選舉中線上口碑風暴之研究 - 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：尚孝純博士 Dr. Shari S. C. Shang. 選舉中線上口碑風暴之研究 An Examination Of Online Firestorm In Election. 研究生：陳怡臻 Yi Chen, Chen 中華民國 104 年 7 月 .

(2) . Acknowledgement Foremost, words are not enough to express my gratitude to my dear advisor, Dr. Shari S. C. Shang for the great support of my research with her time, knowledge and endless patience. Without her guidance I could not have finished this thesis, I am deeply moved for every effort she did for me. How lucky I am to find a good mentor and advisor in my master career, I sincerely thankful that I can enroll in National ChengChi University to meet my advisor. In addition, I would like to thank my thesis committee: Dr. S. C. Chen and Dr. Y. L. Wu for their continuous encouragement, accurate comments and helpful suggestions. They really help my thesis being more insightful and substantial. Besides, I could not forget to give appreciation to my friends and lab mates. Thanks my lab mates for working together on several projects, I have lots of joyous memory because of you guys. Moreover, my friends in other labs, thank you always accept me for staying in your lab watching animation, murmur during the dinner, working until midnight, and celebrating the new years together. I could not imagine the life without all the fun we have in NCCU, I will preserve the memory in my mind. Furthermore, thanks my best university friends, we still hang out together all the time though we are busy, our precious friendship really enriches my life. Last but not least, my beloved family, especially my parents, thanks for your ultimate love and support to me for my education and growing. I could not be an exchange student to Germany and study in NCCU to pursue my master degree without you, and it is no way for me to experience the life and meet so much friends. Being your daughter is the best thing I can have in my whole life, thank you.. . 1 .

(3) . 摘要在這網路發達的世界，電子口碑(electronic word of mouth)一直以來都是學術界研究的重點。電子口碑分成正面跟負面，尤其以負面電子口碑備受關注，因為這會帶給企業難以估計的傷害。而在 2014 年，開始有學者提出了網路風暴(online firestorm)的新名詞，描述在現代社群網路下，負面口碑在網路上突然大量流傳的現象。關於網路風暴的文獻相當少，但在台灣 2014 年舉辦的九合一選舉中，可以看出網路風暴對選情的影響之大，我們觀察到某些候選人可以藉由應對來有效控制情勢，但有些回應卻會讓負面口碑風暴越趨惡化，因此我們想要探究這之間的問題所在。所以本研究首先透過文獻了解口碑風暴的定義，從而利用文獻推斷出口碑風暴形成過程。並以 Google Trend 設計出衡量口碑風暴的方式，以此找出選舉中符合口碑風暴標準的負面口碑案例。之後把案例分成三種情境，每種情境以兩個不同候選人的個案為例，把從 Opview 收集到的負面口碑資料套用在設計出的公式裡來判斷選舉人的應對有效與否。選舉人的應對則會利用 Benoit 印象修復理論來做進一步說明。最後，由於此篇研究是以選舉中的口碑風暴為主，因此我們會把選舉中的口碑風暴與傳統商業中的口碑風暴做比較，來看兩者間有何不同或相同之處。關鍵字：負面電子口碑，網路風暴，選舉. . 2 .

(4) . Abstract During the political election period in Taiwan in Nov. 2014, this study observed significant effects caused by online firestorms: some candidates used appropriate responses to mitigate their effects or minimize the harm, some turned negative into positive support of their brand, but some candidates were unable to manage the crisis and lost their brand value. To date, few studies have noted the importance of negative WOM (NWOM) management in political election campaigns. Nor have studies noted the effects of online firestorms on brand value. This study undertakes research to seek answers to the following research questions: What is the uniqueness of online firestorms in a political election? And, how do candidates respond the online firestorm to retain brand value? This study initially reviewed the literature on negative WOM and online firestorm, and consolidated these studies to formulate the forming process of an online firestorm. Then the study classified online political election firestorm cases into three categories including fault, defect, and counterforce slander, and analyzed the best responses in each situation. Next, the study applied Google Trend to measure the online firestorm and referred image restoration theory to form an understanding of the response of a brand manager. The study results show that for fault situation, the most useful strategies seem to be mortification, minimization the offensive feeling, and ignoring to respond to most offensive accusations. For defect situation, the most used strategies are insisting the main opinion and using attack strategy to evade the most offensive accusations. Last, for counterforce slander situation, most candidates used full denial, blame shifting and accuser attack strategies to prove their innocence. Finally, the study further distinguished the difference between conventional business online firestorm and political elections online firestorm from several angles which include: online firestorm targets, causes, consumer reactions, initiatives, effects and dissemination. . Key words: Negative word of mouth, Online firestorm, Political election. . 3 .

(5) . Table of Contents ACKNOWLEDGEMENT ........................................................................................................1 摘要 ............................................................................................................................................2 ABSTRACT ...............................................................................................................................3 TABLE OF CONTENTS ...........................................................................................................4 LIST OF TABLES ......................................................................................................................6 LIST OF FIGURES ....................................................................................................................8 CHAPTER 1: INTRODUCTION ...............................................................................................9. CHAPTER 2: LITERATURE REVIEW..................................................................................12 2.1. WORD OF MOUTH (WOM)..................................................................................12. 2.2. ELECTRONIC WORD OF MOUTH (EWOM) ....................................................12. 2.3. THE MOTIVATION OF SHARING ELECTRONIC WORD OF MOUTH ........14. 2.4. NEGATIVE WORD OF MOUTH (NWOM) ........................................................15. 2.5. THE MOTIVATION OF NEGATIVE WORD OF MOUTH .................................16. 2.6. ONLINE FIRESTORM .........................................................................................17. 2.7. STRATEGY TO MANAGE AND RESPONSE TO NEGATIVE EWOM ...........23 2.7.1 What a company (candidate) can do to manage the EWOM ...................................... 23 2.7.2 How a company (candidate) response to the crisis ................................................... 26. CHAPTER 3: METHODOLOGY ..........................................................................................30 3.1. RESEARCH APPROACH ....................................................................................30. 3.2. DATA COLLECTION ..........................................................................................30. 3.3. DATA ANALYZE ..................................................................................................35. CHAPTER 4: DATA ANALYSIS ...........................................................................................38 4.1 FAULT ....................................................................................................................38 4.1.1 Ko was accused of a tendency toward sexism .......................................................... 38 4.1.2 Lien’s father criticized Ko’s grandfather is a japanization person ........................... 42 4.1.3 Case analysis ............................................................................................................... 45. 4.2 DEFECT ..................................................................................................................47 . 4 .

(6) 4.2.1 Ko said that rejecting company “Ting-Shin” was similar to witch burning ............... 47 4.2.2 Lien donated one hundred thousand NT dollars to Kaohsiung gas explosion .......... 50 4.2.3 Case analysis ............................................................................................................... 53. 4.3 COUNTERFORCE SLANDER ..............................................................................55 4.3.1 Ko was accused of involving in human organs trading ............................................ 55 4.3.2 Lien was accused of eavesdrop on Ko’s office ........................................................... 58 4.3.3 Case analysis ............................................................................................................... 63. CHAPTER 5: DISCUSSION ...................................................................................................65 CHAPTER 6: CONCLUSION .................................................................................................73 6.1 RESEARCH CONTRIBUTIONS .............................................................................73 6.2 RESEARCH LIMITAIONS ......................................................................................74 6.3 FURTHER STUDY ...................................................................................................74 REFERENCES .......................................................................................................................76 APPENDIX...............................................................................................................................84. . 5 .

(7) . List of Tables. Table 2.6-1. Difference between NWOM and Online firestorm ............................................17 Table 2.6-2. Online firestorm cases ..........................................................................................18 Table 2.7.2-1. The description of Image restoration discourse ..............................................27 Table 3.2-1. Google Trend analyze on table 2.6-2. online firestorm cases ..............................32 Table 3.2-2. Google Trend analyze of NWOM cases in Taiwan political election ..................32 Table 3.3-1. The categorizes of Image restoration strategies .................................................35 Table 4.1.1-1. Response strategy of sexism tendency online firestorm ...................................39 Table 4.1.1-2. NWOM amounts of sexism tendency online firestorm .....................................40 Table 4.1.1-3. Boiling period and explode period of sexism tendency online firestorm .........41 Table 4.1.1-4. Online firestorm characteristic of sexism tendency online firestorm .............41 Table 4.1.2-1. Response strategy of Japanization person online firestorm ..............................43 Table 4.1.2-2. NWOM amounts of Japanization person online firestorm ................................44 Table 4.1.2-3. Boiling period and explode period of Japanization person online firestorm ....44 Table 4.1.2-4. Online firestorm characteristic of Japanization person online firestorm ..........44 Table 4.1.3. Fault cases analysis ...............................................................................................45 Table 4.2.1-1. Response strategy of “Ting-Shin” online firestorm ..........................................48 Table 4.2.1-2. NWOM amounts of “Ting-Shin” online firestorm ...........................................49 Table 4.2.1-3. Boiling period and explode period of “Ting-Shin” online firestorm ................49 Table 4.2.1-4. Online firestorm characteristic of “Ting-Shin” online firestorm ......................49 Table 4.2.2-1. Response strategy of denoting online firestorm ................................................51 Table 4.2.2-2. NWOM amounts of denoting online firestorm .................................................52 Table 4.2.2-3. Boiling period and explode period of denoting online firestorm ......................52 Table 4.2.2-4. Online firestorm characteristic of denoting online firestorm ............................52 Table 4.2.3. Defect cases analysis ............................................................................................53 Table 4.3.1-1. Response strategy of organ trading online firestorm ........................................56 . 6 .

(8) . Table 4.3.1-2. NWOM amounts of organ trading online firestorm ..........................................57 Table 4.3.1-3. Boiling period and explode period of organ trading online firestorm ...............57 Table 4.3.1-4. Online firestorm characteristic of organ trading online firestorm.....................58 Table 4.3.2-1. Response strategy of eavesdrop online firestorm ..............................................60 Table 4.3.2-2. NWOM amounts of eavesdrop online firestorm ...............................................61 Table 4.3.2-3. Boiling period and explode period of eavesdrop online firestorm ....................62 Table 4.3.2-4. Online firestorm characteristic of eavesdrop online firestorm ..........................62 Table 4.3.3. Counterforce slander cases analysis .....................................................................63 Table 5.1. The best response of different situation ...................................................................65 Table 5.2. Compare conventional online firestorm and political election online firestorm .....67 Table 5.3. NWOM amounts of illegal canvass online firestorm ..............................................71 Table 5.4. Boiling period and explode period of illegal canvass online firestorm ...................71. . 7 .

(9) . List of Figures Figure 1. The forming process of online firestorm .................................................................22 Figure 2. The forecast model of online firestorm ...................................................................23 Figure 3. The response time and NWOM pattern of sexism tendency online firestorm ..........40 Figure 4. The explode period of sexism tendency online firestorm .........................................41 Figure 5. The response time and NWOM pattern of Japanization person online firestorm .....43 Figure 6. The explode period of Japanization person online firestorm ....................................44 Figure 7. The response time and NWOM pattern of “Ting-Shin” online firestorm .................48 Figure 8. The explode period of “Ting-Shin” online firestorm ................................................49 Figure 9. The response time and NWOM pattern of denoting online firestorm .......................51 Figure 10. The explode period of denoting online firestorm ....................................................52 Figure 11. The response time and NWOM pattern of organ trading online firestorm .............57 Figure 12. The explode period of organ trading online firestorm ............................................58 Figure 13. The response time and NWOM pattern of eavesdrop online firestorm ..................61 Figure 14. The explode period of eavesdrop online firestorm ..................................................62 Figure 15. The response time and NWOM pattern of illegal canvass online firestorm ...........71. . 8 .

(10) . Chapter 1: Introduction There are tremendous numbers of people surfing the Internet and using social platforms to express their opinions every day. Many comments, opinions, and experiences spread on the Internet; these comments have positive, negative or even no effects. Because information on the Internet is transparent to the public, everyone can see the information at any time; the information can affect people’s cognition. The process of conveying information from person to person plays a major role in customer buying decisions and is characterized as word of mouth (WOM) (Richins & Shaffer, 1988). Web technology provides another avenue for consumers to communicate with other consumers and affect other people, called electronic word of mouth (EWOM) (Goldsmith & Horowitz, 2006 ; Chu & Kim, 2011). EWOM also can be defined as any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet (Hennig-Thurau et al., 2004). Positive WOM and negative WOM are spread online every day of our life. According to Bone (1995), this study might find that negative word of mouth is more influential than positive word of mouth. Studies also have shown that up to two-thirds of dissatisfied consumers do not complain to the marketer, but rather switch brands or engage in negative WOM (Andreason, 1985; Richins, 1983). Kotler (1991) found that dissatisfied customers spread negative WOM to eleven people, whereas satisfied customers may only spread positive WOM to three people. There is also some evidence that consumers place more weight on negative information in making product evaluations (Weinberger & Dillon, 1980). As the Internet was increasing in popularity, Stauss (1997) mentioned that the virtual opinion platform posed a risk to a company because it can spread negative information rapidly to many people. In addition, unlike traditional word of mouth, the negative information will remain available to other people any time they search for it. Academics started to focus on the phenomenon called ’online firestorm’ in 2014 (Pfeffer et al., & Mochalova & Nanopoulos). An ‘online firestorm’ is ‘the sudden discharge of large amounts of information containing negative WOM and complaint behavior against a politician, company and its brands, government institution or a celebrity in social media networks’ (Pfeffer et al., 2014). An online firestorm has affective influences and therefore poses severe threats to companies or celebrities, with huge waves of . 9 .

(11) . criticism and complaint harming brand reputation without warning, causing customer losses and creating or exacerbating a crisis (Pfeffer et al., 2014). An online firestorm starts due to various reasons, such as a marketing campaign or dissatisfaction of customers (Mochalova & Nanopoulos, 2014). However, to date, there is a lack of a clear understanding of the process and type of online firestorms and a lack of any deep analysis of the strategies appropriate to responding to or managing an online firestorm. Therefore, research on it is called for. Taiwan is a democratic country; every Taiwanese over the age of 20 owns anonymous voting qualifications. Taiwan recently (on Nov. 29, 2014) held political elections for mayor, alderman, and chief of village. Specifically, two candidates drew the most attention in the Taipei mayoral political election; one was Sheng-Wen Lien (Lien), and the other was Wen-Je Ko (Ko). There are approximately 2,160,000 voters in Taipei (ETtoday, 2014), but with only a 67.9% turnout, the total number of votes was 1,494,046 (Liberty Times, 2014). During the political election, many negative WOMs and even online firestorms occurred to harm the candidates. Some candidates addressed the crisis very well, mitigating the effect of online firestorms, minimizing harm, or even turning negative into positive support of their brand. For example, Wen-Je Ko was initially accused of being involved in the organ trade and money laundering (Zhen-Kai, Tzu, 2014;Shu-Zhen, Chen, 2014), but an opinion poll indicated that his approval increased after these crises (United Times, 2014). However, there were also some instances of negative WOM that the candidates could not appropriately address; thus, there were more attacks. Because the online WOMs during this political election demonstrate a complete business case of WOM’s effects on brand management and contain various tactics of clarifying, debating, neglecting, verifying and convincing customers, who are the voters, to accept the brand, it is valuable to examine the negative WOM effects in this special brand campaign. This study assume that for every candidate, whether a product or a company, how they behave and what they do relevant to their image is a type of brand management. The voter is the consumer or people who share information with others and evaluate candidates from opinions discussed online. People talk about images or news of candidates on social platforms in their daily lives by continuously communicating opinions with other people; everyone is affected by others’ comments and until they finally confirm opinions in their own minds. Through the discussion, people . 10 .

(12) . determine which candidate they appreciate, and they eventually vote for the preferred candidate in the political election. Therefore, the research questions of this study are as follows: s. What is the uniqueness of online firestorms in a political election?. s. How do candidates respond to online firestorm effectively to retain brand value?. . 11 .

(13) . Chapter 2: Literature Review 2.1 Word of mouth (WOM) There have been many studies on word of mouth; many researchers have different definitions of WOM. Generally, word of mouth (WOM) can be defined as a process of transmitting informal communications directed at other consumers concerning information about goods and services or their sellers. WOM can affect consumer behavior (Hennig-Thurau & Gianfranco, 2003; Richins & Root-Shaffer, 1988; Anderson, 1998). In marketing, East, Hammond, and Wright (2007) mentioned that WOM is used to describe advice from other consumers; its characteristics of interaction, speed and lack of commercial bias make it a powerful source of information for consumers. What is the influence of WOM that appeals to many researchers studying it? According to previous studies, word of mouth (WOM) has proven to have a critical role in the customer buying decision because it can influence customer choices (Arndt, 1967; Richins, 1983; Richins & Root-Shaffer, 1988). WOM also can be considered an effective force in marketing because it can affect consumer attitudes (Bone, 1995), preferences and buying intentions (Charlett et al., 1995; Herr et al., 1991), and even decision-making behavior (Wangenheim & Bayón 2004). Because WOM is created by other consumers rather than a company, consumers consider that WOM is more trustworthy than company-generated information (Feick & Price 1987). Believing the WOM, consumers often depend on it to inform their decision to reduce the uncertainty risk before buying (Murray, 1991). Some research also indicates that WOM is more effective than traditional marketing tools such as personal selling and conventional advertising such as newspaper, magazine and radio advertising (Katz & Lazarfeld, 1955; Goldsmith & Horowitz, 2006). 2.2 Electronic word of mouth (EWOM) The traditional oral, person-to-person WOM has developed into WOM via the Internet, that is, electronic word of mouth (EWOM) (Chu & Kim, 2011; Cheung et al., 2009). The emergence of web technology provides a virtual avenue to consumers to communicate with other consumers; it occurs over a wide range of electronic media, such as blogs, social networking sites (SNS), consumer review websites and forums, electronic bulletin board systems, and emails (Goldsmith & Horowitz, 2006; Chu & . 12 .

(14) . Kim, 2011). Hennig-Thurau et al. (2004) define EWOM as ‘any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet.’ EWOM via electronic media allows consumers to gain information not only from people with whom they are familiar but also from the tremendous group of unknown people who previously had experience of using the product or the service (Ratchford et al., 2001). Compared with WOM, EWOM is less personal but more effective because it communicates immediately and contacts many people (Hennig-Thurau et al., 2004). Although EWOM provides the virtual platform for consumers to express their opinion, it also poses a risk to a company because negative information directed at products or services can be spread rapidly to countless people, and the negative information will exist on the Internet for access by any customer (Stauss, 1997). There are some differences in characteristics between WOM and EWOM. Cheung et al., (2009) identify three. (1) The communication network scale of EWOM is larger than WOM because the Internet can involve the discussions of more contributors or consumers, and it can reach more people than direct personal communication can. (2) EWOM eliminates the limitations of time and location. Consumers can read the information they are interested in and join the discussion at their pace. (3) The credibility of information is critical to EWOM receivers. However, the group discussion on an online forum provides additional resources for other consumers to evaluate whether the information is believable, whereas the credibility of traditional WOM cannot easily be evaluated when opinions are received from friends or relatives. Goldsmith and Horowitz (2006) also state that cyberspace provides (1) a variety of avenues to consumers to exchange information, (2) online anonymity and confidentiality, allowing consumers to conceal their identities when seeking and giving opinions, (3) a lack of physical clues used to evaluate the identity of others, (4) freedom from geographic and time constraints, and (5) permanence of online conversations. Therefore, EWOM can be considered an extension of traditional WOM to a new medium, that is, the Internet. The communication scale of EWOM is larger than traditional interpersonal communication, and the anonymity characteristic allows consumers to conceal their identities but also prevents them from assessing the identities of others. The online forum breaks the limitation of location and time in the world, and the information will be preserved so that everyone can find it. . 13 .

(15) . This study define word of mouth (WOM) as a consumer conveying discourse or informal information about a particular product or service to other consumers, a process that will influence participating consumer buying behavior. Because this research is focused on negative EWOM behavior in a political election, this study can consider the voters as consumers who can communicate opinions and messages with others voters via the Internet consumer opinion platform, and the product or service they talk about can refer to candidates in the political election. Based on our definition of WOM, informal communication among consumers will influence their buying decisions. In this research, the buying decision can be explained as whether the voters want to vote for the particular candidate; the outcome is seen at the polls or as the result of the political election. 2.3 The motivation for sharing electronic word of mouth (EWOM) In this part, this study will discuss why people engage in EWOM. Generally, EWOM behavior can be categorized in three aspects: opinion seeking, opinion giving and opinion passing (Chu & Kim 2011). Hennig-Thurau et al. (2004) identify the motivations for giving opinions or sharing experience in EWOM as (1) social benefits, meaning a desire for social interaction; (2) economic incentives; (3) concern for other consumers; and (4) enhancing their self-worth. These are the main reasons consumers want to publish their experiences online. Additionally, why do other consumers seek to share with others online? Goldsmith and Horowitz (2006) revealed eight reasons for consumers wanting to seek opinions via the Internet. The motivations of seeking behavior include (1) reducing uncertainty risk, (2) occasionally searching for information and stumbling on other unplanned information by accident, (3) the hedonic motive, such as “because it is cool”, (4) stimulation by off-line inputs such as TV, (5) reducing price consciousness, (6) ease of access—the Internet is the easiest means of obtaining different views on a subject, (7) influence by others such as friends, and (8) obtaining information for exploration purposes. Some researchers also propose that consumers read online articulations primarily to save decision-making time and to make better buying decisions (Hennig-Thurau et al. 2003).. . 14 .

(16) . Passing/ forwarding online content is one means of communication in EWOM, but it can easily be ignored (Norman & Russell 2006; Sun et al. 2006). Opinion passing is a unique characteristic of the Internet, which facilitates the spread of information to multitudes of people online with a few easy clicks (Norman & Russell 2006). Ho and Dempsey (2010) use Fundamental Interpersonal Relations Orientation (FIRO) to identify four potential motivations for consumers to forward online information: (1) the need to be part of a group, (2) the need to be individualistic, (3) the need to be altruistic, and (4) the need for personal growth. EWOM provides a virtual avenue for consumers to express their opinions, but it also poses a risk to companies because the negative information can be spread rapidly and exists on the Internet, available to countless people (Stauss, 1997). Therefore, many companies and researchers focus closely on it; some companies also use electronic consumer platforms such as Facebook and Twitter to connect with their target customers. Electronic WOM may be positive, negative or neutral; thus, it can help the company but could also hurt it. The following sections will elaborate these points further. 2.4 Negative word of mouth (NWOM) In general, this study see that the amount of positive WOM is more than that of negative WOM, according to research by Resnick and Zeckhauser (2002). They determine that 99.1% of customer feedback on eBay in the late 1990s was positive; negative and neutral WOM only occupied for 0.6% and 0.3%, respectively. Although the amount of positive WOM is far greater than that of negative and neutral WOM, this study cannot ignore the importance of negative WOM. Most of the literature indicates that negative WOM is more influential than positive WOM (Bone, 1995); moreover, some studies show that up to two-thirds of dissatisfied consumers do not complain to the marketer but instead engage in negative WOM (Andreason, 1985; Richins, 1983). Some studies mention that NWOM originates from dissatisfied customers approximately twice as frequently as does PWOM from satisfied customers (Goodman & Newman, 2003). Mulpuru (2007) also concluded that negative reviews occupy a smaller portion of total reviews but are considered helpful to consumers. Wetzer et al. (2007) defined NWOM as involving all negative recommendations and informal communication between private parties about the evaluation of products or . 15 .

(17) . services. Richins (1984) defines NWOM as a communication method that dissatisfied customers use to damage the product, service or company to others. She (1983) also determined that negative word of mouth is how dissatisfied customers express their complaints. Potential responses of dissatisfied customers include (1) switching brands or refusing to consume any more, (2) complaining to the seller or third party, and (3) telling other people about the unsatisfactory experience. Although complaints would help marketers identify their disadvantages, negative WOM will harm the company because it reaches numerous potential customers. Based on the social sharing literature, this study know that people tend to share information shortly after occurrence of an event; over 50% of such communications even occur on the same day (Rimé et al., 1992). 2.5 The motivation of negative word of mouth Based on several studies, Wetzer et al. (2007) identified eight basic motivations for NWOM: (1) Comfort search—seeking others’ comfort, support or understanding; (2) Venting—release pressure by expressing emotion; (3) Advice search—engaging in NWOM to gain cognitive clarity; (4) Bonding—decreasing interpersonal distance with others and strengthening social ties; (5) Entertaining—keeping the conversation going and amusing the participants; (6) Self-presentation—managing others’ impression or image of oneself; (7) Warning—helping the receiver make a good purchase decision; and (8) Revenge—harming the company or marketers who make one unhappy. Consumers engage in NWOM for different goals; however, these different goals can also result from consumers experiencing a variety of negative emotions. For example, angry consumers engage in NWOM to vent feelings or to take revenge; disappointed consumers engage in NWOM to warn others; and regretful consumers engage in NWOM to strengthen social bonds or to warn others (Wetzer et al., 2007). From the above discussion, this study defines NWOM as a communication method that consumers use to express their unsatisfactory experience or recommendations to others. NWOM is induced by several emotions and different motivations. Large numbers of people spreading NWOM on the Internet will eventually start an online firestorm. This study will discuss online firestorms to identify what characteristics they have and how dangerous they are.. . 16 .

(18) . 2.6 Online firestorm Although public media often discussing the online firestorm phenomenon, this study can only find a few studies that specifically research this topic. Fortunately, Pfeffer et al. (2014) define online firestorm explicitly: online firestorm is the sudden discharge of large quantities of messages containing negative WOM and complaint behavior against politicians, companies and their brands, government institutions and celebrities in social media networks. Social media users can create huge outrage in reaction to a questionable statement or activity within only a few hours. During an online firestorm, angry emotion is often expressed without targeting any specific criticism; most messages in an online firestorm are personal opinions instead of fact, thereby having highly effective influence. Following Pfeffer et al. (2014), this study define the difference between NWOM and an online firestorm in the below chart. This study can see that negative WOM can trigger the generation of an online firestorm. Table 2.6-1. Difference between NWOM and Online firestorm Definition. Difference. NWOM as a communication way that. A “noun” describing a. dissatisfied customer complaint or. communication method used by. damage about the product, service or. people to express their. company to others.. dissatisfaction.. Online. Online firestorm as the sudden. A “phenomenon” characterized. firestorm. discharge of large quantities of. by sudden large quantities of. messages containing negative WOM. NWOM being discussed on. and complaint behavior against a. social media network. Sudden. politicians, company and their brands,. great amounts of NWOM can. government institutions and celebrities. lead to an online firestorm.. NWOM. in social media networks. Pfeffer, et al. (2014) discussed several famous online firestorm cases to illustrate the dangerous of an online firestorm.. . 17 .

(19) . Table 2.6-2. Online firestorm cases Event. Online firestorm. On 2012/1/18, McDonald launched an. Within only 2 hours, McDonald’s. activity to encourage customers aware of. withdrew the activity because many. the heritage of McDonald food by using. customers shared NWOM and insults on. #meetthefarmers on Twitter, but later. the Internet. Although McDonald’s. they changed the hash to #McDStories to. switched back to #meetthefarmers 2. encouraged followers shared their story. hours later, the damage had already been. related to McDonalds.. done. More than 1000 people had posted negative experiences that happened in McDonalds. This online firestorm received coverage from the traditional media, which led to broader propagation.. In January 2012, a German bank. At the beginning of the protest,. ING-DiBa made an advertisement show. comments were posted on the bank’s. a famous basketball player eating. Facebook page every 5 seconds. At first,. sausage in a butcher’s shop. Vegetarian. ING-DiBa ignored the comments and. accused that ING-DiBa promote an. allowed the heated discussion between. unethical (meat) industry on the bank. vegetarians and meat eaters. The. Facebook page.. traditional media instantly propagated the news. Two weeks later, ING-DiBa found the heated discussion had not yet faded, so they announced that from then on, all comments about meat would be deleted in the future. Thereby, the online firestorm finally ended.. At 2011/11/22, a Australian airline. Only 2 hours later, #QantasLuxury was. Qantas invited customers to attend an. listed in the Twitter Trending Topics list. activity to win the First Class gift pack,. in Australia, with approximately 100-150. and use the tag of #QantasLuxury. The. tweets every 10 min. Qantas did not. prize was a pair of pajamas and an. consider that a few weeks previously, it. amenity kit.. had left thousands of passengers stranded at different airports. Therefore, the activity drew many negative comments. . 18 .

(20) . attacking Qantas. However, Qantas ignored all the comments and claimed it received a large amount of positive feedback. This behavior led to further propagation and gained Qantas more negative reputation outside Australia. From the above examples, the most terrible phenomenon this study can observe in an online firestorm is that it can induce intense, huge waves of criticism and complaint without warning to harm the reputation of a company or brand. It could lead the company to lose money. Because to date there are few studies on online firestorms, this study can reasonably assume that companies fear huge volumes of NWOM comments appearing on the Internet and causing negative effects. This study will discuss later how the online firestorm happened to candidates. It is important for us to recognize clearly the factors causing online firestorms. Pfeffer et al. (2014) summarized seven factors that affect opinion spreading in social media. This study includes all of them because they bear on this research. 1.. Speed and volume. The constant flow of real-time messages in social media causes fast information transportation. Compared to the traditional media need of approximately one day to transport information, an online firestorm can let highly attractive news reach many people within a short time through social media; a company should react in hours or even a few minutes. In an online firestorm, the news will temporally dominate what tremendous numbers of people are talking about, thereby causing huge volumes of communication. 2.. Binary choices. Articulate and sophisticated opinions are not what appear on most social media sites. People who use Social media such as Facebook or Twitter tend to ‘like’ or ‘+1’ the information; were they to want to write more, they would be restricted by technical length limitations of messages. Therefore, there are limits on forming gradual or complete opinions. The absence of comment interaction is an important factor for an online firestorm. Deciding whether to support or oppose the opinions is a binary choice because they are ’either-or situations’ (Schelling, 1973). . 19 .

(21) . 3.. Network clusters and echo chambers. Interpersonal communication networks have obvious local clustering. Network clustering refers to transporting link creations (Heider, 1946). In other words, if user A is a friend of user B, and user B is a friend of user C, then user A and user C also have a high chance of being friends. The high intensity of connected friends and local clustering will lead to “echo chambers” (Key & Cummings, 1966). These “echo chambers” create a phenomenon in which information is received from different people and directions of their social network; it feels as though everyone is talking about the same topic or has the same opinion (Sunstein, 2001). 4.. Unrestrained information flow. Granovetter (1973) mentions the strength of a tie as a combination of time, emotional intensity, intimacy and reciprocal service. In general, weak ties play a bridge role for transporting information within different groups; strong ties play a role in effecting message communication in a social network (Brown & Reingen, 1987). One can be a friend or follower of a large number of people in social media. The intense number of social network neighbors can amplify and echo information and messages during transmission. 5.. Lack of diversity and a filter bubble. Pariser (2011) refers to a concept of a filter bubble, which posits that people on social media tend to overemphasize a specific topic. The filter bubble can function in two ways. One way is connection via social media based on homogeny (McPherson et al., 2001); people tend to connect with others with similar characteristics such as age, gender and socioeconomic status, so that they are likely to have similar interests and opinions. The second way is that social media always present information based on previous user experience or information that also interests a user’s friends. Therefore, people become limited and receive filtered information based on their specific preferences. 6.. Cross media dynamics. Social media have gradually become an important information source for traditional media (Diakopoulos et al., 2012). Traditional media use social media such as Twitter and Facebook as a type of ‘radar’ to seek new news or information. This phenomenon . 20 .

(22) . also happened in Taiwan; television or new publishers often find a story on the largest bulletin board system, PPT. Traditional media then broadcasting such a story would cause much more online activity. Myers et al., (2012) found that approximately one-third of the information on social media is caused by external events or factors outside the network. 7.. Network triggered decision processes. The process of a network-triggered decision has four separate parts. 1) ‘Knowledge’ refers to the moment people encounter the first information, but the filter bubble affects the information people receive. 2) ‘Persuasion’ implies that people receive positive and negative opinions of events, and the echo chamber causes people to feel as though everyone is talking about it. 3) ’Propagation’ indicates that people decide to support or decline the opinion, talk about the decision to others, and allow it to affect their own decision making. 4) ’Affirmation’ suggests that people would confirm their own opinion when meeting another with the same opinion; otherwise, their opinion may destabilize. This step would also be affected by the echo chamber. Having explained the seven factors of opinion spreading, this study combine all the factors in a network-triggered decision model to illustrate the process of an online firestorm:. . 21 .

(23) . • Cross media dynamic help the information spreading further. This step affect by Eilter bubble, so customer would only attain limited knowledge information online. • The echo chamber would made user feels like everyone are talking about the news so they would Eind more informaiton to Eigure out the accident. Many decision makers are not passive, they actively seek persuasion information(Haywood, 1989). . • In this step, people will try to express the opinions to persuade others, this action would effect by unrestrained information Elow to propagation effect other people online. . afEirmation. • The affection of echo chamber would still effect in this step to help users Eind others with same opinion to afEirm their mind.. Figure 1. The forming process of online firestorm As this study mentioned, a filter bubble can affect the first time people receive the information. After people know the news, the echo chamber effect would lead them to feel as though everyone is talking about it; therefore, they would seek more information to help them determine what is actually happening. After people in the ‘persuasion’ step receive a great deal of information from others, they would finally make a decision and try to persuade others of their opinion. Unrestrained information factors lead to the amplification effect when transporting information and messages to others. EWOM implies that people express their opinions to others online; therefore, this study can see that some people in “persuasion” steps would be affected by those people who have already made a decision in the “propagation” step. After these people finalize their opinion, some would also spread the information to influence others. Therefore, this study can observe a continuous virtual circle between the ‘persuasion’ and ‘propagation’ steps; this study names this a virtual circle. An online firestorm is defined as the sudden discharge of large quantities of messages containing negative WOM and complaint behavior against a company in social media networks (Pfeffer, et al.,2014). Therefore, this study can model the situation. Suppose . 22 .

(24) . that person A first encounters information online indicating that company B is responsible for the crisis. The echo chamber effect leads A to want more information to understand the event, but the filter bubble effect causes A to gather more opinions against company B. Gradually, A is affected by many people who feel company B is immoderate; therefore, A decides against company B. Moreover, A tries to spread negative information about company B to persuade others. According to unrestrained information flow factors, A can be a friend or follower of tremendous numbers of people in social media. Therefore, the negative information from A potentially can spread to many people. Those people affected by A will enter the virtual circle and later affect other people. Because of the fast transportation speed and great information volume characteristics of social media, many people engage in the virtual circle within a short time, finally resulting in an online firestorm.. knowledge persuasion propagation Virtual Circle . afEirmation Figure 2. The forecast model of online firestorm As the model shows, this study can see information repeatedly spread by people on the Internet affecting many people’s potential decisions. Therefore, this study wants to determine what a company can do to manage online WOM or how to address the crisis after encountering an online firestorm. 2.7 Strategy to manage and response to negative EWOM 2.7.1 What a company (candidate) can do to manage the EWOM Because there did not have any research talking about manage online firestorm, this . 23 .

(25) . study selected the research about managing EWOM to observe the possible answer. Litvin et al. (2008), focusing on hospitality and tourism, refer to two major categories to manage electronic WOM: informational and revenue generation. (1) From an informational perspective, they mentioned that marketers should gather discussion and feedback created online. Gathering information helps marketers enhance customer satisfaction by improving the product, solving problems, discovering the experience of customers, analyzing competitive strategies and monitoring company image. (2) Conversely, from a revenue generating perspective, they mentioned that marketers spread good WOM of a company through email, websites, blogs, virtual communities, newsgroups, or product review sites for themselves, allowing potential customers to see good comments about their company. Additionally, they can spread negative WOM about their competitors. Because the technology is easily accessible to everyone and is difficult for customers to detect, some companies will choose to use the Internet to harm others or improve themselves. Although such revenue-generating strategies are positive proactive marketing activities, these behaviors create some ethical concerns. Many such marketing activities can be classified as ‘stealth marketing’, which Neisser (2004) defines as ‘employing tactics that engage the prospect without them knowing they are being marketed’. Haywood (1989) creates a comprehensive list of what a company can do to manage word of mouth: (1) Listen actively and question effectively Talking with customers face-to-face is the most useful approach for a company to understand their problems. (2) Take appropriate action A senior manager should contact customers or suppliers on weekly basis to provide the impression that they are respected by the company. (3) Focus on a customer/constituent orientation Customer/ constituent orientation is based on a commitment to service and quality. (4) Deliver on promises When a company advertises, it makes promises. Staff must be thoroughly aware and prepared to meet customer expectations. (5) Manage after the service Businesses should consider all customers potential opinion leaders and ask regular . 24 .

(26) . customers to join a customer panel or even build personal connections with them, making them feel that they are important. (6) Target opinion leaders Opinion leaders often have many people who read their comments, so it is important to focus closely on their comments. (7) Work with suppliers Suppliers often contact client companies, so businesses should avoid addressing such transactions casually. (8) Cooperate with competitors Friendly competition can generate positive customer reactions and WOM. (9) Help people who are seeking information A company must ensure that customers can obtain correct and precise information every time they seek it. (10) Generate interest and discussion through advertisements Make customers feel that advertisements are linked with their life and can appeal to their sympathy. (11) Train employees and managers to become more effective communicators Managers and employees should constantly be gathering information inside and outside the company and proactively spread it to interested people. (12) Plug communication leaks The entire company should have a comprehensive system to prevent confidential information leaking to the public media or competitors. (13) Determine what others are saying A company should be open minded to gather information from different people and domains such as advisers, consultants, colleges, attending industry conferences or creating development programs. There are some similarities between the findings of the studies by Haywood (1989) and Litvin et al. (2008). This study decided to use the framework provided by Litvin et al. (2008) because it is more structured and facilitates concise analysis. This study can see that what Haywood mentioned is categorized in a framework of ‘informational perspectives’ (Litvin et al., 2008). All the suggestions help a company gather opinion from customers to help it make improvements to enhance customer satisfaction and company reputation. In addition, revenue generating is another strategy to enhance the company image with customers. Marketers can automatically . 25 .

(27) . spread good WOM through email, websites, blogs, virtual communities, newsgroups, or product review sites to help potential customers encounter positive WOM. From an information perspective, during the campaign, candidates in Taiwan usually would have their own electoral office, Facebook account, advertising and press conferences to offer information, gather suggestions from voters or answer voters’ questions. They would use these opinions to adjust their public political views, and do what supporters want them to do while avoiding negative opinion or competing with competitors. In addition, from a revenue generating perspective, candidates in Taiwan spread good WOM through newsgroups and virtual communities to let potential voters see good comments from and about them. However, some candidates tend to spread negative WOM about their competitors. Such behavior is of some ethical concern; therefore, most candidates would not admit to such behavior to avoid angering voters. 2.7.2 How a company (candidates) respond to a crisis Per our definition, NWOM means that dissatisfied customers share their complaints to others to harm the company. Complaining or telling others about an unsatisfactory experience through the Internet will reach many potential customers; furthermore, if NWOM triggers an online firestorm, it will cause a crisis for the company; for example, customers switching brand loyalty, and company image will suffer tremendous harm, possibly even causing a failure. Therefore, the question is the following: how can a company address negative WOM to restore or maintain its image? Similar to Coombs’s crisis communication strategy (CCS) (1998), Benoit (1997) refer to research by Dutton (1986) and Fink et al. (1971) indicating that image is critical to corporate, government or any non-profit groups; therefore, he created an image restoration discourse to help corporations create responses during image crises. The basics of understanding the image restoration discourse include clearly defining an ‘attack’ that induces a response or a corporate crisis. There are two characteristics of ‘attack’: 1. People’s perception is more important than reality. Whether a company is truly responsible for an offensive action does not matter; if people believe the company . 26 .

(28) . should be responsible for it, then the company must face the resulting crisis. 2. Whether the action is offensive is decided by the public. Therefore, unless people believe that an action is offensive and should be blamed on a specific corporation, the corporate image would not be at risk. Image restoration discourse (Benoit, 1997) focuses on what message a corporation can convey when it suffers a crisis. The theory provides five categories of image restoration strategies to help a corporation address a crisis and repair its image. Table 2.7.2-1. The description of Image restoration discourse Strategy. Definition. Variant. Definition. Denial. Refuse to admit the. Fully deny. Comprehensive deny. corporate had done the. that corporate has do any. accused offensive. offensive action.. action.. Shifting the. Claim that another. blame. corporate should be responsible for the offensive action.. Evasion of. Evading responsibility. Evade. Corporate can claim that. responsibilit. to reduce the pressure. their behavior is. y. or the image damage. response to another’s. of the company.. offensive action to rationalize their behavior. Defeasibility. Corporate claim that they lack the relevant information or fail to control essential element of event to illustrate their mistake.. Accident. Corporate would try to convince that their action is by accidentally to. . 27 .

(29) . lighten the image harm. Good intention. Corporate could claim that their behavior is based on good intention to reduce the people’s dissatisfaction.. Reduce. If the corporate is. offensivenes. already accused by. positive bolstering to. s. offensive action, they. offset the negative image. could still try to. or WOM of company.. reduce the negative. Bolstering. Minimize. affection on corporate.. Corporate could use. Corporate could try to minimize the affection of the offensive action they done.. Differentiation. Corporate could compare the behavior with some similar but more offensive action to make people reduce the negative image of company.. Transcendence. Corporate could explain their action to more positive content to change the impression of people.. Attack. Corporate could attack. accusers. accusers to reduce the conviction of accusers.. Compensation. Corporate could offer money, service or product to compensate people and offset the negative image.. . 28 .

(30) . Corrective. Corporate could make. action. a commit that they. None. None. None. None. would fix the situation before crisis and promising the event would not happened in the future. Mortificatio. Corporate could. n. apologize for their fault and beg for forgiveness of public.. This study chose to use Benoit’s image restoration discourse to analyze what strategy a candidate might use to respond to an online firestorm because image restoration discourse clearly classifies responses. Thus, it can help the content of our analysis be more accurate.. . 29 .

(31) . Chapter 3: Methodology 3.1 Research approach This research used content analysis as the research methodology. This study collected data on several online firestorm events that happened during the nine-in-one political election to understand how candidates reacted in responding to online firestorms. Content analysis is a research method that facilitates making valid inferences from collected data, then surfacing knowledge, new insights, or a practical guide as a research contribution (Krippendorff, 1980). Hsieh and Shannon (2005) mentioned that content analysis is a widely used method for qualitative research; it has been used in communication, journalism, sociology, psychology and business (Neuendorf, 2002). 3.2 Data collection 1. Electronic news The data were primary selected from electronic news and celebrity Fan pages or websites. Most of the news can be accessed online, and most electronic news provides a forum for people to discuss. Therefore, through searching the news online, this study can obtain not only news of the candidates but also opinions from the viewers. In addition, some candidates do not only respond to online firestorms through the public media; some have their own fan pages or websites to respond to the public. Therefore, this can be one data resource. 2. Online firestorm events This research focuses on the online firestorm events that happened during a nine-in-one political election launched in Taiwan; therefore, this study must collect data on the online firestorm events. However, this study initially wants to ensure that the event with NWOM actually triggered the online firestorm. Therefore, this study needs a measurement to help us determine which events are actually online firestorms. For such a measurement, this study must clearly examine which factors cause an online firestorm. In a review, Pfeffer et al. (2014) summarized seven factors that cause opinions to spread: speed and volume, binary choices, network clusters and echo chambers, unrestrained information flow, lack of diversity, cross-media dynamics and network-triggered decision processes. Because ‘binary choices’ is a design characteristic of social platforms, it is not a helpful factor for our measurement. In addition, ‘Network clusters and echo chambers’, ‘unrestrained information flow’, . 30 .

(32) . ‘lack of diversity’ and ‘network-triggered decision processes’ phenomena are seen only in people’s social platforms, they are also difficult to measure. Therefore, this study decided to use ‘cross-media dynamics’ and ‘speed and volume’ to design our measurement. After experiencing the online firestorm event, if there were dissatisfied people spreading NWOM online, then an increasing number of people would encounter the information. As the network-triggered decision processes this study discussed occurred, people who encountered the news would want to ascertain what really happened and make up their mind. Many decision makers are not passive; they actively seek information (Haywood, 1989), and one of their primary information resources is the Internet. According to our research, INSIGHTXPLORER (2014) and Global Views Monthly (2013) both indicate that the primary Internet behavior in Taiwan is searching for information or news. Therefore, this study can believe that an event keyword being searched for many times provides the important characteristics ‘speed and volume’; thus, this study can consider this event the trigger of the online firestorm. To determine the searching popularity of the specific keywords, this study use Google Trend to see the situation about how specific keywords be searched during the specific timespan and period. Google Trends is a service provided by Google to observe the volume of searches and news articles handled by their search engines. Choi and Varian (2012) noted that the information Google presents is not the real searching volume; it is a relative number that provides a comparison with the other days a keyword has been searched. Google uses the numbers 1 to 100 to present keyword search volumes; the day that was searched the most frequently in a particular timespan is assigned a value of 100. Based on the social sharing literature, this study know that people tend to share information shortly after an event happens—over 50% even on the same day (Rimé et al., 1992). Therefore, this study assumes that an online firestorm will form within no more than least five days. In addition, there are many news stories during the political election; therefore, to avoid news overlap altogether, five days is suitable for the measurement. This study analyzes Google Trend of the online firestorm cases mentioned in table 2.6-2. to make our measurements. . 31 .

(33) . Table 3.2-1. Google Trend analyze on table 2.6-2. online firestorm cases Event. Keyword. Average Google. Top two highest. Trend values five. Google Trend values. days after events. during the five days after events. 2012/1/18. McDonald. 49.6. 30,28,30,60,100. McDonald Twitter event. twitter. 2011/12/31. ING-DiBa. 53.6. 25,30,100,67,46. Qantas. 74.6. 82, 79,76,71,65. ING-DiBa Vegetarian event 2011/11/21 Qantas Twitter event From the table, this study can see that the lowest average Google Trend value of the cases is 49.6, and the lowest Google Trend value happened during an online firestorm is 60. Therefore, this study wants to make a measurement to measure the online firestorm: 1.. The Google Trend value of the keyword five days after the events should have at least one day up to 60.. 2.. The average Google Trend values five days after events should be at least 49.. This study assumes that if the event satisfies one of the conditions, then it can be considered an online firestorm. Therefore, this study selected some famous NWOM events that happened during the nine-in-one political election to check whether the NWOM triggered an online firestorm. The search area was set to Taiwan, and the search time limit was set to include the month of the event. Table 3.2-2. Google Trend analyze of NWOM cases in Taiwan political election Events. Date of. Keyword Average. events. Google Trend. Google. value five days. Trend. after the events. values five days after events The Taipei candidate, . 2014/9/6. Ko 32 . 37.2. 25,29,60,46,26.

(34) . Wen-Je Ko (Ko), who was previously a doctor, was accused of a tendency toward sexism. Ko was accused of money. 2014/9/10. Ko. 38.2. 26,31,61,34,39. Ko’s wife is a docter. 2014/10/. Ko. 58.6. 59,50,62,60,62. worked in pubic hospital,. 12. Ko. 64. 54,60,47,76,83. Ko. 82.8. 84,99,73,92,71. Lien. 35.8. 28,54,38,37,22. laundering (MG149 account).. helped Ko ‘s political election activity several times. However, government officer could not attend election activity so the behavior is illegal. Ko said that rejecting. 2014/10/. company “Ting-Shin” was. 14. similar to witch burning in medieval times. The witch-burning choice of words was attacked as being a poor analogy that did not express people’s anger. Ko was accused of. 2014/10/. trafficking in human. 27. organs. A Taipei candidate,. 2014/7/15. Sheng-Wen Lien (Lien) made an embarrassing mistake, telling the media that Neihu and Nangang districts are poor; actually, these districts are . 33 .

(35) . prosperous. Lien donated one hundred. 2014/8/1. Lien. 48.8. thousand NT dollars after. 42,100,46,27, 29. the Kaohsiung gas explosion; he was accused of being too stingy. Lien made a controversial. 2014/8/7. Lien. 34.2. 34,46,40,28,23. 2014/9/11. Lien. 57.8. 63,74,45,55,52. 2014/9/18. Lien. 76.4. 69,66,87,62,98. Lien said that Buddha was a 2014/11/2. Lien. 23.8. 22,41,19,18,19. advertisement, “A Taipei resident’s day”; he was accused of not being able to understand living with hardship. Lien posted a video asking that if we were a rich man, what would we want to do? The video mentioned that some people would want to have fun every day, and mentioned if we were as rich as Lien we would not want to run for Taipei mayor. However, this video caused him to be accused of not understanding citizens at all. Lien performed a working activity in the night-market. He was accused of showing off. prince before, but he chose to liberate all sentient beings. This statement resulted in him being . 34 .

(36) . accused of being shameless by describing himself as Buddha. Lien was accused of. 2014/11/5. Lien. 58.2. 24,23,97,89,58. 2014/11/. Lien. 45.6. 40,61,39,45,43. eavesdrop on Ko’s office. Lien’s father criticize Ko’s. grandfather is a japanization 16 person According to our measurements, the grey-shaded NWOM did not trigger online firestorms. In the next section, this study will focus on the events that conformed to the online firestorm measurement. 3.3 Data analysis 1. Image restoration strategies This study would use the image restoration strategies from Benoit (1997) to analyze the respond strategy of candidates. Table 3.3-1. The categorizes of Image restoration strategies Strategy. Variant. Denial. Fully deny Shifting the blame. Evasion of. Evade. responsibility. Defeasibility Accident Good intention. Reduce offensiveness. Bolstering Minimize Differentiation Transcendence Attack accusers Compensation. Corrective action . None 35 .

(37) . Mortification 2.. None. Using Opview insight clowd service to collect the online comments. Opview social WOM database service launched by eLand technologies corporate from Taiwan, this software can collect the comments from the several social platforms including Facebook, Plurk, news, blogs, BBS, and analysis the comments or news to judge the EWOM is positive or negative. This software can help us analyze the negative and positive WOM patterns of specific key word everyday, and show the data in graphic form. This software service has been used by several public media in Taiwan (Opview, 2013), therefore this study can believe the credibility. 3.. Using Opview data to the formula to analyze the online firestorm. Rimé et al.(1992) mentioned that over 50% people tend to share information shortly after an event happens, some even spread the news on the same day. Since there are many news stories during the political election, in order to avoid news overlap altogether, five days is suitable for the measurement. Therefore, this study assumes that an online firestorm will form within no more than least five days. This study assumes that some online firestorm cases need a “boiling period” to cause the online firestorm; later, events come to a peak that this study calls the “explode period”. After the “explode period”, events gradually receive little interest from the public; this study called this time the “faded period”. However, some online firestorm cases are too intense; they skipped the “boiling period” and jumped directly to the “explode period”. This study develops a formula to describe the developing pattern of an online firestorm. Given specific keyword, Opview would collect amounts of NWOM and PWOM from several social platforms and calculate the quantity. This study use “Wen-Je Ko” as a keyword in the cases of Ko; and use ”Sheng-Wen Lien” as a keyword in the cases of Lien. This study assumes that NWOM amounts of the day i is di , assume i is a positive integer. The average NWOM comments value of the first five days after events is A .. A= . d1 + d2 + d3 + d4 + d5 5 36 .

(38) . . If d1 < A , then day one is “boiling period”.. . If di < A and di−1 to d1 < A, then i is “boiling period”. If di ≥ A, then i is “explode period”. Otherwise, i is ”faded period”.. The effectiveness of responses would be judged by the exploded days, this study assumes that an effective response strategy can help the events exploded for only several days and faded soon. However, the inappropriate response would deteriorate the situation and exploded for long time, causing the online firestorm be more severe and harm the brand of candidate.. . 37 .

(39) . Chapter 4: Data Analysis Online political election firestorms can result from three factors: fault, defect, and counterforce slander. To conduct the analysis, this study thus select six cases, which are presented in table 3.2-2, that correspond to the online firestorm standard. These six cases include: Ko was accused of a tendency toward sexism, Ko said that rejecting company “Ting-Shin” was similar to witch burning and Ko was accused of involving in human organs trading; Lien’s father criticized Ko’s grandfather is a japanization person, Lien donated one hundred thousand NT dollars to Kaohsiung gas explosion and Lien was accused of eavesdrop on Ko’s office. This study separates these six cases in three situations: fault, defect and counterforce slander. Faults occur when candidates lie, break the law or do something that society deems unacceptable. Defects occur when candidates do or say something inappropriate and offend the public, and such situations are shaped by candidate personalities and opinions. Counterforce slander is defined as accusatory remarks from a competitor. In this section, this study describes online firestorm events, candidate responses and use a formula to show how long each event exploded to determine which responses are most effective. In addition, this study also examines the cases to identify which features differ from online conventional business firestorms. 4.1.. Fault. In this part, this study selected two cases both candidates say something society unacceptable to see which responses is more effective in this situation. 4.1.1.. Ko was accused of a tendency toward sexism. On 6 September 2014, Ko stated that a female Jiayi candidate, Chen Yie-Chen, was beautiful enough to be a receptionist but not qualified to be a mayor. The next day, 7 September, during a speech, Ko mentioned that he did not want to be an obstetrician because these professionals only see “one hole.” The speech appalled numerous people, and he was criticized for disrespecting women and thus for being unqualified to be mayor (Liu, K. Y, 2014; Appendix 1-1). However, supporters argued that Ko was not ill-intentioned and that people must have a better sense of humor. During a major online firestorm, Ko made announcements to the public.. . 38 .

(40) . l. On 9/7 Ko interviewed by the media clarified that his speaking want to emphasize the important characteristic to be a mayor is good ability not beautiful looks. He admitted that did not convey his thinking correctly, he needs to do self-reflection (Tang, S., 2014). s. Admitted that did not convey his thinking correctly àEvasion of responsibility: good intention. s l. He needs to do self-reflectionàCorrective action. On 9/8 Ko post on FB admit that he made a mistake and will improve, emphasized most of his colleague are female, and the way he interactive with his wife can tell he never had sexism tendency (Appendix 1-2). s. Emphasized most of his colleague are female, and the way he interactive with his wife àReduce offensiveness: Bolstering. l. s. Admit he made a mistake and will improveàCorrective action. s. He never had sexism tendency àDenial: Fully deny. On 9/8 Ko’s wife, Chen Pei-Chi, also made a post on Facebook mentioned that Ko’s image be played up by somebody, and emphasized Ko is a person who work seriously and only have lowest requirements of life quality (Appendix 1-3). s. Ko’s image be played up by somebodyà Reduce offensiveness: Attack accusers. s. Emphasized Ko is a person who work seriously and only have lowest requirements of life quality à Reduce offensiveness: Bolstering. Table 4.1.1-1. Response strategy of sexism tendency online firestorm Time. Image restoration discourse. Sponsor. 9/6. No. No. 9/7. s. Evasion of responsibility: good intention. s. Corrective action. s. Reduce offensiveness: Bolstering. s. Corrective action. s. Denial: Fully deny. s. Reduce offensiveness: Attack accusers. 9/8. 9/8 . 39 . Ko Ko. Ko’wife.

(41) . s. Reduce offensiveness: Bolstering. Following this study use Opview insight software to see how many NWOM raised up after the event. In this case, this study uses “Wen-Je Ko” as the keyword to collect data and do the analysis. This study will mark the day when candidates respond to the events by yellow dot (Figure 3), and draw the yellow dotted line and yellow shadow to present the explode period (Figure 4). 1400 NWOM . amount of comments. 1200 . PWOM . 1000 800 600 400 200 0 . Figure 3. The response time and NWOM pattern of sexism tendency online firestorm Data source: Opview insight software data Following this study use the Opview insight data to the formula this study mentioned in chapter three to find out how many time the online firestorm needed to boil up and explode, and when it faded is related to the candidate’s response. Opview Insight data : A= (348+788+672+772+750)/5=666 Table 4.1.1-2. NWOM amounts of sexism tendency online firestorm Date. 9/6. 9/7. 9/8. 9/9. 9/10. 9/11. 9/12. 9/13. NWOM. 348. 788. 672. 772. 750. 905. 894. 700. Date. 9/14. 9/15. 9/16. 9/17. 9/18. 9/19. 9/20. 9/21. NWOM. 692. 884. 701. 777. 1228. 1020. 861. 743. amounts. amounts . 40 .

(42) . Date. 9/22. 9/23. NWOM. 699. 617. amounts The green part is boiling period; the orange part is exploding period; the blue part is the day start faded. Data source: Opview insight software data. Table 4.1.1-3. Boiling period and explode period of sexism tendency online firestorm Period. Boiling period. Explode period. Days. 1 day. 16 days. 1400 NWOM . amount of comments. 1200 . PWOM . 1000 800 600 400 200 0 . Figure 4. The explode period of sexism tendency online firestorm Following this study summarize the online firestorm characteristic of this case in table 4.1.1-4. Table 4.1.1-4. Online firestorm characteristic of sexism tendency online firestorm Online firestorm Target. s. Ko’s behavior: unsuitable speaking.. Cause. s. Fault: tendency toward sexism is society unacceptable.. Consumer Reaction . s. Complaint: criticized Ko did not respect to women at all. 41 .