The study reported here raises some questions about the conventional wisdom that theInternet creates a “level playing field” for large and small retailers and for retailers with andwithout an established reputation. In our study, consumers recognized differences in sizeand reputation among Internet stores, and those differences influenced their assessments ofstore trustworthiness and their perception of risk, as well as their willingness to patronizethe store. After describing our research methods and results, we draw some implicationsfor Internet merchants.

1.Introduction
    Internet commerce is claimed to reduce the advantages of scale of large retailers,to lower the costs of entering international consumer markets, and perhaps to reducethe strength of established retailers by allowing new merchants to enter and leavequickly [62,24]. But these speculations appear to overlook the importance of the rela-tionship between the consumer and the merchant in this new form of direct marketing.Quelch and Klein [49, p. 70] note that “Trust is a critical factor in stimulating pur-chases over the Internet, especially at this early stage of commercial development”.Keen [37] argues that the most significant long-term barrier for realizing the potentialof Internet marketing to consumers will be the lack ofconsumer trust, both in themerchant’s honesty and in the merchant’s competence to fill Internet orders. Trust isa critical factor in any relationship in which the trustor (e.g., consumer) does not havedirect control over the actions of a trustee (e.g., merchant or store), the decision isimportant, and the environment is uncertain [20,44]. Building on the social psychologyand industrial marketing tradition, we define “trust” as “a trustor’s expectations aboutthe motives and behaviors of a trustee” [21, p. 37].

　　Since the inception of commercial activity on the Web, security has been per-ceived by some to be a significant barrier to the emergence of a consumer mass marketon the Internet [36]. However, Peterson et al. [48] argue that the issue of transactionsecurity is a short-term technological problem. Rather, the substantive long-term issueis: “How do you know whom to trust? In a virtual world, the issue of trust getsmagnified” [24, p.18].
　　The two major questions in this research are:
　　Is trust a significant antecedent of consumer attitudes toward an Internet store andintended shopping behavior?
　　What influences a consumer to have trust in an Internet store?
　　This paper explores the relationship perspective of Internet consumer commerce.At the center of this relationship perspective is the concept of trust. To date, researchon Internet-based consumer behavior has addressed primarily the flow aspects of theshopping experience to build customer loyalty [31], the consumer’s information anddecision support requirements [1], and site attributes intended to increase store traf-fic [62] and sales [41,32]. This paper is complementary to these existing perspectives.In the next section, we review literature on trust and present a research model andhypotheses. The third section outlines the research methodology, the fourth sectionreports the research results, and the final section discusses the limitations and impli-cations of the research.

Conceptual framework: Trust in buyer–seller relationships
　　The focus of this paper is the antecedents and consequences of consumer trust inan Internet store. This paper focuses solely on the consumer’s perceptions of trust ina commercial store on the Internet, not trust in intermediaries or in third parties thatmight mediate between the consumer and the store.
　　Trust is a governance mechanism in exchange relationships that are character-ized by uncertainty, vulnerability, and dependence [12]. Developmentally, relationshipamong parties who have had no prior association is expected to emerge incrementallyand to begin with small actions that initially require little reliance on trust. If theactions are reciprocated, trust tends to spiral upward; if they are not reciprocated, trustspirals downward [55]. Trust has been found to affect the behavior of consumers [51]as well as industrial buyers, even in situations where the buyer’s switching costs arelow [16].
　　In the marketing literature, trust is traditionally studied both in terms oftrust in the salesperson and in terms of trust in the seller organization [46]. Whenthe salesperson is absent from or peripheral to the selling and buying process, as isgenerally the case with Internet stores [41], then the primary target of the consumer’strust is the merchant organization itself [16].Figure 1 presents the research model for the study. The model hypothesizes thatconsumers’ trust in an Internet store affects their willingness to patronize the store.The model is consistent with exchange theory [60] and balance theory [29], as well asthe theories of reasoned action [4] and planned behavior [2,3]. According to exchangetheory, individuals form associations on the basis of trust, and try to avoid exchangerelationships that are likely to bring more pain than pleasure. Balance theory suggeststhat people tend to develop positive attitudes towards those with whom they havesome prior association. The theories of reasoned action and planned behavior assertthat behavior is influenced by behavioral intention, and that intention is determined byattitudes. Attitudes mediate between beliefs and intention, although beliefs can alsohave a direct effect on intention. Also consistent with these fundamental theories, ourmodel assumes that the consumer–merchant relationship is voluntary.
　　Figure 1 qualifies much of the discussion about Internet commerce in both thepopular and academic press, which deems the Internet suitable for consumer marketingwhere sellers are small, geographically dispersed, and unknown to the buyers. Figure1 suggests that willingness to buy from such sellers (i.e., intention to behave) iscontingent on the sellers’ ability to evoke consumers’ trust (i.e., belief). Consumerswill be less likely to patronize stores that fail to create a sense of trustworthiness. Trustis associated with lower perceived risk of shopping at the site, and trust is expected tobe affected by the consumer’s perceptions of the size and reputation of the store.
　　For trust to exist, the consumer must believe that the seller has both the abilityand the motivation to reliably deliver goods and services of the quality expected by theconsumer. This belief may be more difficult for an Internet merchant to engender thanit is for a conventional merchant. For example, selling via the Internet reduces theresources required to enter and exit the marketplace. From the literature on industrialmarketing we learn that trust is built in the eyes of the customer when the seller investsin dedicated resources for the relationship, and when there are frequent one on oneinteractions between the buyer and representatives of the seller’s organization [21]. InInternet commerce, consumers rarely deal directly with any sales people. Essentially,customers depend on an impersonal electronic storefront to act on their behalf [17].A consumer’s trust in an Internet store may therefore be conceptualized as the con-sumer’s trust directly in the store, or the store’s trustworthiness. We will use the terms“consumer trust” and “trustworthiness of the store” interchangeably in this paper.
　　How can an Internet site evoke trust in the eyes of the consumer? Size andreputation have been most frequently named as factors evoking buyer trust towardsseller organizations in traditional industrial buyer–seller relationships [21]. Do per-ceived size and reputation effect consumer trust in an Internet store? Some Internetmerchants certainly seem to think so. They publish stories and customer testimonialson their sites attesting to their reputation, and invest in web-page banners boasting oftheir size:“ Welcome to the earth’s biggest bookstore::: 1.5 million books in print & 1 millionout-of-print books”. ( Amazon.com)“Welcome to CDnow, the world’s largest music store”. (Cdnow.com)

2.1. Perceived size
　　The literature suggests that a store’s size assists consumers in forming their im-pressions regarding the store’s trustworthiness. What matters in forming those impres-sions is the consumer’s perception of the store’s size, rather than the store’s actualsize (for example, its sales volume or the number of products for sale). Thus, in thisstudy, we are interested in the consumers’ perception of the store’s size, and how thisperception affects their trust in the store. In traditional marketing channels, a buyer(i.e., trustor) uses size as a signal that a seller (i.e., trustee) can be trusted [21]. Theperception of large organizational size implies that other buyers trust the organizationand conduct business successfully with it. This experience of others is taken as areason to trust that an organization will deliver on its promises [21]. Large size alsosignals that the firm should have the necessary expertise and resources for supportsystems such as customer and technical services; the existence of these systems en-courages trust [16]. Large size might be also used to signal that the store is able toassume the risk of product failure and to compensate buyers accordingly. In addition,large sellers should be able to control their suppliers, again increasing the perception ofproduct or service reliability and credibility. Finally, large sellers have more resourcesinvested in their business and hence are perceived by a trustor to have more to losethan smaller firms by acting in an untrustworthy way. Although not hypothesized here,the effect of size on trust might be contingent on the merchandise type. The moreuncertainty, ambiguity, or ongoing dependence on the merchant (e.g., for after salessupport) inherent in the type of merchandise, the more importance the consumer mightplace on the store’s resources, and hence the greater the influence of the perceived sizeof the store in determining its trustworthiness. Our simple hypothesis is:
　　H1: A consumer’s trust in an Internet store is positively related to the store’sperceived size.

2.2. Reputation
　　Reputation, like size, is conceptualized as the consumer’s perception of a store’sreputation, where “reputation” is defined as the extent to which buyers believe a sellingorganization is honest and concerned about its customers [21]. Again, the marketing lit-erature argues that reputation is a valuable asset that requires a long-term investment ofresources, effort, and attention to customer relationships; a good reputation also signalspast forbearance from opportunism [57]. Firms with a good reputation are perceivedto be reluctant to jeopardize their reputational assets by acting opportunistically1[14].The costs of untrustworthy behavior are perceived to be higher for firms that alreadyhave a good reputation, particularly if the network of buyers is small or there is a highchance of communication or interaction among the buyers [8]. In the industrial buyercontext, the seller’s reputation has been positively related with the buyer’s trust inthe seller [6,27]. In the Internet marketing context, Quelch and Klein [49] argue thatInternet consumers will favor sites that represent a merchant with which the consumeris already familiar from traditional channels. Lohse and Spiller [41] speculate that thereputation of the physical store will influence the perceptions of an online site. Hence,we hypothesize:
　　H2: A consumer’s trust in an Internet store is positively related to the store’sperceived reputation.
　　Perceived size and reputation are expected to be related. As noted above, largerstores might be perceived as being more reputable. Because of natural growth limita-tions, larger stores might have been around longer, and longevity might increase thechances that the consumer has had prior experience with the merchant in other channelsor has heard of the merchant in the context of the new channel. Prior association tendsto increase positive affect as well as positive cognition of the other party [29]. Hence,the larger the store’s size, the more likely the consumer might associate a favorablereputation with the store. Equally, stores having favorable reputations might attractmore business than stores offering similar merchandise and prices, but not having fa-vorable reputations. Hence, the more the consumer associates a favorable reputationwith a store, the larger the store might become. Therefore we hypothesize:
　　H3: An Internet store’s perceived size is related to the store’s perceived reputation.

2.3. Risk perception and attitudes
　　Trust is interwoven with risk [45], and both are based on perceptions [28]. Al-though some level of risk is inevitable if there is a need to trust, trust has also beendefined as the expectation that an exchange partner will not engage in opportunisticbehavior [12]. Hence, one of the consequences of trust is that it reduces the con-sumer’s perception of risk associated with opportunistic behavior by the seller [27].Risk perception refers to the “trustor’s belief about likelihoods of gains and lossesoutside of considerations that involve the relationships with the particular trustee” [44,p. 726].
　　High levels of trust by buyers have been found to stimulate favorable attitudesand behavior [5,51]. Macintosh and Lockshin [43] found that a consumer’s trust in astore impacted the consumer’s attitudes towards that store. Exchange partners who aresocially bonded have highly favorable attitudes towards each other [34]. We argue thatthis is because trust reduces the perceived risk of being mistreated by the store [6], andthe low perception of risk in turn influences the attitudinal orientation of the consumertoward the store. Hence, we hypothesize that the following will take place:
　　H4A: Higher consumer trust towards an Internet store will reduce the perceivedrisks associated with buying from that store.
　　H4B: Higher consumer trust towards an Internet store will generate more favorableattitudes towards shopping at the store.
　　H4C: The lower the consumer’s perceived risk associated with buying from anInternet store, the more favorable the consumer’s attitudes towards shopping at thatstore.

2.4. Purchase intentions
　　The theory of reasoned action (TRA) and the theory of planned behavior (TPB)presume that volitional behavior is determined by intentions to act (see, for example [2,4,9]). A major determinant of intentions, in turn, is the actor’s attitudes towards thebehavior2. TRA and TPB have been evaluated and supported in many contexts [2],including IT usage behavior [59]. Internet shopping behavior shares the volitionalnature of the phenomena that TRA tries to explain and predict. Thus, for the purposeof comparing various Internet shopping sites, we assume that the degree to whichpeople express their intentions to buy from a certain site relative to other sites is areasonable predictor of actual purchase behavior from this site relative to the others.According to the TRA, a shopper’s intention to buy is preceded by the shopper’sattitudes toward the purchase. Thus, we expect to find positive relationships betweenattitudes towards an Internet store and willingness to buy from that store.
　　H5A: Favorable attitudes towards an Internet store will increase the consumer’swillingness to purchase from that Internet store.
　　The theory of planned behavior also suggests that a consumer may be willingto buy from an Internet store which is perceived as low risk, even if the consumer’sattitudes towards that merchant are not highly positive. Conversely, a consumer maynot be willing to buy from a merchant perceived as being high risk, even in thepresence of positive attitudes towards that merchant. The direct influence of perceivedrisk on intention is related to the notion of perceived behavioral control in the theoryof planned behavior [2,3]. Perceived behavioral control reflects the degree to which anindividual feels that successfully engaging in the behavior is completely up to them.In the Internet shopping context, the perceived risk associated with shopping in thestore may reduce the consumers’ perception of control, and the extent to which thisoccurs might negatively influence willingness to shop. In summary, we hypothesizethat perceived risk helps shape attitudes but also has an independent, direct influenceon the intended behavior.
　　H5B: Reduced perceived risks associated with buying from an Internet store willincrease a consumer’s willingness to purchase from that Internet store

2.5. Control variables
    Although the study focuses on the antecedents and consequences of consumertrust in an Internet store, previous research suggests that several other variables mightaffect attitudes and willingness to shop in an Internet store. Studies involving TRAhave shown that past behavior has both a mediated effect through attitudes and a directeffect on intentions and behavior [10]. A consumer’s past experience on the Internetin general, or shopping on the Internet specifically, might have generated knowledgeand consequences that reinforce the consumer’s behavior and shape and moderate theconsumer’s beliefs, attitudes, and willingness to shop in Internet stores. Prior studieson electronic shopping systems have found the predisposition towards computers ingeneral to be a significant determinant of adoption and use of new forms of shopping(e.g., [23,38,39,52,53]). Research on new technology adoption has consistently foundattitudes towards computers to be a significant determinant of adoption and use, al-though some studies have found the relationship to weaken as users gain experiencewith the technology [61]. In traditional retail channels, shopping orientation has beenfound to be among the most influential predictors of consumer patronage behavior [18].Research on direct shopping modes (e.g., catalog shopping) suggests that those whogenerally do not have positive shopping orientation will have more positive attitudestowards direct shopping modes [19,33], although some studies have found just theopposite [54]. Finally, the perceived risk of a store might be attributable to a con-sumer’s general risk attitudes towards the Internet. Hence, as control variables, weincluded in the study the frequency of Internet usage, the frequency of Internet shop-ping, shopping enjoyment in general, attitudes towards computers, past direct shoppingexperience (e.g., catalog and TV shopping), and web-shopping risk attitudes.

Methodology
　　To assess the research model in figure 1, we used an experiential survey approachto collect data from a group of undergraduate and MBA students in Australia. Theparticipants were recruited in two ways. First, one hundred and twenty students re-sponded to announcements made in undergraduate and MBA information systems andcomputer science courses and to posters placed on campus noticeboards. This grouptook part in supervised sessions held in a computer lab at the Australian universitywhere they were enrolled. The lab was equipped with Pentium workstations, 17-inchmonitors, and high speed access to the university backbone network. Each of theseparticipants was paid A$10 (about US$7.00 at the time of the sessions) to spend 2 hoursTable 1Demographic profile of study participants.Sex femalemale66118Average age 22.35range = 18 to 48median = 20mode= 18Average years of post-secondaryeducation completed2.45range = 1to5(note: in Australia most undergraduatedegrees take three years to complete;joint or double degrees typically takefive years)Average years of workingexperience3.86range = 0to30median = 277% had at least 1 year experienceCountry of residence AustraliaIndonesia/MalaysiaOther78.3%14.6%7.1%Home country AustraliaIndonesia/MalaysiaHong KongOther47.822.310.419.5Average number of countriesvisited5.0Median= 4.0Range = 0–3093% traveled to at least one othercountry, 87% to at least two countriesengaged in a variety of World Wide Web shopping activities and to complete severalquestionnaires; in addition, random drawings were held to award one undergraduateand one MBA student a prize of A$100.
　　The second group consisted of sixty-four students at another Australian universitywho completed the shopping activities and questionnaires voluntarily at the request oftheir instructors in electronic commerce courses. These students used facilities eitherin their university computer lab or at home. The volunteers received no pay forparticipating, but were given a follow-up briefing by one of the researchers on thegoals and methods of the experiment.
　　Overall, then, the study included 184 participants. The demographic profile ofthe participants is summarized in table 1 and the sample’s World Wide Web experienceis summarized in table 2. The demographic profile reflects the multi-culturalism ofAustralian tertiary education and the relative youth of the sample. Almost all partic-Table 2Degree of Internet experience among study participants.% with Internet access 95.1%% Accessing Internet from ::: home 58.9%office 5.2%university 94.8%other 0%Prior use of Web browsing software 98%Average months of prior use 18Frequency of browser usage Once 0.6%A few times 9.4%About once a month 3.9%A few times a month 10.0%Once a week 7.8%A few times a week 47.8%Every day 20.6%Attitude towards browsing the Internet Very positive 17.2%Positive 57.8%Neither 22.2%Negative 2.2%Very negative 0.6%Prior purchases through the World 10.9% (42% of which only once)Wide Webipants had previous exposure to the Internet although relatively few had engaged incommercial transactions on the Internet.

3.1. Experimental tasks
　　Web pages were created that outlined the shopping tasks and provided linksto various shopping sites. Pointers to the shopping sites and instructions forstudy participants are available at shopping/page1.htm.
　　The participants performed four shopping activities: (1) selecting and buying abook as a gift for a friend, (2) buying a specified book for a course, (3) planninga holiday trip to Helsinki, Finland, and (4) planning a work-related trip to Sydney,Australia. For each activity, the participants were offered four different Internet sites.For the book buying activities, the participants were offered a choice of Amazon.com,the UK-based Internet Bookshop, Glee Books in Sydney, and DA Information Servicesin Melbourne, while for the travel activities the choices were Finnair, the Australianairline Qantas , the well-known Australian travel agent Flight Centre, and the Internet-based travel serviceTravelWeb. It was recommended, but not required, to visit eachof these sites. After each of the four shopping activities, the participants were askedto fill out a brief questionnaire describing their experiences. Participants were toldthat they were not required to make a purchase, although they could do so if theywished using their own credit card. The shopping tasks were meant to be typical ofthose that a consumer might routinely perform. We included both book and travel sitesbecause these two merchandise categories are argued to be suitable for Internet-basedretail commerce [48]. The selection of two different types of stores also allowed usto explore the effects of store type on the model variables. It might be argued thatthe travel task is not realistic for young students with little or no income. However,as the data in table 1 show, the study’s sample is made up mostly of young personswith both work and travel experience. We believe that these people are experienced atbuying both books and travel packages and are therefore well suited to the shoppingtasks included in the research.
　　The participants’ assessments of perceived size and reputation were based ontheir own impressions of the sites as well as on the summary page that was availablefor each site. The summary page captured sales figures, number of products available,location, and the date when the business had been founded (see appendix B). Thisstudy did not systematically control “size” and “reputation” of the store sites althoughthe summary information was intended to generate perceptions of size and reputation.All of the information on the summary pages was extracted from the sites themselvesand summarized in the separate web site that participants visited before visiting thestore itself. For example, among the book sites, only Amazon.com and the InternetBookstore divulged their sales (US$15 million and US$ 850,000 respectively); theother two sites did not. Amazon.com, the Internet Bookstore, and DA InformationServices listed the number of available titles, Amazon claiming over 2.5 million titles,the Internet Bookstore nearly a million, and DA Information Services “hundreds ofthousands”. Glee Books did not list the number of titles.