購買前搜尋與持續性搜尋對購買後認知失調的影響 - 政大學術集成

全文

(1)國立政治大學資訊管理學系. 碩士學位論文指導教授：管郁君博士. 立. 政治大. ‧ 國. 學. 購買前搜尋與持續性搜尋. ‧. sit. y. Nat. 對購買後認知失調的影響. n. er. io. Investigating the Effect of Pre-Purchase Search and a l on Post-Purchase v Dissonance Ongoing Search i n Ch engchi U. 研究生：郭家瑋中華民國一○四年六月. 1.

(2) 誌謝研究所對於我，只是逃避畢業即失業的避風港，原本並不期待在新的環境中得到甚麼驚喜。但是我錯了，這是不輸大學四年的精采兩年。首先感謝管老師，帶給我許多引導與啟發。從初次遇見老師起，就被老師與眾不同的思維所吸引，一直到現在都還是。這兩年待在老師身邊，聽老師說故事，聽老師談研究，真的讓我重新思考很多事。老師對我的肯定也大大的影響了我對自己未來的想像，希望之後能夠有能力走上那條路。再來感謝淑瓊學姊與國華學長。謝謝您們每周都來參加 meeting，給我們. 政治大. 許多寶貴的指導。特別感謝淑瓊學姐，帶著我見識國際研討會。我們一起探索. 立. 新加坡，一起品嘗肉骨茶的時光，我會永遠懷念。. ‧ 國. 學. 也非常感謝遠道而來進行口試的勝為學長與善斌老師，感謝您們細細閱讀. 上一層樓。. ‧. 我們的論文。您們給予的意見都非常寶貴，帶給我很多啟發，也讓這篇研究更. sit. y. Nat. io. er. 也特別感謝同 lab 的佳宏跟瑞祥。每次我都在死線前極限脫出，帶給你們不少麻煩，但是你們一直都只有包容沒有抱怨，超級好的。. n. al. Ch. engchi. i n U. v. 最後是一起在 lab 度過快樂時光的所有學長學姊同學學弟學妹們。感謝你們幫助我在這個新環境，找到新的歸屬感。能在政大遇見各位，是我的榮幸。. 一路走來，除了感謝，還是感謝，謝謝你們。. 郭家瑋謹誌於政大資管中華民國一零四年十二月 2.

(3) 摘要本研究旨在了解不同類型的資訊搜尋對購買後的心理狀態的影響。首先，我們探討消費者的產品知識如何影響他們的搜尋量，包括線下購買前搜尋、線上購買前搜尋和持續性搜尋，再來探討這些搜尋行為如何影響消費者的購買後認知失調。本研究共收集了 542 份有效問卷，我們使用 PLS 結構方程模型來驗證研究模型。研究結果顯示，消費者產品知識對線下購買前搜尋、線上購買前搜尋和持續性搜尋有正向影響，對購買後認知失調有負向影響。在三種類型的搜尋中，只有線上購買前搜尋對購買後認知失調有負向影響。購買後搜尋對購買後失認知調. 政治大期望本研究的發現能為他們未來的研究帶來啟發與幫助。立. 亦有負向影響。對消費者搜尋行為以及消費者搜尋後心理狀態感興趣的研究者們，. ‧ 國. 學. 關鍵詞：消費者產品知識、資訊搜尋、持續性搜尋、購買前搜尋、購買後搜尋、. ‧. 購買後認知失調. n. er. io. sit. y. Nat. al. Ch. engchi. 3. i n U. v.

(4) Abstract This study aimed to understand how different types of information search influences consumers’ post-purchase mental state. First we investigated how consumers’ product knowledge influence their amount of information search, including offline prepurchase search, online pre-purchase search, and ongoing search. Then we investigate how these searching behaviours influence consumers’ post-purchase dissonance. Total of 542 valid online questionnaires were collected and PLS structural equation model was used to verify our research model. The result showed that consumer product. 政治大 search, ongoing search, and had negative influence toward post-purchase dissonance. 立 knowledge had positive influence on offline pre-purchase search, online pre-purchase. ‧ 國. 學. Among three types of information search, only online pre-purchase search had negative influence on post-purchase dissonance. Post-purchase search also had negative. ‧. influence on post-purchase dissonance. For those who are interesting in consumers’. sit. y. Nat. searching behaviours and post-search state of mind, these findings could be referenced. io. n. al. er. in their future research.. Ch. engchi. i n U. v. Keywords: consumer product knowledge, information search, ongoing search, prepurchase search, post-purchase search, post-purchase dissonance.. 4.

(5) INDEX. 1.1. Background and Motivation .......................................................................... 9. 1.2. Research Objectives ..................................................................................... 11. 1.3. Research Questions ...................................................................................... 12. LITERATURE REVIEW ..................................................................................... 13 2.1 2.1.1. Types of product class knowledge............................................................ 13. 2.1.2. Prior product knowledge and information search ................................... 14. 2.1.3. Negative knowledge-search relationship ................................................. 14. 2.1.4. Positive knowledge-search relationship .................................................. 15. 2.1.5. The inverted-U shaped knowledge-search relationship .......................... 16. 立. 政治大. 學. 2.2. Information Search....................................................................................... 17 External and internal information search ............................................... 17. 2.2.2. Pre-purchase search ................................................................................ 17. 2.2.3. Ongoing search ........................................................................................ 18. 2.2.4. Differences between pre-purchase search and ongoing search .............. 19. y. sit. al. Buyer Regret and Cognitive Dissonance ..................................................... 20. v i n Buyer regret ............................................................................................. 20 Ch engchi U Cognitive dissonance ............................................................................... 21 n. 2.4. io. 2.3.2. Nat. 2.3.1. ‧. 2.2.1. 2.3. 3. Consumer’s Product Knowledge ................................................................. 13. er. 2. INTRODUCTION .................................................................................................. 9. ‧ 國. 1. Post-purchase search .................................................................................... 22. METHODOLOGY ............................................................................................... 24 3.1. Research model ............................................................................................ 24. 3.2. Hypotheses ................................................................................................... 24. 3.3. Questionnaire Design ................................................................................... 30. 3.3.1 3.4. Product studied ........................................................................................ 30 Operational Definitions................................................................................ 31. 3.4.1. Consumers’ product knowledge............................................................... 31. 3.4.2. Pre-purchase search ................................................................................ 33 5.

(6) 3.4.3. Ongoing search ........................................................................................ 34. 3.4.4. Post-purchase dissonance and post-purchase search ............................. 35. 3.5 3.5.1. Testing Results of KMO and Bartlett's test .............................................. 39. 3.5.2. Factor analysis and total variance explained.......................................... 40. STATISTICAL ANALYSIS AND HYPOTHESES TESTING .......................... 42 4.1. Data Collection ............................................................................................ 42. 4.2. Sample Structure Analysis ........................................................................... 42. 4.3. Reliability and Validity Testing ................................................................... 45. 4.5. 治政大 Measurement Model Analysis ..................................................................... 49 立 Structural Model Analysis ........................................................................... 53. 4.6. Results of Hypothesis Testing ..................................................................... 55. 4.7. Discussions .................................................................................................. 60. ‧. Influence of consumers’ product knowledge on post-purchase dissonance. y. Nat. 60 4.7.2. 學. 4.7.1. sit. 4.4. Factor analysis, KMO, Bartlett's test and total variance explained........ 45. Influence of consumers’ product knowledge on online pre-purchase,. io. er. 4.3.1. ‧ 國. 4. Pretesting...................................................................................................... 38. offline pre-purchase, and ongoing search ............................................................ 60. al. n. 4.7.3. Ch. i n U. v. Influence of online pre-purchase, offline pre-purchase, and ongoing. engchi. search on post-purchase dissonance .................................................................... 61 4.7.4. Moderating effect of post-purchase search on post-purchase dissonance. and its antecedents................................................................................................ 61 5. CONCLUSION AND SUGGESTIONS .............................................................. 63 5.1. Conclusion ................................................................................................... 63. 5.2. Limitations and Future Research ................................................................. 64. Appendix ...................................................................................................................... 65 Reference ..................................................................................................................... 69. 6.

(7) TABLE CONTENTS Table 3-1. Subjective Knowledge Items ...................................................................... 32 Table 3-2. Measure of Consumers’ Product Knowledge ............................................. 32 Table 3-3. Pre-purchase Search Scale .......................................................................... 33 Table 3-4. Measure of Pre-purchase Search ................................................................ 34 Table 3-5. Multi-item Indices of Ongoing Search ....................................................... 35 Table 3-6. Measure of Ongoing Search ....................................................................... 35 Table 3-7. POSTDIS: Factor 1 “Correctness” ............................................................. 36 Table 3-8. Measure of Post-Purchase Dissonance ....................................................... 37 Table 3-9. POSTDIS: Factor 2 “Support”.................................................................... 37 Table 3-10. Measure of Post-Purchase Dissonance ..................................................... 38 Table 3-11. Results of KMO and Bartlett's test ........................................................... 40 Table 3-12. Total Variance Explained of all constructs................................................ 40 Table 3-13. Rotated Component Matrix of PPS_ON................................................... 41 Table 4-1. Demographic information of collected samples ......................................... 43 Table 4-2. Results of Exploratory Factor Analysis ...................................................... 45 Table 4-3. Reliability Statistics and Item-Total Statistics ............................................ 47 Table 4-4. Latent variables of measuring model .......................................................... 50 Table 4-5. Discriminate validity & latent variable correlations ................................... 52 Table 4-6. Path coefficient, t-value, p-value ................................................................ 54 Table 4-7. Results of hypothesis testing ...................................................................... 59. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 7. i n U. v.

(8) FIGURE CONTENTS Figure 3-1. Research model ......................................................................................... 24 Figure 3-2. Conceptual graph of knowledge-search relationship ................................ 26 Figure 3-3. Research model with hypotheses .............................................................. 30 Figure 4-1. Structural model and hypotheses .............................................................. 53 Figure 4-2. Result of structural model ......................................................................... 55. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 8. i n U. v.

(9) 1. INTRODUCTION. 1.1. Background and Motivation In order to make a good purchase, people in the past searched through reading. magazines, advertisements or asking friends to collect the needed information. In the era of web 2.0, there are numerous information available online. Information search becomes easier and convenient than in the past. People no longer just check the product information on TV ads or on magazines. Nowadays people can easily get more information using internet. Such as browsing official web sites, forum, blog, social. 政治大. networks or microblogs. Using consumer electronics as example, there are lots of. 立. unboxing report and performance tests available online. If people want to get. ‧ 國. 學. information from interpersonal relationship, using communication software or applications are the way people do other than making a phone call. These are the new. ‧. methods people use to collect information. However, even it’s so easy to get. y. Nat. sit. information, there are people only search a little before purchase. Some people even. n. al. er. io. don’t search at all just make the purchase right away. Why would this happen?. Ch. engchi. i n U. v. Consumer’s searching behaviors had been a hot topic during the past decades. Even there were many researchers investigated in this field already, consumer’s information search had not been thoroughly explored yet. There are still lots of unknown findings waiting for us to explore. Especially with progress of internet and communication technologies, consumers’ searching behavior had changed significantly. With this background, we can raised up unprecedented new topics or reexamine the past studies. The research findings might be surprising.. In marketing, literatures proposed that pre-purchase search could help consumers 9.

(10) to make a good purchase and increase purchase satisfaction (Punj & Staelin, 1983), or help consumer to solve the purchase problem (Bloch 1986). This was almost the common sense in marketing (Keaveney, Huber, & Herrmann, 2007). In fact, psychologists provided different point of view. Thinking about other alternatives may trigger the thinking that lead consumers to experience regret (Kahneman & Miller, 1986). Indeed, recently there was study pointing out that the more consumers search, the more likely they will suffer from regret (Keaveney et al., 2007). There’s also study showing that cognitive dissonance after impulse buying was lower than non-impulse. 政治大 disconfirmation of expectations (George & Yaoyuneyong, 2010). 立. buying. Such result was due to the strategy to avoid the surprises arising out of possible. ‧ 國. 學. Cognitive dissonance had been a popular topic among marketing and consumer. ‧. psychology researchers for many years. Many of the researchers had investigated the. sit. y. Nat. dissonance after purchase decision. Quite a few researchers investigated the effect of. io. er. information search on post-purchase regret. In our study, not only the relationship between information search and post-purchase dissonance, we also want to probe the. al. n. v i n C hfrom ongoing search different impacts on the dissonance and pre-purchase search. engchi U. About consumer information search, researchers focused more on pre-purchase search in previous studies. Literatures mention about ongoing search are far less than pre-purchase search. Same as post-purchase dissonance after ongoing search. Past studies only investigated how pre-purchase search influence post-purchase dissonance. If we don’t search at all before purchase, just buy with existing knowledge, or purchase without any prior knowledge (i.e. impulse buying) would we still going to experience cognitive dissonance? This is a new topic and has not been studied before. Would consumer’s prior product knowledge affect post-purchase search? Does consumer 10.

(11) search after purchase and how does it affect post-purchase state of mind? These are the issues we want to investigate in this study.. 1.2. Research Objectives From research background, we can see that searching extensively may not provide. consumer greater satisfaction, sometimes search might bring negative impact to consumers buying experience, e.g. post-purchase regret and dissonance. Pre-purchase search made up a great portion in information search literatures, but ongoing search. 政治大. was hardly mentioned since its publication. Ongoing search was also a crucial part of. 立. consumer information search. Only examining the consequence of pre-purchase search. ‧ 國. 學. was insufficient. Thence, we are going to investigate the effect of ongoing and prepurchase search on dissonance together. There was study showing that post-purchase. ‧. external search would mediate the relationship between post-purchase dissonance and. Nat. sit. y. post-purchase satisfaction. Hence how post-purchase search affect post-purchase. n. al. er. io. dissonance is one of our concerns. Following are our research objectives:. i n U. v. 1. Investigating the impact of consumer product knowledge on post-purchase cognitive dissonance.. Ch. engchi. 2. Investigating how consumer product knowledge affect consumer searching behaviors on pre-purchase search and ongoing search. 3. Investigating how pre-purchase search and ongoing search affect post-purchase dissonance.. 11.

(12) 1.3. Research Questions 1.. Can pre-purchase search provides consumers with a better mental state than ongoing search after purchase?. 2.. Would not searching better than searching?. 3.. What are people looking for in the post-purchase search?. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 12. i n U. v.

(13) 2. LITERATURE REVIEW. 2.1. Consumer’s Product Knowledge. 2.1.1. Types of product class knowledge Based on Brucks’ review of previous studies, consumer’s product class knowledge. fall in to three categories: objective knowledge, subjective knowledge and the experience with the product (Brucks, 1985). The first category is objective knowledge. It’s defined as what an individual actually knows or what is actually stored in the memory (Kanwar, Olson, & Sims, 1981; Russo & Johnson, 1980; Staelin, 1978). The. 政治大. second is the subjective knowledge, which is an individual’s perception of how much. 立. s/he knows (Gardner, 1984; Park & Lessig, 1981). The third is the amount of purchasing. ‧ 國. 學. or usage experience with the product (Monroe, 1976; Marks & Olson, 1981). Before further discussion, we first eliminate the third category, which is the purchase/usage. ‧. experience. Different people learn different things from similar experience. Even. y. Nat. io. sit. people have experience the same thing, they are likely to behave differently. The effect. n. al. er. of experience toward search behaviors is inconsistent. Moreover, if consumer does not. i n U. v. have any purchase or usage experience with the product, then nothing can be measure.. Ch. engchi. Therefore, experience-based knowledge is not considered as a good measure of product knowledge.. The measures of objective knowledge and subjective knowledge are conceptually and operationally distinct (Brucks, 1986). Since it’s difficult for people to perceive how much they actually know precisely, the differences between objective knowledge and subjective knowledge are predictable. Subjective knowledge can provide better understanding of decision maker’s systematic bias and heuristics (i.e. cognitive shortcuts) than objective knowledge (Park & Lessig, 1981). Furthermore, measures of 13.

(14) subjective knowledge can represent one’s self-confidence as well as one’s objective knowledge (Park & Lessig, 1981). Perceived self-confidence is influential to one’s decision making behaviors. One lack of self-confidence in knowledge may conduct more or less searching activities in despite of one’s actual knowledge level.. Thought. it has not been empirically demonstrated, self-confidence is closely related to one’s decision making abilities (Brucks, 1986). In contrast, objective knowledge only reflects an individual’s actual knowledge level. Despite measuring subjective knowledge is easier than objective knowledge operationally, subjective knowledge indeed is a better predictor of behavior.. 立. Prior product knowledge and information search. ‧ 國. 學. 2.1.2. 政治大. In the field of consumer information search, how prior product knowledge affect. ‧. the amount of external information search was many researchers’ research subject.. Nat. sit. y. Previous research findings related to the relationship between consumers’ prior product. n. al. er. io. knowledge and information search were inconsistent. The research results could be. i n U. v. roughly categorized into three types, which are positive, negative, and inverted U-. Ch. engchi. shaped relationship. Every research finding was supported by reasonable explanations. Till now the relationship between prior knowledge and search remains unclear.. 2.1.3. Negative knowledge-search relationship A number of studies showed that product knowledge and amount of information. search is in negative relationship (Anderson, Engledow & Becker, 1979; Katona & Mueller, 1955; Moore & Lehmann, 1980; Newman & Staelin, 1971, 1972; Swan, 1969; Punj & Staelin, 1983; Beatty & Smith, 1987). That is the higher the knowledge they have, the less likely to search the product. For this result, one of the explanation was 14.

(15) the de-motivation effect (Fiske, Luebbehusen, Miyazaki, & Urbany, 1994). Knowledgeable consumer already has enough knowledge, so they don’t need to acquire more information before purchase (Punj & Staelin, 1983). Since these consumers perceive themselves having sufficient product knowledge, usually they are confident to their purchase decision as well (Bloch, 1986). In contrast, consumer lack of product knowledge would often lack of confidence to their knowledge and decision. This type of consumers are likely to engage information search (Bloch, 1986). Therefore the negative knowledge-search relationship was caused. Another explanation was that. 政治大 information, thence they search more efficiently. They abbreviate the searching process 立 knowledgeable consumers search more efficient. They know what is the useful. so they can use less time to find out the information they need. Since this kind of people. ‧ 國. 學. can search within a short time, if we use search time to measure the amount of. y. sit. al. n. Positive knowledge-search relationship. Ch. engchi. er. io. 2.1.4. Nat. (Brucks, 1985).. ‧. information search, the knowledge-search relationship would appear to be negative. i n U. v. At the same time, some researchers postulated a positive relationship between product knowledge and external search (Brucks, 1985; Srinivasan & Ratchford, 1991; Selnes & Troye, 1989; Moore & Lehmann, 1980). Positive relationship reflects the facilitating effect of knowledge (Fiske et al., 1994), which means prior knowledge can improve the efficiency of information collection (Alba & Hutchinson, 1987; Chi, 1981; Punj & Srinivasan, 1989) and encourage more information search (Johnson & Russo, 1984; Bettman & Par, 1980; Brucks, 1985; Punj & Staelin, 1983). Comparing with common people, consumers with better understanding of the product can use lower cognitive cost to process information (Alba & Hutchinson, 1987; Fiske et al., 1994). 15.

(16) “To ask a question, one must know enough to know what is not known” (Miyake & Norman ,1979). Consumers with more product expertise also know more about what to ask and where to ask. Prior knowledge allow them to formulate more questions (Brucks, 1985). New information brings more benefit to the knowledgeable consumers, and encourage them to pursue more. In contrast, consumers lack of prior knowledge would not know how to start then only execute small amount of search (Brucks & Schurr, 1990).. 2.1.5. 政治大. The inverted-U shaped knowledge-search relationship. 立. There existed some researches balance the above contrary findings by finding the. ‧ 國. 學. non-linear inverted-U shaped relationship between prior knowledge and information search level (Bettman & Park, 1980; Hempel, 1969; Johnson & Russo, 1984). Inverted-. ‧. U shape expresses a positive knowledge-search relationship from low-to-moderate. Nat. sit. y. knowledge level, and a negative relationship from moderate-to-high knowledge level.. n. al. er. io. It describes consumers lack of product knowledge don’t know how to start and where. i n U. v. to search therefore seek less information, leading to the positive relationship. In contrast,. Ch. engchi. consumers with product expertise, even they don’t pay much effort on searching, they can still make good purchase decision. The inverted-U shaped knowledge-search relationship gives possible explanation for the previous discrepancy, because when a linear line is applied on inverted-U shaped data, a positive or negative relationship, or no relationship at all may occur (Brucks, 1985). Inverted-U shaped knowledge-search relationship makes the divergent results from previous studies seem reasonable. Noted that some research experimentally tested the inverted-U shaped relationship but was not supported by the result (Brucks, 1985). A few studies demonstrated U-shaped relationship (Rao & Sieben, 1992), no relationship (Bennett & Mandell, 1969; Claxton, 16.

(17) Fry, & Portis, 1974), or positive and negative relationship both exist but cancel out each other (Srinivasan & Ratchford, 1991).. Evidently, there existed great disparity among former studies. Given each study had reasonable explanation to their outcomes, the knowledge-search relationship remains uncertain and controversial. However, reexamine this relationship is not our purpose. We want to explore the knowledge-search relationship in different aspect that is the influence of knowledge on pre-purchase search and ongoing search. Maybe it. 政治大. could give new interpretation of the knowledge-search relationship.. 立 Information Search. 2.2.1. External and internal information search. ‧. ‧ 國. 學. 2.2. There are two phases of information search: internal information search and. y. Nat. sit. external information search (Spreng, 1996; Engel, 1990). Internal information search. n. al. er. io. refers to the acquisition of information available in the memory (Bettman, 1979),. i n U. v. whereas external is the acquisition of information outside of memory, such as friends,. Ch. engchi. in-store displays, advertisements, magazine etc. (Bettman, 1979). When the information in memory is deficient, consumers would change from internal to external search by seeking the information outside of memory. Among the external search, we use purchase intention and readiness to purchase to differentiate the external search into two types: pre-purchase search and ongoing search.. 2.2.2. Pre-purchase search. Base on Kelly’s definition, pre-purchase search is ‘‘information seeking and processing activities which one engages in to facilitate decision making regarding some 17.

(18) goal object in the marketplace’’ (Kelly, 1968). The concept is that consumer conducting a search activity in order to solve a specific purchase problem and to make a wise choice. In the view of this perspective, pre-purchase search was the primary focus of the past search-related studies (Claxton, Fry & Portis, 1974; Furse, Punj & Steward, 1984; Newman, 1977; Punj & Staelin, 1983), until the emergence of ongoing search provided a different view of external search.. 2.2.3. Ongoing search. 政治大. The concept of ongoing search was first raised by Bloch at 1983, he further defined. 立. ongoing search and proposed a framework of consumer search at 1986. As we’ve. ‧ 國. 學. mentioned before, many studies about consumer search were focusing on pre-purchase search, but Bloch postulated that ‘‘an orientation focusing solely on pre-purchase search. ‧. was deficient and unable to account for search activity that is recreational or that occurs. Nat. sit. y. without a recognized consumption need’’. When people are conducting window. n. al. er. io. shopping, reading magazines, browsing websites, or chatting with friends, they may not. i n U. v. have any plan to purchase, yet are still obtaining information. If these sources of. Ch. engchi. information are not included in knowledge search, consumers’ actual knowledge level would be underestimate. Since this phenomenon was not included in the traditional search theories, to fill the gap exists in the theories, Bloch defined ongoing search as “the search activities that are independent from specific purchase needs or decisions”. Consumers do not always search to solve a recognized and immediate purchase problem, sometimes they do not have urgent demand of the product or lack of financial buying ability, the intention to buy is absent. They simply search for fun and for the preparation of future purchase (Bloch, 1986).. 18.

(19) 2.2.4. Differences between pre-purchase search and ongoing search According to Bloch’s consumer information search framework, pre-purchase. search and ongoing search are conceptually different in determinants, motives and outcomes. The most remarkable difference is the search motive. The main motive for the pre-purchase search is to enhance the quality of purchase outcome (Punj & Staelin, 1983, Bloch, 1986), whereas the motive for the ongoing search are pleasure and construction of information bank for future use (Bloch, 1986). In fact, it’s hard to distinguish them in practice. Although we use purpose to differentiate pre-purchase. 政治大 Sometimes consumers do not intent to buy at the beginning, but the purchase intension 立. search and ongoing search, often people just buy in the absence of purchase intension.. arise during the ongoing searching process then they would make the purchase.. ‧ 國. 學. Ongoing search suddenly turns out to be pre-purchase search in this case. Impulse. er. io. sit. y. Nat. them from external observation (Furse, Punj, & Stewart, 1984).. ‧. buying blurs the boundary between the two. Furthermore it is difficult to distinguish. Past literature about knowledge-search relationship usually referred external. al. n. v i n C hstudies didn’t evenUmention. Nevertheless, for our search to pre-purchase search. Some engchi research purpose, we have to distinguish the two. Searching purpose is the main thing differentiate ongoing search and pre-purchase search. We use buying intention and buying ability to approximate the purpose. If consumers are ready to buy and search to. make better decision, then they are conducting pre-purchase search. On the other hand, if consumers merely search for fun, or currently lack of ability to buy, then it’s classified as ongoing search (Bloch, 1986). Actually, some researchers had already proposed similar concept way before Bloch. Claxton, Fry and Portis (1974) stated that information search is a continuous process. People continue to search even the purchasing plan is absent. So when they come to make a purchase, relatively little 19.

(20) explicit search is needed (Claxton, Fry and Portis, 1974). That is, under the exposure of information, consumers may already have enough knowledge to make decision. We can see pre-purchase search comprise only portion of consumers’ information search, yet ongoing search has been excluded from external search in many researched. If ongoing search is also part of external search, then only investigate pre-purchase search is insufficient. Ongoing search is as crucial as pre-purchase search. The importance of ongoing search should not be ignored.. 政治大. 2.3. Buyer Regret and Cognitive Dissonance. 2.3.1. Buyer regret. 立. ‧ 國. 學. Regret, an unpleasant feeling we all familiar with, has defined as a cognitively based emotion that motivates one to think about how the negative event could have. ‧. happened and how one could change it, or how one could prevent its future occurrence. y. Nat. sit. (Zeelenberg, 2007). We often confront regret after we made an important decision and. n. al. er. io. perceive loss of better opportunities. Comparing with other emotions, regret is a rather. i n U. v. complex emotional experience, that feeling regret requires the ability to imagine other. Ch. engchi. possibilities than the current state of the world (Zeelenberg, 2007). In past decades, several major literatures about regret were mainly from three areas: cognitive psychology, economic and operation research (Kahneman & Tversky, 1982; Loomes, & Sugden, 1982, Bell, 1982). Now it’s an important issue in marketing and consumer behavior domain. Many marketing researchers were enthusiastic about regret theory, because they were aware of the harmful effect in consumers’ subsequent purchase decision (Engel, Blackwell & Miniard, 1995; Tsiros, 1998). Regret is not only a reaction to bad decision outcomes or processes, but also a powerful force in motivating and giving direction to behavior (Zeelenberg, 2007), buyer’s regret after bad purchase 20.

(21) may contribute significant impact toward the service/product providers (Keaveney et al., 2007). Quite a lot of studies were focusing on the consequences of buyer regret. Buyer regret can cause dissatisfaction (Cooke, 2001; Inman, 1997; Taylor, 1997; Tsiros, 1998; Tsiros & Mittal, 2000; Zeelenberg & Pieters, 1999, 2004), switch of service (Zeelenberg & Pieters, 1999, 2004) and affect the intention to repurchase the same brand (Keaveney et al., 2007). These insightful findings contributed by marketing researchers show their passion toward this topic, and also the importance of this topic.. 2.3.2. Cognitive dissonance. 立. 政治大. Cognitive dissonance is a social psychology theory developed by Festinger (1957).. ‧ 國. 學. Festinger described dissonance as a psychological uncomfortable state due to the inconsistency between attitude and behavior, or the contradictory among values or. ‧. beliefs that motives people reduce this inconsistency (Festinger, 1957). His theory. Nat. sit. y. claimed that individuals always seek consistency between their expectations and their. n. al. er. io. reality, so when people are experiencing mental discomfort, they are motivated to. i n U. v. reduce the inconsistency. For instance, when people are facing contradictions, they may. Ch. engchi. justify their behavior to be consistent with cognition, or justify their cognition in order to rationalize their behaviors. This definition also points out that post-purchase dissonance does not necessarily lead to bad purchase experience. Sometimes dissonant consumer just relieved from dissonance by convinced themselves that they have made the right decisions. Nevertheless, post-purchase dissonance contains uncertainty and unknown consequences. How to prevent consumer post-purchase dissonance is still an important issue to study.. In marketing and consumer psychology, cognitive dissonance was usually referred 21.

(22) as post-purchase dissonance, sometimes also referred as buyer’s remorse. Regret and remorse are synonyms and in some literatures they were regarded as the same thing (Rosenzweig & Gilovich, 2011), but we want to clarify that dissonance and remorse should not be considered as the same concept. We can see that both of these terms often appear in the research discussing post purchase state of mind. Both of them are the unpleasant feelings that people experience after a bad purchase. However, by definition we’ve mentioned above, they are totally different things. Regret is an emotion but dissonance is a psychological state. Comparing with regret, dissonance is more. 政治大 with several other kind of emotions, dissonance does not necessarily experience with 立. complex and obscure. Although dissonance might include a bit of regret accompanied. regret. So in this study, we will not use the term buyer’s remorse to avoid confusion and. ‧ 國. ‧. Post-purchase search. sit. y. Nat. 2.4. 學. misleading.. n. al. er. io. In some studies, post-purchase search had been found to reduce dissonance and its. i n U. v. negative influences (Fischer, Greitemeyer & Frey, 2008; Holbrook, Berent, Krosnick,. Ch. engchi. Visser & Boninger, 2005; Mittal, Kumar & Tsiros, 1999; Schiffman & Kanuk, 2004; Keng & Liao, 2009). On the other hand, some researches demonstrated that dissonant consumers need reassurance to make sure they have made the wise decisions (Donnelly & Ivancevich, 1970; Jacoby et al., 1987). Dissonant consumers are irritated to ask their family and friends to back up decisions (Holloway, 1967; Straits, 1964; Oshikawa, 1970). They also obtain support from advertising which give consumers greater value to the purchase (Engel, 1963). We can see that dissonant consumers try to acquire support through post-purchase search within interpersonal sources and media. Not surprisingly, based on these literatures there’s study seen post-purchase search as part 22.

(23) of dissonance (Montgomery & Barnes, 1993). It’s reasonable to say when people are looking for support to their purchase, they are confronting post-purchase dissonance. This phenomenon is also consistent with the original definition of cognitive dissonance. By definition, dissonant people are eager to get rid of the uncomfortable feeling. Seeking reassurance after purchase could be seen as a way to reduce the dissonance. However, we want to separate dissonance and post-purchase search into two parts. Consumers who conduct post-purchase search are likely to experience dissonance, but consumers may still experience dissonance even without post-purchase search. When. 政治大 mediating effect of post-purchase search. We want to measure the dissonance on the 立 we measure the dissonant level of the two, the result might not be accurate due to the. same basis. So we separate post-purchase search and dissonance, and examine how. ‧ 國. 學. post-purchase search influence post-purchase dissonance after ongoing search and pre-. ‧. purchase search.. n. er. io. sit. y. Nat. al. Ch. engchi. 23. i n U. v.

(24) 3. METHODOLOGY. 3.1. Research model Based on Keaveney’s model of buyer regret (Keaveney et al., 2007), we. simplified the model (see Fig. 3-1) and add three new constructs to verify the following research purposes:. 立. 政治大. ‧. ‧ 國. 學. n. al. 3.2. Hypotheses. Ch. engchi. er. io. sit. y. Nat Figure 3-1. Research model. i n U. v. By conventional wisdom, it’s not difficult to imagine that people with more product expertise would experience less post-purchase dissonance. Since they have better understanding of product. Even they know many alternatives, they can still make their purchase decision without hesitation. This assumption is supported by literature, but in different aspect. As we choose perceived subjective knowledge to measure the product knowledge, we are not only examining the relationship between knowledge and dissonance, but the relationship between confidence and dissonance. Subjective knowledge can imply one’s knowledge level as well as one’s self-confidence. Perceived 24.

(25) self-confidence is influential to decision making abilities (Park & Lessig, 1981) and it can serve as a predictor of dissonance (Bell, 1967; Montgomery & Barnes, 1993). Consumers with high self-confidence are hard to influence and more certain to their decision. In contrast, consumers with less confidence are easier to be persuaded and have some doubt with their purchase decision (Bell 1967). Yet, consumers who are confident with their decision making ability and greater commitment to their decisions are found to be experience less dissonance. Based on these arguments, we hypothesize: H1: Product knowledge is negatively related to post-purchase dissonance.. 政治大 As mentioned in literature review, relationship between product knowledge and 立. amount of external information search remains unclear. Some researchers proposed a. ‧ 國. 學. positive relationship and some proposed negative. There were also researchers. sit. y. Nat. boundary of positive and negative relationship is obscure.. ‧. demonstrating an inverted U-shaped relationship. Assuming they were correct, the. io. er. Based on Bloch’s research, ongoing search is related to the enduring involvement, and heavy ongoing searchers are likely to be heavy spenders. Consumers who conduct. al. n. v i n ongoing search mostly are theC people that interestedUin the product class and they hengchi acquire product expertise through their pleasurable searching. Thus we speculate these. heavy ongoing searchers have high possibility to be knowledgeable consumers. These knowledgeable consumers continue to search on regular basis due to their passion with the product class. However, once they need to make the purchase, pre-purchase search might not be essential because the product information bank is already built. On the other hand, consumers those do not have passion for the product, they don’t search in regular basis since they are not interested in the product. But when they need to make the purchase, they search to ensure a good decision. Due to these assumptions, we propose knowledge has positive influence toward ongoing search and has negative 25.

(26) influence toward pre-purchase search. Past studies did not distinguish pre-purchase search and ongoing search. Thus we speculate that the inconsistence results of knowledge-search relationship from past studies might due to the difference of search. This might also be the main cause that contributed to the inverted-U shaped relationship. The critical point of inverted U-shaped might be the intersection point of ongoing and pre-purchase search (see Figure 3-2).. 立. 政治大. ‧. ‧ 國. 學 er. io. sit. y. Nat. al. n. Figure 3-2. Conceptual graph of knowledge-search relationship. Ch. engchi. i n U. v. Now is the era of Internet, consumers’ searching behavior often occur on the internet, but sometimes still occur with absence of internet. The difference between online and offline searching behaviors is also an interesting topic. Thus we divided prepurchase search into online and offline pre-purchase search. We intended to compare the influence between these two. However, we didn’t intend to differentiate ongoing search. Since ongoing search is an information gathering activity from variety of sources, it already include online and offline search. Noting that product knowledge and information search are influencing each other. So in this study we are going to use cross sectional research method. We want to capture 26.

(27) consumer’s state of mind after purchase and treat information search as independent variable. Hence we come out with following hypotheses: H2a: Product knowledge is negatively related to online pre-purchase search. H2b: Product knowledge is negatively related to offline pre-purchase search. H3: Product knowledge is positively related to ongoing search.. In Keaveney, Huber and Herrmann’s (2007) research, they’ve mentioned that consumers are acquiring product information, as well as other available alternatives. 政治大 other possible alternatives would evoke people’s counterfactual thinking that leads to 立. through the pre-purchase search process. In psychology’s perspective, thinking about. regret (Kahneman & Miller, 1986). After purchase, by comparing with the unchosen. ‧ 國. 學. alternatives, even consumers have made appropriate decision, they might still feel. ‧. inadequate. When recalling the unchosen alternatives, customers may imagine the. sit. y. Nat. unchosen alternatives could have brought them better outcomes, then they start to feel. io. er. regret. The more one seriously considers other options, the more likely for one to experience regret (Keaveney et al., 2007).. al. n. v i n Cdissonance Though regret and cognitive two different concepts, we suspect the U h e n gare i h c. circumstances that cause regret have high probability to cause dissonance as well. Since. dissonance is the contradiction between beliefs, knowing there exists a better option other than the choice have made, certainly qualify the condition of dissonance. Therefore we assert that search is followed by dissonance. In Keaveney’s study (Keaveney et al., 2007), ongoing search had not been discussed at all. The relationship between ongoing search and regret remains unclear. Thus it is one of our research subject. As long as consumers have made comparison between alternatives, there are possibility to experience dissonance. However, we speculate that the dissonance comes with ongoing search would not as severe as the 27.

(28) dissonance comes with pre-purchase search. For the consumers that conduct ongoing search more than pre-purchase search (or do not conduct pre-purchase at all), they build their product information bank through ongoing search and use their accumulated product knowledge to make their choices. This group of consumers might not pay much effort on comparing the alternatives before purchase. They already know what to buy base on their internal information. So we assume people who conduct ongoing search more than pre-purchase search will experience less dissonance. As we’ve mentioned in previous hypothesis, Bloch’s research result demonstrated that heavy ongoing searchers. 政治大 awareness and frequent contact with retailers. So we expect people who conduct 立 are likely to be heavy spenders and heavy spenders are likely to have high new product. ongoing search heavily to be more knowledgeable customers and more knowledgeable. ‧ 國. 學. consumers may experience less post-purchase dissonance. This provide another cue to. ‧. our assumption. Thus we hypothesize that:. sit. y. Nat. H4a: Online pre-purchase search is positively related to post-purchase dissonance.. io. er. H4b: Offline pre-purchase search is positively related to post-purchase dissonance. H5: Ongoing search is positively related to post-purchase dissonance.. n. al. Ch. engchi. i n U. v. Previous studies supported that dissonant consumers tend to conduct postpurchase search to support their decisions (Schiffman & Kanuk, 2004; Fischer et al., 2008; Holbrook et al., 2005; Mittal et al., 1999). During their searching process, if customers only search and recognize the information that conform to their purchase decisions, then they are having confirmation bias toward their searches. Confirmation bias refers to the situation that consumers “seek confirmatory evidence and fail to search for disconfirming information for desired outcomes or strongly held values” (Carter, Kaufmann & Michel, 2007). In other words, when the information they found contradict their beliefs, which are their purchase decisions, they either ignore them or 28.

(29) interpret them in their desired way. In this case, consumers’ confidence toward their decisions are being enhanced and that can ease the dissonance if they have any. We can see that it’s possible for consumers to relieve their dissonance via biased post-purchased search, and biased post-purchased search might also ease the dissonance caused by knowledge level and searching. Hence we propose the following hypotheses: H6: Post-purchase search is negatively related to post-purchase dissonance. H7: The relationship between product knowledge and post-purchase dissonance is moderated by post-purchase search. H8: The relationship between product information search and post-purchase dissonance. 政治大. is moderated by post-purchase search.. 立. dissonance is moderated by post-purchase search.. 學. ‧ 國. H8a: The relationship between online pre-purchase search and post-purchase. H8b: The relationship between offline pre-purchase search and post-purchase. ‧. dissonance is moderated by post-purchase search.. y. Nat. n. al. Ch. engchi. er. io. moderated by post-purchase search.. sit. H8c: The relationship between ongoing search and post-purchase dissonance is. i n U. v. Completed research model with hypotheses are shown in Figure 3-3.. 29.

(30) 政治大. 立. Figure 3-3. Research model with hypotheses. ‧ 國. 學. 3.3. Questionnaire Design. ‧. This study is going to be a quantitative research. We plan to verify our model by. y. Nat. io. sit. collecting and analyzing online survey data. In this section, we will give operational. n. al. er. definitions to each construct. Based on literatures, we are also going to develop the. Ch. i n U. v. questionnaire by introducing existing scales and modified version of those scales. All. engchi. items employed a 5-point Likert-type response format where 1 = strongly disagree and 5 = strongly agree.. 3.3.1. Product studied. Usually there are three conditions need to be satisfied for dissonance to occur. The decisions have to be important, irrevocable, and voluntary (Cummings & Venkatesan, 1976; Oliver, 1997; Souter & Sweeney, 2003). Other than that, dissonance is more likely to happen due to the purchase of durable goods, given the costs are likely to be 30.

(31) higher and the decisions are relatively irreversible (Menasco & Hawkins, 1978; Oliver, 1997; Souter & Sweeney, 2003). To survey the product knowledge acquiring from searching, product’s available information must be high so a relatively wide range of ongoing search could be observed (Bloch, 1986) as well as the pre-purchase search. Based on these criteria, we consider consumer electronics (e.g. computers, tablets, smartphones, cameras…etc.) to be our studied product class. Electronics are relatively costly compare to other daily durable goods. And after the refund period, the purchased product cannot return, which means irreversible. There are also lots of electronic related. 政治大 Furthermore, for the purpose of measuring post-purchase dissonance, at the beginning 立 information available online. Consumers can easily acquire needed information.. of the questionnaire, we are going to ask the respondents to recall the last time they. ‧ 國. 學. purchase electronic products, then fulfill the survey. Consumer electronics exist. ‧. everywhere in our lives and innovated rapidly. Hence we expect respondents will have. sit. y. Nat. electronic purchasing experience not long ago. Some previous studies related to. io. er. dissonance and external search also use electronic products as their study class (Bloch, 1986; Brucks & Schurr, 1990; Keng & Liao, 2009; Beatty & Smith, 1987). Thus we. al. n. v i n C his a fair product class conclude that consumer electronics to study. engchi U 3.4. Operational Definitions. 3.4.1. Consumers’ product knowledge. Brucks (1985) had categorized consumer knowledge into three groups: objective knowledge, subjective knowledge, and experience. Subjective knowledge is what the consumer thinks (s)he knows, and it is also the better predictor to consumers’ decision making behaviors hence we choose it to be our focus. Flynn & Goldsmith (1999) redefined subjective knowledge as “a consumer’s perception of the amount of 31.

(32) information they have stored in their memory”. The definition and measure they used to describe consumer’s knowledge not only include the subjective knowledge but also the knowledge associated with the product class. Flynn & Goldsmith (1999) had developed a 5-items subjective product class knowledge scale with seven-point Likerttype response format. In their scale development, they use fashion clothing as their product category. We adjust the items by replacing fashion clothing with consumer electronic. Table 3-1. Subjective Knowledge Items. 治政大 I know pretty much about fashion clothing. 立 about fashions. I do not feel very knowledgeable Items. 1. 2.. Reference. 4. Compared to most other people, I know less about fashion clothing.. io. sit. y. Nat. 5. When it comes to fashion, I really don’t know a lot.. ‧. ‧ 國. 學. 3. Among my circle of friends, I’m one of the “experts” on fashion clothing.. Flynn & Goldsmith 1999. A consumer’s perception of the amount of product class knowledge they have stored in their memory. al. n. Operational Definition. er. Table 3-2. Measure of Consumers’ Product Knowledge Item #. Ch. i n U. v. Items. engchi. CPK1. I know pretty much about consumer electronics.. CPK2. I do not feel very knowledgeable about consumer electronics.. CPK3. Among my circle of friends, I’m one of the “experts” on consumer electronics.. CPK4. Compared to most other people, I know less about consumer electronics.. CPL5. When it comes to consumer electronics, I really don’t know a lot.. 32.

(33) 3.4.2. Pre-purchase search Pre-purchase search is the search activity that consumers’ conduct in order to solve. their immediate purchase problem and to make a wise choice. To check whether a consumer has search extensively and exhaustively, Patwardhan and Ramaprasad (2005) had provide a good criteria. They specified online pre-purchase search is the use of Internet for acquiring and evaluating product, price, service and other purchase related information before the purchase. We cited Patwardhan and Ramaprasad’s 8-items online pre-purchase scale. Since we intend to compare the difference between online. 政治大 Internet, then ask respondents to answer their search amount in online and offline 立. and offline pre-purchase search, we adjusted the items into pre-purchase search without. 學. Table 3-3. Pre-purchase Search Scale Items. Reference. ‧. ‧ 國. separately.. I visit Web sites to check out the best deals.. 2.. I search for detailed information about the brand or product category.. 3.. I compare several brands online before making a decison.. 4.. I check out relevant Web ads to get more information about a product.. 5.. I check out Web sites for sales and service information.. 6.. I look for product information that is specific to my requirements.. 7.. I check out company information online for products I would like to buy.. 8.. I look for online discounts and bargains.. n. al. er. io. sit. y. Nat. 1.. Ch. engchi. 33. i n U. v. Patwardhan & Ramaprasad 2005.

(34) Table 3-4. Measure of Pre-purchase Search Operational Definition. Items. PPS1 I check out the best deals before purchase. PPS2 I search for detailed information about the brand or product category before purchase. PPS3 I compare several brands before making a decision before purchase. PPS4 I check out relevant ads to get more information about a product before purchase. PPS5 I check out sales and service information before purchase.. 政治大. PPS6 I look for product information that is specific to my requirements before purchase.. 立. 學. PPS7 I check out company information for products I would like to buy before purchase.. ‧ 國. Pre-purchase search is a process of acquiring and evaluating product information, price, services and other purchase related information before the purchase.. Item #. PPS8 I look for discounts and bargains before purchase.. ‧. 3.4.3. Ongoing search. sit. y. Nat. io. er. Ongoing search is the searching activity without purchase intention. Consumers gather product information on a regular basis from variety of sources merely for. al. n. v i n C h (1986) developedUa 4-items scale based on his pleasure and for future use. Bloch engchi descriptions of ongoing search. The items assess level of ongoing search from three. aspects: retail settings, mass media and interpersonal sources. However, at present people tend to search more on internet rather than traditional media. There are mass of electronic related information available online, such as official web pages, product reviews, open box video, buying guides, etc. Internet domain is so important, it should be included in order to have precise measure of consumers’ ongoing search level. So we create an extra item in the aspect of searching online.. 34.

(35) Table 3-5. Multi-item Indices of Ongoing Search Items. Reference. 1. Often visit clothing stores or departments, just to look around or get information rather than to make a specific purchase.. Bloch, Sherrell & 2. Often discuss with retail salespersons to get information during Ridgway (1986) store visit. 3. Often read fashion magazines to get information about trends/styles. 4. Often talk to friends or acquaintances to get information or advice concerning clothing and clothing styles. Table 3-6. Measure of Ongoing Search Operational Definition. Items 政治大. 立 I often visit electronic stores or departments, just to. OGS1. look around or get information rather than to make a specific purchase.. ‧ 國. 學. I often discuss with retail salespersons to get information during store visit.. OGS3. I often read electronics related magazines to get information about new products and technologies.. io. n. al. OGS5. I often talk to friends or acquaintances to get information or advice concerning electronic products.. er. OGS4. sit. y. Nat. OGS2. ‧. The searching activity without purchase intention, merely for pleasure and for future use.. Item #. v i n CI often h echeck i U information online, n g couthelectronics. such as official web sites, product reviews, unboxing video/reviews, buying guides, performance tests, etc.. 3.4.4. Post-purchase dissonance and post-purchase search. Cognitive dissonance is the psychological discomfort due to the inconsistency between attitude and behavior, or the contradictory among values or beliefs, and people are motivated to relieve this uncomfortable state. Cognitive dissonance accompany with a purchase decision is so called the post-purchase dissonance. Post-purchase 35.

(36) dissonance appears when buyers doubt whether they have made the correct decisions (Montgomery & Barnes, 1993). These dissonant buyers often express anxiety and uncertainty toward the purchases and need their decisions to be supported (Montgomery & Barnes, 1993). Montgomery and Barnes (1993) introduced a comprehensive 10items post-purchase dissonance scale. The scale contains two parts: “correctness”, and the “support” to the decision. Items from “correctness” of the decision are generated to reflect the expected satisfaction, anxiety, uncertainty, and forced compliance to the purchase, and items from “support” are generated to reflect the tendency of consumers. 政治大 indicate the doubt and discomfort brought by dissonance and the intention to ease the 立 to seek support from interpersonal or other source of information. These two parts. discomfort respectively. As we discuss in literature review, post-purchase search and. ‧ 國. 學. dissonance are treated as two different constructs. Therefore we apply “correctness” on. ‧. post-purchase dissonance, and “support” on post-purchase search. The scale was. sit. y. Nat. developed in 1993. At the time internet was not popular among households. Thus we. io. n. al. er. add an extra item to measure the level of post-purchase search on internet.. Ch. i n U. v. Table 3-7. POSTDIS: Factor 1 “Correctness”. engchi. Items. 1. I feel that I will happy with the purchase I have just made. 2. I’m sure that I’ll be pleased with the way this product performs. 3. I’m comfortable with the purchase I’ve just made. 4. I’m uneasy about the purchase decision that I just made. 5. I’m confident that I’ve made the “right” choice when I purchased this product. 6. I would probably purchase this product again in the future. 7. I feel that I got a “good deal” when I purchase this product. 8. This product will probably do a good job of meeting my important needs. 36. Reference Montgomery & Barnes 1993.

(37) Table 3-8. Measure of Post-Purchase Dissonance Operational Definition The expected satisfaction, anxiety, uncertainty, and forced compliance that consumers experience after the purchase.. Item #. Items. PPD1. I feel that I will happy with the purchase I have just made.. PPD2. I’m sure that I’ll be pleased with the way this product performs.. PPD3. I’m comfortable with the purchase I’ve just made.. PPD4. I’m uneasy about the purchase decision that I just made.. PPD5. I’m confident that I’ve made the “right” choice when I purchased this product.. PPD6. I would probably purchase this product again in the future.. 政治大. 立 I feel that I got a “good deal” when I purchase this. PPD7. ‧ 國. PPD8. 學. product.. ‧. This product will probably do a good job of meeting my important needs.. y. Nat. Items. al. er. io. sit. Table 3-9. POSTDIS: Factor 2 “Support”. v ni. Montgomery & Barnes 2. I would probably pick up a copy of Consumer Reports to make 1993 sure the product or brand I just bought received high ratings.. n. 1. I’ll probably talk to my friends or family to ask them if they think I’ve made a wise choice with my purchase.. Reference. Ch. engchi U. 37.

(38) Table 3-10. Measure of Post-Purchase Dissonance Operational Definition. Item #. Search within interpersonal sources, media, and online information sources after purchase.. PPS1. I’ll probably talk to my friends or family to ask them if they think I’ve made a wise choice with my purchase.. PPS2. I would probably pick up a copy of Consumer Reports to make sure the product or brand I just bought received high ratings.. PPS3. After purchase, I check out the product information online, such as official web sites, product reviews, unboxing video/reviews, buying guides, performance tests, etc.. 立. 政治大. Pretesting. ‧ 國. 學. 3.5. Items. Since our study attempted to explore the difference between consumers’ online. ‧. and offline pre-purchase search, we expected respondents to have some online. Nat. sit. y. searching ability. Thus our target respondents were the internet users. We used online. n. al. er. io. survey to eliminate non-internet users, assuming the respondents that access the survey were familiar with internet.. Ch. engchi. i n U. v. To prevent invalid samples caused by ambiguous survey items, before releasing formal survey, we discussed the questions with experts to make sure all the literal translations were clear and precise. From April 4th to April 13th 2015, we’ve conducted a 10 days long pretesting. Link of pretesting survey was posted on Facebook and Plurk, yielded 101 responses. After eliminating 9 invalid responses, we got 92 valid responses with response rate 91.09%. The purpose of pretesting was to revise the semantic unclear questions or the questions that are hard to answer. We’ve conduct reliability test and. 38.

(39) factor analysis to eliminate inappropriate items then used Varimax rotation to execute factor analysis.. Before conducting factor analysis, we used KMO (Kaiser-Meyer-Olkin Measure of Sampling Adequacy) and Bartlett's test of sphericity to ensure samples were appropriate for conducting factor analysis. KMO ranges from 0 to 1. The higher the better. High KMO value indicates the correlations between items are high. KMO values greater than 0.8 are considered to be great, means items are highly related therefore. 政治大. suitable for conducting factor analysis. KMO values range from 0.6 to 0.7s are. 立. considered to be good and adequate for factor analysis. KMO values less than 0.5 are. ‧ 國. 學. unacceptable, indicating the correlations between items are below the minimum criteria. Bartlett's Test of Sphericity is using to test whether the correlation matrix is an identify. ‧. matrix, the significance level must be less than 0.05.. n. al. Ch. er. io. Testing Results of KMO and Bartlett's test. sit. y. Nat. 3.5.1. i n U. v. KMO values of each construct were all greater than 0.8 except OGS and PPS_ON,. engchi. which KMO values was only around 0.7’s but still acceptable. All Bartlett tests had achieved significant level. Hence it’s capable to conduct subsequent factor analysis. The results are shown in Table 3-11.. 39.

(40) Table 3-11. Results of KMO and Bartlett's test KMO Construct. 3.5.2. Bartlett's Test of Sphericity. Measure of Sampling Adequacy. Approx. ChiSquare. df. Sig.. CPK. 0.871. 290.939. 10. 0.000. OGS. 0.763. 181.576. 10. 0.000. PPS_ON. 0.791. 394.767. 28. 0.000. PPS_OFF. 0.902. 495.009. 28. 0.000. PPD. 0.905. 648.380. 28. 0.000. 政治大. Factor analysis and total variance explained. 立. Table 3-12. Total Variance Explained of all constructs. 73.515. 1. 2.947. 58.942. 1. 4.377. 54.713. 2. 1.086. 13.569. 5.268. 65.851. 1. al. 73.515 58.942. y. 3.676. Cumulative %. sit. 1. io. v i 1 5.603 70.043 n Ch Extraction Method: ePrincipal h i UAnalysis. n g cComponent n. PPS_OFF. % of Variance. er. ‧ 國. PPS_ON. PPD. Total. Nat. OGS. Initial Eigenvalues. ‧. CPK. Component. 學. Construct. 54.713 68.282 65.851 70.043. Every construct had only extracted out 1 component except online pre-purchase search (PPS_ON), hence those solutions were unable rotate into factor matrix. Every extracted component had eigenvalue greater than 1 and had high percentage of variance. Showing that these components had strong explanatory power toward their constructs (Table 3-12). Pre-purchase search (PPS_ON) had extracted out two components (Table 3-13). Eigenvalue of Component 1 and 2 were 4.377 and 1.086 respectively, both were greater 40.

(41) than 1. However, the percentage of variance were 54.713% and 13.569 respectively (Table 3-12). Showing these two components had strong explanatory power toward online pre-purchase search.. Table 3-13. Rotated Component Matrix of PPS_ON Items. Component 1. 2. PPS_ON2. .848. .292. PPS_ON1. .818. .223. PPS_ON6. .782. .216. .242 .763 治政 PPS_ON8 .746 大.244 PPS_ON5 .161 .878 立 PPS_ON4. .257. .742. PPS_ON7. .326. .710. 學. ‧ 國. PPS_ON3. Extraction Method: Principal Component Analysis.. a. Rotation converged in 3 iterations.. Nat. n. al. er. io. sit. y. ‧. Rotation Method: Varimax with Kaiser Normalization.. Ch. engchi. 41. i n U. v.

(42) 4. STATISTICAL ANALYSIS AND HYPOTHESES TESTING. 4.1. Data Collection We used online survey platform mySurvey (http://www.mysurvey.tw/) to. formulate and release survey. The survey was opened from May 15th to June 4th 2015, 21 days in total. Data was collected via Facebook and PTT (a Bulletin Board System in Taiwan), yielded 570 responses. After eliminated 28 invalid responses, we got 542 valid responses with response rate 95.09%. Survey items are given in Appendix.. 政治大 According to Hoelter (Hoelter, 1983), minimum sample size of 200 is suggested 立. ‧ 國. 學. to conduct confirmatory factor analysis. To be strict and picky, Schwab recommended that the minimum item-to-response ratio for factor analysis should be 1:10 (Schwab,. ‧. 1980). Our questionnaire contained 37 items. Based on Schwab’s standard, at least 370. sit. y. Nat. valid samples were needed. We’ve collected 542 valid responses, accordingly was. io. n. al. er. capable for conducting subsequent analysis.. 4.2. Ch. Sample Structure Analysis. engchi. i n U. v. The sample comprised of 234(43.2%) males, 302 (55.7%) females and 3(0.3%) others. Majority of participants were aged between 19 and 28 years old (70.3%). Most of the participants were students (52.6%). Demographic data is summarized in Table 41.. 42.

(43) Table 4-1. Demographic information of collected samples Measure Male Female Other. 234 302 6. 43.2 55.7 1.1. Under 15 Age 16~18 Age 19~22 Age 23~28 Age 29~35 Age 36~40 Age 41~45 Age 46~50 Age 51~55 Age 56~60 Age 61~65 Over 65. 2 44 159 222 78 20 3 0 7 1 4 2. 0.4 8.1 29.3 41.0 14.4 3.7 0.6 0 1.3 0.2 0.7 0.4. Elementary School Junior High School High School College Master Ph.D.. 0 3 47 341 144 7. io. y. aAgriculture, v i l n C h Pasturage Forestry, 4 U i e h n and Fishery g c. n Occupation. ‧. Nat. Education. 學. ‧ 國. 立. 政治大. sit. Age. Ratio (%). er. Gender. Frequency. Banking and Insurance Traditional Manufacturing High-tech Manufacturing Information Service Communication Software Legal profession Construction Industry 43. 0 0.6 8.7 62.9 26.6 1.3 0.7. 13. 2.4. 18. 3.3. 27 32. 5.0 32. 8 5. 1.5 0.9. 4. 0.7.

(44) Commerce Media Entertainment Catering Tourism and Transport Medical Publishing Military and Police Government and Education Art Designing Housekeeper Freelance Retired Unemployed Student Other. 0.6 2.2 0.6 1.8. 34 0 3 2 21 2 21 285 16. 6.3 0 0.6 0.4 3.9 0.4 3.9 52.6 3.0. 政治大. n. Ch. engchi. 44. y. sit. 294 91 142 11. er. io. North of Taiwan Middle of Taiwan South of Taiwan East of Taiwan Outer islands of Taiwan Overseas. al. 3 12 3 10. ‧. Nat. Residence. 1.7 0.4 1.5. 學. ‧ 國. 立. 9 2 8. i 1 n U 3. v. 54.2 16.8 26.2 2.0 0.2 0.6.

(45) 4.3. Reliability and Validity Testing Factor analysis and reliability testing were conducted using SPSS 20 and. convergent validity and discriminate validity were tested using Smart PLS 2.0.. 4.3.1. Factor analysis, KMO, Bartlett's test and total variance explained Before conducting factor analysis, we used KMO (Kaiser-Meyer-Olkin Measure. of Sampling Adequacy) and Bartlett's test of sphericity to ensure samples were appropriate for conducting factor analysis. KMO ranges from 0 to 1. High KMO value. 政治大. indicates high correlations between items.. 立. KMO values of each construct were all greater than 0.8 except PPS, which KMO. ‧ 國. 學. values was only 0.604 but still acceptable. All Bartlett tests had achieved significant. in Table 4-2.. ‧. level. Hence it’s capable to conduct subsequent factor analysis. The results are shown. y. Nat. OGS. KMO. Bartlett's Test of Sphericity. Eigenvalue. Variance Explained. 3.609. 72.188. er. al. n. CPK. io. Construct. sit. Table 4-2. Results of Exploratory Factor Analysis. i n C 0.878 h e n g c h0.000 i U. v. 0.821. 0.000. 2.910. 58.203. PPS_ON. 0.901. 0.000. 4.660. 58.250. PPS_OFF. 0.938. 0.000. 5.768. 72.104. PPD. 0.919. 0.000. 4.962. 62.028. PPS. 0.604. 0.000. 2.061. 68.690. We used Varimax rotation to execute factor analysis. Since every construct only extracted out 1 component, the solutions were unable to rotate into factor matrix. We examined percentage of variance to see each factor’s explanatory power. Every construct only had 1 eigenvalue that greater than 1, and the percentage of variance was 45.

(46) between 58.203~72.188% (Table 4-2). Showing every component had strong explanatory power toward each construct.. Further, we evaluated the internal consistency reliability by Cronbach’s α. If α is greater than 0.7, indicates the reliability within the construct is high. Cronbach’s α between 0.3～0.7 indicating medium reliability, and α below 0.3 indicating the internal consistency within constructs is low (Nunnally and Bernstein, 1994). High Cronbach’s α implies that the respondent who give high score to one item would also tend to give. 政治大 consistency of responses within the construct. 立. high scores to other items in the same construct, vice versa, which indicates the. ‧ 國. 學. Cronbach’s α if item deleted indicating the Cronbach’s α after removal an item. If Cronbach’s α after removing an item is higher than original Cronbach’s α, it means that. ‧. item would drop the internal consistency. Indicating the correlations between items are. n. al. er. io. sit. y. Nat. low.. Ch. i n U. v. Next we are going to go through total variance explained and Cronbach’s α of each. engchi. construct to assure they have adequate reliability.. 46.

(47) Table 4-3. Reliability Statistics and Item-Total Statistics Construct/ Item. Scale Mean if Item Deleted. Scale Variance if Item Deleted. Corrected Cronbach’s Cronbach’s Item Total α if Item α Correlation Deleted 0.903. Consumer product knowledge(CPK) CPK1. 12.85. 12.382. 0.725. 0.889. CPK2. 13.13. 10.982. 0.777. 0.877. CPK3. 13.31. 11.306. 0.738. 0.886. CPK4 CPK5. 13.01 13.11. 11.571 10.906. 0.749 0.806. 0.883 0.870. 政治大. Ongoing search(OGS). ‧ 國. 12.86. 10.834 9.647 9.953. 0.503 0.518 0.684 0.695. 9.919. 0.646. 0.812 0.806 0.757 0.756 0.769. ‧. OGS5. 立 10.722. 13.14 13.67 13.30 13.12. 學. OGS1 OGS2 OGS3 OGS4. 0.817. 0.892. 27.42. PPS_ON3. 27.37. PPS_ON4 PPS_ON5 PPS_ON6 PPS_ON7 PPS_ON8. 27.98 27.77 27.30 27.93 27.47. n. al. 0.739. 21.897. 0.775. sit. PPS_ON2. 21.527. 0.872. i n C22.118 U h e n g c0.575 i h 21.946 0.672 22.374. 0.705. 22.419 22.093 22.467. 0.725 0.589 0.620. 0.869. er. 27.55. io. PPS_ON1. y. Nat. Online pre-purchase Search(PPS_ON). v 0.876 0.889 0.878 0.874 0.887 0.883. Offline pre-purchase Search (PPS_OFF). 0.895. PPS_OFF1. 22.01. 45.259. 0.786. 0.938. PPS_OFF2 PPS_OFF3 PPS_OFF4 PPS_OFF5 PPS_OFF6 PPS_OFF7. 22.11 21.87 22.15 22.09 21.95 22.34. 44.493 44.824 46.434 45.174 44.303 45.793. 0.844 0.826 0.741 0.813 0.851 0.761. 0.934 0.935 0.941 0.936 0.934 0.940. 47.

(48) PPS_OFF8. 21.75. 45.621. 0.766. 0.939 0.895. Post-purchase dissonance(PPD) PPD1. 14.3672. 14.957. 0.798. 0.872. PPD2 PPD3 PPD4 PPD5 PPD6 PPD7 PPD8. 14.3007 14.1882 14.1089 14.2694 14.1827 14.2989 14.3893. 14.758 14.926 16.563 14.822 15.041 14.594 15.088. 0.805 0.657 0.299 0.794 0.648 0.819 0.734. 0.870 0.884. Post-purchase search(PPS) 7.51. PPS2. 6.81. PPS3. 6.72. 0.757. 政治0.424 大 3.519 0.715. 0.865 0.530. 3.651. 0.603. 4.062. 立. 0.650. 學. ‧ 國. PPS1. 0.923 0.871 0.885 0.869 0.877. ‧. Cronbach’s α of all constructs were all greater than 0.7, reflecting high internal. sit. y. Nat. consistency within each constructs. Each item’s Cronbach’s α if item deleted was less. al. er. io. than overall total Cronbach’s α except PPD4 and PPS1 (Table 4-3). It means removal. v. n. these 2 items would drop the internal consistency. Indicating the correlations between items were low.. Ch. engchi. i n U. Cronbach’s α if item deleted of PPD4 is greater than the than overall total Cronbach’s α 0.945. Cronbach’s α would increase from 0.895 to 0.923 if we remove PPD4 (Table 4-3). The item-total correlation for PPD4 was also lower than others, which means it was not correlate well with other items. It’s reasonable to drop PPD4.. Cronbach’s α if item deleted of PPS1 is greater than the than overall total Cronbach’s α 0.757. Cronbach’s α would increase from 0.757 to 0.865 if we remove 48.