• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

The Effects of Buyer and Product Traits With Seller Reputation on Price Premiums in E-Auction

Posted on: Saturday, 22 October 2005, 03:00 CDT

By Kim, Yongseog

ABSTRACT

This paper examines the effects of buyer and product traits along with seller reputation on price premiums. Three personal traits studied are based on buyers' shopping pattern (analytical vs. impulsive buyers), propensity to trust (trusting vs. skeptical buyers), and attitude toward auction (competitive vs. noncompetitive buyers). Product characteristics - price and homogeneity of quality - are also considered. Data analysis in this study shows that while impulsive buyers consistently pay high premiums, analytical buyers pay greatly different premiums depending on sellers' reputations. However, buyers' propensity to trust and attitudes toward auctions do not significantly contribute to the difference of price premiums when information about sellers' reputations is explicitly provided. Even competitive buyers do not pay higher premiums to sellers of bad reputation for just being a winning bidder. In regard to product characteristics, buyers pay higher premiums for expensive products or products with various options. However, buyers do not pay high premiums for an expensive product with very homogeneous quality.

Key words: Price premiums, e-auction, buyer traits, product traits, seller reputation, trust.

INTRODUCTION

Electronic commerce (e-Commerce) has been very successful in the last decade and is expected to grow continuously in the next decade. The success of e-Commerce is mainly due to the rapid development of the Internet and, more fundamentally, information technology (IT). According to the view of media-determinism (28), the Internet changes the pattern of information flow and creates virtual business communities governed by new rules and interactions among participants. That is, the Internet revolution transforms the way we produce, trade, and consume by creating new value-chains and reorganizing distribution channels (9). However, the acceptance and diffusion of new technologies are dependent on also social necessities and economic motives of community members. Therefore, the success of e-Commerce should be understood in terms of the interaction of IT technology and societal response.

One of the most distinguished features of e-Commerce is the impersonal nature of trading parties. Since most online parties can easily remain anonymous, they often find themselves in a situation in which they initiate online transactions with unknown parties. Online shoppers also have a very limited ability to physically inspect products before purchasing and a limited access to information regarding the true quality of the product. Therefore, trust among trading parties becomes very crucial in the e-Commerce environment, and, under extreme circumstances, buyers' mistrust of online products and services can eventually jeopardize online markets. In an effort to boost trust in online markets, many online services have emerged. A few examples include information provision services (e.g., Bizrate.com and feedback forums at various auction sites), product review services (e.g., Epinions.com), escrow services (e.g., Escrow.com), product appraisal services, insurance services, and so on. In particular, an online feedback forum is unique in the sense that it does not involve third-party partners or services. By allowing one party of a business transaction to publicize his or her transaction experiences with other parties by rating the quality of the service provided by other parties, online feedback mechanisms help build trust among potential trading parties in virtual markets.

Buyers' feedback forums are good examples of reputation systems that influence the overall confidence and behavior of buyers who have no prior experience with a specific seller by spreading information about sellers' behavior. When consumers trust in an Internet store will perceive very low risks involved with online transaction, the perceived low risk will lead to more favorable attitudes toward shopping. However, since perceived risks and reputations are subjective in their nature, buyers with different personal traits may perceive different levels of risks in transactions with the same seller and product. Therefore, personal attributes can affect the price that buyers are willing to pay. It is also possible that the same buyer perceives different levels of risks from different products. For example, buyers who believe it is riskier to buy an expensive product than an inexpensive product are willing to pay high premiums to reliable sellers to assure the quality of the product. Out of many possible research directions in online auctions (30), this study aims to investigate the effects of personal and product traits on price premiums. The specific questions that this study examines are:

(1) How do buyers' personal traits affect perceived reputations of sellers and price premiums? (2) How do product characteristics affect perceived reputations of sellers and price premiums? and (3) What are the mixed effects of personal and product traits on the perceived reputations of sellers and price premiums?

The remainder of the paper is organized as follows. section 2 briefly reviews trust and reputation systems and the relationships between reputation and price premiums. section 3 introduces a research model that examines the effects of three factors - personal and product traits and sellers' reputations - on price premiums in electronic transactions. In Sections 4 though 6, the research methodology and results of hypothesis testings are presented in detail. Section 7 concludes the paper and provides suggestions for future research directions.

TRUST, REPUTATION SYSTEMS AND PRICE PREMIUMS

Trust and Propensity to Trust

Understanding the purchasing patterns of online buyers is essential to develop successful e-commerce marketing programs and improve the relationship with customers (10, 31, 42). In particular, due to the uncertainty and impersonality of online transactions, establishing trust among business partners is crucial for the successful conduct of e-commerce. According to a study (15), almost 95% of online shoppers have declined to provide personal information because many of them (68%) do not trust online merchants collecting the data. Hence, initiating, building, and maintaining trust between buyers and sellers are widely believed to be the key drivers of success for most online firms. However, developing trust in online transactions is substantially more difficult than in traditional relationships because participants are often physically separated and have short history of transactions.

In order to explain the origin, transference, and effects of trust, many theories including balance theory (13), social network theory (12), exchange theory (40), reasoned action theory (2), institution-based trust theory (46), and planned behavior theory (1) have been presented. While individuals tend to avoid relationships that are likely to bring more pain than pleasure (exchange theory), they tend to develop positive attitudes toward those with whom they have previous business engagement (balance theory). According to the theories of reasoned action and planned behavior, behavior is influenced by behavioral intention that is determined by attitudes. Therefore, attitudes mediate between beliefs and intention, although beliefs can also have a direct effect on intention. Institution- based trust researchers maintain that trust reflects the security one feels about a situation because of guarantees or other impersonal structures. According to social network theory, informal channels of communication (e.g., word-of-mouth referral) from one buyer can significantly influence other buyers' trust when the services are particularly complex and difficult to evaluate.

Three sources of trust (familiarity, calculativeness, and values) are also discussed in (45). Among three sources of trust, calculativeness is believed to be the most prevalent source of trust in online auction market (3) because most online auction market transactions are one-time transactions. When calculativeness is the main source to form trust among business partners, a rational judgment about costs and benefits of the other party plays an important role to form one party's trust.

Individual characteristics - propensity or disposition to trust - also influence buyers' trust in online sellers (24). Those with high levels of propensity to trust have a belief, expectancy, or feeling that others can be trusted in general (18). Therefore, it is typical that the higher the buyers' propensity to trust, the higher the level of initial trust in seller. According to McKnight et al. (26), propensity to trust influences buyers' trust both directly and indirectly through its effect on institution-based trust. The propensity to trust is dependent on two components: faith in humanity (where a person believes that others are usually dependable) and trusting stance (where a person believes that business transactions with others who are reliable will be beneficial). In particular, the effects of propensity to trust are manifest when a trust level is initially formed (27). Therefore, individual characteristics can significantly affect shopping behavior when firsthand experience and knowledge of the other party is not available (36). However, when consumers have experienced a web site, individual dispos\ition to trust, while still an important influence, will not directly affect trusting beliefs or trusting intention. Rather, the impact of this variable will be fully mediated through institution-based trust and perceived site quality.

Reputation and Reputation Systems

Many online services have emerged to provide information on sellers and products and to boost trust in online communities (25). For example, buyers can check sellers' reputations at Bizrate.com and online feedback forums at eBay or Amazon.com. Buyers can also get detailed information about products at Epinions.com. In online feedback forums, one party can publicize her experiences with the other party by posting comments and by rating the quality of the service of the other party. Online feedback forums, also called reputation systems, aggregate and distribute feedback about participants' transaction history. Ideally, a good service provider will receive much positive feedback that eventually enhances her reputation. Further, since a good reputation is likely to increase buyers' trust in sellers, sellers with good reputations enjoy higher prices. Therefore, reputation systems provide an incentive for good performance by sellers.

However, currently available feedback systems have shortcomings. For example, the feedback score of a buyer or a seller on eBay is the sum of positive feedback ratings less the sum of negative ratings. Since the overall rating cannot differentiate feedback ratings that a participant has earned as a seller from ratings as a buyer, users can abuse feedback systems (17). Further, most reputation systems are based on feedback that users voluntarily provided after transactions. Therefore, many users usually attempt to take a free ride without providing their own feedback (17), though half of the buyers on eBay, and three fifths of the sellers provided feedback (34). In particular, most feedback is "Pollyanna" (disproportionately positive) (33). Further, there was a high correlation between buyers' and sellers' feedback, suggesting that the players reciprocate and retaliate (33).

Changes in reputation systems also make it hard to use feedback systems for analysis. Initially, in eBay any user could leave comments and a numeric rating (+1 for positive, 0 for neutral, or - 1 for negative) about any other user. However, beginning February, 1999, all negative feedback had to be tied to a particular transaction, and beginning February, 2000, all feedback had to be tied to a transaction: i.e., only the seller and winning bidder can leave feedback about each other. However, the most serious problem is that many reputation systems are not interchangeable because of different scoring systems and ownership issues. Despite their shortcomings, feedback systems are still valuable resources to online buyers and researchers by quantifying reputation of sellers and buyers (17).

Price Premiums and Reputation

There have been many studies about the economic and financial role of price premiums. Economics literature defines a price premium as a price that yields above-average profit (19, 37). According to an economic model (19), the provision of price premiums is one solution to the problem of quality debasement when product quality is unobservable prior to purchase. That is, price premium provides profit-seeking sellers a monetary incentive to deliver high-quality products by offering a price that yields greater long-term profit than profits available by selling low-quality products but claiming high quality (32). Studies have also demonstrated that the price premium represents the return on the initial investment of firms in reputation (37). Many other studies confirm a positive relationship between sellers' reputations and price premiums. In Brynjolfsson and Smith (6), two different sources of price dispersion among Internet retailers (on average, price dispersion rates for books and CDs are 33% and 25%, respectively) were identified: heterogeneity in buyers' awareness and heterogeneity in retailers' branding and trust. Houser and Wooders (16) found that sellers' reputations had a statistically significant effect on the final price of Intel Pentium III 500 MHz processors. They also observed that reputation effects were economical in the sense that increasing the number of positive comments proportionately increased the winning price. Another study (17) showed that sellers with higher feedback scores received an average 8.3% price premium over other sellers, although higher reputation advantages did not yield commensurately higher price premiums.

The relationships between negative comments and selling prices have been also studied. For example, Lucking-Reiley et al. (21) found a seller's negative feedback ratings have a much greater effect (about four times) than positive feedback ratings from a regression model of 461 auctions of Indian head pennies. Lynch and Ariely (22) showed how the amount of product information provided to customers can affect price competition and increase customer loyalty in electronic channels. Ba and Pavlou (3) demonstrated that appropriate feedback mechanisms induced calculus-based trust and generated price premiums for reputable sellers. In addition, they also showed that buyers paid a higher premium to sellers with a certain level of reputation for expensive products. Another study (29) showed that a seller's positive feedback ratings have a significant effect on the final price, but including a photographic image of the item being auctioned does not.

CONCEPTUAL DEVELOPMENT

Personal Traits and Premiums

Individual characteristics - propensity or disposition to trust - have been known to influence buyers' trust in online sellers (24). According to McKnight et al. (26), propensity to trust influences buyers' trust both directly and indirectly through its effect on institution-based trust. An individual's disposition to trust is dependent on good faith in humanity (assuming that others are usually upright and reliable) and trusting stance (assuming that trusting others will bring better outcomes) (27). In particular, the effects of propensity to trust are manifest when a trust level is initially formed. When firsthand experience and knowledge of the other party is not available, propensity to trust is likely to have a significant effect on a person's initial trust (27). Typically, the higher the buyers' propensity to trust, the higher the level of initial trust in sellers (18).

However, disposition to trust is not the only source that buyers use to form trust. Three sources of trust - familiarity, calculativeness, and values - have been identified and presented in previous studies (11, 45). According to Ba and Pavlou (3), calculativeness is the most prevalent source of trust in the online auction market. In calculativeness-based trust, a rational judgment about the costs and benefits of other parties plays a critical role to form trust. Rational buyers utilize all available information resources to minimize risks inherent in the current transaction with a specific online vendor. For example, buyers who completed previous transactions with a seller formulate their trust in the same seller using their own experience in addition to their disposition to trust. Further, if they have access to other information sources (e.g., reputation systems) that provide other buyers' experiences with the same seller, they can use this additional information to adjust sellers' credibility.

In general, buyers will have no additional information for new sellers or sellers who have new pseudonyms. For these sellers, buyers' propensity to trust will solely determine trust in sellers and, therefore, price premiums that they are willing to pay. However, for sellers with whom buyers have firsthand experience or reputation information, the effect of propensity to trust on premiums will be marginal so that price premiums of trusting buyers will not be significantly different from premiums of skeptical buyers. This is mainly because propensity to trust is one of many trust determinants even for trustful buyers.

Hypothesis 1 (H1): The difference of price premiums between trusting and skeptical buyers for the same pair of sellers will not be significant when information about sellers' reputations is available.

Note that Hl is completely different from a well known hypothesis that buyers' propensity to trust significantly affects price premium when information about sellers reputation is available (3). The main focus of Hl is to compare the difference of price premiums of buyers with different levels of propensity to trust for the same pair of sellers. Therefore, the fact that H1 is confirmed does not necessarily mean that either trusting or skeptical buyers are not willing to pay higher premiums to sellers with higher reputation.

Clustering is a process to group records into a fixed number of groups so that intra-group similarity is maximized and intergroup similarity is minimized. Online retailers (42) and online shoppers can be grouped based on their demographic information or, more importantly, based on shopping patterns. For example, Turban et al. (41) presented six different types of online shoppers and Bapna et al. (5) categorized auction bidders into evaluators, participators, and opportunists based on bidding strategies. Based on buyers' shopping patterns, this paper considers two different types of buyers: analytical and impulsive buyers. In this study, analytical buyers are shoppers who purchase products or services after carefully comparing prices and specifications from multiple online sites or traditional markets. They also do not purchase products that they did not plan to buy before or products that they do not need right away. In contrast, impulsive shoppers purchase products without comparing prices or even without reviewing specifications of products carefully. In general, analytical buyers c\an purchase products at cheaper prices than impulsive buyers. This is because analytical buyers have information about the right price of products, and they will not pay more than they believe is the right price. Therefore, the difference in final prices partially comes from information asymmetry between analytical and impulsive buyers. However, when analytical buyers are very quality-sensitive, they are willing to pay higher premiums to credible sellers.

By providing the same amount of information about product price and sellers' reputations, this study intends to measure the difference of final prices solely attributable to personal characteristics of buyers. It is theorized that even when analytical buyers have the same information as impulsive buyers, their rational and analytical characteristics will make them more carefully analyze the information. Therefore, they will rationally determine price premiums in proportion to the perceived risks. In contrast, impulsive buyers tend to irrationally overestimate credibility of sellers with excellent reputation by paying higher premiums than necessary. Further, impulsive buyers may not fully penalize sellers with bad reputation by not carefully analyzing their credibility and, therefore, pay higher premiums than analytical buyers. In an extreme case, impulsive buyers may not consider sellers' information at all, and they are more likely to be a victim of "winner's curse" (8) by paying a price higher than the rational value of the item. This subtle difference will give analytical buyers a wider gap of price premiums between reliable sellers and unreliable sellers than impulsive buyers. These observations are summarized as follows:

Hypothesis 2a (H2a): Impulsive buyers tend to overestimate the credibility of trusted sellers and pay higher premiums than analytical buyers.

Hypothesis 2b (H2b): Analytical buyers are likely to fully consider the - reputation of unreliable sellers in determining price premiums and pay lower premiums to unreliable and new sellers than impulsive buyers.

Hypothesis 3 (H3): Analytical buyers show a greater difference of price premiums for a pair of trusted and unreliable sellers than impulsive buyers do.

Online shoppers also show differences in risk preference and quality preferences. Rivalry is a natural human attribute observed in the world of adults (e.g., competition in sports, business, research, and so on) and young children (e.g., sibling rivalry). A certain level of rivalry can motivate participants of games to make games more interesting and to make participants enjoy win-win games no matter who actually wins. However, extreme rivalry makes participants stick to winning itself rather than enjoying the whole games process. In particular, in a typical win-loss game like an online auction, all participants compete against each other for the same auction item, and other bidders are regarded as rivals (14).

In this study, online auction participants are grouped into competitive and non-competitive buyers based on their attitude toward auctions. Competitive buyers are those who love to take risks and enjoy "winning" the auction by outbidding the highest bid, which, in turn, makes them happy (23). Therefore, they will end up paying high premiums. In contrast, non-competitive buyers typically use the auction system as a convenient and economical way of shopping. Non-competitive buyers will stop raising their bid if the current winning bid is higher than their own reserved bidding price for a product. Other things being equal, competitive buyers are likely to pay higher premiums than non-competitive buyers for the same product.

Hypothesis 4 (H4): If the same amount of information is provided to competitive and non-competitive shoppers, highly competitive buyers are likely to pay higher premiums than non-competitive shoppers for the same product.

Product Traits and Premiums

As discussed earlier, buyers are willing to compensate trustworthy sellers with price premiums to assure safe transactions and to reduce transaction risks in an uncertain environment. Therefore, online transactions involving riskier products should result in higher price premiums for reputable sellers. Product price is one of the most obvious risk determinants in business transactions. Buyers perceive much higher risks from a transaction involving an expensive product (say, > $1; 000) than a transaction involving an inexpensive product. Other things being equal, the more expensive a product is, the less incentive the seller will have to be honest. This is because sellers have a greater chance of maximizing their shortrun profits by cheating the quality of expensive products. Since buyers are confronted with higher potential for loss, they are looking for more reputable sellers and will pay higher price premiums to trustworthy sellers (3).

Hypothesis 5 (H5): The positive relationship between sellers' reputations and price premiums is stronger for expensive products than for inexpensive products.

Another important attribute of products analyzed in this study is how homogeneous products are. The homogeneity of product quality depends on the current state of technology, government regulation, or industry norms (32). Typically, a complex and sophisticated system that consists of many parts would result in greater variation in product quality. Since buyers do not have enough information or expertise to assess the quality of products, they perceive high risks in a transaction involving products with a greater variation of quality. Therefore, buyers are willing to pay higher price premiums for products with a greater quality variation.

Whether or not a product is actually homogeneous is not a key determinant of price premiums. It is the buyers' perceived degree of homogeneity in the product quality (34). For example, buyers may think that sellers can use low-quality parts for a highly complex system that buyers cannot easily detect problems. If this is the case, even though a specific product is highly homogeneous due to industry standardization, buyers will seek sellers with a better reputation to reduce quality-related risks and to pay high premiums if necessary. However, if the product is relatively homogeneous in quality (or if a buyer thinks so), the buyer is likely to perceive minimal or no transaction risks at all. Therefore, buyers have no incentive to pay high premiums even for sellers with trusty reputations.

Hypothesis 6 (H6): The magnitude of price premiums will be positively related to the perceived degree of variation in product quality.

Figure 1 shows a research model of this study illustrating how three different domains - buyers, sellers, and products affect price premiums. Buyers are further refined based on their shopping behaviors, propensity to trust, or attitudes toward auctions. Sellers and products are refined based on reputation level and product characteristics (i.e., price and homogeneity), respectively. Price premiums are determined as an output of interaction of three domains.

PROCEDURES AND EXPLORATORY ANALYSIS

Measures and Procedures

The study is conducted as an online survey to explore the effects of personal and product attributes on price premiums in online auction markets. In addition, the study also examines interactive effects of personal and product attributes along with sellers' reputation on price premiums. Because of the large number of potential influences on trust in a Web-based business, isolating the influence of one or two factors requires a controlled setting through experimental or survey approach.

In this paper the effects of personal attributes are analyzed by considering three major factors: shopping behaviors, propensity to trust, and attitude toward auction. Therefore, buyers are divided into analytical vs. impulsive buyers, trusting vs. skeptical buyers, and competitive vs. non-competitive buyers. Based on reputation, online vendors are also divided into four different types: sellers with excellent, good, and bad reputation and new sellers. For notational convenience, they are denoted as S^sub exc^, S^sub good^ S^sub bad^ S^sub new^.

Four different products are carefully chosen by varying price and homogeneity to analyze the effect of product attributes. First, two expensive products, Windows Server S/W in CD (a product with very limited variety of quality) and Canon Digital Camcorder (a product with a variety of quality) are selected. In general, two products will have different price tags in a real situation. However, they are priced at the same price ($1200) to measure the pure effect of product homogeneity on price premiums while excluding the effect of price difference on price premiums. It is assumed that while all sellers provide the same quality of Windows Server CD, sellers with bad reputations may easily replace some parts of a digital camcorder with bad parts. Therefore, products with various options have higher risks. In the same way, two inexpensive products, a Movie DVD and a Motorola 56K PCI Speakerphone Modem, are chosen and priced at the same price ($20). A modem is regarded as a more heterogeneous and risky product. However, buyers of homogeneous products are also exposed to sellerrelated risks including no-refund policy and delayed shipping. The four products are summarized in Table 1.

FIGURE 1

Research Model with Three Domains

TABLE 1

Descriptive Information of Four Products

In this study, the subjects were first asked to complete a questionnaire designed to measure their disposition to trust, shopping patterns, and attitudes toward auctions. Whenever possible, questionnaire items were adapted from validated measures used in prior research (27, 36). The instrument was pretested on three faculty members, four graduates, and two undergraduates to ensure the clarity and content validity of the items. Based on the results of the pretest, minor modifications were made tomake items clear. The final list of items are listed in Appendix A and Appendix B.

The final version of the questionnaire was administered to students at Utah State University via the author's Web site. The participants were first presented with a Web page that described the main idea of the online survey. On the next page, they were asked about their personal traits including shopping patterns and attitudes toward auctions. Once the participants finished the first part of online survey, they were presented with four different products available from four different sellers. The participants were also given information about the trustworthiness of each seller and risks involved with each different product. Then the participants were asked to choose the highest price range they were willing to pay for a given product associated with each seller. Ten price ranges are provided to participants, and the lowest and highest price ranges corresponds to from 30% to 120% of the original price. The chosen price range is converted into 1 to 10 scales for further analysis.

Exploratory Factor Analysis

A total of 124 participants participated in the online survey and, of the 124 participants, 11 did not complete the survey. Therefore, the following results are based on the analysis of 113 participants who completed the survey. The composition of male and female participants is 65% and 35%, respectively. They are evenly dispersed from freshman to graduate standing, but many of them are sophomore (35) and junior (35) standing. Exploratory factor analysis along with Cronbach's α is used to ensure that the instruments are reliable and have reasonable construct validity (7). The factor solutions and Cronbach's α for the measurement model are summarized in Table 2.

Exploratory factor analysis is conducted using orthogonal (Varimax) rotation to ensure high loadings on hypothesized factors and low cross-loadings. Note that only loading values above 0.3 are shown for clarity in Table 2. For factor analysis all eigenvalues associated with the factors are set to be greater than unity and, as a result, the ten items in the questionnaire are reduced to three principal constructs: analytical, propensity, and competitive factors.

All items are loaded on their hypothesized factors, and the item loadings are all significant at the 0.05 level for each construct. This provides evidence of convergent validity (4). Overall, the factor solution has an acceptable loading pattern and explains 73% of the total variation in the data. High Cronbach's α also indicates that the constructs are reliably measured and are adequate for hypothesis testing. Therefore, the items corresponding to each variable are averaged to create an overall measure for each factor variable. Table 3 presents the means, standard deviations, and inter- correlations of an overall measure for each identified factor

TABLE 2

Results of Factor Analysis with Varimax Rotation

TABLE 3

Correlation Matrix Between Factors

PERSONAL TRAITS AND PRICE PREMIUMS

Overall Effect of Personal Traits on Price Premiums

One of the main research goals of this study is to investigate whether buyers' personal traits can significantly affect levels of price premiums that they are willing to pay for sellers with various reputations. In this study, three types of personal attributes are studied based on buyers' propensity to trust others, shopping patterns, and attitudes toward online auctions. As a pre-analysis step, the average values of price premiums that various types of buyers are willing to pay to various sellers for four different products are computed.

Figure 2(a) shows the average values of price premiums by various types of buyers for P^sub win^. The thresholds to categorize buyers are chosen after considering the average values in Table 3 and the number of buyers in each group as follows. First, online buyers are grouped based on the average value of items to measure propensity to trust. The threshold value for this grouping is set to 4.0, and 70 and 43 trusting and skeptical buyers are identified, respectively. The threshold is determined after considering the average value (4.01) and the number of buyers in each group. Buyers are also grouped based on their competitiveness. Considering the average value of competitiveness (2.65), the threshold value is set to 3.2, and 36 competitive and 77 non-competitive buyers are identified. Finally, the effect of buyers' shopping patterns is analyzed by categorizing online buyers into analytical and impulsive buyers. A buyer is regarded as either analytical or impulsive based on the average score (5.66) of the items designed to measure the analytical shopping factor. With the threshold value (6.0), there are 62 analytical buyers who more carefully compare the suggested price to other prices on the Web than the remaining impulsive buyers.

Figure 2(a) shows that, for P^sub win^, all buyers are willing to pay higher premiums to sellers with higher reputations, confirming findings in Ba and Pavlou (3). Further, none of the differences of price premiums between two buyer groups (trusting vs. skeptical and competitive vs. non-competitive buyers) for all sellers was not significantly different at α = 0.1. However, the difference of price premiums between analytical vs. impulsive buyers for S^sub bad^ was significantly different at α = 0.1. This exception partially supported hypotheses H2a and H2b, stating that impulsive buyers pay significantly higher price premiums to S^ than analytical buyers. Similar patterns were observed for three other products - P^sub cam^, P^sub dvd^, and P^sub mod^ - and hence results are not shown. Therefore, H2a and H2b are partially supported.

Personal Traits and Difference in Price Premiums

Trusting vs. Skeptical Buyers. The first hypothesis (H1) in this study posits that the effect of propensity to trust on the price premium difference (PPD) will not be significant when information about sellers' reputations is accessible to rational buyers. In order to validate H1, the PPD values of each buyer for a given pair of sellers are computed and shown in Figure 2(b) based on information in Figure 2(a). The PPD is the difference in price premiums that an online shopper would pay for sellers with different levels of reputation. For example, PPO(S^sub exc^, S^sub bad^) of a buyer^sub i^ represents the difference in price premiums that a buyer^sub i^ is willing to pay for S^sub exc^ and S^sub bad^. Since there are four different types of products, PPD is computed for each product category. Note also that PPD can be computed and compared among all possible pairs of sellers. However, only PPDs that include S^sub bad^ as a base-line seller are computed and shown in Table 4. The differences in PPDs between the two buyer groups are compared using paired t-tests (one-tailed).

FIGURE 2

Buyer Types and Price Premiums

TABLE 4

Propensity to Trust and Price Premium Difference (PPD)

The series of t-tests indicates that PPDs of two buyer groups over all four products are not significantly different at α=0.1, supporting a hypothesis (H1). Typically, buyers' propensity to trust is an important determinant of trust in sellers and, therefore, price premiums. However, studies have shown that the effects of propensity to trust are manifest when a trust level is initially formed (27). Therefore, in situations in which buyers have enough information about the reputation of online vendors, propensity to trust is unlikely to have a significant effect on PPDs between trusting and skeptical buyers. Note that in this study information about sellers' reputations is explicitly provided to the participants. Therefore, buyers will be more dependent on their analytical reasoning for determining price premiums after studying different types of product and inherent risks involved with information of sellers' reputations.

Analytical vs. Impulsive Buyers. A hypothesis (H3) in this study posits that analytical buyers will more seriously consider the sellers' reputations and show a greater difference between price premiums for various sellers. In order to validate H3, PPDs for analytical and impulsive group are computed and shown in Table 5.

TABLE 5

Shopping Patterns and Price Premium Difference (PPD)

In general, Table 5 supports the hypothesis (H3) that analytical shoppers more seriously consider the reputation levels of sellers when they determine price premiums than impulsive shoppers. In particular, for P^sub win^ (expensive and homogeneous) and P^sub mod^ (inexpensive and non-homogeneous), the PPDs between analytical and impulsive shoppers are significantly different. Further, the mean values of PPDs among three pairs of sellers indirectly support the view that sellers' reputations are an important determinant of price premiums. For example, for a given product, PPD(S^sub exc^S^sub bad^) should be greater than PPD(S^sub avg^S^sub bad^) and PPD(S^sub avg^,S^sub bad^) should be greater than PPD(S^sub new^, S^sub bad^). Table 5 supports this view for both analytical and impulsive buyers. However, the mean values of PPD(S^sub exc^S^sub bad^) between analytical and impulsive shoppers are not significantly different for P^sub cam^ even though its p-value (0.1004) is close to 0.1. The mean values of PPD(S^sub new^, S^sub bad^) between two different types of shoppers are not significantly different (p-value=0.2431). One possible explanation for this exception is that it is difficult even for analytical users to determine a "reasonable" price premium of an expensive but heterogeneous product for a seller who has no reputation information. Therefore, they will avoid transactions with S^sub new^ and are willing to pay low price premiums. This claim is supported by the fact that PPD(S^sub new^, S^sub bad^) of analytical buyers is lowest for P^sub cam^. However, it warrants further investigation with new data.

Competit\ive vs. Non-competitive Buyers. Table 6 shows the average and standard deviation of PPDs over four different types of products indicated by competitive and non-competitive buyers. Although both competitive and non-competitive buyers are willing to pay higher premiums to sellers with higher reputations as shown in Figure 2, t-test results indicate that PPDs of competitive vs. non- competitive buyers over all four products are not significantly different at α = 0.2, rejecting a hypothesis (H4). There are two possible explanations for this finding. First, the low average score of competitiveness indicates that most buyers are not very competitive and will not rush into becoming a winner by paying high premiums. Therefore, their analytical reasoning will play a critical role in determining whether it is reasonable to win an auction at the cost of paying higher premiums. Another possibility is that the chosen threshold value is not high enough to clearly categorize two different types of buyers. However, raising the threshold value will result in a very small group of competitive buyers. Given the current threshold, two buyer groups are only marginally different, and they essentially belong to the same group. Therefore, it is very unlikely for two buyer groups to show different bidding patterns.

PRODUCT TRAITS AND SELLER REPUTATION

In this section the effects of product traits and sellers' reputations on PPDs are analyzed. Table 7 shows the means and standard deviations of price premiums given a combination of a seller and a product type. The effects of sellers' reputations on price premiums for a given product along with one-tail t-tests are shown at the bottom of Table 7. For example, the value (13.6823) for a pair of (5^sub exc^, S^sub avg^) is the t-value of a paired t- test (one-tailed) to compare price premiums between S^sub exc^ and S^sub avg^ for a product P^sub win^. Since the t-value is significant at α = 0.01, it is safe to conclude that online buyers are willing to pay higher premiums for S^sub exc^ than S^sub avg^ when they purchase a P^sub win^. The fact that all t-values are significant at α = 0.01 over all products indicates that the sellers' reputations are a critical factor in determining price premiums over all products.

Note that buyers are willing to pay higher premiums for a new seller than a seller with a bad reputation for any product considered in this paper. This notion is consistent with an observation made from Table 5 where both analytical and impulsive buyers pay a higher price premium for S^sub new^ than S^sub bad^. If this is the case, online service providers with bad trust scores prefer to have a new pseudonym. The positive relationship between sellers' reputations and price premiums seems to be stronger for expensive products than for inexpensive products. This is based on the fact that t-values of expensive products for (S^sub exc^, S^sub avg^) and (S^sub avg^, S^sub bad^) are greater than those of inexpensive products, supporting a hypothesis (H5). However, for a (S^sub new^, S^sub bad^) the relative effectiveness of reputation on the expensive and inexpensive products cannot be determined because of mixed signs of t-value comparisons for two pairs, (P^sub win^, P^sub dvd^) and (P^sub cam^, P^sub mod^).

TABLE 6

Auction Attitudes and Price Premium Difference

TABLE 7

Effects of Reputation and Product Type on Premiums

This study theorizes that two major characteristics of products- homogeneity and price-can affect the perceived risks of business transactions and, therefore, affect price premiums. The effects of homogeneity of products are analyzed by comparing the mean difference between two products with different levels of homogeneity. For example, consider two products, P^sub win^ and P^sub cam^. Both of them are priced at $1200, but P^sub win^ is considered more homogenous and, therefore, less risky than P^sub cam^. The effects of product price on premiums are also studied. However, all four products have different levels of homogeneity. Therefore, it is not possible to completely exclude the effect of homogeneity when the ultimate goal is to measure the effect of product price on premiums. To reduce the undesired contribution of homogeneity effect to the price effect, an expensive and an inexpensive product with closer homogeneity are combined, and their mean differences are compared. The effects of homogeneity and product products are summarized in Table 8.

The price effects of homogeneous products (P^sub win^ and P^sub dvd^) for S^sub exc^ and S^sub bad^ are significant at α=0.05 and α=0.1, respectively. For non-homogeneous products, the price effects are significant at α=0.05 for S^sub exc^ and S^sub new^, and at α=0.1 for S^sub avg^ and S^sub bad^. Table 8 also shows that homogeneity effects are stronger for expensive products (P^sub cam^, P^sub win^) than inexpensive products (P^sub mod^, P^sub dvd^). The homogeneity effects of expensive products for S^sub exc^ and S^sub new^ are significant at α=0.05. However, the homogeneity effects of inexpensive products are significant for only S^sub exc^ at α=0.1. Overall, these findings support H6 by confirming that the magnitude of price premiums is positively related to the degree of product quality variation. Note that while product price significantly affects price premiums for S^sub bad^, product homogeneity does not show a significant effect for S^sub bad^. This makes sense because online buyers do not want to buy expensive products from a seller with a bad reputation no matter how homogeneous or non-homogeneous a product is.

TABLE 8

Effects of Product Traits on Premiums

CONCLUSIONS AND FUTURE RESEARCH

Table 9 summarizes hypothesized determinants of price premiums and shows effects of individual determinant or mixed effects of determinants. The fact that most hypotheses are fully or at least partially supported indicates that personal and product traits and sellers' reputations are all important determinants of price premiums. In this sense this study not only confirms the findings of previous research (3, 16, 21), but also extends and generalizes their results on the relationship between trust and price premiums. Further, this study is different from previous studies in the sense that it examines both buyer and product traits to quantify the relationships between sellers' reputations and price premiums. This study also extends theoretical and empirical inferences of the moderating effects of personal characteristics and product traits to various reputation systems. Since most previous works (25, 34) studied only a specific feedback forum, their findings may not be generalized well to auction sites with different feedback systems (e.g., eBay vs. Amazon).

TABLE 9

Summary Table of Results

As is typical with a survey-based research, this research has limitations on external validity. In particular, this research is based on online survey from college students. Although students represent a significant portion of all Internet users and their online shopping experiences in this study may not be biased, it will be interesting to compare results presented in this study to results from a new experiment with different samples. In particular, frequent buyers on online auction sites may show different virtual bidding patterns. Further, frequent online bidders may be able to distinguish price premiums of new and bad sellers to products with great variation of quality. In order to further validate results in this paper, it is also desirable to replicate the same experimental results using different products that represent the same categories based on two dimensional characteristics: price and homogeneity.

Most of all, this research does not observe the actual purchasing behaviors of buyers when they are given information about sellers and products. Therefore, it is possible that the discovered effects of the three main factors and the importance of trust do not truly reflect actual effects and their magnitude. When buyers actually purchase a product at an online auction, they might show different purchasing patterns (38). Note that observing bidding behaviors in real auction sites has its own limitations, too. Note that many auction-related variables such as the length of the auction, the minimum bidding price, the number of bidders, popularity of auction sites, and visual design of the auction page can affect experimental results in an unexpected way. Therefore, the proposed model and real world experiments should be used as a complement, not as a replacement of other methods.

The needs of buyers for a product is one important personal characteristic that is not discussed in this paper. A future study should investigate the relationships between the needs of products and price premiums. Buyers are likely to pay higher premiums for products that satisfy their immediate needs. For instance, a coin collector is likely to pay a high premium for a coin that will complete her collection. In an extreme case, a buyer may not mind paying a very high price even to a seller with a bad reputation. The relationship between bidders' expertise level (35, 44) and premiums will also be considered. Other auction parameters including the timing of the bid and starting bidding price have been known to affect the final auction price premium and hence are worthy to look at.

Another major direction of future research is to develop a model to simulate the online auction process. Utilizing a virtual simulation tool can greatly help researchers understand a complex economic process under a completely controlled environment. While real data sets from real auction sites represent actual bidders' behaviors, they only show the integrated effects of several factors. Researchers also benefit from this new methodology by significantly reducing the amount of time for data collection, pre-\processing, and validation. Most of all, this new system allows researchers to test the revenue equivalence theorem (43) by comparing different types of auctions after setting the control factors to the same level except for structural difference.

REFERENCES

1. Ajzen, I. "From Intentions to Actions: A Theory of Planned Behavior." In Kuhl, J. and J. Beckmann (Eds.). Action Control: From Cognition to Behavior. New York: Springer-Verlag, 1985, pp. 11-39.

2. Ajzen, I. and M. Fishbein. Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, NJ: Prentice-Hall, 1980.

3. Ba, S. and P.A. Pavlou. "Evidence of the Effect of Trust Building Technology in Electronic Markets: Price Premiums and Buyer Behavior," MIS Quarterly, 26:3, 2002, pp. 243-268.

4. Bagozzi, R.P. "Marketing as Exchange: A Theory of Transactions in the Marketplace," American Behavioral Scientist, 21, 1978, ppl. 535-556.

5. Bapna, R., P. Goes, and A. Gupta. "Analysis and Design of Business-to-Consumer Online Auctions," Management Science, 49:1, 2003, pp. 85-101.

6. Brynjolfsson, E. and M. Smith. "Frictionless Commerce? A Comparison of Internet and Conventional Retailers," Management Science, 46:4, 2000, pp. 563-585.

7. Campbell, D.T. and D.W. Fiske. "Convergent and Discriminant Validation by the Multitrait-multimethod Matrix," Psychological Bulletin, 56:2, 1959, pp. 81-105.

8. Capen, B.C., R. V. Clapp, and W.M. Campbell. "Competitive Bidding in High-risk Situations," Journal of Petroleum Technology, 23:6, 1971, pp. 641-653.

9. Castells, M. The Rise of the Network Society, 2nd Ed. Oxford, JK: Blackwell, 2000.

10. Chen, K., IJ. Chen, and H. Paul. "Explaining Online Behavioral Differences: An Internet Dependency Perspective," Journal of Computer Information Systems, 41:3, 2001, pp. 59-63.

11. Coleman, R. Foundations of Social Theory. Cambridge, MA: Belknap, 1990.

12. Granovetter, M. "The Strength of Weak Ties," American Journal of Sociology, 78:6, 1973, pp. 1360-1380.

13. Heider, F. The Psychology of Interpersonal Relations. New York: John Wiley and Sons, 1958.

14. Herschlag, M. and R. Zwick. "Internet Auctions - Popular and Professional Literature Review," Quarterly Journal of Electronic Commerce, 1:2, 2002, pp. 161-186.

15. Huffman, D.L., T.P. Novak, and M. Peralta. "Building Customer Trust Online," Communications of the ACM, 42:4, 1999, pp. 80-85.

16. Houser, D. and J. Wooders. "Reputation in Auctions: Theory and Evidence from eBay." Working Paper, Department of Economics, University of Arizona, 2000.

17. Johnston, T.C. "Reputation Price Premium in Online Auctions." In Suter, T.A. (Ed.). Marketing Advances in Pedagogy, Process and Philosophy. Stillwater, OK: Society for Marketing Advances, 1996, pp. 178-181.

18. Kim, K. and B. Prabhakar. "Initial Trust, Perceived Risk, and the Adoption of Internet Banking." In Proceedings of the 21st Intl. Conf. on Information Systems, Brisbane, Queensland, Australia, 2000, pp. 537-543.

19. Klein, B. and K. Leffler. "The Role of Market Forces in Assuring Contractual Performance," Journal of Political Economy, 89:4, 1981, pp. 615-641.

20. Lucking-Reiley, D. "Using Field Experiments to Test Equivalence between Auction Formats: Magic on the Internet," American Economic Review, 89:5, 1999, p. 1063-1080.

21. Luckiing-Reiley, D., D. Bryant, N. Prasad, and D. Reeves. "Pennies from eBay: The Determinants of Price in Online Auctions." Technical Report No. 1736, Econometric Society World Congress 2000, 2000.

22. Lynch, J.G. and D. Ariely. "Wine Online: Search Cost Affect Competition on Price, Quality, and Distribution," Marketing Science, 19:1, 2000, pp. 83-103.

23. Malhotra, D. and J. Murnighan. "Milked for all Their Worth: Competitive Arousal and Escalation in the Chicago Cow Auction." Working Paper, Kellogg School of Management, Northwestern University, 2000.

24. Mayer, R.C., J.H. Davis, and F.D. Schoorman. "An Integrative Model of Organizational Trust," Academy of Management Review, 20:3, 1995, pp. 709-734.

25. McDonald, C.G. and V.C. Slawson. "Reputation in an Internet Auction Market," Economic Inquiry, 40:4, 2002, pp. 633-650.

26. McKnight, D.H., V. Choudhury, and C. Kacmar. "Trust in E- commerce Vendors: A Two-stage Model." In Proceedings of the 21st Intl. Conf. on Information Systems, Brisbane, Queensland, Australia, 2000, pp. 532536.

27. McKnight, D.H., L.L. Cummings, and N.L. Chervany. "Initial Trust Formation in New Organizational Relationships," Academy of Management Review, 23:3, 1998, pp. 473-490.

28. Meyrowitz, J. "Medium Theory." In Crowley, D. and D. Mitchell (Eds.). Communication Theory Today. Cambridge, CA: Stanford University Press, 1994.

29. Ottaway, T.A., C.L. Bruneau, and G.E. Evans. "The Impact of Auction Item Image and Buyer/Seller Feedback Rating on Electronic Auctions," Journal of Computer Information Systems, 43:3, 2003, pp. 56-60.

30. Pinker, E.J., A. Seidmann, and Y. Vakrat. "Managing Online Auctions: Current Business and Research Issues," Management Science, 49:11, 2003, pp. 1457-1484.

31. Ranganathan, C. and E. Grandon. "An Exploratory Examination of Factors Affecting Online Sales," Journal of Computer Information Systems, 42:3, 2002, pp. 87-93.

32. Rao, A.R. and K.B. Monroe. "Causes and Consequences of Price Premiums," Journal of Business, 69:4, 1996, pp. 511-535.

33. Resnick, P. and R. Zeckhauser. "Trust Among Strangers in Internet Transactions: Empirical Analysis of eBay's Reputation System." Working Paper, University of Michigan, 2001.

34. Resnick, P., R. Zeckhauser, J. Swanson, and K. Lockwood. "The Value of Reputation on eBay: A Controlled Experiment." Working Paper, 2002.

35. Roth, A.E. and A. Ockenfels. "Last-minute Bidding and the Rules for Ending second-price Auctions: Evidence from eBay and Amazon Auctions on the Internet," American Economic Review, 92:4, 2002, pp. 1093-1103.

36. Rotter, J. "A New Scale for the Measurement of Interpersonal Trust," Journal of Personality, 35:4, 1967, pp. 651-665.

37. Shapiro, C. "Premiums for High Quality Products as Returns to Reputations," Quarterly Journal of Economics, 98:4, 1983, pp. 659- 679.

38. Stewart, K.J. "Transference as a Means of Building Trust in World Wide Web Sites." In De, P. and J.I. DeGross (Eds.). Proc. Of the 20th Intl. Conf. on Information Systems, Charlotte, NC, 1999, pp. 459-464.

39. Tan, X., D.C. Yen, and X. Fang. "Internet Integrated Customer Relationship Management - A Key Success Factor for Companies in the E-Commerce Arena," Journal of Computer Information Systems, 42:3, 2002, pp. 77-86.

40. Thibaut, J.W. and H. Kelley. The Social Psychology of Groups. New York: John Wiley and Sons, 1975.

41. Turban, E., D. King, J. Lee, M. Warkentin, and H.M. Chung. Electronic Commerce 2002: A Managerial Perspective, 2nd ed. Upper Saddle River, NJ: Prentice Hall, 2002.

42. Van Vliet, P.J.A. and D. Pota. "Understanding Online Retail: A Classification of Online Retailers," Journal of Computer Information Systems, 41:2, 2000, pp. 23-28.

43. Vickrey, W. "Counterspeculation, Auctions, and Competitive Sealed Tenders," Journal of Finance, 16:1, 1961, pp. 8-37.

44. Wilcox, R.T. "Experts and Amateurs: The Role of Experience in Internet Auctions," Marketing Letters, 11:4, 2000, pp. 363-374.

45. Williamson, O.E. "Calculativeness, Trust, and Economic Organization," Journal of Law and Economics, 36:1, 1993, pp. 453- 486.

46. Zucker, L.G. "Production of Trust: Institutional Sources of Economic Structure, 1840-1920." In Staw, B.M. and L.L. Cummings (Eds.). Research in Organizational Behavior. Greenwich, CT: JAJ, 1986, pp. 53-111.

YONGSEOG KIM

Utah State University

Logan, Utah 84322-3515

APPENDIX A

Questionnaire for Personal Traits

APPENDIX B

Questionnaire for Virtual Auction

Copyright International Association for Computer Information Systems Fall 2005

Source: Journal of Computer Information Systems, The

More News in this Category