT virtual dollars after completing the experiment to enhance the motivation of
participation.
3.6 Measures
3.6.1 Expectations
Part items of the scale (item 1 ~item 5) were developed for the study conducted by Voss, Parasuraman, and Grewal (1998). The items are intended to measure some quality-related aspects of a service provider thought to be true by a potential customer prior to actually making the purchase and/or receiving the service. The items are most appropriate for a hotel but might need
to be adjusted for other types of service providers. Further, they are stated tentatively since respondents would not have actually visited the service provider at the time they completed the scale. The remaining items of the scale (item 6 ~item 9) were developed for the present research to measure the expectations of the suite quality. The complete scale adopted by this research is shown below. Cronbach’s α in the pretest is 0.976.
If I were to stay at this hostel (hotel)1:
1. the hostel (hotel)1 would offer excellent overall service.
2. the hostel (hotel)1 would offer accurate and dependable service. 3. the employees would be courteous at all time.
4. the employees would provide prompt assistance.
5. the employees would provide personal, individualized attention. 6. the hostel would offer a excellent quality suite.
7. the equipments and materials in the suite would be complete. 8. the suite provided by the hostel would be clean.
9. the suite provided by the hostel would be comfortable.
Responses to all items were made using a seven-point scale anchored by definitely would not (1) and definitely would (7).
3.6.2 Performance
Part items of the scale (item1~item5) were developed for the study conducted by Voss, Parasuraman, and Grewal (1998). The items are intended to measure a customer’s quality-related perceptions of a service provider after making a purchase and/or receiving the service. The items are most appropriate for a hotel but might be adjusted for other types of service providers. The remaining items of the scale (item6~item9) were developed for the present research to measure the perceptions about the suite quality. The complete scale adopted by this research is shown below. Cronbach’s α in the pretest is 0.983.
1. The service provide by this hostel (hotel)1 was: very low quality/ very high quality
2. The service provided by this hostel (hotel)1 was: unreliable/ reliable
3. The hostel’s (hotel’s)1 employees were: discourteous/ courteous
4. The hostel’s (hotel’s)1 employees were: not helpful/ helpful
5. The hostel’s (hotel’s)1 employees were: uncaring/ caring
6. The suite provide by the hostel was: very low quality/ very high quality
7. The equipments and materials provided by the hostel were: incomplete/ complete
8. The suite provided by the hostel was: unclean/ clean
9. The suite provided by the hostel was: uncomfortable/ comfortable
Responses to all items were made using a seven-point scale using the anchors shown.
1 the original term in Voss, Parasuraman, and Grewal’s study (1998)
3.6.3 Disconfirmation
The method for establishing the disconfirmation scale originated from Oliver’s suggestions (1997). According to the attributes of expectations and performance scale, the study developed a nine-item measurement to capture the subject’s summary judgment of disconfirmation on a “better than expected-worse than expected” scale. This kind of disconfirmation scale has been adopted by much previous research (e.g., Oliver, 1980; Churchill & Surprenant, 1982). The complete scale adopted by this research is shown below. Cronbach’s α in the pretest is 0.988.
1. The quality of the service provide by this hostel was: 2. The reliability of the service provide by this hostel was: 3. The courtesy of the hostel’s employees was:
4. The helpfulness of the hostel’s employees was:
5. The degree that the hotel’s employees cared for my needs was: 6. The quality of the suite provided by the hostel was:
8. The cleanliness of the suite provided by the hostel was: 9. The comfort of the suite provided by the hostel was:
Responses to all items were made using a seven-point anchored by “much worse than expected” (1), “as expected” (4) and “much better than expected” (7).
3.6.4 Satisfaction
Part items (item1~item3) of the scale were original to the study by Voss, Parasuraman, and Grewal (1998). The items are intended to measure the degree to which a customer of a service provider is satisfied with a service that has been experienced or received. The remaining items of the scale (item4~item6) were developed for the present research to measure the perceptions about the suite quality. The complete scale adopted by this research is shown below. Cronbach’s α in the pretest is 0.966.
1. I was satisfied with the service provided. 2. I was delighted with the service provided. 3. I was unhappy with the service provided. 4. I was satisfied with the suite provided. 5. I was delighted with the suite provided. 6. I was unhappy with the suite provided.
Responses to all items were made using a seven-point Likert-type scale using the following anchors: disagree very strongly (1) and agree very strongly (7).
3.6.5 Perceived Performance Discrepancy with Others
This study developed two items for the purpose of testing the success of post-purchase WOM manipulation. The items are shown below. Cronbach’s α in the pretest is 0.827. 1. I thought the quality of the suite others had experienced was:
2. I thought the service quality others had experienced was:
Responses to all items were made using a five-point anchored by “much worse than I experienced” (1), “as I experienced” (3) and “much better than I experienced” (5).
3.7 Pretest
Prior to the actual study, 10 student samples for each condition were collected (total 90 samples) to examine the reliability of the scale and the manipulations, and discover any problems or misunderstanding of the questions and the design of the questionnaire. After discussions with several participators, some instructions and several questions about individual basic data were modified. No recommendations revealing the main measures of the study were misunderstood. The results of the pretest are presented in appendix 5. The measurements are reliable (Cronbach’s α >0.7). According to the LSD tests, the manipulation of expectations, performance and
disconfirmation are successful. The disconfirmation value in zero disconfirmation is not significantly different from the midpoint, 4 (p=0.120). The value (perceived performance discrepancy with others) in WOMCP is not significantly different from the midpoint, 3. The value in WOMAP is significantly bigger the midpoint, 3. Therefore, the manipulation of post-purchase WOM is successful.
CHAPTER 4: RESULTS
4.1 Background of Participants
Table 4.1 demonstrates the characteristics of the participants. The total sample is 274 participants. From it, 53.6% are female, 70.1% live in northern Taiwan, 84.7% are aged 20-29 years old, 76.6% are students, 81% have a College/Bachelor’s degree, and 80.3% have a
disposable income below NT30,000. Also, 89.1% have the experiences of visiting a hostel, 96% have the habit of searching for product information on the Internet, and 99.6% have the habit of using BBS.
4.2 Reliability Analysis
The reliability of the data is tested with Cronbach’s α. If Cronbach’s α is above 0.7, the study is accepted as reliable. Table 4.2 demonstrates the values from reliability tests of five constructs. The result of the reliability test indicates that the measurements are reliable (all above 0.7).
4.3 Manipulation Checks
Three one-way ANOVAs and multiple comparisons with an LSD test were conducted to test the success of the expectations, performance, and disconfirmation manipulations. From table 4.3, expectations are the highest in high expectation situation followed by moderate expectation situation and then followed by low expectation situation. From table 4.4, performance is the highest in high performance situation followed by moderate performance situation and then followed by low performance situation. From table 4.5, disconfirmation is the highest in positive
disconfirmation followed by zero disconfirmation and then followed by negative disconfirmation. As a result, the manipulation of expectations, performance and disconfirmation are successful. The manipulation of WOMCP is tested to see whether the value was equal to the midpoint, 3 and
the manipulation of WOMAP is tested whether the value was bigger than 3. Table 4.5 shows that
Table4. 1 Frequency Distribution of Sample Characteristics
Category Subjects Percentage(%)
Gender Male 127 46.4% Female 147 53.6% Total 274 100% Residence North 192 70.1% Middle 38 13.9% South 42 15.3% Others 2 0.7% Total 274 100% Age 15-19 32 11.7% 20-24 174 63.5% 25-29 58 21.2% 30-34 10 3.6% Total 274 100% Occupation Students 210 76.6% Others 64 23.4% Total 274 100%
Education Degree College / Bachelor’s 222 81%
Master’s degree 52 19% Total 274 100% < NT 10,000 157 57.3% NT 10,001~20,000 63 23% NT 20,001~30,000 28 10.2% > NT 30,001 26 9.5% Disposal Income (per month) Total 274 100% Yes 244 89.1% No 30 10.9% Experience of Visiting a Hostel Total 274 100% Yes 263 96% No 11 4%
The Habit of Searching for Product Information on
the Internet Total 274 100%
Yes 273 99.6%
No 1 4%
The Habit of Using BBS
Table4. 2 Reliability
Construct Item Cronbach’s α
Expectations 9 0.979
Performance 9 0.981
Disconfirmation 9 0.986
Satisfaction 6 0.973
Perceived Performance Discrepancy with Others 2 0.776
Table4. 3 Manipulation Check for Expectations Dependent Variable: Expectations
Level of Expectations N Mean Standard Deviation
Low 96 2.0995 0.79201
Moderate 93 4.0848 0.99210
High 85 5.2784 0.86876
F(2, 271) = 298.532 (p <0.001)
LSD Testa : High > Moderate*** , Moderate > Low*** , High > Low***
a ***p < 0.001
Table4. 4 Manipulation Check for Performance Dependent Variable: Performance
Level of Performance N Mean Standard Deviation
Low 85 2.1725 0.85286
Moderate 93 4.6965 0.82583
High 96 5.3900 1.04903
F(2, 271) = 301.253 (p < 0.001)
LSD Test a : High > Moderate*** , Moderate > Low*** , High > Low***
a
Table4. 5 Manipulation Check for Disconfirmation Dependent Variable: Disconfirmation
Level of
Disconfirmation N Mean Standard Deviation
Negative 85 1.9242 0.70110
Zero 93 4.2330 0.85630
Positive 96 5.6030 0.88944
F(2, 271) = 455.344 (p < 0.001)
LSD Test a : Positive > Zero*** , Zero > Negative*** , Positive > Negative***
a ***p < 0.001
Table4. 6 Manipulation Check for Post-purchase WOM
Test Value = 3 a
N Mean Standard
Deviation t-value p-value
WOMCP 89 3.0169 0.27644 0.575 0.567
WOMAP 94 3.8511 0.70646 11.680 0.000
a
representing that subjects think other’s experienced performance the same as theirs
4.4 Hypothesis Tests
After confirming the reliability and manipulations of the analyzed data, the study presented the descriptive statistics of satisfaction across situations (table 4.7) and proceeded to conduct ANOVA and multiple comparisons to test the hypotheses.
Table4. 7 Descriptive Statistics for Satisfaction
Mean
(Standard Deviation) (N)
NO WOM WOMCP WOMAP
Marginal Mean for Disconfirmati-on Negative Disconfirmation 1.9321 (0.85767) (27) 2.2778 (0.88661) (30) 2.2738 (0.93757) (28) 2.1667 0.89863 (85) Zero disconfirmation 4.0806 (0.91679) (31) 4.6494 (0.68479) (29) 4.4697 (0.71862) (33) 4.3961 0.80754 (93) Positive Disconfirmation 5.0354 (0.85163) (33) 5.5889 (0.62166) (30) 4.6818 (0.81582) (33) 5.0868 0.85086 (96)
Marginal Mean for Post-purchase WOM 3.7894 1.54397 (91) 4.1667 1.58692 (89) 3.8901 1.33793 (94) 3.9465 1.49418 (274)
4.4.1 The Effects of Disconfirmation and Post-purchase WOM on Satisfaction
In order to figure out satisfaction differences across situations, this study conducted a two-way ANOVA to test the influences of disconfirmation and post-purchase WOM on satisfaction and the results are shown in table 4.8 (ANOVA table) and table 4.9 (test assumptions of error). The main disconfirmation effect (F(2, 265) =313.791, p<0.001), the main WOM effect (F(2, 265)
are significant. Levels of disconfirmation, post-purchase WOM and their interaction would result in satisfaction differences.
Figure4. 1 The Interaction between Disconfirmation and Post-purchase WOM
5.5889 5.0354 4.6818 4.0806 4.6494 4.4697 2.2738 1.9321 2.2778 0 1 2 3 4 5 6 NO WOMCP WOMAP Mean (Satisfaction)
Positive Disonfirmation Zero Disconfirmation
Negative Disconfirmation
Through multiple comparisons with the LSD test (table 4.10), it could be found the mean satisfaction in positive disconfirmation is significantly higher than zero disconfirmation (p<0.001) and negative disconfirmation (p<0.001), and the mean satisfaction in zero disconfirmation is significantly higher than negative disconfirmation (p<0.001). H1 ( satisfaction differences among disconfirmation situations are as follow: positive disconfirmation > zero disconfirmation > negative disconfirmation) is supported. Satisfaction will increase as situations transform from negative disconfirmation to zero disconfirmation, and then to positive disconfirmation. The result is consistent with previous research. Further, the LSD test (table 4.11) shows that only
(p<0.01). There are no significant difference between WOMAP and no WOM (p=0.402).
Therefore, taking satisfaction with no WOM as preliminary satisfaction, customers who received WOMCP after a purchase would increase their satisfaction. However, due to the significant
interaction between disconfirmation and post-purchase, the study would do profile analysis to examine satisfaction differences at each disconfirmation level in the following sections.
Table4. 8 Two-way ANOVA Dependent Variable: Satisfaction
Source Type III Sum of Squares Df Mean Square F p-value Model 433.252 a 8 54.157 81.431 <0.001 Disconfirmation 417.379 2 208.689 313.791 <0.001 WOM 11.576 2 5.788 8.703 <0.001 Disconfirmation x WOM 8.745 4 2.186 3.287 0.0119 Error 176.241 265 .665 Corrected Total 609.493 273 a R2 = 0.711 (Adjusted R2 = 0.702)
Table4. 9 Assumption Tests of Error for ANOVA Tests for Normality of Error
Test Statistic p-value
Shapiro-Wilk W 0.993221 0.2509 Kolmogorov-Smirnov D 0.036942 >0.1500 Cramer-von Mises W-Sq 0.051772 >0.2500 Anderson-Darling A-Sq 0.38286 >0.2500
Levene's Test of Equality of Error Variances
F df1 df2 p-value
1.228 8 265 0.283
Table4. 10 LSD Test for Disconfirmation on Satisfaction Dependent Variable: Satisfaction
Disconfirmation N Mean Standard Deviation
Negative 85 2.1667 0.89863
Zero 93 4.3961 0.80754
Positive 96 5.0868 0.85086
LSD Test a : Positive > Zero*** , Zero > Negative*** , Positive > Negative***
a
***p<0.001
Table4. 11 LSD Test for Post-purchase WOM on Satisfaction Dependent Variable: Satisfaction
WOM N Mean Standard Deviation
No WOM 91 3.7894 1.54397
WOMCP 89 4.1667 1.58692
WOMAP 94 3.8901 1.33793
LSD Test a :
WOMCP > No** (p=0.002), WOMCP > WOMAP * (p=0.023), No= WOMAP (p=0.402)
a
***p<0.001, **p<0.01, *p<0.05
4.4.2 The Effect of Post-purchase WOM on Satisfaction in Negative Disconfirmation
To confirm H4a (when experiencing negative disconfirmation, customers who received WOMCP after a purchase are more dissatisfied than those who did not receive any WOM ) and
H4b (when experiencing negative disconfirmation, customers who received WOMAP after a
one-way ANOVA at negative disconfirmation (table12). The main WOM effect is not significant (F(2, 82) =1.36, p>0.05). H4a and H4b are not supported. In negative disconfirmation, there are no
significant satisfaction differences among different WOM levels, namely, post-purchase WOM does not influence satisfaction. The mean plot is shown in figure 4.2.
Table4. 12 Profile Analysis in Negative Disconfirmation Dependent Variable: Satisfaction
Disconfirmation Post-purchase WOM N Mean Deviation Standard
Negative No WOM 27 1.9321 0.85767
WOMCP 30 2.2778 0.88661
WOMAP 28 2.2738 0.93757
F(2, 82) =1.36, p=0.2625
R2 = 0.032 (adjusted R2 = 0.008)
Figure4. 2 Mean Plot of Satisfaction in Negative Disconfirmation
1.9321 2.2778 2.2738 1 2 3 NO WOMCP WOMAP Mean (Satisfaction) Negative Disconfirmation
4.4.2 The Effect of Post-purchase WOM on Satisfaction in Zero Disconfirmation
Table 4.13 and figure 4.3 present the effect of post-purchase WOM on satisfaction in zero disconfirmation. Post-purchase WOM has a significant influence on satisfaction (F(2, 90) =4.203,
p<0.05). Satisfaction with WOMCP is significantly higher than no WOM (p<0.01). H2b (when
experiencing zero disconfirmation, customers who received WOMCP after a purchase are more
satisfied than those who did not receive any WOM) is supported. However, satisfaction with WOMAP is also significant higher than no WOM (p<0.05). The direction is contrary to H3b
(when experiencing zero disconfirmation, customers who received WOMAP after a purchase are
less satisfied than those who did not receive any WOM). Therefore, H3b is not supported. Customers who received WOMCP or WOMAP after a purchase would feel more satisfied.
Table4. 13 Profile Analysis in Zero Disconfirmation Dependent Variable: Satisfaction
Disconfirmation Post-purchase WOM N Mean Deviation Standard
Zero No WOM 31 4.0806 0.91679
WOMCP 29 4.6494 0.68479
WOMAP 33 4.4697 0.71862
F(2, 90) =4.203, p=0.018, R2 = 0.085 (adjusted R2 = 0.065)
LSD Test:
Figure4. 3 Mean Plot of Satisfaction in Zero Disconfirmation 4.0806 4.6494 4.4697 3.6 3.9 4.2 4.5 4.8 NO WOMCP WOMAP Mean (Satisfaction) Zero Disconfirmation
4.4.3 The Effect of Post-purchase WOM on Satisfaction in Positive Disconfirmation
Table14 and figure 4.4 show the ANOVA results, and these indicates significant
differences among different levels of post-purchase WOM (F(2, 93) =10.902, p<0.001). Moreover,
the LSD test was used to inspect differences between each two groups. As predicted by H2a, participants exposed to WOMCP after a purchase had higher satisfaction than those not exposed to
any WOM (mean=5.589 and 5.0354, respectively; p<0.01). As predicted by H3a, participants exposed to WOMAP after purchase felt less satisfied than those not exposed to any WOM
Table4. 14 Profile Analysis in Positive Disconfirmation Dependent Variable: Satisfaction
Disconfirmation Post-purchase WOM N Mean Standard Deviation Positive No WOM 33 5.0354 0.85163 WOMCP 30 5.5889 0.62166 WOMAP 33 4.6818 0.81582 F(2, 93) =10.902, p<0.001, R2 = 0.190 (adjusted R2 = 0.173) LSD Test:
WOMCP > NO (p=0.006) , WOMAP < NO (p=0.067), WOMC P > WOMAP (p<0.001)
Figure4. 4 Mean Plot of Satisfaction in Positive Disconfirmation
5.0354 5.5889 4.6818 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 NO WOMCP WOMAP Mean (satisfaction) Positive Disconfirmation
4.5 Results of the Tested Hypotheses
Table4. 15 Results of the Hypotheses
Hypotheses Description of the Hypotheses Results
Hypothesis 1 Satisfaction differences among disconfirmation situations are as follow:
positive disconfirmation > zero disconfirmation > negative disconfirmation.
Supported
Hypothesis 2a When experiencing positive disconfirmation, customers who received WOMCP after a purchase are more satisfied
than those who did not receive any WOM.
Supported
Hypothesis 2b When experiencing zero disconfirmation, customers who received WOMCP after a purchase are more satisfied than
those who did not receive any WOM.
Supported
Hypothesis 3a When experiencing positive disconfirmation, customers who received WOMAP after a purchase are less satisfied
than those who did not receive any WOM.
Supported
Hypothesis 3b When experiencing zero disconfirmation, customers who received WOMAP after a purchase are less satisfied than
those who did not receive any WOM.
Not Supported
Hypothesis 4a When experiencing negative disconfirmation, customers who received WOMCP after a purchase are more
dissatisfied than those who did not receive any WOM.
Not Supported
Hypothesis 4b When experiencing negative disconfirmation, customers who received WOMAP after a purchase are more
dissatisfied than those who did not receive any WOM.
Not Supported
CHAPTER 5: DISCUSSION AND IMPLICATIONS
This chapter is constructed by four parts. The first part will be the discussion of the research results. Following that will be the implications and limitations of the research. Finally, several suggestions for further research will be introduced.
5.1 Discussion
First, as stated in much previous research (e.g., Churchill & Surprenant, 1982; Mooradian & Olver, 1997; Oliver, 1977, 1980), this research again confirms the effect of disconfirmation on satisfaction through a hostel service, including product experience (a suite) and service
experience (treatment by employees). Disconfirmation will positively influence satisfaction. Positively disconfirmed customers have the highest satisfaction followed by confirmed customers, and then negatively disconfirmed customers who have the lowest satisfaction. To satisfy
customers, performance must be at least consistent with previous expectations. Therefore, when firms invest huge resources in marketing communications, they have to ensure the product or service quality can correspond to what they say to consumers. Second, previous research has supported the idea that people would modify their satisfaction judgments after they interact with other group members (Bohlmann et al., 2006). Further, this research demonstrates interpersonal influence induced by merely receiving post-purchase WOM also results in changes in satisfaction judgments. WOM not only has an effect on pre-purchase attitudes (Herr et al., 1991) and post-usage perceptions (Bone, 1995) but also makes customers change their satisfaction evaluations. Once customers receive WOM after a purchase, they will change their preliminary satisfaction evaluations formed by discrepancies between expectations and performance. However, after a deep analysis, its effect depends on disconfirmation situations.
