The role of risk and trust on the purchase intention of luxury products E-Commerce:

i

E-Commerce:

The role of risk and trust on the purchase intention of luxury products

Master Dissertation

8

th

of December 2014

Advanced International Business Management and Marketing (Dual Award)



iii

E-Commerce:

The role of risk and trust on the purchase intention of luxury products

Master Dissertation by

E.M. Helder

Abstract

v

Master Dissertation

Which role do risk and trust play on the purchase intention of luxury products

within e-commerce?

Ewoud (Maarten) Helder

Schuitemakersstraat 2-25, 9711 HW Groningen e.m.helder.1@student.rug.nl

Tel: +31(0)6 23874266

student number RUG: S1768883 student number: NUBS: B4017381

Word count: 16.178 (including tables and in-house referencing) December 8th, 2014

DD-MSc. Advance International Business Management & Marketing Course code: EBM091A25

Course code: NBS8199

Supervisor & assessor: Drs. Ad Visscher (University of Groningen) ad.visscher@rug.nl

Supervisor & assessor: Prof. Dr. M. Blut (Newcastle University) markus.blut@newcastle.ac.uk Universities

University of Groningen, Faculty of Economics and Business Nettelbosje 2, 9747 AE Groningen

Tel: +31(0)50 3633741

Newcastle University Business School

vii

Acknowledgements

viii

Table of Contents

1 Introduction ... 1 2 Literature Review ... 4 2.1 Literature section ... 4 2.1.1 Luxury products ... 4 2.1.2 Purchase intention ... 5 2.1.3 Risks ... 6 2.1.4 Trust ... 6 3 Conceptual model ... 17 4 Methodology ... 18 4.1 Research Design ... 18 4.2 Questionnaire Design ... 19 4.3 Measurement of variables ... 21 4.4 Control Variables ... 22 4.4 Data analysis ... 22 4.5 Ethical considerations ... 23 5 Results ... 25 5.1 Descriptive analysis ... 25 5.2 Basic analysis ... 28

5.3 Testing the hypotheses ... 29

5.3.1 Multicollinearity... 29

5.3.2 Independent Samples T-test ... 30

5.3.3 Regression analysis ... 35

6 Conclusion and discussion ... 40

6.1 Research findings ... 40

7 Implications ... 43

7.1 Implications for research ... 43

7.2 Implications for practitioners ... 43

8 Research limitations and further research ... 45

9 References ... 48

Appendix I – Figure A.1 ... 53

Appendix II - Survey ... 54

Appendix III – Background research ... 59

Appendix IV- Output - Results SPSS ... 60

ix

1

1

Introduction

The internet has dramatically changed the industry practices of modern retailing and distribution management (Doherty and Ellis-Chadwick, 2006). Since 1995, electronic commerce (e-commerce) has changed business for firms as well as for consumers and people are increasingly relying on web-based commercial information. Laudon and Traver (2010: 24) define e-commerce as “the process of buying and selling or exchanging products and services besides collecting information via computer networks including the internet”. Today the global e-commerce market is worth $1.5 trillion and is growing rapidly (Economist, June 2014). The national Dutch research agency, CBS, stated that the number of people in the Netherlands buying goods and services online will increase the coming years. In 2013, 83 percent of the internet users reported to shop online and 61% of them are frequent online buyers (Appendix I).

Especially the commercial use of the internet continues to increase and online shopping is becoming an increasing part of our daily lives. This has mostly to do with the fact that it is possible to shop or do transactions 24 hours a day from almost any location. According to Hong, Tong and Tam (2004) this is mainly due to the increasing acceptance of online transactions by consumers. Besides that buying online is time-saving, it has a high ease of use and most of the time it saves costs for consumers (Noort, Kerkhof and Fennis, 2008).

However, one industry in particular has been slow in embracing the digital revolution, the luxury industry, because of its fear of losing exclusivity (Okonkwo, 2007). Recent research about luxury products and the use of the internet raised the question if luxury products are suitable to sell and buy online. The last few years, after the financial crisis of 2008, it became even more important for companies to focus on more trustworthy customer relationships. In order to succeed, the possibility to compare different products online and shorter delivery times became more essential for consumers (Laudon and Traver, 2010).

2

The first important concept that influences this relationship within B2C e-commerce is trust (Liu, Brums and Hou, 2013). It is well known that trust is an important issue for consumers who are involved in e-commerce (Keen, Wetzels, Ruyter and Feinberg, 2004). Gefen (2002) stated that there is a greater degree of trust required in an online shopping environment than in a physical shop, since the process of purchasing products online is intangible and not immediately verifiable for the consumer.

The second important influence on the online purchasing intention of consumers is perceived risk. There are certain issues that are particularly associated with the online buying process. Online consumers are often concerned about fraud, privacy concerns, system security, legal protection and the inability to sufficiently inspect goods prior to purchase (Flanagin, Metzger, Pure, Markov and Hartsell, 2014). This research focuses on the online purchasing intention since risk is expected to be more involved in online purchasing. The main reason for this is that consumers have to interact with technology to purchase the goods and services they need. Instead of focusing on face-to-face personal relationship, the physical shop environment is replaced by an electronic shopping environment (Keen et al., 2004). Payment- and counterfeit-issues are especially relevant today, Alibaba for instance, a Chinese online trading platform, is accused of selling fake luxury items by luxury brands like Gucci and Armani (Economist, 2014). Risk, trust and the online purchase intention are now shortly separately discussed. This research will further investigate the interaction effects between trust and risk, since trust is an essential factor under conditions of uncertainty and risk (Lee and Turban, 2011).

3

In short, this research focusses on the customer perspective and intends to investigate the risks that are associated with the online purchase intention of luxury products and the important factors that influence consumer’s trust. This leads to the following research question:

Which role do risk and trust play on the purchase intention of luxury products within e-commerce?

The aim of this paper is to provide useful insights into the consumer’s attitude towards the online shopping intention regarding luxury products. This study contributes to current literature by enhancing academics’ knowledge about the relatively new field of luxury products within e-commerce. Although the academic and practitioner literature about luxury branding is extensive, the specific relationship between luxury products and the variables risks and trust from the consumer perspective is not yet researched within the field of e-commerce (Whetten, 1989). Up till now these relationships are not integrated, which indicates there is a need to analyse this research gap. The findings of this study enable managers of luxury brands to gain a better insight in the purchasing behaviour (intention) of their customers. This research provides practical implications for merchants of luxury brand that wish to build or improve their online business by increasing the trust and decreasing the risk of their consumers. This study will elaborate on the managerial and academic contributions in the implications section.

4

2

Literature Review

This section reviews the existing research and provides descriptions related to the developments and characteristics of luxury products, risks and trust within e-commerce. This section is divided in two sections. First it provides a literature section (2.1 and 2.2) and after that the hypotheses section follows.

2.1 Literature section

This section will discuss what is currently written in the literature regarding (1) luxury products, (2) the purchase intention, (3) risk and (4) trust.

2.1.1 Luxury products

Currently, the luxury industry is largely dominated by large multi-brand groups that are publicly listed. The growth of luxury-businesses in the twentieth century broadened the customer base. The reputation for exceptional quality, design, durability and performance transformed brands into well-established brands (Brun and Castelli, 2013). Only just, luxury brands were the preserve of wealthy people from privileged backgrounds, however due to rising incomes and better availability of credit nowadays, luxury brands have become more affordable to a wider range of consumers than before (Wu, Chen and Chaney, 2013). Luxury is defined as a concept of things that are desirable but not “essential”, next to the fact that the difference between ordinary goods and luxury goods is determined by every consumer individually (Goody, 2006). However the degree of expensiveness is not the only important element of a luxury good; the strong brand name, the quality and emotional benefits of prestige, status and exclusivity of a product are important aspects as well (Grossman & Shapiro, 1988).

5

convenience, price, product availability and have an online shopping attitude with a desire to search for variety. While online trust in-store luxury product buyers focus points are aesthetic appeal, store trust, shopping experience, customer service and sense of power. Since online consumers who are interested in buying luxury products are focused on the quality, price and design of the product, it is interesting to investigate if there is a difference between this type of products and more ‘ordinary’ products.

Why are luxury products different from ‘ordinary’ products? Liu et al. (2013) claims that in the case of selling luxury products there must be a bigger focus on the establishment of trust. For instance the existence of counterfeit products plays a much bigger role within the luxury product industry. This could lead to consumer’s suspicion of the authenticity of certain websites or regarding the products they offer (Wu et al., 2013). Besides the high price, people appraise and value these kind of products more. Luxury products convey a sense of status, wealth, and exclusivity and therefore the perceived risks are higher. Consumers purchase luxury goods for a variety of reasons, sometimes the type of feelings are unique compared to more basic products (Brun and Castelli, 2013). Luxury product consumers exhibit different kind of consumer behaviour since they possess other shopping motivations (Liu et al., 2013). This research focuses on aspects that have to with perceived trust and risks that are associated with luxury products.

2.1.2 Purchase intention

6

Prior research confirmed a strong correlation between behavioural intentions and actual behaviour (Venkatesh and Davis, 2000). According to Fishbein and Ajzen (1975) a behavioural intention measure will predict the performance. They state that as long as no radical changes prior to the performance occur, the behavioural intention is strongly correlated with the actual behaviour of the consumer (Sheppard, Hartwick and Warshaw, 1988). Therefore this research conducts research concerning the purchase intention.

Table 1.1 summarises the existing literature and provides an overview of articles that have already been conducted in this field. It shows that a large base of literature concerning e-commerce, luxury products, the purchase intention, risks or the role of trust on the buying process of the customer exists. However, this study especially investigates the relationship between trust and risk and how they influence the online purchase intention regarding luxury products. In this way it aims to attribute to the existing literature. This study uses for instance the research of Kim, Rao and Ferrin (2007), Lowry et al. (2008) and Mayer, David and Schoorman (1995) as a foundation for the hypotheses.

2.1.3 Risks

According to Gillet (1976) previous research related to shopping-behaviour suggested that perceived risk is also affected by how a product is purchased and not only by what kind of products are purchased. Consumers that are engaged in e-commerce transactions are confronted with certain risks. They must “assess the credibility of information provided online, as well as the trustworthiness of the internet as a commercial medium” (Flanagin et al., 2014: 12). There are different levels of perceived risk for the consumer and according to Hawes and Lumpkin (1986: 38) these risks increase “the more the consumer is separated from the physical presence of the retail store”. This research elaborates on these different levels of risks in the hypotheses section.

2.1.4 Trust

7

Table 1.1: Overview of the literature used in this research

Subject Authors Title/ subject

Luxury products

Brun and Castelli (2013) The nature of luxury: a consumer perspective

Kim and Ko (2010) Impacts of luxury fashion brand's social media marketing on customer relationship and purchase intention

Okonkwo (2009) Sustaining the luxury brand on the internet Vigneron and Johnson (1999) Measuring perceptions of brand luxury Wu, Chen and Chaney (2013) Luxury Brands in the Digital Age

E-commerce

Chaffey (2009) E-business and e-commerce management Horrigan (2008) Online shopping in the U.S.

Hong, Tong and Tam (2004) Online shopping behaviour

Laudon and Traver (2010) E-commerce 2010, the history of e-commerce and guideline for companies how to deal with issues related to e-commerce

Purchase intention

Park, Lee and Han (2007) Effect of online consumer reviews on consumer purchasing intention

Summers, Belleau & Xu (2006)

Predicting purchase intention of a controversial luxury apparel product

Van der Heijden, Verhagen and Creemers (2003)

Understanding online purchase intentions: contributions from technology and trust perspectives

Zhang and Kim (2013) Luxury fashion consumption in China: Factors affecting attitude and purchase intent

Trust

Kim, Chung and Lee (2011) The effect of perceived trust on electronic commerce: Shopping online for tourism products and services in South Korea

Kim, Ferrin, and Rao (2007) Trust and satisfaction, stepping stones for successful e-commerce relationship

Koufaris, Hampton-Sosa (2004)

The development of initial trust in an online company by new customers

Lowry, Vance, Moody, Beckman and Read (2008)

Explaining and predicting the impact of branding alliances and web site quality on initial consumer trust of e-commerce web sites

Lowry, Vance, Moody, Beckman and Read (2008)

The impact of branding alliances and website quality on initial consumer trust of e-commerce websites

Lee and Turban (2001) A trust model for consumer internet shopping Mayer and Schoorman (1995) Integrative model of organizational trust McKnight, Choudhury and

Kacmar (2002)

Developing and validating trust measures for e-commerce

Metzger (2006) Effects of site, vendor, and consumer characteristics on web site trust and disclosure (focus on music industry) Song, Hur and Kim (2012) Brand trust and affect in the luxury brand–customer

relationship

Won-Moo Hur (2014) The role of brand trust in male customers' relationship to luxury brands

Perceived risk in online shopping

Chang, Cheung and Lai (2005)

Literature derived reference models for the adoption of online shopping

Flanagin, Metzger, Pure, Markov and Hartsell (2014)

Mitigating risk in e-commerce transactions: perceptions of information credibility and the role of user-generated ratings in product quality and purchase intention

Hawes and Lumpkin (1986) Perceived risk within retail sector

Wilcock and Boys (2014) Reduce product counterfeiting: an integrated approach

8

the ability to monitor or control that other party”. This definition is relevant since consumers who operate in the online environment need to consider the relative credibility of commercial information sources on the web instead of traditional offline places (Lowry et al., 2008). Online commercial transactions often lack elements that are used to ensure trust and credibility among exchange parties (Flanagin et al., 2014). Besides that, trust could moderate risk in the buying process (Lee, Park and Han, 2011).

2.2 Hypotheses section

This section provides the hypotheses based on the literature section above.

Relationship between perceived risks and trust

A consumer’s perceived risk is an important barrier for online consumers who are considering whether to make an online purchase (Kim et al., 2007). There are different kind of risks consumers have to deal with, for instance regarding security protection, probability of counterfeit products and delay in delivery. Therefore, consumers will be attentive to risks in online transactions, and such risks may influence their decisions whether or not to purchase from an online vendor (Kim et al, 2007). Therefore the first hypothesis is defined as follows: H1a: A consumer's perceived risk in general negatively affects a consumer's intention

to purchase on the internet.

The general risks are commonly measured by asking respondents to assess whether buying goods online can be considered as risky. Jacoby and Kaplan (1972) measurement scale is widely used in measuring perceived risk and a consumer’s risk aversion. They identified six different types of general risks related to purchasing products in general (see Table 1.2). This study measures which risk components influence the purchase intention and will elaborate on this in the data analysis section. In the case of web shopping, three types of risk are expected to be predominant; financial risk, functional risk and information risk (Bhatnagar, Misra and Rao, 2000). This research argues that physical risk is less relevant in because the safety aspect does not differ from buying offline. The social and psychological risk are also less relevant, since these risks are the same offline as online (Kim, et al., 2009).

9

of functional risk, which exists when transactions are done over the internet and when the customer’s identity could not be established (Kim et al, 2007). For example when consumers use credit card payments over the internet. Other reasons for payment disputes can arise from goods that are paid but do not reach the customer. Fulfilment risk refers to the delivery of luxury goods and is also part of financial risk (Bhatnagar, Misra and Rao, 2000).

Table 1.2: Operational definitions of types of risks

Type of risk Operational definition Measure

Functional risk

The actual design, appearance, and functionality of the luxury good (risk of counterfeit-products is part of functional risk), and the risk that a product needs to be repaired or is not delivered properly.

1= high functional risk 5 = low functional risk

Social risk Regarding the opinion of family and friends when buying a product

1 = low social risk 5 = high social risk

Financial risk Concerning (online) transactions, e.g. risk of losing _{money); ‘fulfilment risk’ also part of financial risk} 1 = low financial risk _{5= high chance financial risk}

Physical risk Safety of a product 1= very safe

5 = very unsafe

Psychological risk Risk capital includes affiliations and status 1= low psychological risk _{5= high psychological risk}

Information risk Loss of private information/ security issues credit card fraud.

1 = low information risk 5 = high information risk

Second, functional risk is higher within internet shopping because it involves more uncertainty and risk than traditional shopping. This has to do with fact that a consumer cannot physically check the quality of the product before making a purchase (Lee and Turban, 2011). According to Bhatnagar et al., (2000) product risk is especially relevant for products in which the price, feel, touch and technical complexity are important. This is regarding situations where goods are defective or not matching the description or it refers to delays in delivery (Jacoby and Kaplan, 1972). Kedar (2014) argues that since purchases are performed online, consumers are also more vulnerable to risks such as hijacking or duplication (counterfeit). Uncertainty about the authenticity of the products is one of the major reasons that consumers are a little bit hesitant towards purchasing luxury products online. In this case the influence of this risk on the purchase intention of luxury products is relevant to research.

10

(e.g. credit card numbers) through the internet is important, since there are parties whose behaviours and motives may be hard to predict (Lee and Turban, 2011). Sometimes client information gets lost (e.g. credit card details), this is something consumers fear. Security and privacy risk are important elements of information risk. Viruses and hacker attacks are examples of risks that can arise and could lead to trust damage and confidence of the consumer. Forms of damage are for example, the loss of data, stolen information, denial of service or illegal manipulation of information content (Kedar, 2014). This is often measured with the perceived privacy protection (PPP) and perceived security protection (PSP) measurements designed by Kim et al. (2007). These three risks are considered to be most relevant regarding the online purchase intention. Therefore, the effect of these three risks together on the general perceived risk will be investigated:

H1b: A consumer's perceived financial risk (H1c), functional risk (H1d) and information risk (H1e) negatively affect a consumer's perceived risk in general.

It is important for consumers to assess the risks that threaten the potential purchasing process. The aim of this research is to examine if certain risks are particular relevant in relation to the online purchase intention of luxury products. In short, this study examines which risks influence the customer online purchase intention the most within the luxury product segment. Therefore the following hypotheses are proposed:

H1c-1e: A consumer's perceived financial risk (H1c), functional risk (H1d) and information risk (H1e) negatively affect a consumer's online purchase intention.

11

H1f: A consumer’s perceived information risk has a stronger negative effect on the

online purchasing decision than perceived financial risk. Relationship between perceived risks and trust

An important variable that influences perceived trust, which will be discussed now, is perceived risk. Consumers are focused on risks in online transactions, and such risks may influence the willingness of a consumer to depend on a person or a company (McKnight et al., 2007). When risks exist, faith in the process of purchasing declines, which results in a lower disposition to trust considered by the customer. Consequently this can have impact on the consumer’s decisions whether or not to purchase from an online vendor (Kim et al, 2007). Therefore risks play a vital role on the development of insecurity and trust (Lowry et al., 2008). Resulting in the following hypothesis:

H2a: A consumer's perceived risk in general negatively affects a consumer's perceived trust on the internet.

When risk is present, trust is important and needed to make transactions possible. In short, this research examines which risks influence the customer perceived risk the most. Subsequently, the following hypotheses are proposed:

H2b-2d: A consumer’s perceived financial risk (H2b), functional risk (H2c) and information risk (H2d) negatively affect a consumer's perceived trust on the internet.

Relationship between trust and the online purchase intention

12

H3a: A consumers’ perceived trust in general positively affects the consumer's intention to purchase on the internet.

Existing literature is focused on trust (i) in specific brands, (ii) in specific industries (like apparel or online tourism), (iii) in websites (e.g. specifications and designs) or in (iv) the role of gender or culture for instance. This research focusses on the role of trust in one specific industry: luxury products. Since trust is a strong and positive predictor of a consumer’s intention to purchase (Ba and Pavlou, 2002). In the case of luxury products there are four categories (see Table 1.3) relevant for trust (adapted from Kim et al., 2007). The different trust categories will be discussed next with the accompanying hypotheses.

Table 1.3: Operational categories for trust

Type of trust Operational definition Measure

Cognition-based trust The reliability of a website (Lee and Turban, 2011) 1= high perceived trust 5 = low perceived trust

Affect-based trust Reputation of the luxury brand/ vendors website 1= high perceived trust 5 = low perceived trust

Experience-based trust _{Online purchasing experience (McKnight et al., 2002)} Familiarity with the luxury brand- or vendors website 1= high perceived trust 5 = low perceived trust

Personality- oriented

trust Trusting Belief (McKnight et al., 2002)

1= high perceived trust 5 = low perceived trust

Cognition-based

Reliability refers to the trustworthiness of a website of a luxury brand. This operational

definition refers to the extent to which consumers can rely on promises of vendors (Lee and Turban, 2011). It is often very difficult to tell how frequently the information of websites are updated and whether the facts have been checked (Kim, Ferrin and Rao, 2007). Thus, potential buyers on the internet are likely to be particularly focused on the quality of information of a website since this influences the purchasing intention (Kim et al., 2007). Affect-based

13

H3b-3c: The perceived reliability (H3b) and perceived reputation (H3c) of a luxury-brand website are positively related to the online purchase intention.

Experience-based

Customers gain experience by active participating in the online buying process (Walzuch and Lundgren, 2003). Familiarity, satisfaction and communication are three important aspects that are related to this experience. The attitude towards online shopping depends on the fact if an experience was successful or not. Ganesan (1994) argues that trust will develop over time as consumers build trust-relevant knowledge through experience with e-commerce, since experience derives from active interaction with a process. People who are experienced in buying products online and had positive experiences in the past, will perceive e-retailing as trustworthy (Walczuch and Lundgren, 2004).

H3d: A consumers’ online purchasing experience (H3d) positively affect the consumer's intention to purchase on the internet.

Other researchers like McKnight et al. (2002) and Kim et al. (2009) also examined online purchasing experience. There are some variables that influence experience-based trust, such as the familiarity with a vendors brand or website. According to Kim et al. (2007) familiarity is a precondition of trust. When consumers are unfamiliar with a vendor’s website, they are less likely to trust it and are therefore less likely to purchase products from it (Gefen, 2002). Online consumers are highly influenced by the fact and the way they are familiar with the product or with the brand of the product (Mauldin and Arunachalem, 2002). Gefen (2002) adds that familiarity with a vendors brand or website influences trust. Since consumers are often familiar with luxury brands, they have high brand awareness. This study focusses on the familiarity of the consumer with a luxury brand website.

H3e: A consumer's familiarity with a luxury brands or vendors website positively affects the consumer's online purchase intention.

Personality-oriented

14

This research will extend the research of Kim et al. (2007) by adding trusting belief to the personality oriented trust pillar and extend certain variables by the use of important articles, as some studies were published after the initial research by Kim et al. (2007). According to Gefen (2002) trusting belief is defined as to which extent consumers believe another party will act with benevolence, integrity and competence. These three factors are determining factors regarding trusting belief. This study investigates if consumers belief that luxury brand websites will act with benevolence, integrity, and competence toward the customers.

H3f: A consumer's trusting belief positively affects the consumer's intention to purchase on the internet.

Relationship between trust and risk

This section elaborates on the relationship between risk and trust. This research already discussed the influence of risk on trust and argued that it is expected that a consumer’s perceived general risk (H2a), financial risk (H2b), functional risk (H2c) and information risk (H2d) negatively affect a consumer's perceived trust on the internet. However, there is no consensus yet on the relationship between risk and trust. Some argue that risk is an antecedent to trust, others argue it is an outcome of trust (Mayer, Davis and Schoorman, 1995). According to Mayer et al. (1995) the need for trust only arises in a risky situation. Internet shopping involves more uncertainty and risk than traditional shopping since in most cases a consumer cannot physically check the quality of the product before making a purchase. Therefore, it is relevant to investigate the interaction effects between risk and trust, since trust is only relevant under certain (risky) circumstances. Since trust and risk are closely interrelated (Mayer et al., 1995) this study will examine if multicollinearity, high inter-correlation between these interdependent variables, takes place. This will be discussed in chapter 5.

15

consumers are likely to perceive less risk than if trust is absent (Kim et al, 2007). Henceforth, the following hypothesis examines this relationship between trust and risk:

H4a: A consumers’ perceived trust in general, measured by the disposition to trust,

negatively affects the consumer's perceived risk.

When risk is present, trust is needed to make transactions possible. According to Antony, Lin and Xu (2006) a positive reputation of the selling party is considered as a key factor to reduce risk. Therefore the perceived reliability and reputation of luxury brands websites are considered important. Since these influence the customer’s esteem regarding a selling party (Kim et al, 2007). A positive perceived reliability and reputation by consumers is therefore expected to affect the perceived risk negatively.

H4b-4c: The perceived reliability (H4b) and perceived reputation (H4c) negatively affect the consumer's perceived risk.

People who are experienced in buying online and have had positive experiences will consider e-retailing as less risky (Walczuch and Lundgren, 2004). When web vendors are known it reduces the uncertainty (Kim et al., 2007). This results in the following hypothesis:

H4d: A consumers’ online purchasing experience negatively affects the consumer's

perceived risk.

For consumers, familiarity leads to a better understanding of a vendors current actions (Gefen, 2000). When a consumer is familiar with the vendor, he or she gains knowledge about the vendor and understanding of its (i) relevant procedures such as the search for information or (ii) the search for products and becomes familiar with ordering on the website, the consumer experiences less risk since there is less uncertainty (Kim et al., 2007). In short, it simplifies the relationship between the buyer and the seller. Therefore this study expects that familiarity has a negative effect on risk.

H4e: A consumer's familiarity with a luxury brand or vendors website negatively affects the consumer's perceived risk.

16

ability to trust the vendor. Benevolence as the goodwill and responsiveness of the vendor and integrity covers the morality of the vendor. Since these three beliefs reduce uncertainty, this research expects these beliefs negatively affect the consumers perceived risk.

H4f: A consumer's trusting belief is negatively related to the consumer's perceived risk.

An overview of the different hypotheses is provided that are discussed and developed in the literature section, with their expected effect (see Table 1.4).

Table 1.4: Overview of hypotheses

Hypotheses Explanation Expected effect

H1a A consumers’ perceived risk in general, negatively affects a consumer's

intention to purchase on the internet. -

H1b

A consumer's perceived financial risk (H1c), functional risk (H1d) and information risk (H1e) negatively affect the consumer's perceived risk in general

-

H1c-1e

A consumer's perceived financial risk (H1c), functional risk (H1d) and information risk (H1e) negatively affect a consumer's online purchase intention

-

H1f A consumer’s perceived information risk has a stronger negative effect on

the online purchasing decision than perceived financial risk -

H2a,b,c,d

A consumer’s general perceived risk (H2a), financial risk (H2b), functional risk (H2c) and information risk (H2d) negatively affect a consumer's perceived trust on the internet.

-

H3a

A consumers’ perceived trust in general, measured by disposition to trust, positively affects the consumer's online purchase intention to purchase on the internet

+

H3b,c,d,e,f

The perceived reliability (H3b), perceived reputation (H3c), a consumer's familiarity (H3e) of a website and consumer’s online purchasing experience (H3d) and trusting belief (3f) positively affect the consumer's intention to purchase on the internet

+

H4a A consumers’ perceived trust in general, measured by disposition to trust,

negatively affects the consumer's perceived risk -

H4b,c,d,e,f

The perceived reliability (H4b), perceived reputation (H4c), a consumer's familiarity (H4e) of a website and the consumer’s online purchasing experience (H4d) and trusting belief (H4f) negatively affect the consumer's perceived risk

17

-

-

-

3

Conceptual model

The variables discussed in the theoretical framework combined with the proposed hypotheses lead to the following conceptual model (Figure 3.1). The conceptual model depicts that it is expected that the role of risk negatively influences the online purchase intention and perceived trust. The role of trust is expected to positively affect the online purchasing intention and negatively influence risk in buying luxury products via e-commerce. With the different antecedents of trust and different forms of risks it offers a detailed description of the theoretical framework discussed in the literature section.

Figure 3.1: Conceptual model

H3

Role of trust

Online purchase intention concerning luxury products via e-commerce

18

4

Methodology

The previous chapters discussed the hypotheses, the research question and the conceptual model. Empirical research is conducted to test the conceptual model. This section will discuss the sample, data collection and data analysis techniques that are used. This section will elaborate on the way the research is conducted, explain the research design and discuss the design of the questionnaire. Concluding with explaining how the variables will be measured and subsequently how the results will be analysed.

4.1 Research Design

Before an answer to the research question can be provided and the hypothesis can be tested, the data must be collected and analysed. As already mentioned, the research design is quantitative by nature, by using structured online questionnaires with the sampling of data. Bryman and Bell (2013) have provided arguments that supported the validity of this measurement approach. By using self-completion questionnaires it was possible for the respondents to complete the questionnaires in their own time and it allows data gathering via the internet (Bryman and Bell, 2013). The survey strategy will be used as the research method to measure trust, perceived risk and the purchase intention. The reason for selecting data by means of an online survey is due to the fact that it enables to gather a large amount of information in a relatively short time span (De Leeuw, 1996).

Quantitative research questionnaires (primary data) have been conducted in the Netherlands and England. The target group for this survey will include adults in the Netherlands and England who make use of the internet. According to Thomas (2003) between 120 and 150 people would provide a sufficient sample-size scope for analysis. To increase participation, the participants are offered the outcomes of the study after the research is finished. The online survey program Qualtrics is used to design and distribute the questionnaire, this collaboration with Qualtrics is made possible by the University of Groningen.

The cross sectional research exists of a heterogeneous group of people with different ages, education levels, differing from students to employees with different backgrounds. The survey will be available online, to increase the speed of data collection and to reach a potentially larger number of respondents to complete the questionnaire. The questionnaire was spread via email, social media (e.g. LinkedIn) and other online channels.

19

online. This information provides an insight in how many people do and do not shop online within the sample. Respondents are asked if they are familiar with buying a (luxury) product online. If not, they are asked to assume they are intending to buy a luxury product in the future (so-called hypothetical question). The answers are still very relevant since each individual’s personality, purchasing ability and preferences determines which items will be valued and perceived as luxury items and what their motives are.

The study analyses the perception of the respondents and investigates their purchase intention and attitude towards online shopping of luxury products. This research consists of three main variables (see Table 4.1). One dependent variable (online purchase intention) and two independent variables (role of trust and rust of risk).

Table 4.1: Overview of dependent or independent variables

Variable Type

Online purchasing intention Dependent

Perceived trust Independent

Perceived risk Independent

The two independent variables will be separately measured. This research also applies a few control variables, such as gender and age, the next section provides all the variables. The questions are measured using five-point Likert scales, ranging from e.g. a very small risk (1) to a very high risk (5). Regarding trust questions, e.g. it is relatively easy for me to trust a person/thing; (1) strongly disagree till strongly agree (5). The next section will discuss the design of the questionnaire followed by the statistical analysis.

4.2 Questionnaire Design

Before the main data gathering takes place, a small scale preliminary study was conducted. In the start phase of this research a small number of people were interviewed to get an indication how people think about purchasing online and more specifically about luxury products. As part of this study a variety of consumers and retailers are interviewed, for instance the owner of a jewellery store is interviewed (see Appendix III). The aim of the interviews was to investigate if the designed questions regarding the survey would work and if the research instrument measures the right data the research requires to answer the research questions (Bryman and Bell, 2003).

20

Moreover it was not mandatory for consumers to answer this question. After the socio-demographic questions, respondents are asked about their online purchasing experience, questions like: have you ever purchased a product on the internet (Q7.1) or have you ever bought a luxury product on the internet? (Q7.2), are used (adapted from Walczuch and Lundgren, 2004).

Next, the online purchase intention is measured based on the research of Zhang and Kim, (2012). Examples of questions are how likely is it that you would buy a luxury good on the internet (Q8.2) or would you recommend this to your friends (Q8.3)? Furthermore, Appendix II consists of all the questions that are part of the questionnaire.

Then general perceived risk is measured by investigating if purchasing from a website would involve more when compared to more traditional ways of shopping (Q.9). Subsequently different perceived risks are measured; functional, financial and information risks (five point Likert scale, very low risk - very high risk) based on research from Kim et al. (2007). Functional risks are measured by e.g. asking participants if they believe the product may fail to meet their expectations (Q12.1), are not delivered properly (Q12.2) or the possibility of they buy counterfeit products (Q12.4, adapted from Jacoby and Kaplan, 1972). Corbitt and Thanasankit (2003) provide examples of questions to measure financial risk: I believe that online purchases are risky because there is the risk that a technological error can occur (Q13.2) or the product will not be delivered (Q13.4). Lastly, information risk is measured by asking if respondents are worried website retailers are collecting too much personal information from them (Q14.1) or website vendors will share or sell personal information with other entities without approval of the consumers (Q14.3) (adapted from Kim et al., 2008).

Subsequently, general trust is measured by measuring the disposition to trust (adapted from Koufaris et al. (2004). General propositions like, it is relatively easy for me to trust a person/thing (Q23.4) or I tend to trust a person/thing, even though I have little knowledge of it (Q23.5) were used to measure the general trust factor.

21

by reverse code questions to control and keep the attention of the respondents (e.g. question 16.3: anyone trusting internet shopping is asking for trouble (reverse).

The questionnaire ended by measuring familiarity (based on Kim et al, 2007) and benevolence, integrity and competence all part of trusting belief (adapted from McKnight, Choudhury and Kacmar, 2002). When respondents are searching for luxury items online, they were asked about their familiarity with the site (Q21.1), familiarity with searching for items online (Q21.2) or familiarity with the process of purchasing from this site (Q21.3). To measure benevolence respondents were asked if they assumed luxury brand webshops act in the best interests of the consumers (Q22.1). Or if they would do their best if consumers needed help (Q22.2) and if they were interested in the well-being of their consumers (Q22.3). Integrity was measured by asking if luxury brand webshop keep its commitments (Q22.4), are honest (Q22.5) and are sincere and genuine (Q22.6). Finally, competence was measured by asking if respondents believed luxury brand webshops are competent and effective (Q23.1), reliable (23.2) and competent in their area of expertise. These three variables, integrity, honesty and competence (all part of trusting belief) are based on research of McKnight et al. (2002).

4.3 Measurement of variables

Scale questions are often used to collect data on attitudes and beliefs of the respondents (Saunders, Lewis and Thornhill, 2011). Since categorical (ordinal) variables are used, this research uses the five-point Likert scale, (1 = strongly disagree, 2 = disagree, 3 = Neither agree nor disagree, 4 = agree, 5 = strongly agree). This scale creates enough possibilities for the respondent to express their opinion, it is measurable and the odd number gives respondents the possibility to answer neutral (3) (Malhotra and Birks, 2007).

The survey consists of questions used to measure the variables mentioned in the conceptual model. The relevant causalities are based on the relationship between the online purchase

intention (dependent variable) and the direct effects of the role of trust (independent variable)

22 4.4 Control variables

In addition to the independent and dependent variables this research made use of control variables. Control variables are used to make sure the right variables are tested and to specifically look at the relationship between the variables (see Table 4.2). Socio-demographic characteristics like gender (nominal variable), age (nominal variable; 18-29, 29-45, 46-64 or 65+ ) and highest level of education (Primary or High School, Vocational Education (in Dutch: MBO), University of Applied Sciences (in Dutch: HBO), Bachelor’s or Master’s degree (University) were measured. Besides that the country of residence (interval) and gross annual income (nominal) of the respondents were considered as relevant. Appendix I provides more information about all the variables and survey questions used for this measurement.

Table 4.2: Overview of measurement scales for the different variables

Variable Type Example

Gender Nominal Male, female

Age Nominal 18-29, 29-45 years

Highest level of education Nominal High School, Bachelor’s

Country of residence Interval The Netherlands, UK

Gross annual income Nominal € 20.000 - € 29.999

4.4 Data analysis

In order to answer the research question and the formulated hypotheses, several analyses are performed on the obtained data; this section explains the plan of analysis. After providing the descriptive statics, the data are checked for their reliability and validity using Cronbach’s Alpha. Next, the nature of the distribution will be checked using a normality test with Skewness and Kurtosis statistics, to understand how each variable in the dataset is distributed. Before conducting a regression analysis it is important to check for multicollinearity between the independent variables perceived risk and trust, to make sure there is not a state of high inter-correlation between the independent variables (Malhotra et al., 2007).

23

external variables regarding purchasing behaviour used t-tests or analysis of variance (ANOVA) to determine these influences (Summer and Belleau (2006).

First the normal distribution of the groups is measured. Since the scores are not all equally distributed, a median-split is conducted. By creating two groups of 79 respondents the difference between the low-risk and the high-risk-received groups are shown. Liu et al. (2013) also applied this T-test to give a clear view of the (mean-) differences within one sample. Subsequently, the hypotheses are analysed using the regression analysis to examine the significance of the relations in order to measure the strength of the relations.

4.5 Ethical considerations

This paragraph discusses the ethical concerns and the theoretical and philosophical assumptions behind the methodological choices of this research. Ethical concerns throughout the research process are regarding the “appropriateness of the researchers’ behaviour in relation to the rights of those who become the subject of the research, or are affected by it” (Saunders et al, 2011:178). According to Corley and Gioia (2011) there are four main areas concerning ethical principles in business research: (i) the privacy of the respondents have to be respected and (ii) the participants may not be harmed. (iii) A lack of informed consent and deception should be avoided. Moreover, it must be certain that (iv) the researcher is unaffiliated with the companies that cooperate with the research, in this case this concerns the companies within the luxury industry. Because it is an exploratory research it was not necessary to have contact with particular luxury companies or webshops.

Because (electronic) questionnaires are conducted, the confidentiality of data and anonymity are essential. Since this research involves human participants, the privacy of the participants should be respected. Participants that want to participate have to agree with the informed consent ethics form (first page questionnaire: By ticking ‘Yes’, participants confirm that they

24

25

5

Results

This chapter discusses the results of the analyses. First the descriptive analysis is shown, after which the reliability analysis is executed to test the internal consistency by determining the Cronbach’s Alpha (α). To gain more insights in the distribution of the data a normality test is performed. This research used the Independent-Samples T-test to compare the means between two groups on the dependent and independent variables. Subsequently, in order to test the hypotheses, the results from the regression analyses are needed to determine the significance of each relationship.

5.1 Descriptive analysis

The online questionnaire was started by 192 respondents, after which 34 respondents were excluded since they did not complete the entire questionnaire, resulting in a sample of 158 respondents. The sample is first described with the use of five demographic variables: gender, age, nationality, highest level of education and income based on the dataset (see Table 5.1).

Figure 5.1: Country of residence respondents

Figure 5.1 depicts the variety of residences of the respondents in which it is noticeable that 89,2% of the respondents are from the Netherlands. Although, the sample also contains several other countries, the majority has the nationality of the two international universities that are part of the program and subsequently of this research.

26

Table 5.1: Socio-demographic characteristics of the sample.

Demographic variable Sample Percentage

Sample size (n) 158 100% Gender Male Female 106 52 67,1 32,9 Age 18 – 29 years 29 – 45 years 46 – 64 years 65 + years 111 18 25 4 70,3 11,4 15,8 2,5 Level of education Primary School High School

Vocational Education/ MBO

University of Applied Sciences (HBO) Bachelor’s Degree (University) Master’s Degree (University) Doctoral Degree 1 18 1 28 64 41 5 0,6 11,4 0,6 17,7 40,5 25,9 3,2 Country reside Belgium China Germany Netherlands Slovenia Spain

United Kingdom and North. Ireland United States of America

Uruguay 3 4 2 141 1 2 3 1 1 1,9 2,5 1,3 89,2 0,6 1,3 1,9 0,6 0,6 Gross income under € 20,000 € 20,000 - € 29,999 € 30,000 - € 39,999 € 40,000 - € 49,999 € 50,000 - € 59,999 € 60,000 - € 69,999 € 70,000 - € 79,999 € 90,000 - € 99,999 € 100,000 - € 119,999 € 120,000 - € 139,999 € 140,000 - € 159,999 € 140,000 +

Not willing to answer

27

Figure 5.2: Highest level of education completed by the respondents

Many of the respondents have a background in academics or applied sciences (87,6%). Moreover, the percentage of 70,3%, which shows that 111 respondents are between 18-29 years old (see Figure 5.3) is explainable since the research was conducted mainly at the University of Groningen and Newcastle where many students completed the online survey (see Figure 5.2). This is also noticeable in the age-distribution of figure 5.3.

Figure 5.3: Age distribution of the respondents

28 5.2 Basic analysis

The basic analysis for this research consists of conducting a (i) reliability and (ii) normality test. The reliability analysis is performed to discover whether the constructs incorporated in this study are internally consistent. This is tested by calculating the Cronbach’s Alpha (α). At least three items-scales and sometimes four items-scales (containing one reverse coded item) were used, to make sure that Cronbach’s Alpha was above the threshold of 0,7 (Nunnally, 1967). Than the items can be considered as internally consistent (see Table 5.2).

Table 5.2: Overview of Cronbach alpha’s per variable

Variable Cronbach’s Alpha (α) Number of items

Benevolence α 0,739 3 Competence α 0,724 3 Disposition to trust α 0,740 3 Familiarity α 0,885 4 Financial risk α 0,706 4 Functional risk α 0,749 4 Information risk α 0,853 4 Integrity α 0,816 3

Online buying experience α 0,575 2

Online purchasing intention α 0,774 3

Perceived reputation α 0,766 3

Perceived risk α 0,704 3

Perceived security α 0,719 3

Reliability α 0,746 3

To gain better insights in the distribution of the data, a normality test is conducted for the independent and dependent variables (see Table 5.3). The Skewness and Kurtosis statistics are used to see how each variable is applied and distributed in the dataset. Blumberg, Cooper and Schindler (2011) stated that this statistical test measures the flatness and peakedness of the distribution. The Skewness statistic is used to see whether the data is symmetrical and normally distributed, and the Kurtosis statistic shows how flat or how peaked the distribution is. The Skewness statistics between -0.5 and 0.5 are interpreted as normal and the Kurtosis statistics are normally peaked between -1.96 and 1.96 (Blumberg et al., 2011) Nonetheless, the closer to zero, the more normally distributed a variable is.

29

indicates that the distribution is peaked. However, based on these scores this research does not have to indicate any violation of the normality assumption (Blumberg et al., 2011).

Table 5.3: Normality test analysis

Variable Skewness Standard error Kurtosis Standard error

Benevolence -,096 ,193 ,402 ,384

Competence -,881 ,194 3,238 ,385

Disposition to trust -,133 ,193 -,269 ,384

Familiarity with the website -,744 ,193 ,480 ,384

Financial risk ,146 ,193 ,372 ,384

Functional risk -,606 ,193 ,339 ,384

Information risk -,429 ,193 -,332 ,384

Integrity -,299 ,193 ,648 ,384

Online buying experience -,119 ,193 -,996 ,384

Online purchasing intention -,182 ,193 -,659 ,384

Perceived reputation ,074 ,193 ,560 ,384

Perceived risk -,556 ,193 ,100 ,384

Perceived security -,798 ,193 1,216 ,384

Reliability -1,303 ,193 3,216 ,384

5.3 Testing the hypotheses

This section discusses the results of (i) multicollinearity, (ii) the Independent Samples T-test, and (iii) the regression analysis.

5.3.1 Multicollinearity

30

Table 5.4: Multicollinearity check

Variables VIF Independent variables Perceived risk 1,482 Functional risk 1,632 Financial risk 1,621 Information risk 1,743

Disposition to trust (general trust) 1,804

Control variables

Age 1,097

Gender 1,161

Country of residence 1,224

Highest level of education 1,132

Gross Income 1,123

5.3.2 Independent Samples T-test

Using the Independent samples T-test, allows this research to compare the means of the dependent variable and the independent variables. In the questionnaire participants answered 23 questions of which most questions existed of sub-questions. Each question is answered on a five-point Likert scale [totally disagree (1 point) - disagree (2 points) - neutral (3 points) - agree (4 points) -fully agree (5 points)]. After measuring the Cronbach coefficient alpha and measuring the normal distribution, the sample is determined based on a median-split instead of the mean. This is performed because the sample scores are not all equally distributed. The median-split allows the research to make a clear distinction between respondents who overall scored low or high in the questionnaire.

Figure 5.5: Median split example of the questions

It is expected that the median per question will be between 2 or 3 depending on the sample group (see Figure 5.5). Taking the entire research into account, with a total of 158 respondents, sample A and B respectively consist of 79 respondents. In which sample A consists of respondents that are perceived as the low-scoring group whereas sample B is observed as the opposite (see Figure 5.6).

31

This T-test compares the means between two groups (high vs low-risk group) on the independent variable (online purchase intention).

Hypotheses related to perceived risks on the purchase intention

Hypothesis 1a expects that a consumers’ perceived risk negatively affects a consumer's intention to purchase online in general. The results of the Independent T-Test (see Appendix V) show the relationship between perceived risk and purchase intention. The mean difference of the low risk perceivers group is (M= 3,55, SD= 0,94) and the high risk perceivers group is (M= 3,32, SD = 0,82). In conditions of t (156) = 1,59, p = 0,11. The results (see Table 5.5) show that the difference between the means is slim, p = 0,11 moreover, shows that this difference is not significant (p > ,05). Table 5.5 shows the means and standard deviations of the low risk perceivers and the high risk perceivers and the t-value which is related to the size of the difference between the means of the two samples that are compared. The larger t is, the larger the difference between the means (Blumberg et al., 2011).

Table 5.5: Overview Independent-Sample T-tests for H1-hypotheses

Hypothesis Mean low-risk SD low-risk Mean high-risk SD high-risk T-value H1a 3,55 ,94 3,32 ,82 t (156) = 1,60 H1b Financial risk Functional risk Information risk 2,89 2,58 2,86 ,81 ,58 ,78 3,47 3,78 3,50 ,61 ,33 ,61 t (156) = -5,16 t (156) = -15,99 t (156) = -5,69 H1c 3,62 ,92 3,26 ,81 t (156) = 2,59 H1d 3,67 ,87 3,20 ,84 t (156) = 3,44 H1e 3,36 ,88 3,51 ,89 t (156) = -1,05

Hypotheses 1b measures if a consumer's perceived financial risk, functional risk and information risk negatively affect the consumer's perceived risk in general (see Table 5.5). It is noticeable that the differences between the means of the high and the low risk group of all three variables are large. Although the relationship between financial, functional and information risk regarding perceived risk in general is (highly) significant (resp. 0,003, 0,000 and 0,009).

32

Comparing the independent variable financial risk to general perceived risk, a large mean difference is noticeable between the low risk perceivers group (M= 2,88 , SD= 0,81) and the high risk perceivers group (M= 3,47, SD = 0,61). In conditions of t (156) = -5,158, p = 0,000, which shows a highly significant difference between the means.

The results furthermore show the relationship between information risk and general perceived risk.There is a large mean difference between the low risk perceivers (M= 2,86, SD= 0,88) and the high risk perceivers (M= 3,49, SD = 0,61). In conditions of t (156) = -5,692, p = 0,000. The results show that p < 0,01, thus the relation is highly significant.

The following hypotheses expected that a consumer's perceived financial risk (H1c), functional risk (H1d) and information risk (H1e) negatively affect a consumer's online purchasing intention. The relationship between functional risk and the online purchase intention is measured using the Independent T-Test.It shows that the mean difference of the low risk perceivers (M= 3,67 , SD= 0,87) and the high risk perceivers (M= 3,20, SD = 0,84) is slim. In conditions of t (156) = 3,44, p = 0,001. The results show that the difference between the means is highly significant (p < ,01).

Comparing the independent variable financial risk and dependent purchase intention, see Table 5.5, there is a small difference noticeable between the mean of the low risk perceivers group (M= 3,61 , SD= 0,92) and high risk perceivers group (M= 3,25, SD = 0,81). In conditions of t (156) = 2,59, p = 0,01, thus the mean difference is significant (p = ,01). The results of the final Independent T-Test of the relationship between information risk and purchase intention, show a mean difference of the low risk perceivers (M= 3,36, SD= 0,88) and the high risk perceivers (M= 3,51, SD = 0,89). In conditions of t (156) = -0,150, p = 0,29. Indicating that the difference between the means are slim and p = 0,29 shows that this relationship is not significant (p < ,05).

33

Hypotheses related to different type of risks on perceived trust

Hypotheses 2A examines the relationship between the consumer’s general perceived risk and expects it negatively affects a consumer's perceived trust on the internet. The Independent T-Test (see Appendix V) shows that there is a slim mean difference between the low risk perceivers group (M= 2,88 , SD= 0,72) and the high risk perceivers group (M= 3,08, SD = 0,76). In conditions of t (156) = -1,612, p = 0,109. The results show that this relationship is not significant (p > .05). Table 5.6 provides an overview of the hypotheses related to H2 and perceived trust.

Table 5.6: Overview Independent-Sample T-tests for H2-hypotheses

Hypothesis Mean low-risk SD low-risk Mean high-risk SD high-risk T-value H2a 2,89 ,72 3,08 ,76 t (156) = -1,61 H2b 3,13 ,69 2,84 ,77 t (156) = 2,50 H2c 2,97 ,72 2,99 ,78 t (156) = -0,11 H2d 2,83 ,66 2,83 ,79 t (156) = 2,65

Next, this research examines if financial risk (H2b), functional risk (H2c) and information risk (H2d) negatively affect a consumer's perceived trust on the internet. Hypothesis 2b is measured by conducting an Independent-Samples T-test to compare the independent variable financial risk and perceived trust on the internet. The mean difference between the low risk perceivers group (M= 3,13, SD= 0,69), and the high risk perceivers group (M= 2,84, SD = 0,77) is considered moderate. In conditions of t (156) = 2,5 and p = 0,013, the mean difference is significant (p < ,05). The results of the Independent T-Test for the relationship between functional risk and perceived trust show no significance. The mean difference of the low risk perceivers (M= 2,97, SD= 0,72) and the high risk perceivers is very small (M= 2,98, SD = 0,78). In conditions of t (156) = -0,107, p = 0,91.

The results of H2d depict a moderate mean difference between information risk and perceived risk among the low risk perceivers group (M= 3,14 , SD= 0,66) and the high risk perceivers group (M= 2,83, SD = 0,79). In conditions of t (156) = 2,65, this result shows that this mean difference is highly significant (p 0,009 < ,01).

Hypotheses related to perceived trust on the online purchase intention

34

reliability positively affects the consumers online purchase intention, since p (,00) the moderate mean difference is significant. The moderate mean difference between perceived reputation and the online purchase intention (H3c) is also significant (p = ,00).

Table 5.7: Overview Independent-Sample T-tests for H3-hypotheses

Hypothesis Mean low-risk SD low-risk Mean high-risk SD high-risk T-value H3a 3,46 ,92 3,42 ,85 t (156) = 2,27 H3b 3,18 ,82 3,69 ,88 t (156) = - 3,85 H3c 3,15 ,83 3,73 ,85 t (156) = - 4,33 H3d 3,56 ,75 3,30 ,98 t (156) = 2,03 H3e 3,24 .80 3,63 ,93 t (156) = - 2,85 H3f Benevolence Integrity Competence 3,24 3,14 3,09 ,88 ,84 ,84 3,63 3,73 3,78 ,86 ,84 ,79 T (156) = - 2,85 T (156) = - 4,48 T (156) = - 5,37

The results expose a small mean difference between information risk and perceived risk (H3d) among the low risk perceivers group (M= 3,56 , SD= 0,75) and the high risk perceivers group (M= 3,30, SD = 0,98). The results show that this mean difference is highly significant (p 0,004 < ,01). In conditions of t (156) = 2,03. Moreover H3e, assumed that consumer's familiarity of a website affects the online purchase intention. Since p (,005) is highly significant, the moderate mean difference is accepted. Trusting belief (H3f), exists of three variables (i) benevolence, (ii) integrity and (iii) competence that all three have a significant moderate mean difference (respectively: ,005; ,00 and ,00).

Hypotheses related to perceived trust on perceived risk

This research discovered that the small mean difference between the low risk perceivers group and the high risk perceivers group is insignificant (p > 0,05) regarding H4a; the disposition to trust and the effect on consumers’ perceived risk. However, the mean difference of H4b (see Table 5.8), which measures if perceived reliability of a website negatively affects the consumer's perceived risk, is found to be highly significant (p = ,000). The small mean difference between the low risk perceivers group (M= 3,12 , SD= 0,79) and the high risk perceivers group (M= 3,24, SD = 0,75).

35

high risk perceivers group (M= 3,02, SD = 0,89). The mean difference is highly significant (p ,008). H4f regarding trusting belief, that consist of the mean differences of three variables benevolence, integrity and competence are all three considered as not significant (respectively: ,091; ,122 and ,08).

Table 5.8: Overview Independent-Sample T-tests for H4-hypotheses

Hypothesis Mean low-risk SD low-risk Mean high-risk SD high-risk T-value H4a 3,12 ,79 3,24 ,75 t (156) = -9,93 H4b 3,42 ,67 2,94 ,80 t (156) = 4,16 H4c 3,27 ,67 3,08 ,85 t (156) = 1,55 H4d 3,17 ,74 3,19 ,81 t (156) = -,172 H4e 3,34 ,58 3,02 ,89 t (156) = 2,70 H4f - Benevolence - Integrity - Competence 3,30 3,27 3,29 ,78 ,76 ,71 3,06 3,08 3,07 ,75 ,77 ,82 t (156) = 2,05 t (156) = 1,56 t (156) = 1,77 5.3.3 Regression analysis

The Independent sample T-test compared the means of two groups (high versus low risk group) concerning the different variables of the conceptual model. However, the T-test only measures if two samples are part of the same population by showing the average or in this case; the mean. In order to measure the strength of a relation the regression analysis is used. The hypotheses and the results showed in the previous subsection are now applied in a regression analysis, where the p (value) depicts the significance.

Table 5.9: Results regression analysis for the purchase intention (H1a)

Purchase intention; R² = 0,265

Variable B t Sig.

Perceived risk -,304 -3,434 ,001***

* Significance at 10% level ** Significance at 5% level *** Significance at 1% level

The R value represents the ‘simple’ correlation, the R (0,265) indicates a low correlation (see Table 5.9). R², in this case is 0,07 (see appendix VII), which indicates how much of the total variation is assigned to the dependent variable. In this case this is very low. Since other factors of trust, experience and different kinds of risks also influence the online purchase intention, this outcome is explainable.

Hypotheses related to perceived risks on the purchase intention