Determining factors influencing consumers’ adoption of different types of user-generated content in an e-commerce environment

(1)

UNIVERSITY OF AMSTERDAM

Faculty of Science

Thesis Master Information Science

Business Information Systems

Determining factors influencing consumers’ adoption of

different types of user-generated content in an e-commerce

environment

Author: Rytis Kajokas

Student number: 10865381

Supervisor: dhr. dr. D. Heinhuis

Signature:

Second examiner: prof. dr. T. van Engers

Signature:

(2)

2

ABSTRACT

The aim of this study was to identify factors influencing consumers’ adoption of different types of user-generated content (UGC) while shopping online. Web 2.0 technologies, which produce UGC, have changed the information production landscape over the networks. Previous research indicates importance that UGC have for consumers and online retailers alike. By the help of UGC consumers are able to find more reliable, unbiased and credible information on product or service they are after, consequentially tackling the information overload problem, which is rampant between consumers nowadays. On the other hand, online retailers are given a chance to have a new communication channel with their customers, and collect vast amounts of information about their clients, their expectations and previous experiences with a product or service. Tendencies from online retailers to incorporate some form of UGC into their e-commerce platforms are visible, but companies seem to lack guidance in choosing the right type, considering the number of initiatives failing to materialize. By specifying the different factors influencing consumers’ adoption of different types of UGC, this study sets to help companies integrating UGC and supplement information studies literature as well. By referring to literature assumptions were made that this is a technological issue, rather than one coming from other fields of academia. Thus, technology acceptance models were reviewed to find a right fit for analyzing research problems. UTAUT2 model was selected, because of its proven history in having high degree of explaining variance in research models, and the fact that it is specifically designed for analyzing consumers’ adoption of technology. Questionnaire was used to gather the data, which was later analyzed using structural equation modelling technique.

Extensive analysis proved the assumptions made in the beginning of the research that factors influencing consumers’ adoption of UGC differ from type to type. The contents usefulness (Performance Expectancy) and consumers’ previous experiences (Habit) with a certain types of UGC proved to be dominant factors influencing adoption of UGC. Adoption of ratings and reviews were solely determined by those two factors. Factors like social influence and trust were consistent with blogs, forums, links to social media and photo/video sharing platforms adoption. Ease of use (Effort Expectancy) and having the necessary resources (Facilitating Conditions) for using certain type of UGC seemed to be of marginal influence to consumers’ adoption. The findings of this study illustrate the different factors influencing consumers’ adoption of certain type of UGC, which should provide guidance for online retailers wanting to integrate one of the Web 2.0 technologies into their e-commerce platforms. Additionally, technology acceptance literature is supplemented and possible reconsideration of extension of UTAUT2 is highlighted.

Keywords: User-generated content, e-commerce, technology acceptance, UTAUT2, structural

(3)

3 Table of Contents ABSTRACT ... 2 LIST OF FIGURES ... 4 LIST OF TABLES ... 4 1 INTRODUCTION ... 6

1.1 PROBLEM STATEMENT/PRACTICAL RELEVANCE ... 7

1.2 RESEARCH CONTRIBUTION & RESEARCH QUESTIONS ... 8

1.3 THESIS OUTLINE ... 8

2 LITERATURE REVIEW... 9

2.1 WEB 2.0, E-COMMERCE & USER-GENERATED CONTENT ... 9

2.2 TECHNOLOGY ACCEPTANCE ... 10 3 RESEARCH MODEL ... 13 3.1 IMPORTANCE OF TRUST ... 14 4 METHODOLOGY ... 16 4.1 MEASURES ... 17 4.3 DATA ANALYSIS ... 17 5 RESULTS ... 17 5.1 DESCRIPTIVE STATISTICS ... 18 5.2 RELIABILITY VERIFICATION ... 18 5.3 VALIDITY TESTING ... 20 5.3.1 CONVERGENT VALIDITY ... 20 5.3.2 DISCRIMINANT VALIDITY ... 22

5.4 HYPOTHESES TESTING USING STRUCTURAL EQUATION MODELING ... 24

5.4.1 BLOGS ... 24

5.4.2 FORUMS ... 26

5.4.3 LINKS TO SOCIAL MEDIA ... 27

5.4.4 PHOTO/VIDEO SHARING PLATFORMS ... 29

5.4.5 RATINGS ... 31

5.4.6 REVIEWS ... 32

5.4.7 OVERALL STRUCTURAL EQUATION MODELLING RESULTS ... 33

6 CONCLUSIONS/IMPLICATIONS ... 34 6.1 CONCLUSIONS ... 34 6.2 IMPLICATIONS ... 34 7 DISCUSSION ... 36 8 LIMITATIONS/FUTURE RESEARCH ... 38 REFERENCES ... 39

(4)

4

Appendix A: RESEARCH’S QUESTIONNAIRE ... 44

Appendix B: MODELS FROM FIRST SEM ANALYSIS ... 45

Appendix C: CORRELATIONS OF MEDIATING FACTORS & SIGNIFICANCE WITH CONSTRUCTS ... 51

LIST OF FIGURES

Figure 1 Original UTAUT Model ... 12

Figure 2 Original UTAUT2 Model ... 13

Figure 3 Proposed Research Model (Modified UTAUT2) ... 15

Figure 4 SEM2 for Blogs ... 25

Figure 5 SEM2 for Forums ... 27

Figure 6 SEM2 for Links to Social Media ... 28

Figure 7 SEM2 for Photo/Video Sharing Platforms ... 30

Figure 8 SEM2 for Ratings ... 31

Figure 9 SEM2 for Reviews ... 32

Figure 10 SEM1 for Blogs ... 45

Figure 11 SEM1 for Forums ... 46

Figure 12 SEM1 for Links to Social Media ... 47

Figure 13 SEM1 for Photo/Video Sharing Platforms ... 48

Figure 14 SEM1 for Ratings ... 49

Figure 15 SEM1 for Reviews ... 50

LIST OF TABLES

Table 1 Web 2.0 Applications Relevant for E-commerce's UGC Production ... 9

Table 2 Models Reviewed for UTAUT Creation ... 11

Table 3 Participants Age ... 18

Table 4 Participants Gender ... 18

Table 5 Participants Usage Experience ... 18

Table 6 Reliability Verification for Blogs ... 18

Table 7 Reliability Verification for Forums ... 19

Table 8 Reliability Verification for Links to Social Media ... 19

Table 9 Reliability Verification for Photo/Video Sharing Platforms ... 19

Table 10 Reliability Verification for Ratings ... 19

Table 11 Reliability Verification for Reviews ... 20

Table 12 Convergent Validity for Blogs ... 20

Table 13 Convergent Validity for Forums ... 21

Table 14 Convergent Validity for Links to Social Media ... 21

Table 15 Convergent Validity for Photo/Video Sharing Platforms ... 21

Table 16 Convergent Validity for Ratings ... 21

Table 17 Convergent Validity for Reviews ... 22

Table 18 Discriminant Validity for Blogs ... 22

Table 19 Discriminant Validity for Forums ... 22

Table 20 Discriminant Validity for Links to Social Media... 23

(5)

5

Table 22 Discriminant Validity for Ratings ... 23

Table 23 Discriminant Validity for Reviews ... 23

Table 24 Hypotheses Testing for Blogs ... 24

Table 25 Second Hypotheses Testing for Blogs ... 25

Table 26 Hypotheses Testing for Forums ... 26

Table 27 Second Hypotheses Testing for Forums ... 27

Table 28 Hypotheses Testing for Links to Social Media ... 27

Table 29 Second Hypotheses Testing for Links to Social Media ... 29

Table 30 Hypotheses Testing for Photo/Video Sharing Platforms ... 29

Table 31 Second Hypotheses Testing for Photo/Video Sharing Platforms ... 30

Table 32 Hypotheses Testing for Ratings ... 31

Table 33 Second Hypotheses Testing for Ratings ... 31

Table 34 Hypotheses Testing for Reviews ... 32

Table 35 Second Hypotheses testing for Reviews ... 33

Table 36 Finalized Hypotheses testing for all types of UGC ... 33

Table 37 Moderating factors for Blogs ... 51

Table 38 Moderating factors for Forums ... 51

Table 39 Moderating factors for Links to Social Media ... 52

Table 40 Moderating factors for Photo/Video Sharing Platforms ... 52

Table 41 Moderating factors for Ratings ... 53

Table 42 Moderating factors for Reviews ... 54

(6)

6

1 INTRODUCTION

The use of World Wide Web as a platform for commercial activities is proving to be successful strategy for retailers around the world. Online retailing is a fast growing business, which saw an increase in sales of $313 billion globally between 2008 and 2012 (MarketLine, 2013). Recent reports by analysts specializing in e-commerce predict that number of people shopping online will increase from 1.09 trillion in 2013 to 2.5 trillion in 2018 (eMarketer, 2014). More and more people choose online shopping instead of brick & mortar shops, because of the convenience provided by e-commerce in the sense of time saved and better value for money offers (Guzzo et al., 2014). Most popular purchases made by consumers online are software, books, music, computer hardware, consumer electronics, apparel, footwear, jewellery, travel, holiday accommodation and tickets for events (Eurostat, 2015). Nevertheless, online shoppers are able to find any niche product or service online that suits them. Considering e-commerce reach and staggering statistics, regards its growth, it is safe to say that internet is one of the fastest growing marketplaces for consumer goods and services (Nielsen Company, 2014). Another reason to why consumers are flocking the internet for their shopping needs is the vast source of readily available information, which is used to make better purchase decisions (Nielsen Company, 2014).

People revise various different sources looking for information that might possibly be useful for them (Bandura, 2001). The internet became the most popular medium for consumers searching for information on products or services. In the early days of e-commerce internet was dominated by producer generated content, which took form in carefully depicted product and service descriptions, marketing campaigns, advertisements and others. The advantage companies saw in using internet as a medium for publishing information is the capacity to provide large amount of information to consumers at a very low cost (Evans & Wurster, 1999). However, this has created an information overload problem for consumers. For example, at the moment ASUS online shop provides a choice of 33 tablet computers each one with an explicit description about its characteristics. In a brick and mortar shop a consumer might receive guidance from a sales assistant, but online consumers are left alone to make these decisions, which might be a bit overwhelming. Actually, theory of information overload argues that when information exceeds certain threshold, consumers need more effort to process it and actually are bound to make poorer decisions (Jacoby et al., 1974).

Nowadays consumers are turning to user-generated content (UGC) to tackle information overload problem. Firstly, because UGC provides consumers with a more direct and understandable description of a product or service. Secondly, consumers seem to trust user-generated content more than information provided by retailers (Cheong, 2008). The term Web 2.0 is used to describe this internet phenomenon, which changed content production landscape over the internet. Definition Web 2.0 refers to user’s ability not only to read, but create content through the web (Balasubramaniam, 2009). Users are attracted to the openness that Web 2.0 provides and the ability to participate, which lead to more user-generated content creation over the networks than ever before (Valcke & Lenaerts, 2010). There is no unified definition for UGC, but three distinctive attributes are used to better understand and define user-generated content: “i) content made publically available over the internet, ii) content which reflects a ‘certain amount of creativity’, and iii) content which is ‘created outside of professional routines and practices’” (OECD/OCDE, 2007). User-generated content can take form of text, audio, video and images, or any combination between the types mentioned (Kim et al., 2010; Jensen et al., 2009).

(7)

7 Previous studies show that consumers prefer and trust user-generated content more than producer generated content (MacKinnon, 2012; Cheong, 2008). People feel that UGC reflects on products and services in a more realistic way, sharing not only positive but negative experience as well (Cheong & Morrison, 2008). In a recent survey 91% of people participating expressed that they use various forms of user-generated content, like blogs, ratings, online reviews, before making a purchase, and further 46% admitted that UGC influence their shopping decisions (Cheung & Thadani, 2012). That does not come as a surprise, previous studies show that UGC on e-commerce sites can reduce perceived risks, that a consumer might have (Cheung et al., 2009; Park & Kim, 2008). Furthermore, UGC is linked to increasing degree of satisfaction for online consumers (Liang et al., 2007), and helping to make more efficient purchasing decisions (Cheung et al., 2009).

Speedy developments of Web 2.0, and specifically UGC, are transforming e-commerce from product/service oriented business to customer centred one (Wigand et al., 2008). User-generated content and e-commerce are closely interrelated with each other. UGC supplements online retailing and should be viewed as beneficial by both – consumers and producers. From retailers perspective UGC content gave way for new way of communication between them and the customers. Thus, in turn, companies are able to collect more information about their customers, which allows having better understanding of their clients’ expectations (Valcke & Lenaerts, 2010). Analyzing UGC on an e-commerce platform not only allows companies to see their clients expectations, but to hear about their previous shopping experiences as well, which helps in creating new or modifying existing business strategies (Constantinides & Fountain, 2008). In Michaelidou et al. (2011) article several other advantages that UGC brings to businesses, are mentioned: increased traffic to retailer’s website, stronger business relationships between enterprise and customers, ability to identify new business possibilities, support to brand and product/service development. On the other hand, consumers also benefit from UGC. Customers have a better reach and expression platform, which they can use freely to express their opinions. UGC on an e-commerce platform not only allows the consumer to reach retailer, but at the same time gives way to create and interact with peer communities (Constantinides & Fountain, 2008). Within these peer communities consumers get access to vast amount of social knowledge, read about previous experiences others had with specific product or service. This information allows consumers to make smarter decisions, while shopping online, because they are better informed (Dennison et at., 2009).

1.1 PROBLEM STATEMENT/PRACTICAL RELEVANCE

The importance of UGC for retailers and consumers is very evident. The influence it has on consumers and their decision making has been studied previously in various works (Liang et al., 2011; Dennison et al., 2009; Dellarocas et al., 2007; Duan et al., 2008). Some companies, like TripAdvisor, Yelp, ConsumerReports, Which?, and numerous others, are capitalizing on this trend, and made hosting of consumers UGC as their business model. As a result, companies whose product or service is reviewed on one of those sites have little or no control over the content published online. Adapting some form of UGC on your e-commerce platform allows company to keep customer on their site for longer and helps to keep a better eye on content published, in turn allowing for quicker minimization of effects of negative reviews (Hills & Cairncross, 2011). The latter is quite important considering, that negative reviews have even greater power on consumer decision making than the positive ones (Park & Lee, 2009).

Tendencies to integrate UGC on e-commerce platforms are already visible among online retailers. However, guidance in choosing the type of user-generated content is missing. This is quite important, considering the amount of UGC initiatives failing to show positive results (Rossdavie, 2014). All the

(8)

8 UGC types differ in technological design and content: design refers to the way information is presented to the consumers and content refers to type and size of information provided (Huizingh, 2000). A careful consideration of factors influencing consumers’ adoption of different types of user-generated content would allow online retailers to better tailor integration of UGC. Based on differences in technological design and content of different types of UGC this study makes an assumption that there will be different indicators for their adoption. Thus, this research aims to deliver such required guide for online retailers, specifying types of UGC in a sense of consumers’ judgment of them.

1.2 RESEARCH CONTRIBUTION & RESEARCH QUESTIONS

Previous studies done in the fields of user-generated content and e-commerce mainly focused on effects UGC has on consumer decision making and shopping behaviour (Riegner, 2007; Cheung & Thadani, 2012; Duan et al., 2008; MacKinnon, 2012; Davis & Khazanchi, 2008). Other studies looked at how companies are using UGC as a communication tool to better connect with their customers (Papathanassis & Knolle, 2011; Bernoff & Li, 2008), why consumers trust user-generated content more than producer-generated content (MacKinnon, 2012), and why trust is such an important factor for people shopping online (Nielsen, 2012; Dellarocas et al., 2007; Duan et al., 2008). Nevertheless, extant literature seems to be missing studies concerned with differentiating different types of UGC from users’ perspective, so the following research question is constructed:

What are the factors influencing adoption of different types of UGC?

Several sub-questions are formulated in order to support the main question of the research. Firstly, there is the need to specify different Web 2.0 technologies that produce user-generated content relevant to e-commerce platforms. Consequentially, the following sub-question is formulated:

What are the different forms of UGC available today for e-commerce use?

Secondly, a model needs to be developed to allow testing if there are different factors influencing consumers’ adoption of different types of UGC while shopping online; and it needs to be tested empirically. Consequentially, the following sub-questions are formulated:

Is it possible to develop a model to test factors influencing consumers’ adoption of different types of UGC?

Is it possible to test the developed model empirically?

In recent times, user-centred and/or customer oriented research is receiving a lot of attention from business people and academics alike. However, current literature is missing studies on consumers’ acceptance of user-generated content. This study aims to fill that gap by creating and testing consumer acceptance of UGC model, and as a result supplementing consumer technology acceptance field and adding to current information systems research in general.

1.3 THESIS OUTLINE

The remainder of this paper is organized as follows. Section 2 looks at literature about Web 2.0, user-generated content in e-commerce context and technology acceptance theories. In section 3 research model is presented and hypotheses are created by combining knowledge gathered from previous research and elements of UTAUT model. Section 4 will be used to describe study’s methodology and

(9)

9 will be followed by presentation of empirically gathered data in section 5. Section 6 of this study will present conclusions and implications of the research. In section 7 study’s results will be discussed in more explicit manner, and this will be followed by limitations/future research in section 8.

2 LITERATURE REVIEW

2.1 WEB 2.0, E-COMMERCE & USER-GENERATED CONTENT

Internet has changed a lot since Tim Berners-Lee developed HTML back in 1990. In the beginning a user was required to have good understanding of HTML to be able to post content on the internet, thus usually leaving the user to be a passive information gatherer (Anderson, 2007). This dynamic has shifted and internet has become a more collaborative space for information sharing and gathering (Richardson, 2010). This change is usually referred to as internet moving from its initial phase of Web 1.0 to Web 2.0.

Web 2.0 can be described as “the second generation of web-based technologies that have gained massive popularity by letting people collaborate and share information online in previously unavailable ways” (Reactive, 2007, p. 3). Web 2.0 is changing the landscape of the internet by enabling users to create and distribute content easier than ever before (Hendler et al., 2008). Wikis, blogs, podcasts, RSS, social networks, virtual worlds, and photo/video sharing platforms are the usual suspects that fall in the Web 2.0 technology cluster (Anderson, 2007). These technologies, which are sometimes referred to as Web 2.0 applications or Web 2.0 tools, have increased communication capabilities for every user of the internet (Anderson, 2007).

With the help of Web 2.0 technologies, more and more user-generated content is produced. In Web 2.0 user takes the central role in producing, reviewing and responding to content that is published online (Gretzel, 2007). This has caught the interest of online retailers and marketers, because now they have better access to information and can analyze it easier and respond quicker to consumers in required manner. In a survey conducted by McKinsey (2007) majority of senior executives, taking part in the survey, explained that adoption of Web 2.0 technologies is in their strategic plans and will see an increase in investment. However, not all Web 2.0 applications that are mentioned before are a good fit for e-commerce platforms. This study proposes to select six different types of Web 2.0 applications, which produce user-generated content valuable in e-commerce context (Table 1). Table 1 Web 2.0 Applications Relevant for E-commerce's UGC Production

Web 2.0 Application for Producing UGC

Description Examples

Blogs A website separated from the online shop. Blogs serve as online journals in which people can contribute. Data is presented in a reverse chronological order (starting with the most recent). Data entries usually consists explicit information about a product or service.

Helm Boots Blog – producer and customers share stories about the brand and their products.

Pure Fix Blog – shares stories about company’s products and bike events. River Pools and Spa Blog – used as an communication platform to answer customers questions

Forums A website separated from the online shop. Forums serve as message boards where people can have broader conversations about wide range of subjects. Data entries can vary from short answers to explicit descriptions of a product or service.

Apple’s Discussion Forum – a message board dedicated to helping customers find and share solutions to problems they have, and express their

experiences with company’s products. Basenotes Forum – hosts discussions

(10)

10 about all subjects related to fragrance. Links to Social

Media

Presented within the ecommerce platform and linked directly to actual post on a social media site. Usually presents experiences a consumer had with a product or service. Most popular forms are posts on Facebook and tweets derived from Twitter.

Costa Rican Vacations – travel company incorporated previous customers experiences collected from Facebook on their index page. PSD2HTML – company has incorporated live customer reviews from twitter into their testimonials page.

Photo/Video Sharing Platforms

Sometimes incorporated into ecommerce platform, but usually linked to a photo or video sharing site. Data is presented in a visual form. This trait is unique when compared to other types of UGC.

The Art of the Trench – Burberry’s campaign allowing customers to upload and comment on pictures of people wearing company’s products. Phillips – allows customers to share video reviews of their products. Ratings Presented on an ecommerce platform next to the

product or service at hand. Ratings are defined in numerical scale (e.g. 1-5) or symbolic marking (e.g. stars).

Ebay Rating System - dedicated to rate vendors on the market place rather than products.

Cabela’s Rating System – dedicated to customer who want to rate the product they bought.

Reviews Presented on an ecommerce platform next to the product or service at hand. Reviews are a written evaluation of the experience consumer had with a product or service.

Amazon Review System – allows customers to leave a text summary about the product they bought. Crate & Barrel – every product has a review section allowing customers to express their opinions.

Blogs, forums, social media and photo/video sharing platforms are commonly considered as being part of Web 2.0 applications (Murugesan, 2007). Ratings and reviews are viewed as aggregation services in the Web 2.0 context. Ratings and reviews help to collect and aggregate consumer data that shows expression of consumers’ attention and/or intentions (Anderson, 2007). To put it simply ratings and reviews shows consumers’ opinions about a service or product they have purchased previously expressed in a written review and/or a symbolic rating.

In the following chapter this study will look at different technology acceptance theories to find the right fit for this research, considering that adoption of different types of UGC or in other words different types of Web 2.0 technologies is a technology acceptance issue.

2.2 TECHNOLOGY ACCEPTANCE

Understading peoples’ acceptance of technology has become one of the central research themes in information systems field. This is due to vast developments of information and communication technologies in almost all aspects of life that occured in the last few decades. Information technology acceptance can be defined as a “demonstrable willingness to employ information technology for the task it is designed to support” (Dillion & Morris, 1996). During the years a lot of academics wrote theories and propossed models to better understand technology acceptance. The mostly used research models adopted in the context of IT acceptance are Theory of Reasoned Behaviour (TRB) (Fishbein & Ajzen, 1975), Social Cognitive Theory (SCT) (Bandura, 1986) Theory of Planned Behaviour (TPB) (Ajzen, 1991), Technology Acceptance Model (TAM) (Davis, 1989) and Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh et al., 2003).

(11)

11 The UTAUT model has proven to be most succesful in atempting to create a framework to determine technology acceptance. Venkatesh et al. (2003) synthesized eight prominent technology acceptance models used in the past to propose this unified model (Table 2). Authors of UTAUT defined four main factors influencing user acceptance of technology: performance expectancy, effort expectancy, social influence and facilitating conditions (Ventkatesh et al., 2003). The eight models examined by Venkatesh et al. (2003) had ability to explain 17-53% variance of users’ intentions to use technology, and the UTAUT model (Figure 1) proved to explain 70% of the same variance.

Table 2 Models Reviewed for UTAUT Creation

Model ` Core Constructs Studies that adopted the model

Theory of Reasoned Action (TRA) Fishbein and Ajzen, 1975

Attitude Toward Behaviour, Subjective Norm

“The Theory of Reasoned Action Applied to Coupon Usage”, T. A. Shimp & A. Kavas – 1984, “Consumer Concern, Knowledge, Belief, and Attitude Toward Renewable Energy”, H. K. Bang et al. - 2000 Technology

Acceptance Model (TAM)

Davis, 1989 Perceived Usefulness, Perceived Ease of Use, Subjective Norm

“Consumer Acceptance of Electronic Commerce”, P. A. Pavlou – 2003, “The Intended and Actual Adoption of Online Purchasing”, X. Cao & P. L. Mokhtarian - 2005 Motivational Model (MM) Davis et al., 1992 Extrinsic Motivation, Intrinsic Motivation

“Internet Shopping Acceptance Examining the Influence of Intrinsic vs Extrinsic Motivations”, M. Suki et al. – 2008, “Extrinsic versus Intrinsic Motivations for Consumers to Shop Online ”, R. Shang et al. – 2005

Theory of Planned Behaviour (TPB)

Ajzen, 1991 Attitude Towards Behaviour, Subjective Norm, Perceived Behaviour Control

“E-tailers vs Retailers: Which Factors Determine Consumer Preferences”, C. Keen et al. – 2004, “Understanding and Predicting Electronic Commerce Adoption”, P. A. Pavlou & M. Fygenson - 2006

Combined TAM and TPB (C-TAM-TPB)

Taylor and Todd, 1995

Attitude Toward Behaviour, Subjective Norm, Perceived Behavioural Control, Perceived Usefulness

“Combination of TAM and TPB in Internet Banking Adoption”, R. Safeena et al. – 2013, “The Exploration of Network Behaviours by Using The Models of TPB, TAM & C-TAM-TPB”, C. Chen - 2013 Model of PC Utilization (MPCU) Thompson et al., 1991

Job-fit, Complexity, Long-term Consequences, Affect Towards Use, Social Factors, Facilitating Conditions

“Influence of Experience on Personal Computer Utilization”, R. L. Thompson – 1994, “Extending the Technology Acceptance Model and the Task Technology Fit Model to Consumer E-Commerce”, Klopping & McKinney - 2004 Innovation Diffusion Theory (IDT) Rogers, 1962

Relative Advantage, Ease of Use, Image, Visibility, Compatibility, Results Demonstrability, Voluntariness of Use

“Diffusion of E-commerce: An Analysis of the Adoption of Four E-commerce

Activities”, M. Eastin – 2002, “Consumer Adoption of Online Grocery Buying: A Discriminant Analysis”, T. Hansen – 2005 Social Cognitive Theory (SCT) Bandura, 1986 Outcome Expectations – Performance, Outcome Expectations – Personal, Affect, Anxiety

“Web vs Campus Store? Why Student Buy Textbooks Online? B. E. Foucalt & D. A. Scheufele – 2002, “An Empirical

Investigation of Consumer Control Factors on Intention to Use Selected Self-service Technologies” A. Oyedele & P. M. Simpson - 2007

Original UTAUT model has four main independent variables: Performance Expectancy (PE) is defined as the extent an individual believes the system will help them to do their task better, Effort Expectancy (EE) relates to how easy an individual believes the system is to use, Social Influence (SI)

(12)

12 relates to whether or not important others’ influence an individual’s intention to use the system and Facilitating Conditions (FC) relates to whether an individual has availability of support and resources in order to use the system (Venkatesh et al., 2003). As seen in the UTAUT model, PE is moderated by age and gender, EE by age, gender and experience, SI by age, gender, experience and voluntariness of use, FC by age and experience. Worth to mention, that the original study by Venkatesh et al. (2003) proved that FC directly influences actual use behaviour.

Figure 1 Original UTAUT Model

The original UTAUT model was constructed while investigating technology acceptance in large organizations and was more concerned with employees rather than ordinary people. Nevertheless, not so long ago Venkatesh, Thong & Xu (2012) developed and empirically tested an extension to UTAUT. The model was simply labelled UTAUT2 (Figure 2) and was more specifically designed to understand technology acceptance in consumer context.

(13)

13 Figure 2 Original UTAUT2 Model

In UTAUT2 three new determinants’ which are more consumer oriented are presented by the authors. First addition is Hedonic Motivation (HM), which relates to enjoyment an individual can derive from using the system (Brown & Venkatesh, 2005). Second addition, Price Value (PV), relates to individual’s belief that monetary costs for using the system are worth the benefits system will bring (Brown & Venkatesh, 2005). Last addition is Habit (HT), which relates to individual’s previous use of the system or individual’s belief that use of system is automatic to them (Venkatesh, Thong & Xu, 2012). Similarly to UTAUT model, individual differences are defined – age, gender and experience. These can moderate relationships between core concepts, behaviour intentions and actual system use. Voluntariness of use is removed from UTAUT2, because in consumers’ context of technology acceptance individuals have free will, when deciding to adopt new technologies (Venkatesh, Thong & Xu, 2012).

3 RESEARCH MODEL

Taking into consideration all the available technology acceptance models, UTAUT2 will be adopted for this study. Several characteristics have influenced this decision. Firstly, as mentioned before UTAUT models have predictive efficiency of 70%, which is a big improvement compared to previous models used in researching technology acceptance. Secondly, UTAUT2 model is specifically constructed to reflect technology acceptance in consumer context. Therefore, it seems as the best fit for researching and testing consumers’ acceptance of different types of user-generated content. In the original article, where UTAUT was first presented, Venkatesh et al. (2003) suggested that the model should always be revised and modified in order to have a good fit with the research at hand. Following these recommendations this study will apply a slightly modified version of the UTAUT2 model to suit the theme and scope of the research better.

(14)

14 To begin with, this study will be conducted in a time frame of twelve weeks; therefore Use Behaviour will not be researched, because it requires studying continuous use of technology. As a result, Behaviour Intention will be selected as the main dependent variable.

There are no monetary costs involved for the consumer, when using user-generated content on an e-commerce platform. Therefore, Price Value (PC) will not be used as an independent variable in the proposed research model.

Even though UTAUT2 model is regarded as one of the best models of technology acceptance, it is not without flaws. Many studies modified and extended the original model, no different from this research. In order to have a more complete picture of possible factors influencing consumers’ adoption of different types of UGC, one addition to original UTAUT2 model is suggested. In the context of online retailing, trust plays an important role for consumers (Corbitt et al., 2003). Thus, the following part will discuss the proposed addition of Trust (T) as an independent variable to the research model.

3.1 IMPORTANCE OF TRUST

Previous studies show that the more people are dependent upon others’ and their own vulnerability regards misconduct of others, the more significant trust becomes (Deutsch, 1958; Luhmann, 1979). Moreover, Fukuyama (1995) argued that trust plays an important role in successful new technology acceptance. Adoption of trust as a variable in technology acceptance is not a new thing. Pearson et al. (2010) included trust in researching internet banking adoption in Jordan, Alharbi (2014) used trust as a determinant for behavioural intention while studying adoption of cloud computing and Gefen et al. (2003) incorporated trust into TAM (Davis, 1989) to research consumers’ adoption of online shopping.

Consumers turning to user-generated content for unbiased information about a product or service are faced with overwhelming dilemma – Is this information trustworthy? Contributors of UGC in any given medium are independent, and from the view of consumer - unpredictable. This combined with our human nature, which requires us to understand actions of others, put consumers in complex situations. Consumers do not have any control over content published by UGC contributors, nor do they have a good understanding about their motivations to publish it. This creates unnecessary complexity towards information in UGC in any given medium. However, trust is regarded as being one of the best complexity reduction methods (Luhmann, 1979), thus vital for fruitful interaction between consumer and UGC contributor. Luhmann (1979) argues that trust takes a central role in complexity reduction in social environments, which are not regulated by rules or customs.

Trust as a concept has many varying definitions, but the following quote from Gefen’s (2003, PP-54) article is adopted for this study: “Trust is expectation that others one chooses to trust will not behave opportunistically by taking advantage of the situation”. In the scope of this study, trust means that consumers who rely on some type of UGC to make purchasing decisions, expect the information provided to be unbiased, truthful and reliable. Trust has been proved to play an important role in most social and economic interactions surrounded by uncertainty and involving dependency (Kumar, 1996; Rousseau et al., 1998), thus trust is an important factor in the UGC and e-commerce context. Furthermore, Luhmann (1979) argues that trust is crucial, when consumers are faced with important decisions.

Reflecting on the last three paragraphs, this study proposes adding trust as a determinant of consumers’ choice to rely on information provided in certain type of UGC. This refers to willingness

(15)

15 of consumers to follow advice received from different mediums of UGC, which consequentially increases his/hers behavioural intentions towards that type of UGC. As a result, research model (Figure 3) for testing consumer acceptance of different types of user-generated content is presented below. This model is a modified version of UTAUT2 designed specifically for this study, considering the theme and scope of the research.

Figure 3 Proposed Research Model (Modified UTAUT2)

In order to test consumers’ acceptance of different types of user-generated content several hypotheses are raised, based on currently proposed research model.

Performance Expectancy (PE) of a certain type of user-generated content is defined as the degree to which individual believes that using that type of UGC will help him or her to attain gains in online shopping performance (Venkatesh et al., 2003) - as in helping individuals to make better purchasing decisions by tackling information overload problem.

Hypothesis 1: Performance Expectancy will have a significant positive influence on Behaviour Intention of using certain type of UGC, while shopping online.

(16)

16 Effort Expectancy (EE) is defined as the degree of ease associated with using a certain type of UGC, while shopping online (Venkatesh et al., 2003). As in finding it easy to navigate and find information on certain type of UGC.

Hypothesis 2: Effort Expectancy will have a significant positive influence on Behaviour Intention of using of certain type of UGC, while shopping online.

Social Influence (SI) is defined as the degree to which an individual perceives that important others believe he or she should use certain type of UGC, while shopping online (Venkatesh et al., 2003).

Hypothesis 3: Social Influence will have a significant positive influence on Behaviour Intention of using of certain type of UGC, while shopping online

Facilitating Conditions (FC) is defined as the degree to which an individual perceives having necessary resources and support to use certain type of UGC (Brown & Venkatesh, 2005).

Hypothesis 4: Facilitating Conditions will have a significant positive influence on Behaviour Intention of using certain type of UGC, while shopping online

Trust (T) in certain type of UGC on an e-commerce platform is defined as the degree to which an individual believes that that type of UGC would provide credible information.

Hypothesis 5: Trust will have a significant positive influence on Behaviour Intention of using of certain type of UGC, while shopping online

Hedonic Motivation (HM) is defined as the degree to which an individual derives pleasure or fun from using certain type of UGC, while shopping online (Venkatesh, Thong & Xu, 2012).

Hypothesis 6: Hedonic Motivation will have a significant positive influence on Behaviour Intention of using of certain type of UGC, while shopping online

Habit (H) is defined as the degree to which using a certain type of UGC, while shopping online, has become automatic, considering previous use of that type of UGC (Venkatesh, Thong & Xu, 2012).

Hypothesis 7: Habit will have a significant positive influence on Behaviour Intention of using of certain type of UGC, while shopping online

Moderating factors presented with UTAUT2 model - age, gender and experience – will not be considered for proposed model of this research due to time frame and scope of this study. However, side analysis was carried out and results can be found in Appendix C.

4 METHODOLOGY

The data gathering for this study was carried out as an anonymous internet based survey (Appendix A), which was launched through Survey Monkey. This research method was chosen for a couple of reasons. Firstly, Newsted et al. (1999) explains that surveys seem to be most widely used method in information systems research. Secondly, online survey allowed collecting necessary data from wider public in effective and time efficient manner. Convenience sampling was used for the data collecting.

(17)

17 This allowed better accessibility to potential participants. Overall 135 responses were gathered by publishing the survey online via social media sites, spreading it through author’s personal network and gathering responses from students in university halls. This sample size was deemed statistically valid (Wolf et al., 2013).

4.1 MEASURES

The survey’s questionnaire was adopted from the original UTAUT2 paper written by Venkatesh et al. (2012) with some additions and slight changes (Appendix A). Overall participants had to answer 30 questions, starting with background questions about their age, gender and previously had experiences with UGC while shopping online. The rest of the questionnaire’s questions were specifically aimed to gather information about possible Behavioural Intention (BI) determinants and BI itself. The research model of this study had 8 constructs, which are reflected in the survey by having 3 to 4 questions about each. Questions for Performance Expectancy (PE), Effort Expectancy (EE), Social Influence (SI), Facilitating Conditions (FC), Hedonic Motivation (HM) and Habit (H), were all taken from Venkatesh et al. (2012) originally suggested questionnaire. Questions for the independent variable Trust (T) were adopted from Koufaris & Hampton-Sosa (2002). As mentioned before in the paper Price Value (PV) and Use Intention (UI) were not included in the research model, thus questions regarding them are not present in the questionnaire. Participants were to measure all the questions for PE, EE, SI, FC, T, HM, HT and BI in 5-point Likert scale, ranging from “Strongly Disagree (1)” to “Strongly Agree (5)”.

4.2 DATA ANALYSIS

Reliability and validity of constructs of the research model had to be confirmed before moving to defining factors influencing behavioural intention. In this research Likert scale allowed to measure participant’s opinion in a scale difference numerically, thus data gathered is considered at interval-level. Cronbach’s alpha coefficient was used to check the internal consistency and reliability of all the constructs. This is essential, when using Likert scales, according to previous research (Gliem & Gliem, 2003). Defining construct validity was another step taken prior to main analysis. This allowed checking and proving the internal consistency of research model’s constructs, which is prerequisite of SEM (Suhr, 2006). Structural Equation Modelling (SEM) technique was chosen to test and validate the proposed research model. This decision was made for several reasons. Firstly, SEM analysis is very common in research using UTAUT and UTAUT2 models. Williams et al. (2014) have conducted an extensive literature review on research using UTAUT model and specified that majority of them used SEM for their data analysis. Secondly, even though SEM is similar to other data analysis techniques such as Multiple Regression Analysis (MRA), it has more advantages over them. In a paper written by Chau (1997) advantages of constructing models using SEM are specified as: “i) they make the assumptions, constructs, and hypothesized relationships in a researcher’s theory explicit; ii) they add a degree of precision to a researcher’s theory, since they require clear definitions of constructs, operationalizations, and functional relationships between constructs; iii) they permit a more complete representation of complex theories; iv) they provide a formal framework for constructing and testing both theories and measures” (Chau, 1997, p. 314). Even though moderating factors: age, gender and experience were not included into this research’s model; they still were considered and analyzed as part of the analysis. Results of these analyses can be found in Appendix C.

5 RESULTS

The results of the study are divided and presented in four subgroups: descriptive statistics, reliability verification, validity testing and hypotheses testing using SEM.

(18)

18

5.1 DESCRIPTIVE STATISTICS

The following three tables provide general information about the people who participated in survey of this research. The demographic information about the participants includes their age, gender and their previous usage experience of different types of UGC.

Table 3 Participants Age

Age Group Frequency Percentage

18-24 53 39.3% 25-34 70 51.9% 35-44 9 6.7% 45-54 1 0.7% 55-64 1 0.7% 65-74 1 0.7%

Table 4 Participants Gender Gender Frequency Percentage

Male 67 49.6%

Female 68 50.4%

Table 5 Participants Usage Experience

Usage Blogs Forums Links to Social Media Photo/Video Sharing Platforms Ratings Reviews FRQ % FRQ % FRQ % FRQ % FRQ % FRQ % Never 31 23.0 52 38.5 32 23.7 31 23.0 0 0.0 0 0.0 Rarely 53 39.3 43 31.9 45 33.3 38 28.1 16 11.9 15 11.1 Occasionally 36 26.7 31 23 37 27.4 34 25.2 47 34.8 41 30.4 Frequently 13 9.6 8 5.9 17 12.6 23 17.0 47 34.8 52 38.5 Very Frequently 2 1.5 1 0.7 4 3.0 9 6.7 25 18.5 27 20.0

5.2 RELIABILITY VERIFICATION

Reliability verification was performed to check the accuracy and precision of the measurement instrument and its procedure (Thorndike et al., 1991). Reliability testing refers to the stability and consistency of the measurement and checks the instrument for random errors. Consistency testing as type of reliability measurement is widely used in the field of information systems (Sekaran, 2003). This study used Cronbach’s alpha coefficients to check the internal consistency of survey items. Literature suggests that Cronbach’s alpha values that fall into range of 0.70 to 0.95 serve as satisfactory proof of internal consistency of the measurement (Tavakol & Dennick, 2011).

The following six tables provide Cronbach’s alpha values for all constructs of the research model and are separated regarding different types of UGC. Overall, Cronbach Alpha’s values proved that measurement instrument was reliable and appropriate. Construct validity was found for all the different types of UGC.

Table 6 Reliability Verification for Blogs Construct Cronbach’s Alpha No. of Items

PE 0.805 3

(19)

19 SI 0.828 3 FC 0.847 4 T 0.803 3 HM 0.801 3 HT 0.804 3 BI 0.798 3

Table 7 Reliability Verification for Forums Construct Cronbach’s Alpha No. of Items

PE 0.735 3 EE 0.798 4 SI 0.767 3 FC 0.778 4 T 0.745 3 HM 0.737 3 HT 0.755 3 BI 0.723 3

Table 8 Reliability Verification for Links to Social Media Construct Cronbach’s Alpha No. of Items

Table 9 Reliability Verification for Photo/Video Sharing Platforms Construct Cronbach’s Alpha No. of Items

Table 10 Reliability Verification for Ratings

(20)

20 PE 0.756 3 EE 0.789 4 SI 0.796 3 FC 0.769 4 T 0.772 3 HM 0.784 3 HT 0.754 3 BI 0.755 3

Table 11 Reliability Verification for Reviews

Construct Cronbach’s Alpha No. of Items

5.3 VALIDITY TESTING

This research performed validity testing to check the validity of the model’s constructs. In other words, it was necessary to check if survey’s items measured the intended construct. In this study convergent validity and discriminant validity were assessed to test validity of the model’s constructs. 5.3.1 CONVERGENT VALIDITY

Validity implies that construct measures what it was designed to measure. Convergent validity for constructs is achieved when composite reliability (C.R.) is above 0.7 mark and average variance extracted (AVE) is above 0.5 mark (Hair et al., 2006). The following six tables provide convergent validity testing results for different types of UGC. Majority of constructs for different types of UGC proved to have convergent validity. However, in analysing survey results for Ratings and Reviews responses for construct of Habit (HT) showed loadings of average variance extracted 0.433 and 0.443 respectively. These loadings do not comprise enough of variance and will be suggested to be dropped from the model for those types of UGC.

Table 12 Convergent Validity for Blogs Construct AVE C.R.

PE 0.828 0.805 EE 0.838 0.854 SI 0.862 0.828

(21)

21 FC 0.946 0.847

T 0.874 0.803

HM 0.866 0.801 HT 0.774 0.804

Table 13 Convergent Validity for Forums Construct AVE C.R. PE 0.844 0.735 EE 0.837 0.798 SI 0.832 0.767 FC 0.941 0.778 T 0.862 0.745 HM 0.873 0.737 HT 0.651 0.755

Table 14 Convergent Validity for Links to Social Media Construct AVE C.R. PE 0.794 0.824 EE 0.811 0.872 SI 0.813 0.836 FC 0.728 0.866 T 0.755 0.831 HM 0.847 0.828 HT 0.754 0.823

Table 15 Convergent Validity for Photo/Video Sharing Platforms Construct AVE C.R. PE 0.786 0.812 EE 0.852 0.871 SI 0.775 0.832 FC 0.682 0.854 T 0.744 0.826 HM 0.852 0.811 HT 0.623 0.816

Table 16 Convergent Validity for Ratings

Construct AVE C.R. PE 0.836 0.756 EE 0.821 0.789 SI 0.774 0.796 FC 0.676 0.769

(22)

22

T 0.679 0.772

HM 0.754 0.784 HT 0.433 0.754

Table 17 Convergent Validity for Reviews

Construct AVE C.R. PE 0.854 0.780 EE 0.833 0.807 SI 0.754 0.808 FC 0.706 0.793 T 0.701 0.790 HM 0.785 0.806 HT 0.443 0.772 5.3.2 DISCRIMINANT VALIDITY

Hair et al. (2006) asserted that if the average variance extracted (AVE) is higher than the squared inter-scale correlation of the construct, then discriminant validity is supported. The following six tables provide discriminant validity testing results for different types of UGC, which show that discriminant validity is supported in all cases.

Table 18 Discriminant Validity for Blogs

PE EE SI FC T HM HT PE 0.828 EE 0.043 0.838 SI 0.130 0.038 0.862 FC 0.008 0.546 0.064 0.946 T 0.460 0.011 0.162 0.012 0.874 HM 0.383 0.024 0.156 0.034 0.453 0.866 HT 0.312 0.042 0.163 0.026 0.465 0.454 0.774

Table 19 Discriminant Validity for Forums

PE EE SI FC T HM HT PE 0.844 EE 0.070 0.837 SI 0.077 0.041 0.832 FC 0.012 0.531 0.080 0.941 T 0.338 0.005 0.484 0.012 0.862 HM 0.255 0.040 0.235 0.008 0.253 0.873 HT 0.212 0.005 0.031 -0.015 0.222 0.417 0.651

(23)

23 Table 20 Discriminant Validity for Links to Social Media

Table 21 Discriminant Validity for Photo/Video Sharing Platforms

Table 22 Discriminant Validity for Ratings

Table 23 Discriminant Validity for Reviews

PE EE SI FC T HM HT PE 0.854 EE 0.263 0.833 SI 0.106 0.019 0.754 FC 0.218 0.531 0.054 0.706 T 0.120 0.034 0.277 0.084 0.701

(24)

24

HM 0.130 0.009 0.119 0.027 0.192 0.785

HT 0.303 0.099 0.125 0.089 0.219 0.260 0.443

5.4 HYPOTHESES TESTING USING STRUCTURAL EQUATION MODELING

The seven hypotheses that were raised in the beginning of the research were tested using structural equation modelling technique. This means that independent variables were evaluated to see their contribution to the explanation of the dependent variable and if it is meaningful (Hair et al., 2006). The following six subchapters provide results of SEM analysis and hypotheses testing for different types of UGC. Usually, the hypotheses with P values below 0.01 and below 0.05 are considered to have very significant influence (Hair et al., 2006). However, considering the small sample size of the research and referring this back to literature (Whitley & Ball, 2002), this study will consider all the P values below 0.15. This will be applied to all the different types of UGC analyzed in this study. The main reason behind this decision is to avoid type II error and to have more conclusive research. Literature states that increased P values will lead to less chance of committing a type II error, which will eventually increase the power, especially when dealing with small sample sizes (Calkins, 2005). Models derived from first structural equation modelling analysis are presented in Appendix B; the models derived and confirmed during the second structural equation modelling analysis are presented in the sub-chapters below according to the type of UGC.

5.4.1 BLOGS

Table 24 Hypotheses Testing for Blogs

Estimate β

SE CR P Value Hypothesis Behavioural Intention  Habit 0.741 0.077 9.622 P<0.01 Supported Behavioural Intention  Performance

Expectancy

0.536 0.054 10.001 P<0.01 Supported Behavioural Intention  Effort Expectancy -0.572 0.066 -8.641 P<0.01 Supported Behavioural Intention  Social Influence 0.093 0.054 1.707 P=0.088 Supported Behavioural Intention  Facilitating Conditions 0.946 0.260 3.640 P<0.01 Supported Behavioural Intention  Trust 0.082 0.051 1.592 P=0.111 Supported Behavioural Intention  Hedonic Motivation -0.027 0.038 -0.718 P=0.473 Not Supported Earlier conducted reliability and validity tests proved that all constructs of research model were possible determinants for Behavioural Intention to use blogs while shopping online. After using structural equation modelling to test hypotheses of the research the following results were constructed: Habit (β = 0.741, P<0.01), Performance Expectancy (β=0.536, P<0.01), Effort Expectancy (β=-0.572, P<0.01) and Facilitating Conditions (β=0.946, P<0.01) proved to be significant indicators of intention to use blogs while shopping online. Social Influence (β=0.093, P=0.088) and Trust (β=0.082, P=0.111) will be considered as indicators of intention to use blogs while shopping online, because of decisions made previously in the paper.

Second structural equation modelling analysis was performed without variables that proved to be insignificant to the use intention of blogs. This was done to test the validity of the model produced with the first structural equation modelling analysis. The second analysis of the model and hypotheses can be found below.

(25)

25 Figure 4 SEM2 for Blogs

Table 25 Second Hypotheses Testing for Blogs

Estimate β

Expectancy

0.533 0.053 9.980 P<0.01 Supported Behavioural Intention  Effort Expectancy -0.575 0.066 -8.681 P<0.01 Supported Behavioural Intention  Social Influence 0.089 0.054 1.643 P=0.100 Supported Behavioural Intention  Facilitating Conditions 0.946 0.260 3.643 P<0.01 Supported Behavioural Intention  Trust 0.075 0.051 1.472 P=0.141 Supported The second structural equation modelling analysis confirmed the results of first SEM analysis. Performance Expectancy (β=0.533, P<0.01), Effort Expectancy (β=-0.575, P<0.01), Facilitating

(26)

26 Conditions (β=0.946, P<0.01) and Habit (β=0.661, P<0.01) are significant indicators of use intention for blogs. Moreover, Social Influence (β=0.089, P=0.100) and Trust (β=0.075, P=0.141) will be considered indicators of use intention for blogs, because of the boundaries set for this study previously.

5.4.2 FORUMS

Table 26 Hypotheses Testing for Forums

Estimate β

Expectancy

0.219 0.051 4.325 P<0.01 Supported Behavioural Intention  Effort Expectancy -0.031 0.046 -0.676 P=0.499 Not Supported Behavioural Intention  Social Influence 0.095 0.050 1.920 P=0.055 Supported Behavioural Intention  Facilitating Conditions 0.225 0.136 1.661 P=0.097 Supported Behavioural Intention  Trust 0.282 0.080 3.523 P<0.01 Supported Behavioural Intention  Hedonic Motivation 0.014 0.048 0.288 P=0.773 Not Supported Earlier conducted reliability and validity tests proved that all constructs of research models were possible determinants for Behavioural Intention to use forums while shopping online. After using structural equation modelling to test hypotheses of the research the following results were constructed: Habit (β=0.980, P<0.01), Performance Expectancy (β=0.219, P<0.01) and Trust (β=0.282, P<0.01) proved to be significant indicators of intention to use forums while shopping online. Social Influence (β=0.095, P=0.055) and Facilitating Conditions (β=0.225, P=0.097) will be considered as indicators of intention to use forums while shopping online, because of decisions made previously in the paper.

Second structural equation modelling analysis was performed without variables that proved to be insignificant to the use intention of forums. After analysis Facilitating Conditions showed P higher than 0.15, thus the analysis was performed again without that independent variable. This was done to test the validity of the model produced with the first structural equation modelling analysis. The second analysis of the model and hypotheses can be found below.

(27)

27 Figure 5 SEM2 for Forums

Table 27 Second Hypotheses Testing for Forums

Estimate β

Expectancy

0.223 0.051 4.412 P<0.01 Supported Behavioural Intention  Social Influence 0.112 0.049 2.267 P=0.023 Supported Behavioural Intention  Trust 0.287 0.080 3.585 P<0.01 Supported The second structural equation modelling analysis confirmed results of the first SEM analysis. Performance Expectancy (β=0.223, P<0.01), Trust (β=0.287, P<0.01) and Habit (β=1.013, P<0.01) are significant indicators of use intention for forums. Moreover, Social Influence (β=0.112, P=0.023) will be considered indicator of use intention for forums, because of the boundaries set for this study previously.

5.4.3 LINKS TO SOCIAL MEDIA

Table 28 Hypotheses Testing for Links to Social Media

Estimat e β

Expectancy

(28)

28 Behavioural Intention  Effort Expectancy -0.201 0.046 -4.348 P<0.01 Supported Behavioural Intention  Social Influence 0.324 0.053 6.072 P<0.01 Supported Behavioural Intention  Facilitating Conditions 0.533 0.384 1.390 P=0.164 Not Supported Behavioural Intention  Trust 0.090 0.057 1.574 P=0.116 Supported Behavioural Intention  Hedonic Motivation -0.062 0.032 -1.925 P=0.054 Supported Earlier conducted reliability and validity tests proved that all constructs of the research models were possible determinants for Behavioural Intention to use links to social media while shopping online. After using structural equation modelling to test hypotheses of the research the following results were constructed: Habit (β=0.673, P<0.01), Performance Expectancy (β=0.405, P<0.01), Effort Expectancy (β=-0.201, P<0.01) and Social Influence (β=0.324, P<0.01) proved to be significant indicators of intention to use links to social media while shopping online. Trust (β=0.090, P=0.116) and Hedonic Motivation (β=-0.062, P=0.054) will be considered as indicators of intention to use links to social media while shopping online, because of decisions made previously in the paper.

Second structural equation modelling analysis was performed without variables that proved to be insignificant to use intention of links to social media. After analysis Effort Expectancy and Hedonic Motivation showed P values higher than 0.15, thus the analysis was performed again without those independent variables. This was done to test the validity of the model produced with the first structural equation modelling analysis. The second analysis model and hypotheses can be found below.

(29)

29 Table 29 Second Hypotheses Testing for Links to Social Media

Estimate β

Expectancy

0.351 0.048 7.389 P<0.01 Supported Behavioural Intention  Social Influence 0.322 0.054 5.973 P<0.01 Supported Behavioural Intention  Trust 0.079 0.051 1.549 P=0.121 Supported The second structural equation modelling analysis confirmed results of the first SEM analysis. Performance Expectancy (β=0.351, P<0.01), Social Influence (β=0.332, P<0.01) and Habit (β=0.628, P<0.01) are significant indicators of use intention for forums. Moreover, Trust (β=0.079, P=0.121) will be considered indicator of use intention for links to social media, because of the boundaries set for this study previously.

5.4.4 PHOTO/VIDEO SHARING PLATFORMS Table 30 Hypotheses Testing for Photo/Video Sharing Platforms

Estimate

β SE CR P Value Hypothesis

Behavioural Intention  Habit 0.352 0.081 4.325 P<0.01 Supported

Behavioural Intention  Performance Expectancy 0.332 0.068 4.907 P<0.01 Supported

Behavioural Intention  Effort Expectancy -0.350 0.148 2.360 P<0.05 Supported

Behavioural Intention  Social Influence 0.320 0.059 5.458 P<0.01 Supported

Behavioural Intention  Facilitating Conditions -4.979 5.003 -0.995 P=0.320 Not Supported

Behavioural Intention  Trust 0.300 0.091 3.308 P<0.01 Supported

Behavioural Intention  Hedonic Motivation 0.206 0.056 3.665 P<0.01 Supported

Earlier conducted reliability and validity tests proved that all constructs of research model were possible determinants for Behavioural Intention to use photo/video sharing platforms while shopping online. After using structural equation modelling to test hypotheses of the research the following results were constructed: Habit (β=0.352, P<0.01), Performance Expectancy (β=0.332, P<0.01), Effort Expectancy (β=-0.350, P<0.01), Social Influence (β=0.320, P<0.01), Trust (β=0.300, P<0.01) and Hedonic Motivation (β=0.206, P<0.01) proved to be significant indicators of intention to use photo/video sharing platforms while shopping online.

Second structural equation modelling analysis was performed without variable that proved to be insignificant to use intention of photo/video sharing platforms. After analysis Effort Expectancy showed P value higher than 0.15, thus the analysis was performed again without that independent variable. This was done to test the validity of the model produced with the first structural equation modelling analysis. The second analysis model and hypotheses can be found below.

(30)

30 Figure 7 SEM2 for Photo/Video Sharing Platforms

Table 31 Second Hypotheses Testing for Photo/Video Sharing Platforms Estimate β

Expectancy

0.353 0.046 7.162 P<0.01 Supported Behavioural Intention  Social Influence 0.332 0.046 7.162 P<0.01 Supported Behavioural Intention  Trust 0.152 0.050 3.062 P=0.002 Supported Behavioural Intention  Hedonic Motivation 0.195 0.035 5.575 P<0.01 Supported The second structural equation modelling analysis confirmed the results of the first SEM analysis. Performance Expectancy (β=0.353, P<0.01), Social Influence (β=0.332, P<0.01), Hedonic Motivation (β=0.195, P<0.01) and Habit (β=0.372, P<0.01) are significant indicators of use intention for photo/video sharing platforms. Moreover, Trust (β=0.152, P=0.002) will be considered indicator of use intention for photo/video sharing platforms, because of the boundaries set for this study previously.

(31)

31 5.4.5 RATINGS

Table 32 Hypotheses Testing for Ratings

Estimate

β SE CR P Value Hypothesis Behavioural Intention  Habit 0.814 0.144 5.669 P<0.01 Supported Behavioural Intention  Performance Expectancy 0.254 0.067 3.769 P<0.01 Supported Behavioural Intention  Effort Expectancy 0.062 0.047 1.314 P=0.189 Not Supported Behavioural Intention  Social Influence 0.036 0.036 1.006 P=0.315 Not Supported Behavioural Intention  Facilitating Conditions 0.047 0.160 0.296 P=0.767 Not Supported Behavioural Intention  Trust 0.022 0.060 0.365 P=0.715 Not Supported Behavioural Intention  Hedonic Motivation -0.041 0.046 -0.891 P=0.373 Not Supported

Earlier conducted reliability and validity tests proved that all constructs of the research model besides Habit were possible determinants for Behavioural Intention to use ratings while shopping online. After using structural equation modelling to test hypotheses of the research the following results were constructed: Only Performance Expectancy (β=0.332, P<0.01) proved to be significant indicator of intention to use ratings while shopping online. SEM analysis indicated that Habit (β=0.814, P<0.01) should be a determinant for Behavioural Intention, but it was discredited because it did not pass validity testing.

Second structural equation modelling analysis was performed without variables that proved to be insignificant to use intention of ratings. This was done to test the validity of the model produced with the first structural equation modelling analysis. The second analysis model and hypotheses can be found below.

Figure 8 SEM2 for Ratings

Table 33 Second Hypotheses Testing for Ratings

Estimate β

Expectancy

0.310 0.070 4.398 P<0.01 Supported The second structural equation modelling analysis confirmed results of the first SEM analysis. Performance Expectancy (β=0.310, P<0.01) and Habit (β=0.506, P<0.01) are significant indicators of