以基於新型多準則模式之技術接受模型探討隱私悖論於使用者對社群信任之影響

全文

(1)國立臺灣師範大學工業教育學系碩士論文指導教授：黃啟祐博士. 以基於新型多準則模式之技術接受模型探討隱私悖論於使用者對社群信任之影響 A Hybrid Framework Based TAM for Deriving Influences of Privacy Paradox on User’s Trust on Social Network. 研究生：吳信宏中華民國 104 年 6 月.

(2) 致謝論文得以順利完成，首先感謝恩師黃啟祐教授於授業期間的指導與鼓勵，讓我在研究所生涯中，不論是知識的啟發及真理的追求皆有所領悟。過去這幾年來，感謝國立臺灣師範大學提供優良的教學環境、師資以及豐沛的圖書資源，在求學階段無論在專業領域中有所收穫，更透過師大豐沛的資源增廣自己的見聞。此學，學生吳信宏亦感謝國立台灣師範大學工業教育學系所有師長們的付出和教導，才得以對研究方向有所認知。感謝 99 工業教育學系科技管理組的所有同學，因有你們共同的學習與分享，讓我的視野更加開闊，並在學習的過程中一起激發出無限可能。學習過程中，同學的相互扶持，成為此生難忘的經驗。此外也要謝謝在學習期間曾經幫助過我的每一個人，因為有你們的鼓勵，才能讓我順利完成論文。最後，感謝小孩在求學階段的體諒極支持，更衷心感謝老婆英秋學習階段的體諒與鼓勵，讓我能無後顧之憂的情況下完成學業獲取豐沛的知識，順利完成學業。僅以這本論文獻給我最愛的家人。. 吳信宏謹致民國一百零四年八月 I.

(3) 摘要社群網站在全球掀起一股熱潮，成為人際聯繫互動之新平台；隨著社群網站功能日增，使得社群網站更容易收集、儲存和發送個人的資訊，隱私也成為使用者與網站經營者關心的議題。經營者為追求利潤最大化，需考量如何充分運用使用者的資訊於行銷和其它網站功能上，同時，使用者也關心於揭露詳細個人資料於社群網站的同時，個人資訊被濫用的程度；由於社群網站經營者亟欲充份利用資訊之企圖與使用者對於資訊揭露之顧慮往往存在差距，因此產生「隱私悖論」相關議題，而隱私悖論於使用者對社群網站之信任與進一步接受該社群網站亦產生直接影響。雖然「隱私悖論」議題對於社群網站之經營管理日益重要，但少有學者探討「隱私悖論」如何影響社群使用者對社群網站之信任與接受，因此，本研究擬探討社群網站中使用者與網站經營者間對使用者的隱私關注程度是否存在顯著差異，並分析此關注程度的差異是否將進一步影響使用者未來持續使用社群的意願。為達成此目的，本研究擬導入技術接受模型 (Technology Acceptance Model，TAM)，以使用者對社群網站之隱私顧慮、制度信任(含情境常態與結構保證)、計算性信任、與熟悉度做為外在變項，探討「隱私悖論」是否將影響使用者未來持續使用某社群網站；此外，研究發現「認知有用性」與「認知易用性」亦影響新型資訊科技產品之使用意圖，因此，本研究亦導入此二因素作為研究變項。為分析影響使用者信任並進而接受社群網站，持續使用的因素，本研究擬針對社群網站之一般使用者與專家分別偏最小平方法 (Partial Least Square，PLS)與基於決策實驗室之網路流程法(Decision Making Trial and Evaluation Laboratory Based Network Process，DNP) 建構影響社群網站技術接受因素與使用者行為間之關係架構，並比較專家與一般使用者對於「隱私悖論」於使用者信任社群網站並進一步接受網站、持續使用之重要因素。本研究之結果將可作為社群網站經營者於行銷與服務策略之用。. 關鍵字: 社群網站、技術接受模式、隱私悖論、偏最小平方法、決策實驗室法、基於決策實驗室之網路流程法、多準則決策分析。. II.

(4) Abstract Social network sites surged recently all over the world and have become new platforms for intimate communications. As the functionality of social networks was enhanced, users’ own information can be collected, stored, and manipulated much more easily. Privacy concern has thus become the most concerned issue by both users and website operators. The operators intend to maximize the profits and need to consider how users’ confidential information can be fully utilized in marketing and aspects of social network operations. At the same time, users usually concern over the misuse of private information by the website operations at the moment when disclosing individual details on social networking sites. Apparently, a significant gap exists between the website operators’ intention to fully utilize the private information as well as the users’ privacy concerns about disclosing information on the social networking sites. Such cognition gap, or the "privacy paradox", influences users’ trust on a specific social networking site directly and further influence users’ acceptance and continuous usage of the site Albeit such privacy paradox issues have been becoming daily important for the social network operators as well as users, very few scholars tried to uncover how privacy paradox influences users’ privacy concern and the social network sites structural assurance, which will influence the user whether to accept of social network sites or not. In this study, TAM (Technology Acceptance Model) is the theoretical basis, applying users' private disclosure behavior; disclosure risks perception, and the extent of privacy settings in the social networking sites as main variables. Whether it influences users continued interactive in the social networking sites in the future. In addition, in the past study found that perceived usefulness, perceived ease of use and the interaction strength for modern technology III.

(5) services or products influencing use intention, so adding these factors as research variables. It conducted a questionnaire survey targeted at end users and experts, respectively, with Partial Least Square (PLS) and "new multi-criteria model" (Decision Making Trial and Evaluation Laboratory Based Network Process, DNP) as the construction and analysis. To explore which variables influence users in social networks in the future continued interaction intention, and comparing the two methods, end users and experts to that site's privacy concerns for the social networking site, privacy setting mechanism, and technology acceptance at the similarities and differences. Results of this study, the author hopes to provide the site operators in the marketing and service strategies to develop privacy and competitive ref.. Keywords: Social Networking Sites, TAM (Technology Acceptance Model), Privacy Paradox, Decision Making Trial and Evaluation Laboratory (DEMATEL), Analytic Network Process (ANP), DANP (Decision Making Trial and Evaluation Laboratory Based Network Process). IV.

(6) 目錄致謝………………………………………………………………………I 摘要……………………………………………………………………II Abstract………………………………………………………………..III 目錄....................................................V List of Table............................................................................................ VII List of Figure…………………………………………………………….IX Chapter 1 Introduction …………………………………………………...1 1.1. Research Background…………………………………………………….…1. 1.2. Research Motivation………………………………………………………..2. 1.3. Research Objectives. ……………………………………………………….4. 1.4. Research Procedure…………………………………………………………4. 1.5. Research Contribution and Limitation……………………………………..5. 1.6. Research Framework……………………………………………………….6. Chapter 2 Literature Review…………………………………………….7 2.1. Social Networking Sites Definition…………………………………………7. 2.2. Privacy Concerns……………………………………………………………8. 2.3. Factors Influencing Users’ Trust of Social Network Sites………………..12. 2.4. Technology Acceptance Model (TAM)……………………………………15. 2.5. Research Hypothesis……………………………………………………....17. Chapter 3 Research Methodology……………………………………...19 3.1. Modified Delphi Method…………………………………………………..20. 3.2. Decision Making Trial and Evaluation Laboratory (DEMATEL)…………22. 3.3. DEMATEL based Network Process (DNP)……………………………….26 V.

(7) 3.4. Partial Least Squares (PLS)……………………………………………….30. Chapter 4 Empirical Study…………………………………………….35 4.1. Industry Background and Research Problem Description……………….35. 4.2. Deriving Factors for Social Networking Site by the Modified Delphi…36. 4.3. The Casual Relationships and Weight Derivations by the DNP………..38. 4.4. Basic Data Analysis and Questionnaire Response Rate (Mass Users)…42. 4.5. Description Statistics Analysis (Mass Users)…………………………...44. 4.6. Statistics and Analysis Confirmatory Factor Analysis and Reliability Analysis (Mass Users)…………………………………………………......45. 4.7. Verification of Hypotheses and Discussion of Empirical Results (Mass Users).........................................................................................................49. Chapter 5 Discussion…………………………………………………51 5.1. Managerial Implication………………………………………………….51. 5.2. Advanced in Research Methods…………………………………………58. Chapter 6 Conclusions………………………………………………..61 6.1. Conclusions………………………………………………………………61. 6.2. Research Limitations and Directions for Future Researches……………62. References…………………………………………………………….63 Appendix A: Expert List…………………………………………....68 Appendix B 1: Questionnaire………………………………………..69 Appendix B 2: Questionnaire………………………………………..73. VI.

(8) List of Table Table 2-1 Previous TM Studies…………………………………………………….. .16 Table 4-1 The evaluative results of criteria based on the modified Delphi method…38 Table 4-2 ( ri+ci ) and (ri−ci) versus each dimension……………………………….39 Table 4-3 The weights versus each dimension……………………………………….41 Table4-4 The effective questionnaire response rate……………………………….....42 Table4-5 Sample structure……………………………………………………........... 43 Table4-6 Measurement Items, Means, Standard Variances of Main Measured Variables……………………………………………………………...........44 Table4-7 Cronbach's α value metrics……………………………………………..….45 Table4-8 Composite Reliability, AVE, Cronbach’s Alpha, Communality, Redundancy, 2. GoF and R .................................................................................................48 Table 4-9 Path Coefficient and Effective Size……………………………………….49 Table 4-10 Total Effects………………………………………………………….......50. VII.

(9) List of Figure Figure 1-1 Research Procedure………………………………………………………..5 Figure 2-1 Technology Acceptance Model…………………………………………...16 Figure 2-2 Measuring Model…………………………………………………………18 Figure 3-1 Research Model………………………………………………………......19 Figure 3-2The Analytical Process……………………………………………………20 Figure 3-3 An example of the directed graph………………………………………...24 Figure 3-4 Example of a PLS Path Model…………………………………………....31 Figure 4-1 The Influence Diagram…………………………………………………...41 Figure 4-2 The Influence Diagram…………………………………………………...50 Figure 5-1 The original model versus the construct being derived by DEMATEL.....53. VIII.

(10) Chapter 1 Introduction. 1.1 Research Background Due to the fast growing of the social networking sites, users share a huge amount of personal information and data from the site. When the site provides more and more personal information, it also increases information misuse and identity theft on other privacy issues. Privacy groups list several online risk behaviors as being harassed by unknown persons, tracking, and received a lot of junk mail, etc. Internet users feel that they are safe behind their computer, but in fact, there is no privacy (Nissenbaum, 2009). Information provided by users could be mined and stored for forthcoming reference. Lacking of privacy concern for the user, the information could potentially be used by fated employers or the government for judgment of character (Young et al., 2013). A social networking site has such a vast number of users, and thus to protect the privacy of personal information is facing many challenges. Such as users generally feel that the risks of information sharing. To meet these challenges, social networks offer a number of mechanisms to protect the privacy of users' personal information, studies show that Internet users have found a general lack of concern for the privacy mechanism (Torres, 2012). Social networking site's privacy settings scattered at different places and too complex, the need to simplify the site's privacy settings to provide a unified interface, and allows users to have more control their personal files or information. On the other hand, social networking sites have done many mechanisms to protect users' privacy and also well structure on the site; there still some users feel incomplete privacy protection. 1.

(11) Users are afraid about the misusing of personal information and the disclosing their personal information, but with lack of correlation background or lack of consideration on using the social network sites; the user might provide personal details on social networking sites. There is a gap between privacy concerns and actual privacy settings for the system. It has a "privacy paradox" situation (Trepte et al., 2011). Research and analysis related to social networking sites, the use of the social networking site Facebook has more than 4,000 students who at most a small part of the user to change the original default privacy settings (Stutzman et al., 2013). greater than 20,000 MySpace privacy settings, simply 27% are set to private to see (Thelwall et al., 2012). However, some studies show that privacy concerns would affect the user to make changes privacy settings (Walrave et al., 2012). User does not change privacy setting mechanism, part of poor website design created by the privacy interface (Livingstone et al., 2011). From past research on the social networking sites, privacy is not difficult to see the users expressed great concern for Internet privacy. Users doubt privacy commitment in social networks. User questioned the privacy mechanism of social networking sites, whether the privacy policies have the same as description, giving the user control for privacy, but also to provide them complete protection.. 1.2 Research Motivation With the fast growing of the social networking site, previous research most only explores preliminary behavior of users in social network sites. For example, online users use social sites‟ reasons and privacy concerns in a few information fields (Collins et al., 2012). However, the lack of social networking sites can affect user information privacy concerns and the site structural assurance, or different types of user information 2.

(12) privacy concerns whether there are differences in the literature. In this study, another object is to explore the social networking site's privacy mechanism. In addition to general privacy policy guarantees and third-party certification in social networks, a social networking site also allows users to change the privacy settings. User can determine the level of personal information disclosure. Privacy setting mechanism is unlike an ordinary website. A social network developed privacy mechanism, whether privacy mechanism can increase user confidence in social networks. It will be the purpose of this study. This study explores the reasons behind the formation of trust in social networking sites. Although, past literature has been studied trust in information privacy concerns or other effects of the antecedent for trust (Hong et al., 2013). Nevertheless, the formation of trust lacks integration the information privacy concerns and other factors in that literature. Past studies explore information privacy concerns in e-commerce. It must have the network of trading under the online shopping environment (Brandimarte et al., 2013). However, a social networking site does not involve two-party transactions in the situation. Users trust social networking site, whether he has influenced by privacy concerns. It will be explored. This study explores the user's information privacy concerns. It will affect user's satisfaction and intention of sharing information in social networks. There are also exploring the social networking site's privacy mechanisms, and other factors influence satisfaction. Whether, the user can generate more trust. Exploring users have unusual views for information privacy concerns in different social networking groups. The preference of widespread users will be predicted by Lead User Method. The differences between lead users and mass users towards the social networking site are infrequently discussed. The difference between the Partial Least Square (PLS) and the DEMATEL based Network Process (DNP) are scarcely discussed. 3.

(13) 1.3 Research Objectives Based on the background and motivations mentioned above, this study aims to investigate privacy concerns, structural assurance, perceived ease of use and perceived usefulness whether it influences users‟ trust and satisfaction in social networking sites. The purposes of the research stated as below respectively. First of all, users have different social network groups on the website, whether they have different personal privacy concerns, including gender, age, educational level and occupation. Second, users feel about privacy concerns, whether it influences users‟ satisfaction and trust in social networking sites. Whether it influences users continued the intention of sharing information in social networking sites in the future. Third, whether social networks‟ privacy mechanism can increase users‟ satisfaction and trust, including privacy settings, privacy policy statements, and third-party certification. Fourth, figure out the vital factors which will influence the acceptance of SNS by using PLS, DEMATEL and DNP methodology. Fifth, measure and analyze the preference of lead users and mass users. Sixth, measure and analyze the preference of SNS users. At last, predict the preference of mass users by lead users.. 1.4 Research Procedure As of the research, the author will first set the object and the region of the research. At the second step, this research will do literature review to help to set up the research framework. The fourth step will use the research framework to set up the hypothesis and design for the questionnaire. After designing the questionnaire, the research we search for the expert help by using modified Delphi to verify whether the research framework makes sense or not. With the expert help, the research can issue the mass user 4.

(14) questionnaire and expert questionnaire to calculate the path coefficient by using PLS, using DEMTEL to build up the casual relationship between every dimension, and then DNP method will be used to figure out the weight when the user selecting the social network site. The entire research procedure will show in Figure 1-1.. Figure1-1 Research Procedure Note. This Study. 1.5 Research Contribution and Limitation The four expected main contributions of this research are: (1)Probing into the question that whether different social groups would influence users‟ privacy concerns to share personal information; (2) Probing into the effect of privacy concerns (collection, control and awareness) to users‟ satisfaction and trust on using social networking sites are our targets; (3) Probing into the effect of privacy mechanism (privacy settings, privacy policy statements, and third-party certification) to users‟ satisfaction and trust 5.

(15) on using social networking sites are our targets; and (4)The suggestions and recommendation will be made as operators and development teams‟ reference based on the result of the present research. However, the limitation of this research will be the amount of present experts and end users due to the limited number of attainable Taiwanese experts and end users who could be gotten into contact in the related fields. The foreign experts, lead users and mass users couldn‟t be investigated in this research. Only the experts and users in northern Taiwan could be investigated in this research.. 1.6 Research Framework The rest of this research is organized as follows. In chapter 1, introduction will be explained the research background, motivations and objectives. In chapter 2, literature will be reviewed on social networks, definition private paradox, trust and satisfaction for SNS. In chapter 3, research methods will be shown research framework, the definition of the variables measured way, and describes the design of a questionnaire study and process, and which tools to use for data analysis. In chapter 4, hypothesis development and data analysis and results will be shown data analysis process and results. Finally, according to the findings from this research, conclusions and recommendations will be presented in chapter 5. Review the background and derive the criteria by literature review. Elicit the users‟ requirements by the modified Delphi method. The lead users‟ preference will be demonstrated by the DNP method. The mass users‟ preference will be demonstrated by the PLS method. The difference between lead users and mass users will be compared.. 6.

(16) Chapter 2 Literature Review The main purpose of this study was to understand the social networking site as a platform environment, the user's privacy concerns, privacy mechanisms, antecedents of trust and technology acceptance, whether it influences users' trust and satisfaction in social network sites, and whether users continue to share personal information. This chapter will discuss relevant theoretical basis and previous literature. Section 1 includes the definition of social networking sites. Section 2 explores privacy paradox from privacy concerns and privacy settings in social network sites. Section 3 explores influencing users' satisfaction and trust from antecedents of trust and factors of satisfaction. Section 4 explores the construction of relationship quality that it influences users' satisfaction and trust. Section 5 explores influencing users to share the intention of personal information and improvement strategy.. 2.1 Social Networking Sites Definition Social network sites provide a group for people with similar interests and activities, to establish and consolidate the friendship on the social networks. „„Social network sites‟‟ is the personal construct their own background information (profiles), and public display that he and his knowledge of the same to participate in the site's members link to an online environment, and using this to see the link network to make new friends, dating , and other business(Ellison et al., 2013). In the use of terminology is often used „„Social Network Sites‟‟ or „„Social Networking Sites‟‟, Scholars have pointed out difference between two words that "Networking" more emphasis on establishing the first link in a strange relationship. Although social networking sites have such a 7.

(17) situation, this is not its main function. So most people use the „„Network‟‟ than to use the „„Networking‟‟ (Ellison et al., 2013).Previous literature defines social network sites as web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system; (2) articulate a list of other users with whom they share a connection; and (3) view and traverse their list of connections and those made by others within the system. The nature and nomenclature of these connections may vary from site to site(Ellison et al., 2014).Scholars also define for the public display of social networks represent the meaning of the following points: (1) things to categories together, can you open the friends list guess your socio-economic status, political rationale, music taste and so on; (2) public display of these friends worked for you within the data field in the background information provided by endorsement; (3) public displays of these friends are willing to represent you in your social network friends with the joys of trouble and woe friendship; (4) The endorsement of these has a friend for your personal information is worry-free to move the future of people has deliberately fake; (5) As the new technology brought more and more inexpensive and slight communication opportunities, personal social network will expand into a large and heterogeneous network. The individual's self-presentation needs to face the viewer, I. not compatible.. 2.2Privacy Concerns Information privacy concerns are the organization how to use and protect personal information of the general concerns. Previous research has raised concerns about several dimensions of information privacy measure, namely, collection, unauthorized secondary use; improper access and errors are described as follows (Tan et al., 2012). Collection 8.

(18) covers a wide range of concerns that consumers will be a large number of identifiable personal information is collected if appropriate. 1.. Errors refer to the consumer concerns about whether the company can be intentional or accidental errors protect their personal information. The other is the consumers using the wrong information may cause harm to the individual concerns.. 2.. Unauthorized secondary use reflects the concerns of consumers for the purpose within the organization in order to collect some personal information, but within the organization or outside with other purposes. This concern can be divided into external or internal was used again to discuss this merger.. 3.. Improper access is an individual for personal information may be obtained by non-authorized unit or visits the concerns. Past study mentioned above consumers' information privacy concerns (CFIP). dimension of proven and successful network of direct sales in the non-application of the market (Preibusch, 2013). However, with time evolution, the rise of the Internet in recent years, the development of e-commerce market is booming, scholars have proposed Internet users for personal information concerns, most likely with non-line consumers may be different. So check these dimensions of information privacy concerns can move to e-business environment is very important. Smith et al. (1996) also show that "these dimensions are not absolute nor are fixed, the advocates of these ideas, and academic customers might change over time.. "Previous study, the information privacy concerns in all dimensions by the literature review later, and according to the characteristics of Internet users as Internet users to re-organize the information privacy concerns, and define the dimensions of the three operable, were collection, control and awareness, respectively, as described below (Preibusch, 2013): 1.. Collection is internet users‟ information privacy concern the first dimension, referring to the collection of information, whether legal or illegal actions. The 9.

(19) dimensions of information privacy concerns are defined as "personal care on individual specific information to be collected relative to the degree of benefit they receive" (Staddon, Acquisti, & LeFevre, 2013). In the direct sales market, many customers are willing to share their information even though in exchange for something, but they still want to limit access to information about them. (Zhou &. Li, 2014). 2.. Control is the second dimension, is also an important dimension, as consumers when they provide personal information, is often accompanied with high risk. If the individual's information in some areas has been given control through the power, they think this is a fair process (Zhou & Li, 2014). If a particular set of resources to allow people to collect their personal information and give them to opt out the opportunity, they will be more concerned about data collection.. 3.. Privacy Policy perception (Awareness of privacy practice) is IUIPC third dimension refers to the "consumers care about him aware of the privacy policy of the organization" (Al Sawah et al., 2015). After the majority of online users in the uncertain status of use of these data, it will refuse to disclose their personal information. (Platenius, Arifulina, Petrlic, & Schäfer, 2014).. 2.2.1. Privacy Settings. Social network site's privacy settings, most of which revolve around the following four levels to provide the user fine-tuning options. Users can share permissions are set to "friends," "friends of friends" or "owner," in addition to users who do not want to disclose their personal information; you can also select "personalize" the sharing of personal data permissions set to "own or limited users"(Liu et al., 2011). Social network site will share users' privacy rights, divided into the following main blocks, namely, "basic information list," "share on the site," "applications and websites," and "blacklist" four privacy setting blocks (Facebook Data Use Policy, 2014). 10.

(20) Privacy setting blocks provide users to understand what they are setting relevant content, but also allow users to do the privacy settings for each adjustment detail. The following is the privacy settings block of information shows: 1.. List of fundamental information, to allow users to identify each other and keep in touch, sharing some primary information access is set to "all." Such as name, ID photos, and sex is to make real friends to identify when a user message in graffiti on the wall when, in order properly display (Liu et al., 2011).. 2.. Sharing on the site, to allow users to save past too many steps, social networking sites offer four different privacy permissions (everyone, friends of friends, only friends and recommended setting) for the user quickly apply the default privacy settings in order to save more cumbersome procedure. There is also personal to do setting for small items. However, the basic core remains the user to control the privacy settings; it determines which users and which applications can see the personal data and share content (Nosko et al., 2012).. 3.. Applications and sites, User can adjust what data type is open to including search engine applications, including use at the site (Madden, 2012). These applications can access to basic public information, and other settings for "Everyone" to share content. From here, users can also view applications currently in use are the list, and remove any unwanted application that re-use, or completely shut down the use of social networking applications, enabling any user applications is not available information (but not while the withdrawal of any application and website).. 4.. Blacklist, privacy settings control who can contact through the social networking site users can choose to block the name and email address. Users can reject the blockade between the households included in your blacklist to prevent these people in the community Web site and your contacts, or browse your data, are you blocked will not be able to interact with you on social networking sites. You also can specify some friends of mine reject an application or game to send the invitation, and can use this setting to block specific applications and trying to get your information and contact you (Chenju et al., 2014). Social network sites provide these privacy rights mechanisms, the statement also. said they would not share the user and the advertiser personal information, and explain 11.

(21) the lock advertisers target is completely anonymous. Advertisers can choose to target certain features as advertising, social networking site merely for those ads automatically with the users. Advertisers will not get any personal information that they will at most receive the anonymous data reports (Facebook Advertise, 2012).. 2.3 Factors Influencing Users’ Trust of Social Network Sites This section explores influencing users' trust from antecedents of trust and the definition of trust. It includes trust and privacy relationship between users' trust on social networking sites.. 2.3.1. Trust. Trust is defined as “the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party”(Holahan et al., 2014). For in person, trust is a critical determinant of sharing information and developing new relationships (Van Lange, 2015). Trust is also important for successful online interactions (Sherchan et al., 2013).. 2.3.2. Antecedents of Trust. The antecedents of trust refer to trust or foundation trust has led the process of trust antecedent‟s influence whether individuals trust each other's inner feelings. Here, trust can also be seen as the antecedent of trust external variables. Antecedents of trust over the past have been defined by many scholars. In this study only explored Gefen (2015) that he brings the Institute's confidence in the antecedent together, described as follows: 12.

(22) 1.. Personality-Based trust or propensity to trust refers to the others believe or not believe and trust in other people's tendencies. If the introductory stage of the consumer, not to go through the experience and business knowledge, familiarity, or assess the business, the antecedent is particularly important at this stage, is the first phase of the trust. However, when consumers and businesses began to have real interaction in the process, they will be influenced by the tendency of the trust will begin to become smaller (Gefen, 2015).. 2.. Cognition based trust test viewed by the first impression of the trust, rather than through personal experience as the interaction. The first impression is through categorization and illusions of control are formed. Type of process is a fresh man on the face of things; it will taste Wei's grouping, according to modern things about the past or the reputation of their own stereotypes for new things to build trust in faith. The illusion of control refers to the uncertain environment; it will confirm whether, Wei was under control, and through this process to determine whether to trust each other. However, due to perceived control obtained during the course may not be practical, this process will also affect the formation of trust (Gefen, 2015). The cognitive basis of trust and character-based trust is the trust and the initial formation of the more relevant. As the object of study is the desire of those who have used Facebook users, Facebook has started with the user interaction; the two antecedents of trust beliefs about the impact of the trust will gradually diminish (Gefen, 2015).. 3.. Knowledge-based trust in this case familiarity, is for something, when and where, and how will the experience of place. In e-commerce environment, consumers are familiar to the business, and trading is due to the experience of the past accumulation. Familiarity for trusted web sites, can effectively increase consumer confidence, as consumers can successfully interact at the site prior to experience, accumulated more knowledge to increase their familiarity with the site, when the cumulative more successful trading experience, users will experience because I believe these websites (Gefen, 2015).. 4.. Calculative-based trust by the formation of a rational assessment of costs and benefits, the resulting co-operative relationship, or to deceive the other party. In this view, under the assumption that although other people may not necessarily is good, but they are rational, and will calculate the way in line with their interests, so 13.

(23) they do not hurt themselves. It can also be calculated upon the basis of trust, which has been generated by the economic analysis of the relationship that consumers believe that their businesses to take behavior if it is not worth it, engaged in speculation under the cost will be greater than interest, and therefore, choose to believe them. E-commerce environment, consumers believe that online businesses such as, they had to deceive, contrary to promise the price will be bigger than the punishment they received a profit, so the assessment will be selected through trusted network businesses will keep their promises (Gefen, 2015). 5.. System-based trust (institutional-based trust): a symbol of trust in e-commerce as the aforementioned study (Mcknight, Carter, Thatcher, & Clay, 2011), this is a personal trust in the antecedent or through third-party organizations to ensure the security felt by the formation. Gefen et al. (2015) study, this construct and then dismantle, namely, the normal situation (situational normality) and structural assurance (structural assurances), are described as follows: (1) "Situational normality" refers to consumer assessment of the success of the transaction. The transaction will be successful is based upon the situation is normal or line used. In e-commerce environment, because unlike traditional Internet business as a business entity, can give consumers the feeling of peace of mind, so they must be through a common interface design, and have a fixed process, so that consumers feel that all settings and processes are then (Gefen, 2015). (2) "Structural assurances" refer to consumers believe that the success of the transaction is created by security measures, such as law, business assurance, and special environment specification and so on. If the site has a certification mark, such as toll free, legal norms, or a detailed description of the privacy policy, etc., all help to strengthen customer trust (Gefen, Gefen, & Carmel, 2015).. 14.

(24) 2.3.3. Trust and Privacy Relationship between Users’ Trust on Social. Networking Sites Electronic-commerce research has found trust to be strongly related to information disclosure (Taddei et al., 2013). Trust is also a central component of social exchange theory (Nunkoo et al., 2012). Social exchange theory presents a cost-benefit analysis with respect to social interaction. If the exchange is perceived to be beneficial, then the individual is likely to enter an exchange relationship. Trust be used in the calculation of perceived cost. High trust would lead to a perception of low cost, and vice versa. Studies of interpersonal exchange situations confirm that trust is a precondition for self-disclosure, because it reduces perceived risks involved in revealing private information (Taddei et al., 2013). Social networking sites record all interactions, and retain them for potential use in social data mining. Offline, most social transactions leave behind no trace. Therefore, these sites need explicit policies and data protection mechanisms in order to deliver the same level of social privacy. Since online social privacy is harder to guarantee, does a higher level of concern for Internet privacy affect the use of collective networking sites?. 2.4 Technology Acceptance Model (TAM) The TAM provides a theoretical basis for determination of the external variables that affect users‟ internal beliefs. (Davis et al., 1989), attitudes, and intentions, thereby. affecting users‟ information technology usage behaviors. (Chang et al., 2015). It stated. that user attitude toward information system/information technology was determined by two particular beliefs: perceived usefulness (PU) and perceived ease of use (PEU). The 15.

(25) user attitude in turn leads to behavioral intention (BI) to use (accept) technology, and then generate the actual usage behavior. Venkatesh and Davis (2000), and Venkatesh and Morris (2000) observed that PEU and PU were influenced, to some extent, by external variables, and extensions of the TAM have been introduced by comprehensive study on the determinants of PEU. Furthermore, in distinctive research applications, or when predicting or interpreting the acceptance of technology through different theories and studies, external variables should be elaborated to expand the discussion about the degree of acceptance of technologies. The model proposed at Figure 2-1 and the previous studies proposed at Table 1-1.. Figure 2-1 Technology Acceptance Model Note. This Study. Table 2-1 Previous TAM Studies Authors Teo et al.. Context. Internet. Extended perceived variables Perceived enjoyment (PET). Result PU->all usage dimensions PET->frequency of Internet usage, daily Internet usage PE->all usage dimensions. Luo and Strong. Groupware. Perceived critical mass (PCM). PE, PU, and PCM-> BI. Moon and Kim. WWW. Perceived playfulness (PP). PE, PU, and PP-> BI. Mathieson and Chin. Bulletin board system. Perceived user resources (PR). PU and PR-> BI. Chen et al.. Virtual store. compatibility. PE, PU, and compatibility-> BI. Venkatesh et al.. System. Intrinsic motivation (IM). PE, PU, and IM-> BI. Legris et al.. Literature review. Voluntariness, experience, subjective. TAM2. norm, image, job relevance, output quality, and result demonstrability. Note. This Study.. 16.

(26) 2.5 Research Hypothesis According to the research been proposed at 2007, the research has shown that privacy concern (PC) will influence structural assurance (SA), privacy concern (PC) influence perceive ease of use (PEU), structural assurance (SA) influence perceive ease of use (PEU), perceive ease of use (PEU) affect perceive usefulness (PU), perceive ease of use (PEU) influence intend to use (IU), and perceive usefulness (PU) will influence intend to use (IU ) (Davis, Bagozzi, & Warshaw, 1989; Gefen et al., 2015; Kwon & Wen, 2010). According to the previous research, we made six hypotheses. Within the hypothesis, we will assure whether privacy concern and structural assurance these two external variables will influence the TAM model or not. All of the hypotheses are listed below and the measuring model proposed Figure 2-2. H1: “Privacy Concern (PC)” will influence “Structural Assurance (SA) has a positive effect” (Kwon & Wen, 2010). According to the past research, the user selecting for the SNS will consider whether the platform can provide more security place. As of this reason, the SNS designer will try to reinforce the site structural assurance, to provide a more security place for the user to use. H2 : “Privacy Concern (PC)” will influence “Perceive ease of use (PEU)” and has a positive effect (Kwon & Wen, 2010): As for the pervious study, the result has shown that the for the SNS user, they will consider their privacy when using on the social networking site. For the user consider about their privacy, the SNS developers will reinforce the platform to protect the user personal information which disclose on the Internet 17.

(27) H3: “Structural Assurance (SA)” will influence “Perceive ease of use (PEU)” has a negative effect (Gefen et al., 2015; Kwon & Wen, 2010): According to the previous, the result has shown that SA will have a negative effect toward PEU. The reason is because if the SNS being more structural, it will be more complicated for the user to use. As a consequence, these two dimensions has a negative effect. Hypothesis 4: “Perceive ease of use (PEU) will influence Intended to use (IU)” has a positive effect (Davis et al., 1989): According to the previous research, the PEU will have a positive effect toward IU. Hypothesis 5: “Perceive ease of use (PEU) ” will influence “Perceive usefulness (PU)” has a positive effect (Davis et al., 1989: According to the previous research, the PEU will have a positive effect toward PU. Hypothesis 6: “Perceive usefulness” will influence “Intended to use” has a positive effect (Davis et al., 1989: According to the previous research, the PU will have a positive effect toward IU.. Figure 2-2 Measuring Model Note. This study 18.

(28) Chapter 3 Research Methodology This study is divided into two phases conducted the survey, the first phase of Technology Acceptance Model questionnaire, the second phase of DNP survey. This modification Gefen et al. (2014) antecedents of trust, privacy concerns and a combination of recent years has been verified in the "IT use" (usage) has considerable explanatory power of the technology acceptance model (Technology Acceptance Model, TAM) (Davis, 1989), and modify the physical environment of the service dimensions for the virtual environment, the "website usability (usefulness)" and "Web site usability (ease of use) "into the higher-level" network service "construct in, and additional" asset specificity (asset specificity) "and" product disconfirmation, "these two exogenous variables, made Figure 3-1 chart shown in research, an attempt to privacy concerns, antecedents f trust (Gefen et al., 2015) and the Technology Acceptance Model (Davis et al., 1989) point to explore users‟ intension of sharing personal information on social networking sites.. Figure3-1 Research Model Note. This Study.. The analytical framework (Figure 3-2) consists of five phases: (1) reviewing related literature in social network sites; (2) establishing evaluation criteria for social 19.

(29) networking sites by using the Modified Delphi; (3) building the structure between evaluation criteria by using the DEMATEL; (4) analyzing and ranking the weight among the factors by using the ANP and finally, (5) deciding the enhancing strategies for users‟ trust on social networking sites. The analytical framework consists of four phases: (1) Elicit consumers‟ requirements through the concept of the TAM theory and the Modified Delphi method, and then construct the questionnaire. (2) The mass users‟ preference will be demonstrated by using Partial Least Square method. (3) The lead users‟ preference will be demonstrated by using the DNP method. (4) The acceptance model will be analyzed by the result of the Partial Least Square, DEMATEL and DNP.. Figure 3-2 The Analytical Process Note. This Study.. 3.1 Modified Delphi Method The Delphi method is a structured of analyze opinions inductively technique. It originally developed as a criteria interactive forecasting method which relies on experts‟ questionnaire. Collect experts‟ questionnaires in two or more rounds, and the researcher 20.

(30) provides an anonymous summary of the experts' the reasons they provided for their judgments after each round. Thus, experts can revise their earlier answers after got other member's information. It is believed by this process; the range of the research topic will be narrow, and the result will be towards the "correct or best" answer. Finally, the questionnaire round will be stopped after a pre-defined stop criterion when the number of rounds or achievement of consensus (Heiko, 2012). The Delphi method was developed from the beginning of the United States and Russia Cold War to forecast the technology impact on war. In 1944, General Henry H. Arnold creates the report which the future technological capabilities that might be used by the military purposes (Heiko, 2012). After a period of evolution, Dalkey and Helmer (1963) designed the Delphi method. After, Murry and Hammous (1995) tried to identify problems and issues that were collected from a group of technology education professionals using the Modified-Delphi Technique. The modified Delphi simplified the first round of a survey and replaced the conventionally adopted open style survey (Snape et al., 2014). The purpose of it is to save time, and experts can focus on research themes, and to improve the response of the main topic (Bryman, 2012). Murry and Hammons (1995) modified the traditional Delphi Technique by eliminating the first-round questionnaire containing unstructured questions. It is simplified to replace the conventionally adopted open style survey; doing so is commonly referred to as the modified Delphi method (Snape et al., 2014). The major modification consists of beginning the process with a set of carefully selected items. These pre-selected items may be drawn from various sources, including related competency profiles, synthesized reviews of the literature, and interviews with picked out content experts. The primary advantage of this modification to the Delphi is that it 21.

(31) (a) typically improves the initial round response rate, and (b) provided a solid grounding in previously developed work. Additional advantages related with the use of the modified Delphi technique include reducing the effects of bias due to group interaction, assuring anonymity, and providing controlled feedback to participants. (Dalkey, 1972). Brooks (1979) noted. that three mailings were usually sufficient in order to arrive at a consensus.. 3.2 Decision Making Trial and Evaluation Laboratory (DEMATEL) Decision Making Trial and Evaluation Laboratory (DEMATEL) technique was proposed by Fontela and Gabus at 1971. Considering experts‟ Opinion to solve many global complex problems in scientific, Industry, political, and economic. On verification, DEMATEL method has been applied to the relationship between the criteria and to find out criteria which play a central utility in research systematic to represent the effectiveness (Falatoonitoosi, Leman, Sorooshian, & Salimi, 2013). Moreover, the hybrid model DEMATEL combinations with other methods have been widely used in different areas such as green supplier evaluation, airline security evaluation (Tzeng & Huang, 2012), and e-learning assessment (Lu, Lin, & Tzeng, 2013). Then DEMATEL method was developed by the Battelle Geneva Institute (Chen & Chen, 2010) to analyze complex „world problems‟ dealing mainly with interactive man-model techniques; and to evaluate qualitative and factor-linked aspects of societal problems(Gabus & Fontela, 1972). The applicability of the method is widespread, ranging from industrial planning and decision-making to urban planning and design, regional environmental assessment, analysis of world problems, and so forth. It has also been successfully applied in many situations, such as marketing strategies, control 22.

(32) systems, safety problems, developing the competencies of global managers and group decision-making. Furthermore, a hybrid model combining the two methods has been widely used in various fields, for example, e-learning evaluation (Tzeng, Chiang, & Li, 2007), airline safety measurement, and handset design for next generation handset. Therefore, in this paper we use DEMATEL not only to detect complex relationships and build a Network Relation Map (NRM) of the criteria, but also to obtain the influence levels of each element over others. To apply the DEMATEL method smoothly, the authors refined the definitions based on above authors, and produced the essential definitions indicated below. The DEMATEL method is based upon graph theory, enabling us to plan and solve problems visually, so that we may divide multiple criteria into a relationship of cause and effect group, in order to better understand causal relationships. Directed graphs (also called digraphs) are more useful than directionless graphs, because digraphs will demonstrate the directed relationships of sub-systems. A digraph typically represents a communication network, or a domination relationship between individuals, etc. Suppose a system contains a set of elements,. S  s1 , s2 ,. , sn  , and particular pair-wise relationships are determined for modeling,. with respect to a mathematical relationship (MR). Next, portray the relationship MR as a direct-relation matrix that is indexed equally in both dimensions by elements from the set. S. . Then, extract the case for which the number 0 appears in the cell (i, j ) , if the. entry is a positive integral that has the meaning of: the ordered pair ( si , s j ) is in the relationship MR; it has the kind of relationship regarding that element such that si causes element s j . The digraph portrays a contextual relationship between the elements of the system, in which a numeral represents the strength of influence (Figure. 3-3). The elements s1 , s2 , s3 and s4 represent the factors that have relationships in Fig. 3-3. The 23.

(33) number between factors is influence or influenced degree. For example, an arrow from. s1 to s2 represents the fact that s1 influences s2 and its influenced degree is two. The DEMATEL method can convert the relationship between the causes and effects of criteria into a structural model (Huang, Shyu, & Tzeng, 2007). s3 3. 3. s1 1. 2. s4. s2. Figure 3-3An example of the directed graph Note. This Study.. Definition 1: The pair-wise comparison scale may be designated as eleven levels, where the scores 0, 1, 2,…, 10 represent the range from „no influence‟ to „very high influence‟. Definition 2: The initial direct relation/influence matrix A is a n  n matrix obtained by pair-wise comparisons, in terms of influences and directions between the criteria, in which. aij. th th is denoted as the degree to which the i criteria affects the j. criteria..  a11  a A   21    an1. a12 a22 an 2. a1n   a2 n    ann . Definition 3: The normalized direct relation/influence matrix N can be obtained through Equations (1) and (2), in which all principal diagonal elements are equal to zero. 24.

(34) n z  1 max  aij 1in j 1. (1). N  zA .. (2). In this case, N is called the normalized matrix, since lim N k k .  [0] .. Definition 4: Then, the total relationship matrix T can be obtained using Equation (3), where I stands for the identity matrix. T  N  N 2  ...  N k  N ( I - N )-1. where. k . (3). and T is a total influence-related matrix; N is a direct influence. . . lim N 2   N k stands for a indirect influence k  n n n n matrix and 0   xij  1 or 0   xij  1 , and only one  xij or  xij equal to 1 j 1 j 1 i1 i 1 matrix and N  [ xij ]nn ;. for i, j . So. lim N k  [0]nn . The (i, j ) element tij of matrix T denotes the k . direct and indirect influences of factor i on factor j . Definition 5: The row and column sums are separately denoted as r and c within the total-relation matrix T through Equations (4), (5), and (6). T  [tij ], i, j  {1, 2,..., n}. (4).  n  r  [ri ]    tij  n1  j 1 . (5).  n1. . n  c  [c j ]    tij  1n i 1  1n where the. r. and. (6). c vectors denote the sums of the rows and columns,. respectively. Definition 6: Suppose ri denotes the row sum of the ith row of matrix T . Then,. ri is the sum of the influences dispatching from factor i to the other factors, both 25.

(35) directly and indirectly. Suppose that c j denotes the column sum of the j th column of matrix T . Then, c j is the sum of the influences that factor i is receiving from the other factors. Furthermore, when i  j (i.e., the sum of the row sum and the column sum (ri  ci ) represents the index representing the strength of the influence, both dispatching and receiving), (ri  ci ) is the degree of the central role that factor i plays in the problem. If (ri - ci ) is positive, then factor i primarily is dispatching influence upon the strength of other factors; and if (ri - ci ) is negative, then factor i primarily is receiving influence from other factors.. 3.3 DEMATEL based Network Process (DNP) The DNP, an integrated MCDM which consists of the Decision Making Trial and Evaluation Laboratory (DEMATEL) and Analytic Network Process (ANP) was proposed by Prof. Tzeng, in which the network process for deriving the impact of each criterion on others as the weight. The DEMATEL technique was developed by the Battelle Geneva Institute: 1.. To analyze complex “real world problems” dealing mainly with interactive map-model techniques (Gabus & Fontela, 1972).. 2.. To evaluate qualitative and factor-linked aspects of societal problems (W.-Y. Chiu et al., 2010).. The DEMATEL technique was developed with the belief that the pioneering and proper use of scientific research methods could help to illuminate specific and intertwined phenomena and contribute to the recognition of practical solutions through a hierarchical structure. The DEMATEL has been successfully applied in many situations 26.

(36) such as e-business model definitions, policy definitions, global manufacturing system optimization, marketing strategies, safety problems and environment watershed plans (W.-Y. Chiu et al., 2010). The ANP is general form of the analytic hierarchy process (Saaty, 1980) which has been used in multi criteria decision making (MCDM) to can release the restriction of hierarchical structure. DNP combines the DEMATEL and ANP method, which had been interdiction in this chapter, the steps of this method can be summarized as follows: Step 1: Calculate the direct-influence matrix by scores. Based on experts‟ opinions, evaluations are made of the relationships among elements (or variables/ attributes) of mutual influence using a scale ranging from 1 to 5, with scores representing “no influence” (1), “low influence” (2), “medium influence” (3), “high influence” (4), and “very high influence” (5). They are asked to indicate the direct effect they believe a factor i will have on factor j, as indicated by dij .The matrix D of direct relations can be obtained. Step 2: Normalize the direct-influence matrix based on the direct-influence matrix. D , the normalized direct relation matrix N is acquired by using Eq. (7) n n N  vD; v  min{1 / max  dij ,1 / max  dij }, i, j  {1, 2,..., n} i j 1 j i 1. (7). Step 3: Attaining the total-influence matrix T . Once the normalized direct-influence matrix N is obtained, the total-influence matrix T of NRM can be obtained.. T  N  N 2  ...  N h  N ( I - N )-1. (8). when h   and T is a total influence-related matrix; N is a direct influence matrix and N  [ xij ]nn ;. . lim N 2  h.  Nh 27.  stands for an indirect influence matrix.

(37) n n n n 0   xij  1 or 0   xij  1 , and only one  xij or  xij equal to 1 for j 1 j 1 i1 i 1. and. i, j . So. lim N h  [0]nn . The (i, j ) element tij of matrix T denotes the direct and h. indirect influences of factor i on factor j . Step 4: Analyze the result. In this stage, the row and column sums are separately denoted as r and c vectors within the total-relation matrix T through Equations (11), (4), and (5). T  [tij ], i, j {1, 2,..., n}. (11).  n  r  [ri ]    tij  n1  j 1 . (12).  n  c  [c j ]  tij  1n i 1  n1. (13). .  n1. Where r and c vectors denote the sums of the rows and columns, respectively. Suppose ri denotes the row sum of the ith row of matrix T . Then, ri is the sum of the influences dispatching from factor i to the other factors, both directly and. th indirectly. Suppose that c j denotes the column sum of the j column of matrix T . Then, c j is the sum of the influences that factor i is receiving from the other factors.. . Furthermore, when i  j (i.e., the sum of the row sum and the column sum) ri  ci. . represents the index representing the strength of the influence, both dispatching and receiving),.  ri  ci  is the degree of the central role that factor i plays in the problem. If.  ri  ci  is positive, then factor i primarily is dispatching influence upon the strength of other factors; and if.  ri  ci . is negative, then factor i primarily is receiving. 28.

(38) influence from other factors (Huang et al., 2007; Liou et al., 2007). Therefore, a causal graph can be achieved by mapping the dataset of (ri  si , ri  si ) providing a valuable approach for decision making (Chiu et al., 2010). Now we call the total-influence matrix T  tij  obtained by criteria and C   nxn obtained by dimensions (clusters) from T . Then we normalize the T  tijD  C D   nxn ANP. T D. weights.  D  t 11  11     D   t i1  i1     Dm1 tm1 . of. dimensions. (clusters). by. using. matrix TD .. influence. D t 1j 1j. D  t 1m  1m . D tij ij.  m Dij m D Dim   , di   tij ij , i  1,..., m tim   di   tij j 1 j 1 . m D1 j d   t  1   j 1 1 j .      dm  Dmm tmm  . D tmjmj. m D   tmjmj j 1. (14) Step 5: The original supermatrix of eigenvectors is obtained from the total-influence matrix T  [tij ] . For example, D values of the clusters in matrix T , as D Eq.(14). Where if tij  D , then tijD  0 else tijD  tij , and tij. is in the. total-influence matrix T . The total-influence matrix TD needs to be normalized by dividing by the following formula. There, we could normalize the total-influence matrix and represent it as TD .. T D.  D  t 11 / d 1  11     D   t i1 / d i  i1     Dm1 tm1 / dm . D t 1j / d 1 1j.   D D t 1m / d    11 1   11 1m.  1j. D tij ij / di. D timim. ij ij. D tmjmj / dm. Dmm tmm.         D / di     i1   i1         D / dm   m1   m1. 29. D. 1j. D. D.  mjmj.       Dim  im      Dmm   mm  . D 1m.  1m. (15).

(39) Dij Dij where ij  tij / di . This research adopts the normalized total-influence matrix TD (here after abbreviated to “the normalized matrix”) and the unweighted supermatrix W using Eq. (15) shows theses influence level values as the basis of the normalization for determining the weighted supermatrix.  D   11  W 11  11  D   12  W 21  12  * W       D1m 1m  Wm1. D. D.  2121  W12. .  m1m1  W1m . D  2222  W22 D  ji ji  Wij D.  2m2m  Wm2.      D  mimi  Wim      Dmm  mm  Wmm  . (16). Step 6: Limit the weighted supermatrix by raising it to a sufficiently large power. h , as Eq. (16), until the supermatrix has converged and become a long-term stable supermatrix to get the global priority vectors or called ANP weights.. lim. h. (W * )h. (17). 3.4 Partial Least Squares (PLS) Partial least square (PLS) is a soft-modeling and classical statistical method. It originally proposed by Wold (1966) for econometrics and integrated path analysis and causal modeling method to resolve complex economic problems. PLS is a recent technique that combines features from and generalizes principle component analysis (PCA) and multiple linear regression (Abdi, 2010). In the chemo metrics fields, PLS regression first gained a huge success. Currently, PLS has been extensively applied in industrial applications, such as computer science and information, technological prediction, marketing, management, and human behavior (Berghman, Matthyssens, & Vandenbempt, 2012; Chiang, 2013; Y.-T. H. Chiu, Lee, & Chen, 2014; Land Jr et al., 30.

(40) 2011; Martínez, De Andrés, & García, 2014; Moon, Park, Jung, & Choe, 2010; Venkatesh, Thong, & Xu, 2012). For example, Joe F Hair, Ringle, and Sarstedt (2011) proposed the comparison between structure equation model (SEM) and PLS, and used PLS techniques to estimate causal models in many theoretic models and empirical data situations. The purpose of PLS is to understand or forecast a set of dependent variables from a set of independent variables. By obtaining from the variables a set of orthogonal factors called latent variables which have the best predictive power can realize the prediction and analysis (Abdi, 2010). Besides, PLS has the ability to model latent constructs that are uncontaminated by measurement error (Joseph F Hair, 2009) under conditions of non-normality and small to medium sample sizes (Chiang, 2013). In other words, it can avoid the small sample size problem in linear model analysis. Therefore, it offers some analytical advantages over techniques such as regression assuming error-free measurement (Chiang, 2013). In general, two applications of PLS approach being used are possible (Chin, 1998): (1) theory confirmation; (2) theory development. Due to the above advantages of PLS, it is often considered by researchers to be a good alternative to traditional covariance-based techniques in SEM. In order to simplify the notation of the model and in line with conventional descriptions of PLS, the latent and manifest variables being assumed are standardized so that the location parameters can be discarded in the following equations based on literature referenced by Henseler, Ringle, and Sinkovics (2009).. 31.

(41) x11. x12. 1. 1. x31. 3. x41. x13 x21. x32. 2. x22. Inner  2 Model. 4. x42 x43. x23 Outer Model (Formative Mode). Outer Model (Reflective Mode). Figure 3-4Example of a PLS Path Model.. The inner model of relationships between latent variables can be illustrated as:   B  . (18). where  is the vector of latent variable, B denotes the matrix of coefficients of their relationships, and  stands for the inner model residuals. The basic PLS design assumes a recursive inner model that is subject to prediction specification. Therefore, the inner model constitutes a causal chain system. Predictor specification reduces equation (18) to:.  |    B. (19). PLS path modeling comprises of two main outer modes: reflective modes and formative mode (see Figure 3-4). The reflective model has causal relationships from the latent variable to the manifest variables in its block. Each of manifests thereby in a certain measurement model is assumed to be generated as a linear function of its latent variables and the residual  :. X x   x   x. (20). Where  means the loading coefficients. The outer relationships are also subject to predictor specification  implying that there are no correlations between the outer residuals and the latent variable of the same block  that reduces equation (20) to:.  X x |     x. (21). The formative mode of a measurement model has causal relationships from the 32.

(42) manifest variables to the latent variable. For those blocks, the linear relationships are given as follows:.   x X x   x. (22). In the formative mode, predictor specification is also in effect, reducing equation (22) to :.  | X x    x X x. (23). Urbach and Ahlemann (2010), who suggest that reflective and formative indicators‟ measurement. First, reflective model, which is termed as an outer model, likely the functions as the measurement model in covariance-based SEM, is often employed to obtain the construct validity and reliability of measurement items. The four stages are shown below: 1. Undimensionality refers to use exploratory factor analysis (EFA) with eigenvalue to understand whether only one constructs in multi-construct empirical researches. 2. Reliability is used to conduct the assessment of the internal consistency in a construct and there are two common indexes to fit including composite reliability and Cronbach‟s alpha. 3. Convergent validity, which is used to measure the correlation of a construct‟s multiple indicators. Convergent validity is acceptable if the following criteria are met (Joseph F Hair, 2009): (i) the statistical significance of each factor loading is confirmed by a p-value of 0.5, (ii) construct reliability exceeds 0.7, and (iii) average variance extracted (AVE) is greater than 0.5 (Fornell & Larcker, 1981). 4. Discriminant validity concerns the degree to which the various constructs are distinct from one another. The formative model, which is termed as the inner model, is also often used to 33.

(43) measure and check the model of research outcomes. To effectively assess the validity of the structural model, the four criteria being suggested by Urbach and Ahlemann (2010) are shown below: 1. Coefficient of determination ( R2 ), stands for attempting to assess the explained variance of a latent variable relative to its total variance. Values of approximately 0.670 are viewed as substantial, values around 0.333 moderate, and values around 0.190 weak (Chin, 1998; Ringle, 2004). 2. Path coefficients between the latent variables should be analyzed in terms of their algebraic sign, magnitude, and significance (Huber, Herrmann, Meyer, Vogel, & Vollhardt, 2007). 3. Effect size (. f 2 ),. which is described to assess whether an independent latent variable. has a substantial impact on a dependent latent variable. Values of 0.020, 0.150, 0.350 indicate the predicator variable‟ low, medium, or large effect in the structural model (Chin, 1998; Cohen, 1988; Ringle, 2004). 4. Predictive relevance ( Q 2 ), refers to the. Q2. statistic is a measure of the predictive. relevance of a block of manifest variables. The proposed threshold value is. Q 2  0.. The predictive relevance‟s relative impact can be assessed by means of the measure q2. (Fornell & Cha, 1994; Geisser, 1975; Stone, 1974). In summary, according to above researchers‟ perspectives for using PLS as the. analytic technique to execute data analysis of structural model including (Urbach & Ahlemann, 2010): PLS makes fewer demands with regard to sample sizes than other approaches; PLS does not require normal-distributed input data; PLS can be used to complex SEM with a great number of dimensions; PLS is able to handle both reflective and formative constructs; PLS is better suited for theory development than for theory testing; and PLS is especially useful for prediction 34.