Determinants of online purchasing behavior: An empirical investigation using an extension of the Theory of Planned Behavior

Transcription

1 BUSINESS AND SOCIAL SCIENCES Department of Business Administration Master Thesis Master of Science in Marketing Determinants of online purchasing behavior: An empirical investigation using an extension of the Theory of Planned Behavior Author: Vania Daniela Vera Velarde Academic Supervisor: Athanasios Krystallis Krontalis Number of characters (with spaces): 139,890 November 2012

2 Acknowledgement First I would like express my gratitude to Athanasios Krystallis, my supervisor, for his guidance and help during the entire working process. I also want to thank my Danish family for their constant support and for their help with the sample collection. Last but not least, I would like to express my appreciation to my parents, my sister and my husband who are always there to give me the inspiration to achieve my goals. Daniela Vera

3 CONTENTS 1. Introduction Background and Research Justification Problem Statement and Research Questions Contribution Methodology and Delimitations Structure Literature Review Online Consumer Behavior Major Research Models The Theory of Reasoned Action (TRA) The Technology Acceptance Model (TAM) Theory of Planned Behavior (TPB) Decomposed Theory of Planned Behavior (DTPB) TAM vs. TPB vs. DTPB Model Extensions Trust Online Experience Product Types Online Product Classifications Online Behavior and Product Types Conceptual Model and Hypotheses The Conceptual Model Behavioral Intention and Behavior... 22

4 3.1.2 Attitude and Behavioral Intention Subjective Norm and Behavioral Intention Perceived Behavioral Control, Behavioral Intention and Behavior Decomposing Attitude Decomposing Subjective Norm Decomposing Perceived Behavioral Control Product Type and Purchase Intention Control Variables Methodology Research Approach Instrument Development Sampling and Data Collection Data Analysis and Results Data Analysis Sample Characteristics PLS Analysis Measurement Model Results Reliability Validity Structural Model Results Variance Explanation Path Analysis Effect Size Predictive Relevance Hypotheses Testing Bootstrapping One-Way ANOVA... 48

5 ANOVA Assumptions ANOVA Results Additional Analysis Trust Summary of Results of Hypotheses Testing Discussion How do perceived usefulness, perceived ease of use, compatibility and trust impact attitude toward online shopping? How do interpersonal and external influences impact the subjective norm regarding online shopping? How do self-efficacy and facilitating conditions impact perceptions of behavioral control regarding online shopping? How do attitude, subjective norm and perceived behavioral control impact online purchase intentions and consequently, online purchasing? To what extent, if any, is product type related to online purchase intention? Summary and Conclusion Summary of Findings Managerial Implications Limitations Further Research References Appendices... 70

6 List of Figures Figure 1: Thesis Structure... 6 Figure 2: The Technology Acceptance Model... 9 Figure 3: The Theory of Planned Behavior Figure 4: The Decomposed Theory of Planned Behavior Figure 5: The Conceptual Model Figure 6: The Conceptual Model Results of Path Significances List of Tables Table 1: Peterson et al. (1997) Product and Service Classification Grid Table 2: Construct Definitions Table 3: Sources of Questionnaire Items Table 4: Sample Characteristics Table 5: Products and Services Purchased Online Table 6: Products and Services Selected to Answer the Survey Table 7: Reliability Results Composite Reliability and Cronbach s Alpha Table 8: Validity Results AVE Table 9: Variance Explanation Results Table 10: Effect Size Results Table 11: Predictive Relevance Results Table 12: Tests of PLS Paths with Bootstrap Table 13: Products and Services Classification in This Study Table 14: Test of Homogeneity of Variances Product Types Table 15: ANOVA Behavioral Intention and Product Types Table 16: Behavioral Intention and Product Type Statistics Table 17: Low and High Trust Groups Table 18: Test of Homogeneity of Variances Trust Groups Table 19: ANOVA Behavioral Intention and Trust Groups Table 20: Robust Tests of Equality of Means Trust Groups Table 21: Summary of Results of Hypotheses Testing... 52

7 List of Appendices Appendix 1: Questionnaire Measures and Sources Appendix 2: Web-Based questionnaire Appendix 3: Purchase Frequencies Appendix 4: Danish Online Shoppers Appendix 5: Latent Variables Correlations Appendix 6: Latent Variables Squared Correlations and AVE Appendix 7: Cross Loadings Appendix 8: SmartPLS Output Path Analysis Appendix 9: SmartPLS Output Bootstrapping Appendix 10: P-P Plots and Q-Q Plots Appendix 11: Post Hoc Tests... 91

8 Introduction 1 1. INTRODUCTION 1.1 Background and Research Justification Internet & Retailing The number of internet users has been steadily increasing all over the world, including developing countries. In Asia and Eastern Europe internet access rates were considerably high during the last year (Euromonitor International, 2012a). In addition, more than 70 percent of the population has access to internet from home in Western Europe, North America and Australia, and by 2015 the number will be closer to 80 percent in these regions (Euromonitor International, 2012a). As the number of internet users increases, the number of consumers who shop over the internet is also growing. For instance, the number of EU consumers buying products and services online doubled to 40 percent in 2010 from 20 percent in 2005 (Muller et al., 2011). However, not only does the number of online shoppers is growing, the volume of their purchases is also increasing as online shopping is becoming more popular across the globe. This is evidenced by global online retail sales, which reached billion in 2011, a growth of 22 percent from In 2011 online retail sales in Europe were billion, an increase of 18 percent from 2010 (Euromonitor International, 2012d). Although the forecast for the sector is to decelerate with an annual growth rate of 13.7 percent, online retail sales are expected to continue growing to 670 billion by 2015 (Datamonitor, 2011). The increase in online retail sales is influenced not only by pure players but also brickand-mortar companies that have taken in the opportunities of e-commerce and are implementing multi-channel strategies. It is clear that e-commerce has created opportunities for both, small and large companies and a wide range of benefits for consumers as well. Compared to traditional brick-and-mortar stores, online stores have many advantages for consumers. An important benefit for consumers is convenience, since consumers are able to purchase a 1 Excluding sales tax

9 Introduction 2 product or service without incurring in time and transportation costs; moreover, online stores are available to consumers anytime and anywhere. Another benefit is that consumers have easy access to product and service information; and many online stores provide tools for product comparison and help in making purchasing decisions. Through online stores, consumers have access to products in foreign countries that are not available in their own countries. Lastly, online stores have reduced operation costs compared to traditional brick-and-mortar stores, cutting on labor and store rental costs which allows for lower prices offered to consumers (Euromonitor International, 2012b). Further evidence is found in a Danish survey which revealed that the three main reasons for shopping online were lower prices in online stores compared to physical stores, convenience and quick price comparisons (Euromonitor International, 2012c). Although online stores have many advantages, they also have disadvantages over brickand-mortar stores. One of the most important disadvantages is the fact that consumers cannot touch, feel, taste or smell the products; this prevents consumers from assessing product quality and increases risk perceptions. Other disadvantages are related to delivery delays, security and privacy concerns which can affect consumers trust on online stores. A web retailer s storefront is its website, meaning that when interacting with a web retailer, consumers become IT users and face new challenges such as navigating the website, information overload, unfriendly user interfaces, or complex ordering processes. These challenges reduce consumer s perceptions of control and confidence over online activities. Online retail sales show that online shopping remains popular for certain types of products, for instance according to a study sponsored by the European Parliament, EU consumers mainly buy clothes and travel related products and services online, whereas computers and electronic products are the least likely to be bought online (Muller et al., 2011). Despite the fact that the figures show increasing online sales, many online consumers use information gathered online to make purchases offline, this is evidenced by the high abandon rates of shopping carts (Kiang et al., 2011). Consumers use online stores to gain market knowledge, they learn about price levels and product differences, yet they

10 Introduction 3 don t make the final transaction with the online store (Broekhuizen and Huizingh, 2009). Online Consumer Behavior Research Given the widespread proliferation of online shopping, online consumer behavior has become an important topic among researchers, this is illustrated by the great number of publications on different fields such as information systems, marketing, management and psychology (Cheung et al., 2005). Researchers have been exploring online consumer behavior for many years and two widely accepted views stand out in e-commerce literature: consumer-oriented and technology-oriented view. The consumer-oriented view places focus on consumer s salient beliefs about online shopping, whereas the technology-oriented view studies the impact of website design and usability on consumer s behavior (Zhou et al., 2007). The findings support both views and it is apparent in the literature that both views complement each other. Prior research shows that there are numerous factors that affect online consumer behavior, nonetheless there are mixed findings in literature and many factors that influence online consumer purchasing behavior have yet to be explored, especially considering the dynamics of technology and consumer needs, which are constantly evolving, and as a result significant factors five years ago may differ today as consumers become more experienced internet users. Furthermore, most of the previous online shopping research is focused on one specific type of product such as books (Gefen et al., 2003, Lin, 2007), clothing (Ha and Stoel, 2009, Hansen and Møller Jensen, 2009, Kim and Kim, 2004, Tong, 2010, Kim et al., 2003, Yoh et al., 2003), groceries (Hansen et al., 2004), financial services (McKechnie et al., 2006, Suh and Han, 2003) and car insurance (Broekhuizen and Huizingh, 2009). Previous research has also investigated product characteristics and online behavior using a conventional product classification scheme, exploring how search, experience, and credence goods vary in their impact on purchase intentions (Brown et al., 2003, Girard et al., 2003, Korgaonkar et al., 2006, So et al., 2005, Soopramanien et al., 2007). Few studies explore different product types and online purchasing intentions using a classification scheme that takes into account the specific features of internet (Ian and

11 Introduction 4 Sui Meng, 2000, Vijayasarathy, 2002) and the findings show mixed results which call for further investigation. 1.2 Problem Statement and Research Questions From the preceding introduction, it is apparent that the fast technological progress is changing consumer shopping habits. Research on online consumer behavior is becoming more prominent in literature and prior studies have set the foundation into the factors that influence online consumers, however, it is still not clear what drives consumers to shop online. Moreover, e-commerce has become an important marketing and sales channel, complimenting traditional channels, thus, it is important for retailers to understand the determinants of online purchasing and what type of products or services are more suitable to be marketed online, as knowing these factors will enable retailers to meet consumer s needs and for marketers to target consumers effectively. The fundamental problem that motivated this study is what factors determine online purchasing behavior. In consideration of this problem, the purpose of this study was to understand online consumer behavior by empirically testing a model based on an extension of the theory of planned behavior (Ajzen, 1991). The conceptual model is based on Taylor and Todd (1995) approach with decomposed belief structures and it is built upon literature findings; the model will help to predict and examine online consumer behavior and identify key factors which determine online purchasing. The overall purpose of the thesis will be reached by attempting to answer the following research questions: RQ1. How do perceived usefulness, perceived ease of use, compatibility and trust impact attitude toward online shopping? RQ2. How do interpersonal and external influences impact the subjective norm regarding online shopping? RQ3. How do self-efficacy and facilitating conditions impact perceptions of behavioral control regarding online shopping?

12 Introduction 5 RQ4. How do attitude, subjective norm and perceived behavioral control impact online purchase intentions and consequently, online purchasing? RQ5. To what extent, if any, is product type related to online purchase intention? 1.3 Contribution The results from this study will make a positive contribution to the online consumer behavior literature by providing a deeper understanding of consumer beliefs about online shopping and how these impact attitude and intention to shop online. The knowledge from this study will provide valuable information for retailers about the relevant factors that drive consumers to shop online and the products and services that are more likely to be purchased online. The information could help retailers adapt their strategies to fit customer needs and attract and retain customers. From the marketing point of view, gaining useful insight into online consumer behavior is fundamental and the knowledge from this study could help create marketing strategies tailored to respond to online consumer specific requirements and needs. The figures and statistics show that e-commerce is full of opportunities for businesses of any size and at the same time low barriers to entry are making the market more competitive (Datamonitor, 2011), thus understanding online consumer behavior and what drives them to shop online is crucial for any business that wants to be competitive. 1.4 Methodology and Delimitations In order to fulfill the aim of the research, a survey was conducted online, where a link led to a web-based questionnaire distributed by means of snowballing technique via social networks. A sample of 138 respondents was generated. The survey results were analyzed with SPSS (IBM Corp.) and SmartPLS (Ringle et al., 2005). The respondents who completed the survey online were not restricted by country of origin. Throughout this paper internet shopping, online shopping, and online purchasing refer to the use of online stores or web retailers by consumers up until the transactional stage of purchasing and logistics, hence, it did not include the behavior of browsing for information.

13 Introduction Structure This paper is divided into seven sections: introduction, literature review, conceptual model and hypotheses, methodology, data analysis and results, discussion, summary and conclusions (Figure 1). After the current introductory part, the literature review is presented in the second section, where relevant theoretical models applied in the e-commerce literature are discussed as well as prior research findings. The third section presents the conceptual model and hypotheses. The fourth section outlines the research methodology, where instrument development and data collection procedures are discussed. The fifth section includes data analysis and results. The sixth section provides discussions concerning the results. Finally, summary of findings, managerial implications, limitations and further research are presented. 1. Introduction 2. Literature Review 3. Conceptual Model and Hypotheses 4. Methodology 5. Data Analysis and Results 6. Discussion 7. Summary and Conclusion Figure 1: Thesis Structure

14 Literature Review 7 2. LITERATURE REVIEW This section consists of three main parts. The first part reviews major conceptual research models from both the perspective of traditional consumer behavior and information systems. The second part reviews research findings and extensions to the major conceptual models in the literature. The third part reviews literature relating to product classifications and research findings regarding purchase intention in the context of different product types. 2.1 Online Consumer Behavior Traditional offline consumer behavior and the drivers that take consumers into action have been studied from different perspectives and disciplines like marketing, psychology and the economic view, thus making the research of consumer behavior quite rich and diverse. However, the development of the internet and e-commerce have made an impact on consumer s lives, the way they transact and their decision making process. Online consumers are using a computer and getting cues from a virtual environment, thus information technology has great influence on online consumer behavior and the drivers that motivate their actions, therefore creating differences between consumer online behavior and traditional offline behavior (Pavlou, 2003, Pavlou and Fygenson, 2006). Online consumer behavior has been studied from two widely accepted views: consumer-oriented and technology-oriented. The findings in previous studies support both views (Cheung et al., 2005, Lin, 2007, Monsuwé et al., 2004, Taylor and Todd, 1995), furthermore the two views compliment and reinforce each other (Zhou et al., 2007).The consumer-oriented view focuses on consumer s salient beliefs about online shopping and the influence of such beliefs on purchase channel selection (Zhou et al., 2007). For example, online consumer behavior research has been examined from the perspective of shopping orientations, shopping motivations, personal traits, internet experience, among others (Monsuwé et al., 2004, Zhou et al., 2007). On the other hand, the technology-oriented view focuses on predicting consumer acceptance of online

15 Literature Review 8 shopping by studying web site design and content as well as system usability (Zhou et al., 2007) Major Research Models In order to understand consumer s online behavior and the determinants of online purchasing, researchers have relied on the theory of reasoned action (TRA), technology acceptance model (TAM), the theory of planned behavior (TPB), expectationconfirmation theory (ECT), innovation diffusion theory (IDT) and transaction cost theory (TCT), however Cheung et al. (2005) found that in most research studies the backbone for understanding online behavior was based mainly on TAM and TPB with the other theories integrated into these research models The Theory of Reasoned Action (TRA) A very well established theory from the social-psychology discipline, proposed by Fishbein and Ajzen (1975), TRA postulates that an individual s behavior is determined by the individual s behavioral intention. In TRA behavioral intention is a function of two primary determinants: attitude towards the behavior, and subjective norm, i.e. an individual s perception of normative social pressure to perform the behavior. Attitude towards the behavior is measured by the combination of salient beliefs about the behavior and an individual s evaluation of the outcome resulting from the behavior. Additionally, subjective norm is measured by the combination of salient beliefs regarding a relevant reference group opinion about the behavior and an individual s motivation to comply with the reference group (Fishbein and Ajzen, 1975). TRA is concerned with rational, volitional behaviors i.e. behaviors over which the individual has control (Fishbein and Ajzen, 1975), nonetheless some researchers are interested in situations where the behavior is not completely under the individual s control, which has been the main reason for some of the critics to the model (Hansen et al., 2004). In the context of online consumer behavior, TRA has been used in empirical research, for example, Kim and Kim (2004) explored online clothing purchase intention yielding results with relative low predictive power, while Yoh et al. (2003) used TRA and incorporated aspects of innovation diffusion theory, thereby increasing the explanatory power of their research model.

16 Literature Review The Technology Acceptance Model (TAM) Developed by Davis (1989), TAM seeks to explain users adoption of information technology. Based on TRA, TAM adopts the belief attitude intention behavior causal relationship to explain the adoption of computer-based technologies in the workplace. TAM postulates that behavioral intention to use a new technology will lead to actual system use. Furthermore, behavioral intention to use a new technology is determined by an individual s attitude toward using the new technology. The model posits that there are two determinants that influence attitude toward using a new technology: perceived usefulness (PU) and perceived ease of use (PEOU) (Davis, 1989). PU is defined as the degree to which a person believes that using a particular system would enhance his or her job performance and PEOU is defined as the degree to which a person believes that using a particular system would be free of effort (Davis 1989, p. 320). Additionally, an improved version of TAM (Davis, 1993) suggests that PU is influenced by PEOU and not the other way around, the rationale behind it is that easyto-use technology is more useful than hard-to-use technology and useful technology may not necessarily be easy to use (Figure 2). Figure 2: The Technology Acceptance Model Source: Davis, 1993 TAM has been widely adopted in information systems (IS) research and it has been successfully applied as a theoretical framework to predict online purchasing behavior (Chen and Tan, 2004, Hernández et al., 2010, Pavlou, 2003, Vijayasarathy, 2004), moreover, researchers have applied TAM to predict online purchasing behavior in the context of books (Gefen et al., 2003, Lin, 2007), clothing (Ha and Stoel, 2009, Tong,

17 Literature Review ) and financial services (Suh and Han, 2003). Although TAM has proven to have valid constructs, the explanatory power of TAM has been enhanced by the aggregation of other constructs into the research model, as the model has been seen as too parsimonious by researchers (Venkatesh and Davis, 2000). One of the extensions of TAM was proposed by Venkatesh and Davis (2000), referred as TAM2, the model includes subjective norm as it was found to have significant influence on PU and behavioral intention Theory of Planned Behavior (TPB) Ajzen s TPB is an extension of TRA (Fishbein and Ajzen, 1975). TPB takes into account conditions where individuals do not have complete control over their behavior. In addition to an individual s attitude towards the behavior and the subjective norm proposed in TRA, TPB integrates perceived behavioral control (PBC) into the model. PBC is defined as an individual s perception of how easy or difficult would be to carry out a behavior (Ajzen, 1991). TPB postulates that the actual behavior is determined by both, the behavioral intention and PBC. Behavioral intention, in turn, is predicted by subjective norm, attitude toward the behavior and PBC ( Figure 3). Figure 3: The Theory of Planned Behavior Source: Ajzen (1991)

18 Literature Review 11 There are a number of studies that focus on understanding and predicting an individual s intention to engage in a particular behavior in a variety of application areas based on this popular and widely accepted model. Furthermore, empirical research has shown the appropriateness of this model to understand consumer s behavior in the context of online shopping (George, 2004, Hansen et al., 2004). Hansen et al. (2004) tested both TRA and TPB and found that TPB provided better explanation to online consumer behavior than TRA did. However, like TAM, many researchers have added constructs to the model to better reflect the characteristics of consumer online behavior (Cheung et al., 2005). Pavlou and Fygenson (2006) tested one of the most comprehensive TPB model extensions in a longitudinal study, which explored two behaviors: getting information and actual product purchasing. The study findings confirmed the significance of technology adoption variables (perceived usefulness and perceived ease of use) for the prediction of e-commerce adoption; additional significant constructs were trust, consumer skills, time and monetary resources, and product characteristics (product diagnosticity and product value) Decomposed Theory of Planned Behavior (DTPB) Taylor and Todd (1995) introduced the idea that TPB beliefs can be decomposed into multidimensional constructs. They argued that the aggregation of beliefs to create measures of attitude, subjective norm and PBC, proposed by Ajzen and Fishbein, does not identify specific factors that might predict a particular behavior. Moreover, Taylor and Todd argue that the decomposed TPB model has advantages similar to TAM in that it identifies specific salient beliefs that may influence IT usage (Taylor and Todd, 1995, p.147). According to Taylor and Todd (1995), in the decomposed TPB (DTPB) attitudinal, normative and control beliefs are decomposed into multidimensional beliefs constructs (Figure 4). The decomposition of attitude beliefs has three characteristics of innovation that influence behavioral intentions; these are based on the diffusion of innovation theory proposed by Rogers (1995): relative advantage, complexity and compatibility. Relative advantage can be defined as the degree to which an innovation provides benefits which supersede those of its precursor and may incorporate factors such as economic benefits, image enhancement, convenience and satisfaction (Rogers, 1995). Considering that PU in TAM is the degree to which a person believes that using a

19 Literature Review 12 particular system would enhance his or her job performance (Davis 1989, p. 320), Taylor and Todd (1995) suggest that PU, as defined in TAM, is equivalent to Roger s relative advantage, since both constructs refer to a relative improvement in performance and their measures have been operationalized in terms of their relative impact on performance. According to Rogers (1995), complexity represents the degree to which an innovation is perceived to be difficult to understand, learn or operate. Taylor and Todd (1995) suggest that PEOU (the degree to which a person believes that using a particular system would be free of effort) is analogous to Rogers complexity construct, although in an opposite way. Compatibility refers to the degree to which the innovation fits with the potential adopter s existing values, previous experiences and current needs (Rogers, 1995). Previous studies have suggested the decomposition of subjective norm into two dimensions: interpersonal influence and external influence (Bhattacherjee, 2000, Hsu and Chiu, 2004, Lin, 2007). Interpersonal influence refers to word-of-mouth influence by friends, family, colleagues, while external influence is related to mass media reports, experts opinions and other non-personal information. Ajzen (1991) decomposed the PBC component into two dimensions: self-efficacy and facilitating conditions. The dimension of self-efficacy is defined as an individual s perception of his or her individual capabilities; in the context of online shopping it refers to consumer s self-assessment of his or her capabilities to shop online. The second dimension, facilitating conditions, is concerned with external resource constrains that may influence on engaging a particular behavior, such as time, money and technology; in the context of online shopping the issue of technology constrains is related to the availability of supporting internet equipment (Ajzen, 1991, Ajzen, 2002, Lin, 2007). DTPB has been successfully applied as research model in online shopping to predict purchasing behavior, repurchase intention and as a model to understand the relation of two behaviors such as getting information and actual online purchasing (Chen, 2009, Hsu and Chiu, 2004, Lin, 2007, Pavlou and Fygenson, 2006).

20 Literature Review 13 Figure 4: The Decomposed Theory of Planned Behavior Source:Lin (2007), Taylor and Todd (1995) TAM vs. TPB vs. DTPB It is important to understand the contribution and differences of each model to the understanding of online consumer behavior. Although TAM and TPB derive from TRA, there are differences between these models approach to understanding behavior. Empirical tests comparing TAM, TPB and DTPB, showed that the three models achieve similar fit to the data and that all three models are comparable in terms of their ability to explain overall behavior, although subjective norm and PBC added slightly to the prediction of behavior. However, the results also showed that DTPB had better explanatory power over TPB and TAM, when behavioral intention is considered (Lin, 2007, Taylor and Todd, 1995).

21 Literature Review 14 It can be said that TAM is more parsimonious than TPB, thus it can be useful in research focused on achieving an overall understanding of behavior (Lin, 2007). On the other hand, DTPB sacrifices parsimony but provides better insight into the determinants behavioral intention and actual behavior (Lin, 2007). Therefore, it is reasonable to conclude that DTPB is appropriate for the purpose of the present study. 2.2 Model Extensions Online consumer behavior literature demonstrates that TAM and TPB are valid, reliable models for understanding online behavior. Nonetheless, researchers concur with the idea that there are other relevant factors, besides the ones in the models, that help in understanding online consumer behavior, thus many studies have extended TAM and TPB by identifying major antecedents or mediator factors that have improved the understanding of the determinants of online purchase intentions. Furthermore, there are a number of research models that have taken elements of TAM and/or TPB and integrated them with other theories like expectation-confirmation theory or innovation diffusion theory. For example, Wen et al. (2011) extended TAM with constructs drawn from expectation-confirmation theory and explored consumers intention to continue shopping online. Since online consumer behavior has been studied from either a consumer-oriented or a technology-oriented perspective, the types of determinant factors in online purchase intention identified in previous studies are diverse: demographic, personality traits, online service quality, website quality, brand effect, shopping orientation, shopping motivation, trust and perceived risk, internet experience, prior online shopping experience and product types (Cheung et al., 2005, Monsuwé et al., 2004, Zhou et al., 2007). Considering that the purpose of this study is to gain insight into the factors that determine online consumer behavior and considering the parsimony principle, it is not possible to include all possible factors in one research model. Furthermore, the literature review suggests that there is one particular element that has great influence on online purchase intentions: trust. In addition, another factor that seems to stand out in the literature is consumer s online experience, and finally, given that prior research shows that product characteristics have an impact on online consumer behavior, and that few

22 Literature Review 15 studies have explored this factor, the relation between product types and online behavior is also discussed in detailed Trust Trust can be defined as the expectation that other individuals or companies with whom one interacts will not take advantage of a dependence upon them. It is the belief that the trusted party will behave in an ethical, dependable, and socially appropriate manner and will fulfill their expected commitments. (Gefen et al., 2003, p. 308). According to McKnight et al. (2002), trust is defined as the belief that allows consumers to willingly become vulnerable to web retailers after having taken the retailer s characteristics into consideration. These definitions are consistent with the three trusting beliefs that are used most often in literature (Bhattacherjee, 2000, McKnight et al., 2002, Pavlou, 2003, Pavlou and Fygenson, 2006): competence, integrity and benevolence. Competence is the belief in the trustee s ability to perform as expected by the trustor. Integrity is the belief that the trustee will be honest and keep its promises. Benevolence is the belief that the trustee will not act opportunistically. Trust is a central element in exchange relationships that are characterized by uncertainty and vulnerability (McKnight et al., 2002). Prior research confirms that trust plays a relevant role in consumer behavior, both online and offline. (Chen, 2009). However, the importance of trust increases in the online context because perceptions of uncertainty may be especially significant on an e-commerce environment, where certain cues that evoke trust cannot be fully assessed (e.g. product characteristics, physical store, sales person) (Jarvenpaa et al., 2000, Pavlou, 2003, Verhagen et al., 2006). Furthermore, lack of trust has been credited as one of the main reasons preventing consumers from engaging in e-commerce (Jarvenpaa et al., 2000, Monsuwé et al., 2004). Several researchers have demonstrated that trust positively influences attitudes towards online purchasing (Chen and Tan, 2004, Delafrooz et al., 2011, Ha and Stoel, 2009, Kim, 2012, Ling et al., 2010, Suh and Han, 2003, Zarmpou et al., 2012). For instance, Jarvenpaa et al. (2000) empirically showed that trust has a significant effect on consumer s attitudes toward online purchasing in multiple cultures. Suh and Han (2003) integrated trust with TAM constructs and found comparable results, namely, trust is a significant factor influencing attitude toward online shopping Pavlou and Fygenson

23 Literature Review 16 (2006), as mentioned previously, conducted a longitudinal study using TPB, their findings confirmed that trust influences online purchase intentions through attitude. Some researchers have explored the direct impact of trust on purchase intentions. For example Gefen et al. (2003), integrated trust with TAM and found that trust is as influential as the two core TAM variables (perceived usefulness and perceived ease of use), the results also support the importance of trust in purchase intentions among potential customers and repeat customers, although repeat customer trust web retailers more than potential customers. Pavlou (2003), integrated trust with a TRA-based model and found that trust was the most influential predictor of intention and Wen et al. (2011) found similar results regarding trust in the prediction of online repurchase intention. In literature, trust has been studied from three major perspectives: online vendor characteristics (merchandise value, order fulfillment, etc.), web site characteristics (navigation, privacy, security, etc.), and customer characteristics (personality, online expertise, etc.) (Benedicktus et al., 2010, Chiu et al., 2009). From the three perspectives, research supports online vendor characteristics as the driver with the biggest impact on trustworthiness (Benedicktus et al., 2010). Trust and Risk perceptions Risk perception refers to a consumer s perceptions of the uncertainty and adverse consequences of engaging in an activity (Hsu and Chiu, 2004). Uncertainties related to online transactions create different risks, Pavlou (2003) distinguished economic risk (monetary loses), seller performance risk (transaction fulfillment), privacy risk (illegal disclosure of personal information) and security risk (theft of credit card information). Bhatnagar et al. (2000) investigated how risk affects online shopping, the study differentiated two types of risks: product category risk, which is associated with the product itself, and financial risk, which is associated with security concerning credit card information over the internet, the results showed that as consumers become more knowledgeable, their perceptions of product and financial risks decrease. Barkhi et al. (2008) considered only one aspect of risk related to security (information integrity) and combined elements of TAM and TPB to explain purchase decisions from a online store, the empirical findings showed that PU, PBC and subjective norm impact

24 Literature Review 17 attitude toward purchasing from an online store, while security did not have a significant effect. Vijayasarathy (2004) extended TAM with subjective norm, privacy risk and security risk in order to predict consumer s online purchase intentions, the empirical test proved all constructs significant but privacy concerns. The concept of risk has been widely studied in literature from different perspectives, but in general terms it can be said that trust and risk are interwoven (Jarvenpaa et al., 2000), as trust is needed in uncertain situations and this means assuming risks and becoming vulnerable to trusted parties. Thus, trust plays a central role in helping consumers overcome perceptions of risk; this applies to the e-commerce context, when a web retailer can be trusted to show competence, benevolence, and integrity, there is much less risk involved in engaging and interacting with the web retailer since trust makes consumers comfortable sharing personal information and making purchases. According to Pavlou (2003), trust is one of the most effective tools for reducing uncertainty and risks. Several studies incorporated both perceived risk and trust in their research model while exploring online consumer behavior, their findings support the significance of both constructs, however, trust shows stronger significance in influencing attitude and purchase intention (Chen, 2009, Jarvenpaa et al., 2000, Pavlou, 2003, Verhagen et al., 2006). Furthermore, if a consumer decides to trust on a web retailer, the decision to transact inseparably entails an interaction with its website interface, hence, it is fair to say that trust in a web retailer implicitly encompasses trust in the integrity of the transaction medium (i.e. web retailer s infrastructure) (Pavlou, 2003). Web retailers can affect trust in their infrastructure by facilitating encrypted transactions, installing firewalls, using authentication mechanisms, and establishing privacy seals and disclosures. By implementing these privacy and security mechanisms it can be argued that both, seller performance risks, privacy and security risks can be decreased. Based on the findings regarding trust and perceived risks and the fact that both constructs are interwoven, the present study will focus on trust as a major determinant of online consumer behavior, assuming that seller performance risk and privacy and security risks are implicitly considered when a consumer decides to trust on a web retailer s competence, benevolence and integrity.

25 Literature Review Online Experience Online experience is an important element found in e-commerce literature, studies have explored the role of consumer s prior online shopping experience and its relationship with purchase intentions. Research shows that prior online shopping experience strongly influences purchase intentions (Broekhuizen and Huizingh, 2009, Brown et al., 2003, Gefen et al., 2003, Hernández et al., 2010, Jayawardhena et al., 2007, Ling et al., 2010, So et al., 2005).The findings suggest that previous experiences purchasing online assist in reducing consumer s uncertainties. Potential and Repeat Customers Based on their online shopping experience consumers can be differentiated. Gefen et al. (2003) explored the importance of trust using TAM, their research showed that there are two distinct populations: potential customers and repeat customers. Potential customer s initial trust on a web retailer is greatly influenced by their disposition to trust because there is not much else on which to base this trust. For repeat customers, after interactions with the web retailer, trust is influenced by the nature of the previous interactions, in other words, their trust is influenced by prior experience; TAM constructs (PEOU and PU) were more significant among repeat customers. Hernández et al. (2010) explored the moderating effects of experience using TAM as a research model, they argued that the perceptions that induce individuals to purchase online for the first time are not the same as those that induce repeat purchasing behavior, their findings demonstrated that consumer s perceptions evolve as they acquire online shopping experience; the findings suggest that PEOU has a weaker effect on experienced online shoppers while PU is stronger among this group. Online purchasing behavior is influenced by different factors and these are perceived differently between those consumers with prior online shopping experience, for instance Broekhuizen and Huizingh (2009) found decreased perceptions of risks and more concerns about time and effort saving among consumers with online purchasing experience.

26 Literature Review 19 An empirical study by Jayawardhena et al. (2007) focusing on the relationships between shopping orientations, purchase intention and prior experience found that although shopping orientation had no significant effect, prior experience had a significant effect on consumer s propensity to shop online. The role of experience has been proven significant from different perspectives, either directly on purchase intention or its moderating effect on factors preceding purchase intention; it is therefore evident from the literature review that potential customers and repeat customers are differentiated, since the factors that influence their behavior are not perceived equally among these two populations due to their online shopping experience Product Types Online Product Classifications Online shopping for a product or service incorporates different considerations, as consumers may engage in different purchasing decision processes in different product categories (Lowengart and Tractinsky, 2001). In an attempt to better understand consumers, different approaches of classification for products have been proposed. Nelson (1974) first classified products into search and experience goods; this distinction is based on how product quality can be determined. For search products, quality can be determined prior to purchase and use, these products physical characteristics are known prior to purchase. For experience goods, there is some uncertainty with respect to their quality or likelihood of physical malfunctioning, thus they need to be personally tried and examined in order to assess their quality. Since many products are not purely search or experience, Nelson later generalized the classification and defined search goods as those whose full information for dominant product attributes can be known prior to purchase; while experience goods are those that are dominated by attributes that cannot be known prior purchase or when acquiring information is more costly and/or difficult than direct product experience (Kiang et al., 2011). Norton and Norton (1988) extended Nelson s classification by dividing experience goods into durable and non-durable and by adding credence goods to represent goods whose quality is hard to assess even after consumption. One weakness of Nelson s classification scheme is that it does not take into account product characteristics related to the online context, as nowadays it is possible for

27 Literature Review 20 consumers to download music and movie clips as well as demo software and games for evaluation before purchase (Kiang et al., 2011). Based on the special characteristics of the internet Peterson et al. (1997) proposed a specially designed classification system for online products and services. The system includes a three-dimension classification scheme that distinguishes online and offline channel impacts: cost and frequency of purchase, value proposition, and degree of differentiation. The cost and frequency dimension ranges from inexpensive and frequently purchased to expensive and infrequently purchased goods. According to Peterson et al. (1997), individuals tend to avoid purchasing inexpensive and frequently purchased goods online. The second dimension, value proposition, distinguishes between tangible and intangible products. The third dimension, degree of differentiation is related to the degree of product customization that creates competitive advantage. The three dimensions are illustrated in Table 1. Dimension 1 Dimension 2 Dimension 3 Low outlay, frequently purchased goods High outlay, infrequently purchased goods Value proposition tangible or physical Value proposition intangible or informational Value proposition tangible or physical Value proposition intangible or informational Table 1: Peterson et al. (1997) Product and Service Classification Grid Source: Peterson et al. (1997) Differentiation potential high Differentiation potential low Differentiation potential high Differentiation potential low Differentiation potential high Differentiation potential low Differentiation potential high Differentiation potential low There are several different product classifications and contributions in the context of online shopping, for example De Figueiredo (2000), based on information asymmetry between sellers and buyers in e-commerce, classified products into a spectrum of four categories that include: commodity (its quality is easily determined by its description), quasi-commodity, look and feel, and look and feel with variable quality. De Figueiredo (2000) found that the biggest increase in e-commerce occurred in product categories which are near the commodity product side of the spectrum. Perceived risk is an important element in online shopping, as mentioned previously in this paper, and the effect of perceived risk may be subject to product characteristics (Zhou et al., 2007). Bhatnagar et al. (2000) found that product risk is higher for

28 Literature Review 21 technologically complex products like electronics and ego-related products like sunglasses; furthermore, the study showed that risk perceptions increase with higher expenditure levels and the effect of financial risk varies across product categories. More recently, Kiang et al. (2011) integrated different approaches and proposed an online product classification that distinguishes degree of standardization and brand recognition, cost and purchase frequency, and digitizability (digital and physical) Online Behavior and Product Types Using Nelson (1974) classification scheme, several studies explored the influence of product type on online behavior and found that consumers prefer to use the internet to buy search products rather than experience products (Brown et al., 2003, Girard et al., 2003, Korgaonkar et al., 2006, Lee and Tan, 2003, So et al., 2005). Researchers have also studied online behavior and product types based on Peterson (1997) classification. For instance, Vijayasarathy (2002) suggests that tangibility of the product has a significant effect on intention to shop online but cost does not have an effect. Studies concerning attitudes toward online shopping show that cost, purchase frequency and product tangibility have an important influence on consumer s attitude towards online shopping (Keisidou et al., 2011, Lian and Lin, 2008). Furthermore, Ian and Sui Meng (2000) found that product type influences consumer choice between physical and virtual stores, the results suggest that products and services that have a low outlay, are frequently purchased, have intangible value proposition, and relatively high on differentiation are more likely to be purchased online. Grounded on perceived risk in product purchase, Soopramanien et al. (2007) explored shopping channel preference, the results showed that perceived product-specific risks of purchasing products online do not reduce the intention to shop online for consumers who have previously experienced online shopping. Lowengart and Tractinsky (2001) found the existence of differences in terms of the risk dimensions considered by consumers when buying high vs. low risk goods; the study found that when purchasing experience goods online, aspects of uncertainty and risk were more salient than when purchasing search goods.

29 Conceptual Model and Hypotheses CONCEPTUAL MODEL AND HYPOTHESES In this section, a conceptual model based on TPB is developed. The conceptual model draws upon the idea credited to Taylor and Todd (1995) that TPB beliefs can be decomposed into multidimensional constructs, additionally trust beliefs are integrated into the model based on the literature review and empirical findings that support trust as a major determinant of consumer online behavior; finally product type and its influence on behavioral intention is incorporated in the analysis. 3.1 The Conceptual Model TPB (Ajzen, 1991) postulates that behavior is determined by both, behavioral intention and perceived behavioral control (PBC). Behavioral intention in turn, is determined by attitude, subjective norm and PBC. The conceptual model extends TPB based on Taylor and Todd (1995) multidimensional beliefs approach, where attitudinal, normative and control beliefs are decomposed in order to provide a more specific understanding of online consumer purchasing behavior. This decomposition approach provides several advantages. First, it is noted by Taylor and Todd (1995) that it is unlikely that antecedents of intention will be consistently related to monolithic belief structures representing a variety of dimensions. Second, by decomposing beliefs, the relationships between dimensions and intention will be clearer to understand. Third, decomposing beliefs helps to overcome some of the disadvantages in operationalization. Finally, gaining insight into specific beliefs makes the model more managerially relevant Behavioral Intention and Behavior According to Ajzen (1991), behavioral intentions are motivational factors that capture how much effort a person is willing to make in order to perform a behavior. TPB suggests that behavioral intention is the most influential predictor of behavior. The following hypothesis is therefore proposed: H1: A consumer s behavioral intention to purchase online positively affects his/her actual online purchase (BI B).

30 Conceptual Model and Hypotheses Attitude and Behavioral Intention According to Ajzen (1987) there are two different kinds of attitudes: attitudes toward objects and attitudes toward behaviors, based on this distinction, the present study considers attitude toward a behavior; that is, online purchasing. An individual s attitude toward online purchasing is defined as the individual s favorable or unfavorable evaluation of using the internet to purchase products or services from a web retailer. Attitude influences behavioral intentions (Fishbein and Ajzen, 1975). This relationship has received considerable empirical support. Thus, the following hypothesis is proposed: H2: A consumer s attitude toward online purchasing positively affects his/her behavioral intention to purchase online (A BI) Subjective Norm and Behavioral Intention Subjective norm can be described as an individual s perceived social pressure to perform or not to perform the behavior (Ajzen, 1991). Prior studies suggest that there is a positive relationship between subjective norm and behavioral intention (Barkhi et al., 2008, Bhattacherjee, 2000, Hansen et al., 2004, Vijayasarathy, 2004, Yoh et al., 2003). Applied to e-commerce, subjective norm reflects a consumer s perceptions of whether online purchasing is accepted and encouraged by important referent others; these perceptions influence online purchase intention. Therefore, the following hypothesis is presented: H3: A consumer s subjective norm in relation to online purchasing positively affects his/her behavioral intention to purchase online (SN BI) Perceived Behavioral Control, Behavioral Intention and Behavior Perceived behavioral control (PBC) is defined as an individual s perception of how easy or difficult it would be to carry out the behavior. Furthermore, PBC denotes a degree of control over the performance of the behavior and not the likelihood of a behavioral outcome (Ajzen, 2002). PBC plays a dual role in TPB, as it is a direct determinant of behavioral intention along with attitude and subjective norm, and it is also a determinant of actual behavior along

31 Conceptual Model and Hypotheses 24 with intention. There is support for the dual role of PBC by Taylor and Todd (1995), Lin (2007) and Pavlou and Fygenson (2006), who argued that neglecting PBC could lead to an incomplete study of consumer online behavior. Hence, the following hypotheses are proposed: H4: A consumer s PBC over online purchasing positively influences his/her behavioral intention to purchase online (PBC BI). H5: A consumer s PBC over online purchasing positively influences his/her actual online purchase (PBC B) Decomposing Attitude There is good evidence that TAM beliefs (PU and PEOU) significantly influence online purchasing behavior (Gefen et al., 2003, Pavlou, 2003). Drawing upon innovation diffusion theory proposed by Rogers (1995), three characteristics of innovation have been found to influence IT adoption: relative advantage, complexity and compatibility (Taylor and Todd, 1995). Rogers (1995) defined relative advantage as the degree to which an innovation provides superior benefits to those of its precursor. The relative advantage construct, as defined by Rogers, is considered to be analogous to perceived usefulness in TAM. According to Rogers (1995), complexity represents the degree to which an innovation is perceived to be difficult to understand, learn or operate. Complexity is considered to be the opposite of perceived ease of use in TAM (Chen and Tan, 2004, Lin, 2007, Taylor and Todd, 1995). Compatibility is defined as the degree to which the innovation fits with the potential adopter s values and needs (Rogers, 1995). Taking into account that online shopping can be considered a service innovation, and the empirical findings by Taylor and Todd (1995) in their DTPB model, three attitudinal belief dimensions are proposed: perceived usefulness, perceived ease of use and compatibility. Based on the literature review, trust is an important determinant of online consumer behavior. In addition to perceived usefulness, perceived ease of use and compatibility, trust is proposed as an attitudinal belief in the conceptual model.

32 Conceptual Model and Hypotheses 25 Perceived Usefulness PU is defined as the extent to which an individual believes that using a system will enhance his or her performance (Davis, 1989). In the e-commerce context, PU refers to the extent to which a consumer believes that online purchasing would enhance his or her effectiveness in the purchase of product or services. There is strong evidence that PU influences behavioral intention through attitude (Barkhi et al., 2008, Chen and Tan, 2004, Ha and Stoel, 2009, Hernández et al., 2010, Pavlou and Fygenson, 2006, Suh and Han, 2003, Taylor and Todd, 1995, Vijayasarathy, 2004). Therefore, the following hypothesis is proposed: H6: A consumer s perceived usefulness of online purchasing positively affects his/her attitude toward online purchasing (PU A). Perceived Ease of Use PEOU is the extent to which an individual believes that using a system will be effortless (Davis, 1989). In terms of online shopping, PEOU can be defined as the extent to which a consumer believes that online purchasing would be free of effort. Similar to PU, the role of PEOU has been found to be significant in affecting behavioral intention through attitude (Barkhi et al., 2008, Chen and Tan, 2004, Ha and Stoel, 2009, Hernández et al., 2010, Pavlou and Fygenson, 2006, Suh and Han, 2003, Taylor and Todd, 1995, Vijayasarathy, 2004). Consequently, the following hypothesis is proposed: H7: A consumer s perceived ease of use of online purchasing positively affects his/her attitude toward online purchasing (PEOU A). Compatibility In the context of e-commerce, the compatibility construct is evaluated by assessing the compatibility between a consumer s needs and lifestyle with online shopping. Prior research supports that compatibility with online purchasing influences consumer s attitude toward online shopping (Chen and Tan, 2004, Lin, 2007, Taylor and Todd, 1995). Hence, the following is proposed:

33 Conceptual Model and Hypotheses 26 H8: Compatibility between online purchasing and a consumer s lifestyle and needs positively affects his/her attitude toward online purchasing (COM A). Trust Trust can be defined as the belief that the trustee will act cooperatively to fulfill the trustor s expectations without exploiting its vulnerabilities (Pavlou and Fygenson, 2006). According to McKnight et al. (2002) trust is the belief that allows consumers to willingly become vulnerable to web retailers after having taken the retailer s characteristics into consideration. These definitions imply that trust in both the web retailer and online technologies underlie consumers' beliefs about the safety of shopping online. The literature on e-commerce has focused on 3 dimensions of trust: competence (web retailer s ability to do what the consumer needs), benevolence (web retailer s motivation to act in the consumer's interests), and integrity (web retailer s honesty and promise keeping). In order to place trust in a TPB-based model, trust must be defined with respect to a behavior through a specified target, action, context and time frame (Ajzen, 2002). In this case, the target of trust is the web retailer, the action is purchasing, the context is the online environment, and the time frame is the window of time during which, consumers are making their decisions. In light of the literature review, consumer s trust on a web retailer is a key factor that influences consumer s attitude toward online purchasing (Chen and Tan, 2004, George, 2004, Ha and Stoel, 2009, Jarvenpaa et al., 2000, Pavlou and Fygenson, 2006, Suh and Han, 2003). According to Pavlou and Fygenson (2006), trust enables favorable expectations that a web retailer will fulfill its promises and no harmful outcomes will occur if a consumer engages in the behavior, thus creating positive attitudes. Therefore, the following hypothesis is proposed: H9: A consumer s trust in a web retailer positively influences his/her attitude toward online purchasing (TR A) Decomposing Subjective Norm Although the literature suggests a strong relationship between subjective norms and online purchase intention, George (2004) and Pavlou and Fygenson (2006) did not find

34 Conceptual Model and Hypotheses 27 such relationship significant. Bhattacherjee (2000) suggested two forms of influence as part of subjective norm: interpersonal influence and external influence. Interpersonal influence refers to the influence of friends, family members, colleagues or superiors known to the individual, while external influence refers to mass media reports, expert opinions and other non personal information considered by the individual in performing a behavior. A possible reason for George (2004) or Pavlou and Fygenson (2006) not finding a link between subjective norm and online purchase intention may be related to not including external influences in their studies. Therefore subjective norm is decomposed into two components: interpersonal influence and external influence. Interpersonal Influence Within the online context, a consumer s relevant referent groups include family and friends, as the online purchasing behavior is not engaged in an organizational setting, colleagues and superiors are not considered relevant in the present study. Interpersonal influence can be described as a consumer s belief that online purchasing is accepted, encouraged and promoted by his/her social circle of influence. If social expectations support online purchasing it is more likely that a consumer will shop online. Accordingly, the following hypothesis is presented: H10: A consumer s perception of interpersonal influence is positively associated with his/her subjective norm about online purchasing (II SN). External Influence External influences refer to any relevant factors which are not related in a personal way to an individual. In the online setting, a consumer may consider mass media reports and press reports where the online shopping is encouraged and promoted, thus the following hypothesis is proposed: H11: A consumer s perception of external influence is positively associated with his/her subjective norm about online purchasing (EI SN) Decomposing Perceived Behavioral Control According to Ajzen (2002) PBC is composed by beliefs associated to the resources and opportunities needed to perform a behavior, this notion is represented by two components: self-efficacy and facilitating conditions. The first component is related to

35 Conceptual Model and Hypotheses 28 the internal notion of perceived ability, while the second component is related to external constrains. Prior research has found significant links between these constructs and PBC, either using the construct controllability or facilitating conditions as similar concepts (Chen, 2009, Hsu and Chiu, 2004, Lin, 2007, Pavlou and Fygenson, 2006, Taylor and Todd, 1995). Self-Efficacy Self-efficacy is defined as an individual s self-confidence in his/her ability to perform a behavior. In the context of the present study, self-efficacy is defined as a consumer s self-assessment of his/her capabilities to shop online as proposed by Vijayasarathy (2004). It is expected that higher levels of self-efficacy will cause higher levels of PBC, thus the following hypothesis is presented: H12: A consumer s positive self-efficacy positively influences his/her perceived behavioral control over online purchasing (SE PBC). Facilitating Conditions Facilitating conditions reflect the availability of resources needed to engage in a behavior. Online purchasing requires resources such as time and money, thus more resources available to the consumer lead to higher PBC. Hence, the following hypothesis is proposed: H13: A consumer s positive facilitating conditions positively influence his/her perceived behavioral control over online purchasing (FC PBC) Product Type and Purchase Intention Several studies on online behavior focus on one product, for example books (Gefen et al., 2003, Lin, 2007), clothing (Ha and Stoel, 2009, Hansen and Møller Jensen, 2009, Kim and Kim, 2004, Tong, 2010, Kim et al., 2003, Yoh et al., 2003), groceries (Hansen et al., 2004), financial services (McKechnie et al., 2006, Suh and Han, 2003) and car insurance (Broekhuizen and Huizingh, 2009). However, e-commerce literature shows that different product types play an important role in online consumer behavior, whether on consumer shopping preferences or attitudes toward online purchasing (Cheema and

36 Conceptual Model and Hypotheses 29 Papatla, 2010, Girard et al., 2003, Ian and Sui Meng, 2000). Some researchers focusing on online purchase determinants also explored the influence of product types. Brown et al. (2003) and So et al. (2005) found significant results between search and experience products and online purchase intentions; Cha (2011) compared the factors that facilitate or hinder purchase intention of real and virtual items and found significant results; and Vijayasarathy (2002) whose study was based on Peterson et al. (1997) product classification, suggests that tangibility of the product has a significant effect on intention to shop online. This study employs two of the dimensions of the product classification proposed by Peterson et al. (1997): cost and value proposition. The third dimension, namely degree of differentiation is not included due to practical reasons. The first dimension can be distinguished between low cost products (e.g. CDs/DVDs) and high cost products (e.g. refrigerator). Value proposition is an indication of the tangibility of the product and can also be differentiated at two levels, tangible (e.g. clothing) and intangible (e.g. software). The combination of the two dimensions yields the following four product types: (1) Low cost, tangible (2) Low cost, intangible (3) High cost, tangible (4) High cost, intangible Based on the aforementioned literature review, the following hypothesis is presented: H14: A consumer s behavioral intention to purchase online differs by product type (Product type BI). 3.2 Control Variables In order to be able to assess the impact of the proposed hypotheses, the following variables are controlled for in the present study: internet experience and shopping enjoyment. These variables were selected because of their potential impact on online purchasing as suggested by the extant literature.

37 Conceptual Model and Hypotheses 30 Internet experience Past research found a relationship between consumer s internet experience and online consumer purchasing behavior (Bhatnagar et al., 2000, Yoh et al., 2003); although it is important to note that some of the recent studies did not find the relationship significant (Zhou et al., 2007). The effect of internet experience among consumers online behavior tends to decrease as the penetration of the internet into their daily life increases. This is confirmed by a recent study by Hernández et al. (2010) that found that the effect of internet experience decreases once individuals acquire more online shopping experience. A consumer s experience involving the use internet may have generated knowledge that might reinforce the consumer s willingness to shop online. Hence this study controls for the role of consumer s internet experience in terms the time they dedicate to its use, frequency and level of importance they attribute to it. Shopping enjoyment Researchers have investigated the role of shopping orientation on online consumer purchasing behavior, with mixed results. Hansen and Møller Jensen (2009) and Seock and Bailey (2008) found a significant relationship between shopping orientations and online purchase behavior. Research by Girard et al. (2003) and Brown et al. (2003) suggest that multiple orientations exist among the population of internet users who have previously made purchases via the Internet, however, the studies did not find a significant relationship with purchase intention. One common orientation identified by these studies is shopping enjoyment, which refers to consumers whose motivation is the pleasure of shopping itself, for these people, shopping is a recreational pursuit. Hence, shopping enjoyment in general is included as control variable. Figure 5 illustrates the conceptual model and presents the hypotheses proposed in this study. Table 2 summarizes the conceptual model s constructs and their definitions.

38 Conceptual Model and Hypotheses 31 Figure 5: The Conceptual Model

39 Conceptual Model and Hypotheses 32 Construct Definition Reference Perceived Usefulness Perceived Ease of Use Compatibility Trust Interpersonal Influence External Influence Self-efficacy Facilitating Conditions Behavioral Intention Attitude The degree to which a person believes that using a particular system would enhance his or her job performance The degree to which a person believes that using a particular system would be free of effort The degree to which an innovation is perceived as being consistent with existing values, needs, and past experience of potential adopters The belief that the trustee will act cooperatively to fulfill the trustor s expectations without exploiting its vulnerabilities Influence by friends, family members, colleagues, superiors, and experienced individuals known to the potential adopter Mass media reports, expert opinions, and other non-personal information considered by individuals in performing a behavior An individual's self-confidence in his/her ability to perform a behavior Reflects the availability of resources needed to engage in a behavior, such as time, money or other specialized resources Motivational factors that capture how much effort a person is willing to make in order to perform a behavior An individual s evaluation of the outcome resulting from performing a behavior An individual s perception of normative social pressure to perform a behavior Subjective Norm Perceived Behavioral Control Table 2: Construct Definitions An individual s self-assessment of his or her capabilities to perform a behavior Davis, 1989 Davis, 1989 Rogers, 1995 Pavlou and Fygenson, 2006 Bhattacherjee, 2000 Bhattacherjee, 2000 Taylor and Todd, 1995 Taylor and Todd, 1995 Ajzen, 1991 Ajzen, 1991 Ajzen, 1991 Ajzen, 1991

40 Methodology METHODOLOGY This section presents the research methodology of this study. First, the research approach is outlined; this is followed by instrument development, sampling and data collection method. 4.1 Research Approach In order to find the determinants of consumer s online purchase behavior, the present study proposes a conceptual model which is based on current literature and builds upon empirically tested findings; therefore the research and data collection takes a quantitative approach. A quantitative approach allows the generalization to a wider segment of population, furthermore, it provides a basis for analysis and interpretation (Adams et al., 2007). An online survey was developed to validate the conceptual model and the proposed research hypotheses, a method that is suitable for collecting data from large geographical areas, contact a large amount of people and collect statistically significant data. (Adams et al., 2007). The survey was conducted using a structured and standardized questionnaire. The unit of analysis includes individuals who had at least purchased online once in the last twelve months at the moment of data collection. The reasoning behind this choice lies on the findings on e-commerce literature that suggest two populations, potential and repeat customers, based on their online shopping experience. 4.2 Instrument Development A questionnaire was developed in order to gather in-depth information for the measurement of the conceptual model s constructs: Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Compatibility (COM), Trust (TR), Interpersonal Influence (II), External Influence (EI), Self-efficacy (SE), Facilitating Conditions (FC), and TPB constructs: Attitude (A), Subjective Norm (SN), Perceived Behavioral Control (PBC), Behavioral Intention (BI) and Behavior (B). The items used to operationalize constructs are based on literature review and have been validated in previous studies.

41 Methodology 34 Table 3 summarizes the constructs and their respective sources and Appendix 1 presents the detailed item information. Constructs Items Sources Internet Experience (INT) 3 Barkhi et al. (2008) Shopping Enjoyment (ENJ) 3 Jarvenpaa et al. (2000) Perceived Usefulness (PU) 3 Lin (2007) adapted from Davis (1989) Perceived Ease of Use (PEOU) 3 Lin (2007) adapted from Davis (1989) Compatibility (COM) 3 Vijayasarathy (2004) adapted from Taylor and Todd (1995) Trust (TR) 3 Gefen et al. (2003) Interpersonal Influence (II) 2 Bhattacherjee (2000) External Influence (EI) 3 Bhattacherjee (2000) Subjective Norm (SN) 2 Pavlou and Fygenson (2006) Self-Efficacy (SE) 2 Lin (2007) adapted from Taylor and Todd (1995) Facilitating Conditions (FC) 2 Lin (2007) adapted from Taylor and Todd (1995) Perceived Behavioral Control (PBC) Table 3: Sources of Questionnaire Items The questionnaire was developed in English. The approach to testing the conceptual model was based on the one used by Taylor and Todd (1995) to test a TPB model with decomposed belief structures. 2 Pavlou and Fygenson (2006) Attitude (A) 4 Pavlou and Fygenson (2006) Behavioral Intention (BI) 2 Pavlou and Fygenson (2006) In order to gather the right group of respondents for the purpose of the study, the questionnaire was established to start with a screening question: During the last 12 months, have you purchased any product/service using the internet?, those individuals who answered no were excluded because they did not fit the target population. Self-report method is used in this study where respondents are asked to report their previous purchase experiences and behaviors. Respondents are asked to indicate which products/services they have purchased online; they had the option to mark up to three products/services. Actual purchasing behavior was measured with a single item, Please indicate how many times you have purchased each product/service in the last 12 months, where a 7-point Likert scale included: Once, 2-3 times a year, 4-5 times a year, once per 1 or 2 months, 2 times a month, 3 times a month, more than 3 times a month. Prior research measured actual behavior in a similar way by asking purchase frequency during a determined period of time with a Likert scale either indicating

42 Methodology 35 number of times purchased or frequency ranging from seldom to often (Barkhi et al., 2008, Chen and Tan, 2004, Pavlou, 2003, Suh and Han, 2003). At this point, respondents are requested to choose one product/service (from those selected in the first part of the survey) in order to answer the rest of the questionnaire with that product/service in mind. A program replaced the name of the product/service selected by the respondent as this approach ensured that the respondent was consistent in his or her answers to the questions. All measurements used 7-point Likert scales ranging from strongly disagree to strongly agree, except for item INT1 regarding internet experience with a scale ranging from limited to significant, item INT2 where respondents indicated how many hours per week they use the internet 2, the scale for item SE2 ranged from difficult to easy and attitude 7-point Likert scales ranging from bad idea to good idea, foolish idea to wise idea, dislike to like and unpleasant to pleasant. The questionnaire s last section included demographic questions such as gender, age, country, education and marital status. 4.3 Sampling and Data Collection Data were obtained through social networking sites, mainly Facebook, using the snowball sampling technique, which is a special type of non-probability sampling where a link is sent to acquaintances, who are asked to answer and share the web-based questionnaire to other people they know and so forth. Appendix 2 shows the web-based questionnaire. A pilot study was conducted (n=15) to test the web-based questionnaire and examine questions and wording. The feedback from respondents was used to make the decision to modify the question regarding products/services purchased during the last twelve months, since the question included other as an additional option where respondents could type the missing product/service themselves. It was noticed that some answers were sentences rather than one word. Considering that a program replaced the name of the product later in all questions, as explained previously, this would make the questions confusing. Hence, the option other was removed. 2 Based on the data set, INT2 was later transformed into groups, using SPSS -Recode variable functionforming a 7-point Likert scale in order to execute the analysis.

43 Methodology 36 During data collection all respondents participated voluntarily, however it is important to mention that halfway the collection process, it was necessary to offer an incentive in view of the fact that participation was considerably reduced. The incentive consisted of a gift certificate for the itunes store with a value of 200 Kr. for the winner of a draw among all of those who filled in the complete questionnaire.

44 Data Analysis and Results DATA ANALYSIS AND RESULTS This section presents data analysis and discusses the results. Sample characteristics are presented first, followed by measurement validation. Then model assessment, path analysis and hypotheses testing are provided. Finally, the results are discussed. 5.1 Data Analysis SPSS version 19.0 was used to perform some of the preliminary analysis, data screening prior model estimation and analysis. In order to estimate the structural equation model (SEM), Partial Least Square (PLS) method was chosen and the software application used was SmartPLS 2.0. (Ringle et al., 2005). PLS provides a powerful method for assessing a structural model and can be used not only for theory confirmation, but also for suggesting where relationships might or might not exist (Hair et al., 2012). Compared to covariance-based SEM, PLS has minimal demands on measurement scales, sample size and residual distributions (Chin, 2010). Furthermore, PLS has become a very popular technique in research, as its use has increased over time (Ringle et al., 2012). Unlike covariance-based SEM, PLS focuses on maximizing the variance of the dependent variables explained by the independent ones instead of reproducing the empirical covariance matrix (Haenlein and Kaplan, 2004). PLS uses latent variable proxies which are linear composites of the associated observed variables, this calculation is similar to principal component analysis for reflective indicators or regression analysis for formative indicators. In PLS path modeling, parameter estimation is accomplished through a multistage algorithm, the various stages involve a sequence of regressions in terms of weight vectors, with iteration until convergence is achieved on a final set of weights (Haenlein and Kaplan, 2004). In terms of sample size, due to the partial nature of the estimation procedure where only a portion of the model is involved at any one time, only the part that requires the largest multiple regression becomes important (Chin, 1998). According to Chin (1998), using a regression heuristic of ten cases per predictor for the minimum sample size in PLS analysis the sample size is determined by (a) the block with the largest number of formative indicators or (b) the dependent latent variable with the largest number of

45 Data Analysis and Results 38 independent latent variables impacting it. The conceptual model proposed in this study has no formative indicators and the largest number of independent latent variables that impact the same dependent variable is four. Thus, the minimum required sample size for this study is 40. The number of usable responses collected in the present study is 138, therefore satisfying the minimum sample requirement in PLS analysis. For the control variables, the approach by Liang et al. (2007) was used, where the control variables are connected with the independent variable on SEM PLS analysis. 5.2 Sample Characteristics After eliminating 15 non-usable responses due to significant missing data and 12 responses representing those who had not purchased products online, 138 final responses were usable for data analysis. Frequency Percent Gender Male Female Age < and over 9 7 Education No formal schooling 1 1 Primary school 3 2 High school 13 9 College/University Post-graduate degree Marital status Single Married Separated 5 4 Divorced 3 2 In a relationship Hours using the < internet (per week) > Continent Europe North America 13 9 South America 13 9 Other Table 4: Sample Characteristics

46 Data Analysis and Results 39 The sample collected is relatively balanced in terms of gender representation with 54 percent males and 46 percent females. Most of the participants are well educated adults between 21 and 49 years of age (mean age 38), who are either in a relationship or married. The respondents are very familiar with the use of internet as most spend up to 40 hours using the internet on a week. Most of the respondents come from European countries, mainly Denmark, with the rest of the sample including North America, South America and other countries from Asia. Table 4 summarizes some of the demographic attributes of the respondents. Since a screening question was used, all respondents have at least purchased online once (during the last twelve months at the time of the survey), however it appears as most of them have experience shopping online as the majority indicated three products and/or services. Among the most frequently purchased items are travel services, clothing and books. Table 5 presents the products and services purchased online by the respondents. Products/services purchased online Frequency Percent Travel Clothing Books Tickets Software 34 9 Electronics 34 9 CDs/DVDs 24 7 Personal care products 21 6 Sports equipment 13 4 Subscriptions 11 3 Appliances 6 2 Furniture 4 1 TOTAL Table 5: Products and Services Purchased Online In addition, respondents were asked to select one product/service to answer the questionnaire in order to keep their answers coherent and consistent. Table 6 shows the products and services selected by the respondents. Travel, clothing and books are among the top three products selected and also the most frequently purchased online (See Appendix 3: Purchase Frequencies).

47 Data Analysis and Results 40 Frequency Percent Travel Clothing Books Tickets Electronics 13 9 Software 8 6 CDs/DVDs 4 3 Sports equipment 4 3 Personal care products 4 3 Subscriptions 2 1 Appliances 1 1 Table 6: Products and Services Selected to Answer the Survey By using the snowball sampling technique there is a risk that the sample is not representative of the population of online shoppers. Therefore non-response bias was assessed by comparing the study s respondents with online shoppers in previous studies that did not use student populations as their only target in the following dimensions: gender, age and education (Aljukhadar and Senecal, 2011, Broekhuizen and Huizingh, 2009, Burkolter and Kluge, 2011, Verhagen et al., 2006, Vijayasarathy, 2002). In prior studies gender distribution tends to be balanced, additionally, online shoppers tend to be well educated and with a mean age of 33 years. Gender distribution and education in the sample appear to be similar to prior studies samples; however respondents are slightly older in the present study, t-tests showed no significant differences (p > 0.05). Since half the sample comes from Denmark; current Danish online consumer demographics were considered. According to a publication by Posten Norden Distanshandeln i Norden 2010 (Distance Trade in the Nordic Countries in 2010) a typical Danish online consumer is a man between 30 and 49 years old (Euromonitor International, 2012c). In addition, according to Danmarks Statistik, 41 percent of online shoppers are highly educated and 51 percent of shoppers are men; the only parameter that differs slightly with this study s sample is age, namely the younger 16 to 19 group, which represents 18 percent of online shoppers according to Danmarks Statistik (Appendix 4). Based on these results, the sample appears to be representative of the population, although the effect of non-response bias cannot be discounted entirely.

48 Data Analysis and Results PLS Analysis Measurement Model Results The first step in order to present the results of PLS analysis is to assess the reliability and validity of the measurement items or indicators, as it is important to determine that the measures represent the constructs. This section provides an evaluation on how accurate the measures are and also their convergent and discriminant validities Reliability All constructs consist of more than one item, except actual behavior (B), which was measured by purchase frequency of the respondent s selected product/service. Cronbach s alpha was used to assess internal consistency, since it provides an estimate for the reliability based on the indicator intercorrelations (Henseler et al., 2009). Alpha coefficients range from 0 to 1 where higher coefficients indicate higher reliability. The accepted value of Cronbach s alpha is 0.70, whereas a value below 0.6 indicates a lack of reliability (Nunnally et al., 1967). Table 7 shows that all constructs present alpha coefficients higher than 0.70, except for trust (TR) with 0.68 and internet experience (INT) with Construct # of Indicators Composite Reliability Cronbach s Alpha Attitude (A) Behavior (B) Behavioral Intention (BI) Compatibility (COM) External Influence (EI) Enjoyment (ENJ) Facilitating Conditions (FC) Interpersonal Influence (II) Internet experience (INT) Perceived Behavioral Control (PBC) Perceived Ease of Use (PEOU) Perceived Usefulness (PU) Self-Efficacy (SE) Subjective Norm (SN) Trust (TR) Table 7: Reliability Results Composite Reliability and Cronbach s Alpha Cronbach s alpha tends to provide an underestimation of the internal consistency (Henseler et al., 2009), therefore it is also appropriate to apply the composite reliability

49 Data Analysis and Results 42 measure. The composite reliability takes into account that indicators have different loadings, and can be interpreted in the same way as Cronbach s alpha. The accepted value for composite reliability is 0.70 or higher (Henseler et al., 2009). The composite reliability values are shown on Table 7, the values for all constructs are above the cutoff level. The averaged composite reliability for all constructs is 0.90 showing high reliability. Therefore, it can be said that the measurement instrument of this study is reliable Validity For the assessment of validity, convergent and discriminant validities are used. Convergent validity means that a set of indicators represents one and the same underlying construct, which can be analyzed through their unidimensionality. Discriminant validity is a complementary concept, meaning that each indicator should not have a stronger connection with constructs other than the one it attempts to reflect. Construct AVE Attitude (A) 0.72 Behavior (B) 1.00 Behavioral Intention (BI) 0.93 Compatibility (COM) 0.80 External Influence (EI) 0.71 Enjoyment (ENJ) 0.78 Facilitating Conditions (FC) 0.85 Interpersonal Influence (II) 0.85 Internet experience (INT) 0.51 Perceived Behavioral Control (PBC) 0.84 Perceived Ease of Use (PEOU) 0.82 Perceived Usefulness (PU) 0.67 Self-Efficacy (SE) 0.84 Subjective Norm (SN) 0.88 Trust (TR) 0.61 Table 8: Validity Results AVE Fornell and Larcker (1981) suggest using the average variance extracted (AVE) as a criterion of convergent validity. AVE measures the amount of variance that a latent variable captures from its indicators relative to the amount due to measurement error. (Chin, 2010). An AVE value of at least 0.5 indicates sufficient convergent validity, meaning that a latent variable is able to explain more than half of the variance of its

50 Data Analysis and Results 43 indicators on average (Henseler et al., 2009). AVE is only applicable for mode A (outward-directed) reflective constructs or latent variables. The entire measurement instrument in the present study is reflective, thus AVE is applicable to all constructs. AVE values are shown in Table 8, all values are greater than 0.50, achieving convergent validity. There are two measures of discriminant validity: The Fornell-Larcker criterion and the cross loadings (Henseler et al., 2009). The Fornell-Larcker criterion indicates that a latent variable shares more variance with its assigned indicators than with any other latent variable, in other words, the AVE of each latent variable should be greater than the latent variable s highest squared correlation with any other latent variable. The second measure of discriminant validity takes into account the loading of each indicator, where it is expected to be greater than all of its cross-loadings (Henseler et al., 2009). Although the Fornell-Larcker criterion assesses discriminant validity on the construct level, the cross loadings allow this evaluation on the indicator level (Chin, 2010). The results of both Fornell-Larcker criterion (Appendix 6: Latent Variables Squared Correlations and AVE) and cross loadings (Appendix 7: Cross Loadings) suggest that all construct measurements have adequate discriminant validities Structural Model Results Having tested for reliability and validity of the measures, the next step is to focus on the structural model. PLS analysis implies great emphasis on variance explained as well as establishing the significance of all path estimates. PLS algorithm was executed on SmartPLS using 300 as maximum number of iterations, path weighting scheme was selected since Haenlein and Kaplan (2004) suggest that the choice between the different weighting schemes for determining inner model proxies has only a minor impact on the final results Variance Explanation The explanation power of the structural model is assessed by the R 2 values of the endogenous constructs, these values represent the amount of variance in the construct that is explained by the model (Tabachnick and Fidell, 2007).

51 Data Analysis and Results 44 Chin (1998) describes R 2 values of 0.67, 0.33, and 0.19 in PLS path models as substantial, moderate, and weak, respectively. Table 9 summarizes the R 2 values obtained, where attitude, purchase intention, PBC and subjective norm show moderate values, and behavior (i.e. online purchasing) a rather weak value. In other words, the model is able to explain 10 percent of the variance in online purchasing and 38 percent of the variance in behavioral intention Path Analysis Construct R 2 Attitude (A) 0.52 Behavior (B) 0.10 Behavioral Intention (BI) 0.38 Perceived Behavioral Control (PBC) 0.66 Subjective Norm (SN) 0.47 Table 9: Variance Explanation Results The path coefficients of the PLS structural model provide a validation of the theoretically assumed relationships between constructs (Adams et al., 2007). The individual path coefficients measure the magnitude of the causal relation between constructs, they can be interpreted as standardized beta coefficients of ordinary least squares regressions (Henseler et al., 2009). The results of the structural path analysis are depicted in Appendix 8, in which PLS path coefficients and indicators loadings are shown. All path coefficients are positive, except for the path PBC to behavior (B); the negative path coefficient indicates the causal relation is negative. Tests of the path significances in the model are provided in the Hypotheses testing section Effect Size Henseler et al. (2009) recommend that all indirect effects of a particular latent variable on another variable should be evaluated, considering that the standardized inner path model coefficients decline with an increased number of indirect relationships. In order to evaluate the effect size in the path model, Cohen s (1988) 2 was calculated as the increase in R 2 relative to the proportion of variance of the endogenous latent variable that remains unexplained (Henseler et al., 2009):

52 Data Analysis and Results 45 According to Cohen (1988) values of 0.02, 0.15, and 0.35 can be interpreted as small, medium, and large effects at the structural level, respectively. The 2 values were calculated manually for each latent variable. Table 10 presents the results of effect size; all effects on behavior are small, attitude has a medium effect on behavioral intention, PEOU appears to have a medium effect on attitude while the rest of variables have a small effect on attitude. All effects on subjective norm and PBC are above 0.15, representing medium effects. Effects Effects on Behavior: Behavioral Intention 0.03 PBC 0.01 Enjoyment 0.03 Internet experience 0.01 Effects on Behavioral Intention: Attitude 0.23 Subjective Norm 0.02 PBC 0.04 Effects on Attitude: Trust 0.01 Perceived Usefulness 0.04 Perceived Ease of Use 0.20 Compatibility 0.04 Effects on Subjective Norm: Interpersonal influence 0.22 External influence 0.16 Effects on PBC: Predictive Relevance Self-efficacy 0.24 Facilitating Conditions 0.39 Table 10: Effect Size Results Stone-Geisser s Q 2 (Stone, 1974; Geisser, 1975) is used to assess the model s capability to predict. The Stone-Geisser criterion postulates that the model must be able to provide a prediction of the endogenous latent variable s indicators (Henseler et al., 2009). The predictive relevance can be measured using blindfolding procedures (Tenenhaus et al., 2005), which can only be applied to endogenous latent variables that have a reflective measure. The blindfolding procedure omits a part of the data for a particular block of indicators and then attempts to estimate the omitted part using the estimated parameters 2

53 Data Analysis and Results 46 (Chin, 2010). Basically Q 2 represents a measure of how well observed values are reconstructed by the model and its parameter estimates. Q 2 values above 0 indicate that the observed values are well reconstructed and that the model has predictive relevance (Hair et al., 2012). Q 2 was obtained using the cross-validated redundancy measure as suggested by Chin (1998) by running the blindfolding procedure on SmartPLS with omission distance 7 for each latent variable. The choice of omission distance was based on Chin (2010). The results listed on Table 11 show that all endogenous latent variables have predictive relevancy as all values are above Hypotheses Testing Construct Q 2 Attitude (A) 0.35 Behavior (B) 0.09 Behavioral Intention (BI) 0.36 Perceived Behavioral Control (PBC) 0.55 Subjective Norm (SN) 0.41 Table 11: Predictive Relevance Results Bootstrapping Bootstrap procedure was used to estimate the significance of path coefficients i.e. hypotheses H1-H13 in the model. Bootstrapping provides an estimate of the shape, spread, and bias of the sampling distribution of a specific statistic (Adams et al., 2007). Bootstrap procedure creates a number of samples where each bootstrap sample has the same number of cases as the original sample; bootstrap samples are created by randomly drawing cases with replacement from the original sample and PLS estimates the path model for each bootstrap sample. Then, the obtained path model coefficients form a bootstrap distribution, this information allows a student s t-test for the significance of the path model relationships (Henseler et al., 2009). In this study, bootstrap was performed with 138 cases and 500 samples. The significance of path relationships was determined with one tail t-test distribution with 500 degrees of freedom. One tail t-test is used because all hypotheses are directional in this study. According to one tail t-test (df = 500), 95 percent significance level or p <

54 Data Analysis and Results requires t-value > Appendix 9 shows the graphical bootstrap output with t- value for each path and Table 12 summarizes the results. Hypotheses Path Coeff. T Statistics H1: Behavioral Intention Behavior ** H2: Attitude Behavioral Intention ** H3: Subjective Norm Behavioral Intention H4: PBC Behavioral Intention * H5: PBC Behavior H6: Perceived Usefulness Attitude * H7: Perceived Ease of Use Attitude ** H8: Compatibility Attitude * H9: Trust Attitude H10: Interpersonal Influence Subjective Norm ** H11: External Influence Subjective Norm ** H12: Self-efficacy PBC ** H13: Facilitating Conditions PBC ** Note: one-tail * Significant at.05 level ** Significant at.01 level Table 12: Tests of PLS Paths with Bootstrap The results support the proposed relationships between behavioral intention and behavior (H1) (t = 2.553, p < 0.01); attitude and behavioral intention (H2) (t = 4.963, p < 0.01); perceived behavioral control and behavioral intention (H4) (t = 2.260, p < 0.05); perceived usefulness and attitude (H6) (t = 2.051, p < 0.05); perceived ease of use and attitude (H7) (t = 4.608, p < 0.01); compatibility and attitude (H8) (t = 1.986, p < 0.05); interpersonal influence and subjective norm (H10) (t = 4.238, p < 0.01); external influence and subjective norm (H11) (t = 4.210, p < 0.01); self-efficacy and perceived behavioral control (H12) (t = 3.685, p < 0.01); facilitating conditions and perceived behavioral control (H13) (t = 5.279, p < 0.01). The results do not support the proposed relationships between subjective norm and behavioral intention (H3) (t = 1.321, p > 0.05); perceived behavioral control and behavior (H5) (t = 1.000, p > 0.05); trust and attitude (H9) (t = 1.266, p > 0.05). With regard to the control variables included in the model, shopping enjoyment (ENJ) is significantly related with the behavior, namely online purchasing (t = 1.948, p < 0.05) while internet experience (INT) is not significantly related. The total effect size of both control variables is small, only 0.05; this is also the case if behavioral intention is used as the focal variable for the control variables.

55 Data Analysis and Results One-Way ANOVA Since analysis of variance (ANOVA) is a method which determines differences between groups consisting of different population means (Tabachnick and Fidell, 2007), one-way ANOVA was conducted to test if behavioral intention differs by product type (H14). In order to obtain the dependent variable behavioral intention, measures of behavioral intention BI1 and BI2 were transformed into one variable using SPSS Transform Compute variable Mean function. The products selected by the respondents were grouped with SPSS Variable transformation function -recode into different variables. Products were grouped according to the classification proposed by Peterson et al. (1997), Table 13 illustrates the results. Product Classification Product/Service Frequency 1 Low cost, intangible Entertainment tickets, subscriptions 22 2 High cost, intangible Software, travel 48 3 Low cost, tangible Books, CDs/DVDs, sports equipment, clothing 54 4 High cost, tangible Electronics, appliances 14 Table 13: Products and Services Classification in This Study ANOVA Assumptions In order to be able to perform ANOVA analysis, assumptions of normal distribution and homogeneity of variances were tested for the dependent variable behavioral intention. Normal Distribution The dependant variable behavioral intention was screened for normal distribution. The value of skewness was and for kurtosis The values are out of the interval (-1+1), thus a normal distribution cannot be assumed, as a large deviation from normality leads to hypothesis test conclusions that are too liberal and a decrease in power and efficiency (Adams et al., 2007). Data transformation is considered an acceptable remedial measure for non-normality (Tabachnick and Fidell, 2007), therefore the dependent variable behavioral intention was transformed using SPSS Transform Compute variable Natural logarithm

56 Data Analysis and Results 49 transformation (reflected). After the transformation, the value of skewness was and for kurtosis The values are in the interval (-1+1), thus it can be stated that the variable does not deviate significantly from normality. P-P plots and Q-Q plots also confirm that the distribution is closer to normal as the observed standardized residuals are closely located around the 45 degree line (Appendix 10). Homogeneity of variance The second assumption regarding equal variance between the groups on the dependent variable was assessed using Levene s Test of Homogeneity of Variance. The result is shown in Table 14. The p-value is 0.092, meaning that the hypothesis of equal variances cannot be rejected, thus, the homogeneity of variance assumption is satisfied ANOVA Results Levene Statistic df1 df2 Sig Table 14: Test of Homogeneity of Variances Product Types The relationship between behavioral intention and product type was tested using oneway ANOVA. The results in Table 15 reveal that behavioral intention does not differ significantly by product type at the p-value of 0.05 (F = 1.716, p = 0.167). Sum of Squares df Mean Square F Sig. Between Groups Within Groups Total Table 15: ANOVA Behavioral Intention and Product Types Post Hoc Tests In order to gain insight and compare which specific product types may present differences with regard to behavioral intention, Post Hoc tests were performed using LSD test (least significant difference), which explores all possible pair-wise comparisons of means comprising a factor using the equivalent of multiple t-tests. It is important to note that this method controls comparison-wise Type I error, however it does not control experiment-wise error rate (Tabachnick and Fidell, 2007). The output (Appendix 11) shows that there is a significant difference between low cost, tangible products and high cost, tangible products, the mean difference is

57 Data Analysis and Results 50 significant at the level. It can be concluded that online purchase intention with regards to tangible products differs when it comes to low vs. high cost. Therefore, H14 which postulated differences in intentions by product type is partially supported. In addition, it can be inferred from Table 16 that among tangible products, purchase intention tends to be higher for low cost products than high cost products. N Mean Std. Deviation Low cost, intangible High cost, intangible Low cost, tangible High cost, tangible Total Table 16: Behavioral Intention and Product Type Statistics Additional Analysis Trust Since the relationship between trust and attitude proposed on H9 (TR A) was not supported, additional analysis were performed in order to obtain better understanding of the role of trust and online purchasing behavior. It is important to mention that the relationship between trust and behavioral intention was also tested with PLS analysis; the result showed the relation is not statistically significant. It was decided to test for differences in behavioral intention between respondents with low levels of trust and high levels of trust. To be able to obtain the variable trust, measures of trust TR1, TR2 and TR3 were transformed into one variable using SPSS Transform Compute variable Mean function. Variable dichotomization was employed using median split as a method, in this case, any value that is found to be below 5 is categorized as low and any value equal or above 5 is categorized as high. The sample was then divided into low trust and high trust, the result is presented in Table 17. N Mean Std. Deviation Low trust High trust Total Table 17: Low and High Trust Groups In order to statistically test the significance in difference of means, one-way ANOVA was applied and the respective assumptions were checked. The dependant variable behavioral intention was transformed, as explained previously, to obtain a distribution

58 Data Analysis and Results 51 closer to normality. Equal variance between the groups on the dependent variable was assessed using Levene s Test of Homogeneity of Variance; Table 18 illustrates that the p-value is 0.000, meaning that the assumption of equal variances is not satisfied. As a consequence it will be referred to the p-value of the Welch-test which can be conducted if two groups have unequal variances. Levene Statistic df1 df2 Sig Table 18: Test of Homogeneity of Variances Trust Groups The ANOVA result in Table 19 reveals that differences between low and high trust groups in regard to behavioral intention differ significantly at the p-value of (F = , p = 0.000). As the assumption of homogeneity was not fulfilled it is referred to the Welch-test. Sum of Squares df Mean Square F Sig. Between Groups Within Groups Total Table 19: ANOVA Behavioral Intention and Trust Groups The Welch-test in Table 20 has a p-value of and is significant. As < 0.05, H0 is rejected and H1 is accepted. There is evidence to support that consumers with high levels of trust show higher intentions to purchase online, than consumers with lower levels of trust on web retailers. Statistic a df1 df2 Sig. Welch Table 20: Robust Tests of Equality of Means Trust Groups Summary of Results of Hypotheses Testing The overall hypotheses testing results are presented in Table 21 and Figure 6 illustrates the conceptual model with significant paths (p < 0.05) in solid lines and non significant paths (p > 0.05) in dashed lines. The detailed discussions are presented in the next section.

59 Data Analysis and Results 52 Hypotheses Relationship Result H1: H2: H3: H4: H5: H6: H7: H8: A consumer s behavioral intention to purchase online positively affects his/her actual online purchase. A consumer s attitude toward online purchasing positively affects his/her behavioral intention to purchase online. A consumer s subjective norm in relation to online purchasing positively affects his/her behavioral intention to purchase online. A consumer s PBC over online purchasing positively influences his/her behavioral intention to purchase online. A consumer s PBC over online purchasing positively influences his/her actual online purchase. A consumer s perceived usefulness of online purchasing positively affects his/her attitude toward online purchasing. A consumer s perceived ease of use of online purchasing positively affects his/her attitude toward online purchasing. Compatibility between online purchasing and a consumer s lifestyle and needs positively affects his/her attitude toward online purchasing. A consumer s trust in a web retailer positively influences H9: his/her attitude toward online purchasing. A consumer s perception of interpersonal influence is H10: positively associated with his/her subjective norm about online purchasing. A consumer s perception of external influence is positively H11: associated with his/her subjective norm about online purchasing. A consumer s positive self-efficacy positively influences his/her H12: perceived behavioral control over online purchasing. A consumer s positive facilitating conditions positively H13: influence his/her perceived behavioral control over online purchasing. A consumer s behavioral intention to purchase online differs H14: by product type. Table 21: Summary of Results of Hypotheses Testing BI B A BI SN BI PBC BI PBC B PU A PEOU A COM A TR A II SN EI SN SE PBC FC PBC Product type BI Supported Supported Rejected Supported Rejected Supported Supported Supported Rejected Supported Supported Supported Supported Partially Supported

60 Data Analysis and Results 53 *Significant at.05 level ** Significant at.01 level Significant paths in solid lines Figure 6: The Conceptual Model Results of Path Significances

61 Discussion DISCUSSION 6.1 How do perceived usefulness, perceived ease of use, compatibility and trust impact attitude toward online shopping? The results show that perceived usefulness, perceived ease of use, compatibility and trust can explain 52 percent of the variance in attitude toward online shopping. Following, each factor and its impact is discussed. Perceived Usefulness and Perceived Ease of Use: Two important IT adoption factors, PU and PEOU, have been empirically supported by a great number of studies to positively influence attitude and technology acceptance in the information system field. Moreover, PU and PEOU have been also widely used and supported in the e-commerce setting. This study found that PU has a positive impact on attitude toward online purchasing (p < 0.05). The results also show that PEOU has a strong impact on attitude toward online purchasing (p < 0.01). Both findings are consistent with prior research (Barkhi et al., 2008, Chen and Tan, 2004, Hernández et al., 2010, Lin, 2007, Pavlou and Fygenson, 2006, Suh and Han, 2003, Taylor and Todd, 1995, Vijayasarathy, 2004). PEOU showed the strongest effect on attitude, meaning that despite consumers experience shopping online, browsing for products or services free of effort, with access to user friendly interfaces and simple check out processes are important when consumers consider shopping online. In addition, PU significance shows that the ability to get product or service information and compare products or service offerings plays a significant role in forming positive attitudes for those consumers who seek convenience and time saving. Compatibility: It was hypothesized in this study that the compatibility between online shopping and a consumer s existing values, beliefs and lifestyle would have an impact on his/her attitudes toward online shopping, the results show that the hypothesized relationship is statistically significant (p < 0.05), the result is consistent with findings obtained by Chen and Tan (2004), Vijayasarathy (2004) and (Lin, 2007). Thus, it is safe to assume that consumers whose lifestyle and shopping habits are compatible with the

62 Discussion 55 convenience and time saving offered by web retailers are more willing to purchase online. In addition to time saving, consumers who like to shop from home, avoiding big crowds or maintain busy schedules may show a more positive attitude towards online shopping. Trust: The impact of trust on attitude toward online purchasing has been empirically supported in the e-commerce literature (Chen and Tan, 2004, George, 2004, Ha and Stoel, 2009, Jarvenpaa et al., 2000, Pavlou and Fygenson, 2006, Suh and Han, 2003). However, unlike most studies, this study did not provide empirical evidence to support that trust positively impacts attitude. A plausible explanation for this unexpected finding is related to the participants prior experience purchasing online. In a study comparing potential customers versus repeat customers in the online context, Gefen et al. (2003) found that there are two separate populations regarding trust beliefs, since for repeat customers trust is influenced by prior experience with web retailers and for potential customers, trust is mainly based on their disposition to trust. The participants in this study have experience shopping online and may not had have bad experiences from their past online purchasing activities, therefore, trust does not play a significant role in future interactions with web retailers as many may believe that payment by credit card online might be as safe as payment in a physical store. Another possible reason for the finding in this study could be due the fact that half of the sample is formed by Danish respondents, who are probably familiar with Trust Pilot, which is an open communitybased platform for sharing real reviews of shopping experiences, Trust Pilot is ranked among the top visited Danish websites (Alexa, 2012). Danish consumers may be also familiar with the quality stamp E-mark that was developed with the support of the Danish Ministry of Science with purpose of ensuring legal and ethical operations (Euromonitor International, 2012c). Further, it is fair to say that many online stores have established good reputations and in turn, many online consumers may choose to transact with those online stores. To sum up, positive reviews from websites like Trust Pilot and the E-mark quality stamp together with positive prior experiences shopping online may have an effect on trusting beliefs which can render them not critical in future interactions. Low vs. High Trust groups and Behavioral Intention: Given the unexpected result regarding the significance of trust in this study, additional analysis were performed in order to get a better understanding of the role of trust on online behavioral intention.

63 Discussion 56 Many researchers have proposed the relationship between trust and attitude, as in this study; nevertheless e-commerce literature has also found significant results when analyzing the direct impact of trust on purchase intention (Gefen et al., 2003, Pavlou, 2003), thus additional analysis were performed in this study to be able to understand the role of trust on purchase intentions. The results show that consumers who have high levels of trust on web retailers express significantly more willingness to purchase online than those consumers who have lower levels of trust. The results are somewhat aligned with the extant literature that supports the relevant role of trust in consumer online behavior. However, the dual role of trust as an attitude predictor and as an important factor impacting purchase intention needs to be examined. 6.2 How do interpersonal and external influences impact the subjective norm regarding online shopping? This study found empirical evidence supporting interpersonal influence and external influence as significant belief structures that affect subjective norm, such findings are in line with the findings in e-commerce literature (Bhattacherjee, 2000, Lin, 2007). Moreover, interpersonal and external influence account for 47 percent of the variance on subjective norm. Although it is important to note that the findings in this study suggest that subjective norm is not significantly related to behavioral intention; discussion on this matter is presented further on. 6.3 How do self-efficacy and facilitating conditions impact perceptions of behavioral control regarding online shopping? The results show that self-efficacy and facilitating conditions are significantly related to PBC, both at p < In addition, both self-efficacy and facilitating conditions explain 66 percent of the variance in PBC. The findings in this study are aligned with the e- commerce literature (George, 2004, Lin, 2007, Pavlou and Fygenson, 2006, Taylor and Todd, 1995, Vijayasarathy, 2004). The finding regarding self-efficacy confirms that consumers who are confident about purchasing over the internet show more willingness to engage in online shopping. In regards to facilitating conditions, consumers evidently need time and money to be able to make purchases online. According to Taylor and Todd (1995), the absence of facilitating conditions represents barriers to engage in the

64 Discussion 57 behavior; however the presence of facilitating conditions may not, per se, encourage the behavior. 6.4 How do attitude, subjective norm and perceived behavioral control impact online purchase intentions and consequently, online purchasing? This study examined the causal relationships proposed in TPB (Ajzen, 1991) between attitude, subjective norm, PBC and behavioral intention, the relationship between behavioral intention and actual online purchasing behavior, and the relationship of PBC with purchase behavior. The results support all relationships, except the relationship between subjective norm and behavioral intention (p > 0.05) and the relation between PBC and purchase behavior (p > 0.05). In general, it can be stated that this study provides empirical support to the well established TPB model in the context of online behavior, as it further confirms the importance of attitude and PBC as predictors of behavioral intention and behavioral intention as a direct determinant of behavior. Subjective Norm and Behavioral Intention: Many studies in the online setting found the causal relation between subjective norm and behavioral intention significant (Barkhi et al., 2008, Bhattacherjee, 2000, Hansen et al., 2004, Kim et al., 2003, Taylor and Todd, 1995, Vijayasarathy, 2004, Yoh et al., 2003). Although this study s results do not support the relationship between subjective norm and purchase intention, such finding is consistent with findings by George (2004), Lin (2007), Pavlou and Fygenson (2006) and Chen (2009). One possible explanation is that the relative importance of the subjective norm may be related to the phase of implementation of the technology, according to Taylor and Todd (1995), subjective norms have been found to be more important prior to, or in the early stages of implementation when users have only limited direct experience. This can be explained by the sample in this study, which was formed mostly by consumers who had experience purchasing online, in addition, purchase frequencies show that the respondents were not new to online purchasing as a one-time event. Another possible explanation could be the failure to consider all of the relevant social factors, which according to Conner and Armitage (1998), is one of the reasons that contribute to the mixed findings in literature regarding the role of subjective norms. In this respect, virtual communities where consumers share experiences about shopping from certain online stores and product reviews are

65 Discussion 58 acquiring popularity and the word-of-mouth influence by these virtual communities needs more exploring, in order to determine its role on the subjective norm. PBC and Behavior: The dual role of PBC in relation to intention and the behavior has been proven significant in e-commerce literature (Ling et al., 2011, Pavlou and Fygenson, 2006, Taylor and Todd, 1995), the findings in the present study show that PBC has a significant positive influence on behavioral intention, however the direct influence of PBC on online purchasing behavior was not statistically significant, in fact, it was negatively related. One possible reason for the inconsistency of this finding with previous research could be attributed to the measurement of actual behavior. While the present study used purchase frequency as a measure for behavior, Taylor and Todd (1995) and Lin (2007) conducted a 2-stage study in order to measure actual behavior, for instance in Lin s study, participants browsed for books on the first stage and the second stage included two tasks representing the online transaction process: (1) register with an online bookstore, search for the selected book and place it in the shopping cart, (2) fill in certain payment and delivery information. Pavlou and Fygenson (2006), conducted a longitudinal study where a follow up survey confirmed whether participants purchased online. It is worth noting that although many researchers have included behavior as a construct measured by online purchase frequency and/or online purchase within a certain period of time prior to the moment of data collection (Barkhi et al., 2008, Chen and Tan, 2004, Pavlou, 2003, Suh and Han, 2003), the research models in those studies did not include the causal relation PBC and behavior, thus comparisons could not be made in regards with PBC and behavior with studies that used similar measurement of behavior. 6.5 To what extent, if any, is product type related to online purchase intention? Previous research shows that consumers risks perceptions of online shopping are associated with the type of product, as higher costs could imply greater levels of economic risks (Bhatnagar et al., 2000), additionally, risks are enhanced when there are chances that a product or service could perform less than expected, such risk is higher with online shopping because of its limitations regarding touch, feel, smell and access to sales personnel assistance. This study explored whether product type had an effect on behavioral intention, by focusing on two major products characteristics: cost and

66 Discussion 59 tangibility. The results show that cost plays a significant role for tangible products, as consumers indicate higher purchase intentions for low cost, tangible products when compared to high cost, tangible products. No significant differences were found in relation to intangible products, regardless of cost differences. When compared to previous research, the results are mixed. Ian and Sui Meng (2000) found that low cost and frequently purchased goods are more likely to be purchased online than high cost and infrequently purchased goods; the study also found that intangible products are related to higher willingness to buy than tangible products. Vijayasarathy (2002) found that, cost does not have an impact on purchasing intentions while product tangibility does, his study found that intentions to shop online where higher for intangible products than tangible products. One possible explanation for the results in this study is that the internet penetration and spread of web retailers has increased considerably in the last decade, thus web retailers are dealing with a competitive environment where free delivery and easy returns are now common, and as a result, low cost tangible products show similar high levels of purchasing intention as intangible goods. Moreover, internet shoppers are able to try out the demonstration version of computer software, or be given trial periods of online newspapers, video and music samples, before making a purchase decision. This reduces uncertainty in purchase decision and stimulates purchases. However it appears that perceptions of risk related to high cost, tangible products (i.e. financial and performance risks), are still making an impact on consumers, in other words, web retailers of high cost tangible products still face challenges of narrowing the sensory gap that exists between their products and online consumers.

67 Summary and Conclusion SUMMARY AND CONCLUSION This section presents the summary and conclusion of the study. First a summary of findings is presented, then managerial implications, limitations and finally recommendations for future research are presented. 7.1 Summary of Findings The purpose of this study was to understand what factors determine online consumer purchasing behavior. This study tested a model based on the theory of planned behavior. The approach to testing the model was based on the one used by Taylor and Todd (1995) with decomposed belief structures. Beliefs about perceived usefulness, perceived ease of use, compatibility, interpersonal influence, external influence, self-efficacy and facilitating conditions were integrated in the model in order to explain consumers behavior in regards to online shopping and identify key determinants of online purchasing. Moreover, the relationship between product characteristics and online purchase intention was also explored. The model was tested empirically, a survey of 138 online consumers was conducted and the results from PLS path modeling indicated that the model was able to explain 38 percent of the variance in behavioral intention and 10 percent of the variance in purchasing behavior. Of the 13 causal paths specified in the conceptual model, 11 were found to be statistically significant, and the relation between product characteristics and online purchase intention was partially supported. It was hypothesized and empirically supported that perceived usefulness, perceived ease of use and compatibility between online shopping and consumers needs, positively impact attitude towards online purchasing. Therefore, it is reasonable to estate that consumers who perceive online shopping to be advantageous, the websites easy to operate and navigate and believe that shopping online is compatible with their shopping needs; express a positive attitude toward online shopping and a high willingness to shop online. This study also hypothesized that consumers beliefs about their own ability to make purchases online combined with the availability of resources affect their perceptions of control and consequently, their intention to shop online. The empirical results confirm

68 Summary and Conclusion 61 this, as control beliefs become particularly relevant considering the virtual and impersonal nature of online shopping, furthermore, engaging in online purchasing implies the use of technology and consumers who are confident about their capabilities are more willing to make purchases online. It was hypothesized that online purchase intention would differ by product type. In this study products and services were categorized based on Peterson et al. (1997) classification scheme, using cost and tangibility as two major characteristics. The results showed that consumers are more willing to purchase low cost, tangible products and are less willing to purchase high cost, tangible products over the internet, while no significant differences were found in relation to intangible products, regardless of their cost. This study draws from previous research and presents a relatively comprehensive, yet parsimonious model to describe and predict online consumer behavior. The empirical findings not only offer more insight into the factors that impact online purchasing behavior but also further empirically support the theory of planned behavior in the e- commerce context. 7.2 Managerial Implications Internet retail is unlikely to completely replace traditional brick-and-mortar retail (Datamonitor, 2011), in reality, traditional stores are expanding their reach through the internet channel and implementing multi-channel strategies. Thus, the findings in this study may have significant implications for the retailing industry, including pure players and traditional brick-and mortar businesses. The findings suggest that online stores websites should be easy to navigate and interact with, so consumers can concentrate on the purchase experience, rather than dealing with a complex system. Consumers who feel confident about their skills using the internet to shop are more likely to make purchases online, for those less confident consumers, help and assistance tools can be essential in building up their skills and in increasing their willingness to purchase online. Online shopping can be appealing to those consumers who seek convenience and perceive greater advantages in online shopping over shopping in traditional stores. It can be concluded that the design of the online store environment must be able to deliver

69 Summary and Conclusion 62 advantages such as useful product information and ease to compare products and prices online. Higher levels of trust are associated with higher willingness to shop online, maintaining clear shipping and return policies, as well as a secure check out process is fundamental. Marketers must communicate consumers that online shopping can be convenient, safe and simple to use. With regard to product types, the findings suggest that high cost, tangible products are associated with higher levels of risk, thus consumers are less willing to purchase this type of products online. High costs imply a potential financial risk and consumers are often limited in terms of touch, feel and smell when purchasing tangible products over the internet, as a result performance risks are enhanced. The more marketers know about the path to purchase for a particular product type, the more relevant they can make their messages, for example providing consumers with broad and useful information for research-heavy, high cost, tangible products such as electronics. 7.3 Limitations The main limitation of the present study is related to the sampling method, as a result of using the snowball technique, the chances of having participants with similar traits are higher (people tend to associate with those similar to them, thereby sharing the survey with those contacts in their network and so forth). Although there were no significant differences regarding gender distribution, education and age of the participants compared to similar prior studies samples, participants in this study were slightly older in average, hence non-response bias cannot be entirely discounted. This study explored some of the key factors affecting online purchasing, where measurement of purchasing behavior was based on purchasing frequencies, which can be considered a limitation. First, self-report methods have disadvantages related to the accuracy of the information provided by the participants, since recalling information from the last twelve months can be challenging. Second, in order to get information of the actual behavior, longitudinal studies or lab experiments provide more accurate and real information.

70 Summary and Conclusion Further Research E-commerce literature supports trust on web retailers as one of the most influential beliefs associated with online purchasing behavior. Different research models integrate trust as an attitude predictor and other research models integrate trust as a factor directly impacting purchase intention, thus, the role of trust needs to be examined, furthermore, it is important to understand how initial trust is formed on potential online shoppers and how it is compared to trust among consumers with prior online purchasing experience. Further, as many consumers turn to well-known and trusted web retailers, they start to develop loyalty toward certain online stores, thus understanding the factors that affect loyalty in e-commerce is becoming increasingly important. Technology is in constant progress, new devices like tablets are available to browse for products and mobile apps to shop online are becoming popular in developed countries, as technology changes and mobile online sales increase (Internet Retailer, 2012), consumers shopping habits are also changing. Younger generations have great technology assimilation and are growing with an online culture, therefore understanding mobile-commerce and its potential is fundamental. Although the present study explores eight belief dimensions, it is evident that there is more to explore in online consumer behavior and that the significance of factors may differ across different stages of e-commerce acceptance. Thus, it is important that future research investigates additional determinants of online purchasing behavior.

71 References REFERENCES ADAMS, J., KHAN, H. T. A., RAESIDE, R. & WHITE, D Research methods for graduate business and social science students, Sage. AJZEN, I The theory of planned behavior. Organizational behavior and human decision processes, 50, AJZEN, I Perceived Behavioral Control, Self Efficacy, Locus of Control, and the Theory of Planned Behavior. Journal of applied social psychology, 32, ALEXA Trust Pilot, Statistics Summary [Online]. Available: [Accessed November 2012]. ALJUKHADAR, M. & SENECAL, S Segmenting the online consumer market. Marketing Intelligence & Planning, 29, BARKHI, R., BELANGER, F. & HICKS, J A model of the determinants of purchasing from virtual stores. Journal of Organizational Computing and Electronic Commerce, 18, BENEDICKTUS, R. L., BRADY, M. K., DARKE, P. R. & VOORHEES, C. M Conveying Trustworthiness to Online Consumers: Reactions to Consensus, Physical Store Presence, Brand Familiarity, and Generalized Suspicion. Journal of Retailing, 86, BHATNAGAR, A., MISRA, S. & RAO, H. R On risk, convenience and internet shopping behavior. Communications of the ACM, 43, BHATTACHERJEE, A Acceptance of e-commerce services: the case of electronic brokerages. Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on, 30, BROEKHUIZEN, T. & HUIZINGH, E. K. R. E Online purchase determinants: Is their effect moderated by direct experience? Management Research News, 32, BROWN, M., POPE, N. & VOGES, K Buying or browsing?: An exploration of shopping orientations and online purchase intention. European Journal of Marketing, 37, BURKOLTER, D. & KLUGE, A Online consumer behavior and its relationship with socio-demographics, shopping orientations, need for emotion, and fashion leadership. Journal of Business, 2. CHA, J Exploring the Internet as a Unique Shopping Channel to Sell Both Real and Virtual Items: a Comparison of Factors Affecting Purchase Intention and Consumer Characteristics. Journal of Electronic Commerce Research, 12, CHEEMA, A. & PAPATLA, P Relative importance of online versus offline information for Internet purchases: Product category and Internet experience effects. Journal of Business Research, 63,

72 References 65 CHEN, L Online consumer behavior: An empirical study based on theory of planned behavior. Ph.D , The University of Nebraska - Lincoln. CHEN, L. D. & TAN, J Technology Adaptation in E-commerce: Key Determinants of Virtual Stores Acceptance. European Management Journal, 22, CHEUNG, C. M. K., CHAN, G. W. W. & LIMAYEM, M A Critical Review of Online Consumer Behavior: Empirical Research. Journal of Electronic Commerce in Organizations, 3, CHIN, W. W The partial least squares approach for structural equation modeling. In: MARCOULIDES, G. A. (ed.) Modern Methods for Business Research. Mahwah, NJ: Lawrence Erlbaum Associates. CHIN, W. W How to Write Up and Report PLS Analyses. In: ESPOSITO VINZI, V., CHIN, W. W., HENSELER, J. & WANG, H. (eds.) Handbook of Partial Least Squares. Springer Berlin Heidelberg. CHIU, C. M., CHANG, C. C., CHENG, H. L. & FANG, Y. H Determinants of customer repurchase intention in online shopping. Online Information Review, 33, CONNER, M. & ARMITAGE, C. J Extending the theory of planned behavior: A review and avenues for further research. Journal of applied social psychology, 28, DATAMONITOR Industry Profile: Global Online Retail. DAVIS, F. D Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. MIS Quarterly, 13, DAVIS, F. D User acceptance of information technology: system characteristics, user perceptions and behavioral impacts. International Journal of Man-Machine Studies, 38, DELAFROOZ, N., PAIM, L. H. & KHATIBI, A Understanding consumer s internet purchase intention in Malaysia. African Journal of Business Management, 5, EUROMONITOR INTERNATIONAL 2012a. Global Internet Users: Euromonitor International from International Telecommunications Union/OECD/national statistics. EUROMONITOR INTERNATIONAL 2012b. How are the latest technologies changing companies and consumers? EUROMONITOR INTERNATIONAL 2012c. Internet Retailing in Denmark. January 2012 ed. EUROMONITOR INTERNATIONAL 2012d. Internet Retailing: Euromonitor from trade sources/national statistics. FISHBEIN, M. & AJZEN, I Belief, attitude, intention and behaviour: An introduction to theory and research, Addison-Wesley. GEFEN, D., KARAHANNA, E. & STRAUB, D. W Inexperience and experience with online stores: the importance of TAM and trust. Engineering Management, IEEE Transactions on, 50,

73 References 66 GEORGE, J. F The theory of planned behavior and Internet purchasing. Internet research, 14, GIRARD, T., KORGAONKAR, P. & SILVERBLATT, R Relationship of Type of Product, Shopping Orientations, and Demographics with Preference for Shopping on the Internet. Journal of Business & Psychology, 18, HA, S. & STOEL, L Consumer e-shopping acceptance: Antecedents in a technology acceptance model. Journal of Business Research, 62, HAENLEIN, M. & KAPLAN, A. M A beginner's guide to partial least squares analysis. Understanding statistics, 3, HAIR, J., SARSTEDT, M., RINGLE, C. & MENA, J An assessment of the use of partial least squares structural equation modeling in marketing research. Journal of the Academy of Marketing Science, 40, HANSEN, T. & MØLLER JENSEN, J Shopping orientation and online clothing purchases: the role of gender and purchase situation. European Journal of Marketing, 43, HANSEN, T., MØLLER JENSEN, J. & STUBBE SOLGAARD, H Predicting online grocery buying intention: a comparison of the theory of reasoned action and the theory of planned behavior. International Journal of Information Management, 24, HENSELER, J., RINGLE, C. M. & SINKOVICS, R. R The use of partial least squares path modeling in international marketing. Advances in international marketing, 20, HERNÁNDEZ, B., JIMÉNEZ, J. & MARTÍN, M. J Customer behavior in electronic commerce: The moderating effect of e-purchasing experience. Journal of Business Research, 63, HSU, M.-H. & CHIU, C.-M Predicting electronic service continuance with a decomposed theory of planned behaviour. Behaviour & Information Technology, 23, IAN, P. & SUI MENG, P Factors influencing the types of products and services purchased over the Internet. Internet research, 10, IBM CORP IBM SPSS Statistics for Windows ed. Armonk, NY: IBM Corp. INTERNET RETAILER Forrester Research: US Smartphone Commerce Forecast [Online]. Available: [Accessed November 2012]. JARVENPAA, S. L., TRACTINSKY, N. & VITALE, M Consumer trust in an Internet store. Information Technology and Management, 1, JAYAWARDHENA, C., WRIGHT, L. T. & DENNIS, C Consumers online: intentions, orientations and segmentation. International Journal of Retail & Distribution Management, 35, KEISIDOU, E., SARIGIANNIDIS, L. & MADITINOS, D Consumer characteristics and their effect on accepting online shopping, in the context of different product types. Int. Journal of Business Science and Applied Management, 6.

74 References 67 KIANG, M. Y., YE, Q., HAO, Y., CHEN, M. & LI, Y A service-oriented analysis of online product classification methods. Decision Support Systems, 52, KIM, E. Y. & KIM, Y. K Predicting online purchase intentions for clothing products. European Journal of Marketing, 38, KIM, J An empirical study on consumer first purchase intention in online shopping: integrating initial trust and TAM. Electronic Commerce Research, 12, KIM, Y. K., KIM, E. Y. & KUMAR, S Testing the behavioral intentions model of online shopping for clothing. Clothing and Textiles Research Journal, 21, KORGAONKAR, P., SILVERBLATT, R. & GIRARD, T Online retailing, product classifications, and consumer preferences. Internet research, 16, LEE, K. S. & TAN, S. J E-retailing versus physical retailing: A theoretical model and empirical test of consumer choice. Journal of Business Research, 56, LIAN, J.-W. & LIN, T.-M Effects of consumer characteristics on their acceptance of online shopping: Comparisons among different product types. Computers in Human Behavior, 24, LIANG, H., SARAF, N., HU, Q. & XUE, Y Assimilation of enterprise systems: The effect of institutional pressures and the mediating role of top management. MIS Quarterly, 31, LIN, H.-F Predicting consumer intentions to shop online: An empirical test of competing theories. Electronic Commerce Research and Applications, 6, LING, K. C., CHAI, L. T. & PIEW, T. H The Effects of Shopping Orientations, Online Trust and Prior Online Purchase Experience toward Customers Online Purchase Intention. International Business Research, 3, P63. LING, K. C., DAUD, D. B., PIEW, T. H., KEOY, K. H. & HASSAN, P Perceived Risk, Perceived Technology, Online Trust for the Online Purchase Intention in Malaysia. International Journal of Business and Management, 6, LOWENGART, O. & TRACTINSKY, N Differential effects of product category on shoppers selection of Web-based stores: a probabilistic modeling approach. Journal of Electronic Commerce Research, 2, MCKECHNIE, S., WINKLHOFER, H. & ENNEW, C Applying the technology acceptance model to the online retailing of financial services. International Journal of Retail & Distribution Management, 34, MCKNIGHT, D. H., CHOUDHURY, V. & KACMAR, C Developing and Validating Trust Measures for e-commerce: An Integrative Typology. Information Systems Research, 13,

75 References 68 MONSUWÉ, T. P., DELLAERT, B. G. C. & DE RUYTER, K What drives consumers to shop online? A literature review. International Journal of Service Industry Management, 15, MULLER, P., DAMGAARD, M., LITCHFIELD, A., LEWIS, M. & HÖRNLE, J Consumer behaviour in a digital environment. In: POLICY, P. D. E. A. S. (ed.). NUNNALLY, J. C., BERNSTEIN, I. H. & BERGE, J. M. F Psychometric theory, McGraw-Hill New York. PAVLOU, P. A Consumer Acceptance of Electronic Commerce: Integrating Trust and Risk with the Technology Acceptance Model. International Journal of Electronic Commerce, 7, PAVLOU, P. A. & FYGENSON, M Understanding and Predicting Electronic Commerce Adoption: An Extension of the Theory of Planned Behavior. MIS Quarterly, 30, PETERSON, R. A., BALASUBRAMANIAN, S. & BRONNENBERG, B. J Exploring the Implications of the Internet for Consumer Marketing. Journal of the Academy of Marketing Science, 25, RINGLE, C. M., SARSTEDT, M. & STRAUB, D. W A Critical Look at the Use of PLS-SEM in MIS Quarterly. MIS Quarterly, 36, iiv-8. RINGLE, C. M., WENDE, S. & WILL, A SmartPLS. 2.0 (beta) ed. Hamburg, Germany: SmartPLS. ROGERS, E. M Diffusion of innovations, Simon and Schuster. SEOCK, Y. K. & BAILEY, L. R The influence of college students' shopping orientations and gender differences on online information searches and purchase behaviours. International Journal of Consumer Studies, 32, SO, W. C. M., WONG, T. N. D. & SCULLI, D Factors affecting intentions to purchase via the internet. Industrial Management & Data Systems, 105, SOOPRAMANIEN, D. G. R., FILDES, R. & ROBERTSON, A Consumer decision making, E-commerce and perceived risks. Applied Economics, 39, SUH, B. & HAN, I Effect of trust on customer acceptance of Internet banking. Electronic Commerce Research and Applications, 1, TABACHNICK, B. G. & FIDELL, L. S Using Multivariate Statistics, Pearson/Allyn & Bacon. TAYLOR, S. & TODD, P. A Understanding Information Technology Usage: A Test of Competing Models. Information Systems Research, 6, TENENHAUS, M., VINZI, V. E., CHATELIN, Y. M. & LAURO, C PLS path modeling. Computational Statistics & Data Analysis, 48, TONG, X A cross-national investigation of an extended technology acceptance model in the online shopping context. International Journal of Retail & Distribution Management, 38, VENKATESH, V. & DAVIS, F. D A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies. Management Science, 46, 186.

76 References 69 VERHAGEN, T., SELMAR, M. & YAO-HUA, T Perceived risk and trust associated with purchasing at electronic marketplaces. European Journal of Information Systems, 15, VIJAYASARATHY, L. R Product characteristics and Internet shopping intentions. Internet research, 12, VIJAYASARATHY, L. R Predicting consumer intentions to use on-line shopping: the case for an augmented technology acceptance model. Information & Management, 41, WEN, C., PRYBUTOK, V. R. & XU, C An Integrated Model for Customer Online Repurchase Intention. The Journal of Computer Information Systems, 52, YOH, E., DAMHORST, M. L., SAPP, S. & LACZNIAK, R Consumer Adoption of the Internet: The Case of Apparel Shopping. Psychology & Marketing, 20, ZARMPOU, T., SAPRIKIS, V., MARKOS, A. & VLACHOPOULOU, M Modeling users acceptance of mobile services. Electronic Commerce Research, ZHOU, L., DAI, L. & ZHANG, D Online shopping acceptance model a critical survey of consumer factors in online shopping. Journal of Electronic Commerce Research, 8,

77 Appendices APPENDICES APPENDICES

78 Appendices 71 Appendix 1: Questionnaire Measures and Sources Constructs Measures Internet Experience (INT) Shopping Enjoyment (ENJ) Perceived Usefulness (PU) Perceived Ease of Use (PEOU) Compatibility (COM) Trust (TR) Interpersonal Influence (II) External Influence (EI) Subjective Norm (SN) Self-Efficacy (SE) Facilitating Conditions (FC) Perceived Behavioral Control (PBC) Attitude (A) Behavioral Intention (BI) INT1: What is your level of experience with the use of Internet? (limited/significant) INT2: On average how many hours per week do you spend on the Internet? INT3: Using the Internet is very important to me. ENJ1: In general, I view shopping as an important leisure activity. ENJ2: Under the right circumstances, shopping is fun. ENJ3: For me, shopping is a pleasurable activity. PU1: Shopping online for product x makes it easier to compare products. PU2: Shopping online for product x provides access to useful shopping information. PU3: Shopping online saves me time when purchasing product x. PEOU1: Shopping online for product x to me is clear and easy to understand. PEOU2: I find shopping online for product x easy to do. PEOU3: It would be easy for me to become skilled at shopping online for product x. COM1: Shopping online for product x fits well with my lifestyle. COM2: Shopping online for product x fits well with my shopping needs. COM3: Shopping online for product x is compatible with the way I like to shop. TR1: Online stores keep my best interests in mind. TR2: I select online stores which I believe are honest. TR3: Overall, online stores are trustworthy. II1: My friends or family think that shopping online for product x is a good idea. II2: My friends or family encourage me to shop online for product x. EI1: I have read/seen news reports which say that online shopping provides a good way of purchasing product x. EI2: The popular press adopts a positive view towards online shopping for product x. EI3: Mass media reports have influenced me to try online shopping to purchase product x. SN1: People who are important to me would recommend that I purchase product x online. SN2: Most of the people who I value would purchase product x online. Sources Barkhi et al. (2008) Jarvenpaa et al. (2000) Lin (2007) adapted from Davis (1989) Lin (2007) adapted from Davis (1989) Vijayasarathy (2004) adapted from Taylor and Todd (1995) Gefen et al. (2003) Bhattacherjee (2000) Bhattacherjee (2000) Pavlou and Fygenson (2006) SE1:If I wanted to, for me to purchase product x online would be...(difficult/easy) Lin (2007) adapted from SE2: If I wanted to, I am confident I could purchase product x online on my own. Taylor and Todd (1995) FC1: I have the time to purchase product x online. Lin (2007) adapted from FC2: I have enough money to purchase product x online. Taylor and Todd (1995) PBC1: I am able to purchase product x online. Pavlou and PBC2: Using the internet to purchase product x is entirely within my control. Fygenson (2006) A1: Purchasing product x online is... (bad idea/good idea) A2: Purchasing product x online is (foolish idea/wise idea) A3: Purchasing product x online is an idea I... (dislike/like) A4: Shopping online for product x is (unpleasant/pleasant) BI1: I intend to purchase product x online in the near future. BI2: I will purchase product x online in the near future. Pavlou and Fygenson (2006) Pavlou and Fygenson (2006)

79 Appendices 72 Appendix 2: Web-Based questionnaire

80 Appendix 2: Web-Based questionnaire (cont.) Appendices 73

81 Appendices 74 Appendix 2: Web-Based questionnaire (cont.)

82 Appendices 75 Appendix 2: Web-Based questionnaire (cont.)

83 Appendices 76 Appendix 2: Web-Based questionnaire (cont.)

84 Appendices 77 Appendix 3: Purchase Frequencies (During the last twelve months at the time of the survey) Appliances Books CDs/DVDs Clothing Electronics Furniture Personal care Software Sports equipment Subscriptions Tickets Travel Once Books CDs/DVDs Clothing Electronics Personal care Software Sports equipment Subscriptions Tickets Travel 2-3 times a year times a year Books CDs/DVDs Clothing Electronics Personal care Software Sports equipment Tickets Travel

85 Appendix 3: Purchase Frequencies (cont.) Appendices 78 Books CDs/DVDs Clothing Electronics Personal care Software Sports equipment Subscriptions Tickets Travel Once per 1 or 2 months times a month Books Clothing Software Tickets times a month Books Clothing Electronics Software Tickets

86 Appendices 79 Appendix 3: Purchase Frequencies (cont.) More than 3 times a month Books Software

87 Appendices 80 Appendix 4: Danish Online Shoppers Source: Danmark Statistik Source: Danmark Statistik

88 Appendices 81 Appendix 4: Danish online shoppers (cont.) Source: Danmark Statistik

89 Appendices 82 Appendix 5: Latent Variables Correlations

90 AVE (highlighted) Appendices 83 Appendix 6: Latent Variables Squared Correlations and AVE