Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Time and price dimensions in an online auction reputation mechanism Pan, Yin 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-0590.pdf [ 8.5MB ]
Metadata
JSON: 831-1.0091655.json
JSON-LD: 831-1.0091655-ld.json
RDF/XML (Pretty): 831-1.0091655-rdf.xml
RDF/JSON: 831-1.0091655-rdf.json
Turtle: 831-1.0091655-turtle.txt
N-Triples: 831-1.0091655-rdf-ntriples.txt
Original Record: 831-1.0091655-source.json
Full Text
831-1.0091655-fulltext.txt
Citation
831-1.0091655.ris

Full Text

Time and Price Dimensions in an Online Auction Reputation Mechanism By Yin Pan Bachelor of Economics, Renmin University of China, 1999 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN BUSINESS ADMINISTRATION In THE FACULTY OF GRADUATE STUDIES (Sauder School of Business; MIS Division)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA September 2004 © Yin Pan, 2004  Library Authorization  In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  Name of Author (please print)  Date (dd/mm/yyyy)  Title of Thesis:  Degree: M & S T W  0-^  S C A ^ C C  Department of U.(L^ j^pjfr The University of British Columbia Vancouver, BC Canada  grad.ubc.ca/forms/?formlD=THS  rWvhfcfr*  M^^V^^L  l^fry^-frTAyy  ^yVft^v  \V\  page 1 of 1  U&M'  last updated: 20-Jul-04  Abstract  The reputation mechanism in online auctions provides online auction participants with information about the past behaviour of other participants, in the form of a numerical rating. eBay, which is the most popular online auction website, uses a simple aggregation to summarize and demonstrate participants' reputations. This study proposes and investigates designs for multi-dimensional reputation mechanisms, with time and/or related transaction price information. A laboratory experiment was conducted to empirically test whether the time and price dimensions in a reputation mechanism have significant effects on bidding behaviour of online auction participants, and whether these help participants to better judge the trustworthiness of sellers with a higher level of confidence. A design with one treatment of five levels was adopted and 150 university students were recruited for the experiments. The results showed that the mechanisms with a weighted time and/or price dimension have some ability to enable bidders to better distinguish sellers of different reputation levels, and were perceived to be more helpful to bidders in making bidding decisions.  ii  Table of contents Abstract  ii  Table of contents  iii  List of tables  vii  List of figures  viii  Acknowledgements 1  2  3  Introduction  ix 1  1.1  Overview  1  1.2  Online auctions  2  1.3  Fraudulent behaviours in online auctions  3  1.4  Reputation mechanisms in online auctions  4  1.5  Motivation  6  1.6  Organization of the thesis  8  Theoretical foundation and literature review  2.1  Information asymmetry between buyers and sellers in online auctions  2.2  Reputation mechanisms  9  9 11  2.2.1  Time dimension  12  2.2.2  Price dimension  13  2.3  Buyer's trust in online auctions  14  2.4  Uncertainty in online auctions and the bid amount  15  Research framework  3.1  Independent variables - reputation mechanisms 3.1.1  Existing eBay mechanism  18  18 18  iii  3.2  3.3  4  Time-weighted mechanism  19  3.1.3  Price-weighted mechanism  20  3.1.4  Price-time-weighted-double mechanism  21  3.1.5  Price-time-weighted-single mechanism  23  Dependent variables  24  3.2.1  Trust  24  3.2.2  Confidence and helpfulness in bidding judgment  25  3.2.3  Bid amount  25  Hypothesis development  26  3.3.1  Bid amount  26  3.3.2  Trust  27  3.3.3  Confidence  30  3.3.4  Helpfulness  31  Experiment design  33  4.1  Overview  33  4.2  Experiment settings  34  4.3 5  3.1.2  4.2.1  The game  34  4.2.2  Seller profiles  35  4.2.3  Auctions and items  45  4.2.4  Proxy bidding  45  Experiment procedure  46  Results  48  5.1  48  Data analysis overview  iv  5.2  5.3  6  Reliability analysis  5.1.2  Factor analysis upon Trust and Confidence  49  5.1.3  Correlation between Trust and Confidence  50  Maximum bid  48  51  5.2.1  Eliminate "no-bids"  51  5.2.2  Set "no-bids" to C$10  53  Trust  54  5.3.1  Trust in sellers  54  5.3.2  ANOVA results of trust between Random seller and other sellers  55  5.3.3  Tukey of trust between Random seller and other sellers  57  5.4  Confidence  59  5.5  Helpfulness  61  Discussion and conclusion  62  6.1  62  6.2 7  5.1.1  Discussions 6.1.1  Bids  62  6.1.2  Trust  63  6.1.3  Confidence in judgment  64  6.1.4  Helpfulness in decision making  65  Conclusion  66  Limitations and future research  67  7.1  Limitations  67  7.2  Future research  68  Bibliography  69  v  Appendix 1 Experiment instructions for participants  74  Appendix 2 Seller's profiles presented in five mechanisms  91  Appendix 3 Description of the items used in the experiments  103  Appendix 4 Questionnaire  109  vi  List of tables  Table 1 Seller's profiles  36  Table 2 Summary of seller's behaviour patterns  ^ 37  Table 3 Reliability analysis  48  Table 4 Results of factor analysis on Trust and Confidence: total variance explained 49 Table 5 Results of factor analysis on Trust and Confidence: rotated component matrix  49  Table 6 Linear and Quadratic regression statistics with Trust and Confidence  50  Table 7 Bids to sellers under different mechanisms  52  Table 8 Bids to seller under different mechanism ("No bids" are replaced with C$10) 54 Table 9 Mean of trust in sellers  54  Table 10 ANOVA results of trust in sellers  55  Table 11 ANOVA results of difference in trust between Random seller and other sellers  56  Table 12 Tukey results of difference in trust between Random seller and other sellers  59  Table 13 ANOVA results of confidence in judgments  59  Table 14 ANOVA results of confidence in judgments by sellers  60  Table 15 ANOVA results of helpfulness in decision making  61  vii  List of figures Figure 1 eBay ID card Figure 2 Nagel and Holden's economic value analysis  6 16  Figure 3 Kauffman and Wood's analysis of effects on purchases for traditional and e-Commerce markets  17  Figure 4 eBay mechanism  19  Figure 5 Time-weighted mechanism  20  Figure 6 Price-weighted mechanism  21  Figure 7 Price-time-weighted-double mechanism  22  Figure 8 Price-time-weighted-single mechanism  24  Figure 9 Screen capture of the experiment website  33  Figure 10 eBay reputation scores of the six sellers during the most recent six months 37 Figure 11 Positive percentages of the six sellers underfivetreatment mechanisms 38 Figure 12 Behaviour pattern - Honest seller  39  Figure 13 Behaviour pattern - Random seller  40  Figure 14 Behaviour pattern - Time increase seller  41  Figure 15 Behaviour pattern - Price low low seller  42  Figure 16 Behaviour pattern - Price low high seller  43  Figure 17 Behaviour pattern - Price time increase seller  44  viii  Acknowledgements I wish to express my gratitude to Dr. Paul Chwelos and Dr. Michael Brydon for their guidance and supervision in this thesis. Without their encouragement and support, this thesis would not have been possible. In addition, I appreciate the help of Dr. Izak Benbasat, for his advice and insightful inputs. Further, I would like to say thanks to Leon Qiu, Jack Jiang, Weiquan Wang, Dongmin Kim and David Hood for fruitful discussions. Last but not least, I would like to express my special thanks to my parents, for their endless love and support.  ix  1  Introduction  1.1  Overview With the development of information technology, a rapid increase in electronic  commerce (e-Commerce) has been made possible during the past decade. Like all other Internet business, online auctions have experienced phenomenal growth in recent years. Researchers have paid careful attention to online auctions, especially the reputation mechanisms used by online auction websites. eBay (www.ebay.com) is the dominant and most popular online auction website in the cyber world. Boasting tens of millions of registered users, eBay is often the first place to which both online auction buyers and sellers turn. Like other types of online transactions, online auctions are run virtually; buyers and sellers do not meet before or after the transaction. Some opportunistic sellers may take advantage of buyers by behaving dishonestly. To counteract this problem, eBay provides a reputation mechanism, which makes buyers feel safe and informed about the seller's trustworthiness before they bid or while they are bidding. With this mechanism, every member of eBay has an eBay ID card showing a summary of his or her transaction history. In this research, based on the existing eBay mechanism, we propose new reputation mechanisms. We discuss the effectiveness of the eBay reputation mechanism and the proposed mechanisms, in terms of inducing buyers' trust and distinguishing sellers of various reputation levels. We propose four alternative reputation mechanisms, each of which modifies or extends the existing eBay mechanism in different ways: (1) Time-weighted mechanism (TM); (2) Price-weighted mechanism (PM); (3) Price-time-weighted-double mechanism  1  (PTDM); and (4) Price-time-weighted-single mechanism (PTSM). We performed a laboratory experiment using an experimental website and examined whether the proposed mechanisms helped buyers make better judgments on sellers' trustworthiness, and their effects on bidders' bidding behaviours. 1.2  Online auctions  Auctions on the Internet have become a fascinating new type of online exchange mechanism. They display detailed information about a good or service online, with the intent to sell it through a competitive procedure to the highest bidder (Beam and Segev 1998). Bidders view descriptions of auction items online, then submit bids by sending an email or filling out an electronic form. The bidding which may last for a few days, is shown at the auction website and is updated continually to display the current highest bid. Many person-to-person online auction websites use "proxy bidding" method. The auctions take the form of sealed bid, second price Vickery auctions (Lucking-Reiley, 2000a), in which the highest bidder wins, but pays only the amount of the second highest bid plus the minimum bid increment. Every day, hundreds of thousands of different auctions take place online, for goods ranging from "The Matrix" movie posters to a piece of lakeside property. Internet technology has lowered the cost to sellers of organizing an auction, and the cost of participating as a bidder, making online auctions more and more popular. Online auctions first took place on text-based Internet newsgroups and email discussion lists in 1994 (Lucking-Reiley, 2000b). eBay, now the largest online auction site, commenced business in September 1995. It is one of the pioneers in web-based  2  auctions. Initially, most of the items on eBay tended to be collectibles such as Beanie Babies, comic books, coins, and etc. eBay was a first mover in taking advantage of the technologies offered by the web, including the use of automated bids entered through electronic forms, search engines and clickable categories that allow bidders to locate items of interest (Lucking-Reiley, 2000b). eBay has grown very rapidly since its establishment. At the end of year 2003, 95 million registered users from more than 150 countries listed 971 million items on eBay, with the total value of goods sold on the site reached nearly $24 billion (EI, 2003). Following in eBay's steps, other websites, including Yahoo! Auctions, Amazon, ubid and OnSale, launched their own person-toperson auction services. eBay appears to have a first-mover advantage in a market with significant economies of scale: sellers prefer to list their goods where the most buyers visit, and buyers prefer to visit sites with large selections of goods. The Internet has opened a global marketplace to online auction users, providing geographic and temporal convenience and a less costly way for buyers and sellers to perform transactions. For sellers, online auctions make it possible to obtain a much larger audience on relatively short notice, which increases the chance of selling items successfully. For bidders, with the extensive listings and powerful search technologies, it is more convenient and easy to find the goods that fit their needs. 1.3  Fraudulent behaviours in online auctions  Most online auction websites act only as intermediaries, offering hosting services for sellers to post their items for sale, and allowing buyers to bid on those items. They do not verify that the merchandise really exits, or whether it matches the description provided by  3  the sellers. Due to geographical factors, it is usually impossible for bidders to inspect items before they bid. Therefore, online auction buyers are participating in transactions with uncertainty about product quality. Also, sellers do not send the items being sold to the winning bidders until after they have received payment for the items. Moreover, online auction transactions are nearly anonymous. The bidders only know the sellers' email address or their claimed name. The above factors create opportunities for fraudulent behaviour by sellers in online auctions, including misrepresentation of product characteristics, delay in product delivery, and even receipt of payment without product delivery. Online auction fraud is already the most frequently reported form of Internet fraud, topping the FBI's scam list (Kane, 2002). The Fraud Information Center of the National Consumer League (NCL) in the United States found that 90% of online fraud complaint cases in 2002 involved online auctions, with more than $13 million in losses (NCL, 2002a). In 2003, fraud in online auctions accounted for 89% of the complaints they received. According to the 2002 Internet Fraud Report by the Internet Fraud Complaint Center, Internet auction fraud is by far the most commonly reported offense, comprising 46% of referred complaints, three times the 2001 total, with an average loss per victim of US$ 320 (JPCC, 2003; NCL 2002b). 1.4  Reputation mechanisms in online auctions Some online auction websites provide reputation mechanisms to encourage honest  behaviour, while discouraging fraudulent behaviour. The reputation mechanism also provides online auction participants with information about other participants' past  4  behaviour, thereby helping them to distinguish "dishonest" participants from "honest" participants. Most of these mechanisms encourage buyers and sellers to rate or evaluate each other at the close of a transaction. These ratings and comments are summarized, aggregated and published publicly for the online community of the auction sites, usually in the form of member feedback profiles. When viewing an auction listing, a bidder also sees a numeric feedback rating for the seller, which summarizes a seller's previous feedback, which is intended to provide insight into the seller's past transaction behaviours overtime. eBay, the most popular online auction website, allows both sellers and buyers to rate how their trading partners behaved in a transaction. The rating takes the form of a positive (+1), negative (-1) or neutral (0) response, along with text comments. For each positive response, a seller receives one point; for negative, minus one point; for neutral, zero. A seller's reputation is presented with a "feedback profile" (See Figure 1), the socalled "eBay ID card." The eBay ID card displays a summary of the rating statistic (calculated by a simple aggregation of the ratings from unique users). By clicking on the seller's name, bidders can view all feedback received from the seller's previous transactions, including all comments as well as a statistic based on the total number of positive, negative and neutral responses.  5  Member Profile: cocoa37 (2400 & ) 2400 99.8%  Recent Ratings:  2404 4 All positive feedback received: Learn about what these numbers mean:  2726  ©  Past Month.  Past 6 Months  •Past 12 Months  . positive  59  372  811  neutral  0  negative  0  Bid Retractions (Past S months): 0  Figure 1 eBay II) card  With the eBay reputation mechanism, all ratings are treated equally in the calculation of a seller's reputation score. Positive feedback from the transaction of a $20 music CD, or a $100,000 Porsche convertible will be counted in the same way: +1 point in a person's reputation score. Similarly, five-year-old feedback contributes the same weight in the score calculation as feedback received yesterday. The simple aggregation mechanism ignores the value of the transaction related to the feedback and the recentness of the feedback, elements that can be salient in evaluating a seller's performance and predicting his future behaviour. Thus the existing eBay mechanism provides opportunities for fraudulent sellers, because they can build up a very good reputation by selling many low value items, then without warning, act dishonestly in a high value transaction. Even if they receive negative feedback from this high value transaction, they can still masquerade as very reputable sellers with high reputation scores. Bidders have no way to discover this tactic using the information provided by the eBay seller profile ID card. 1.5 Motivation eBay uses the reputation mechanism to help bidders know more about a seller's performance in previous transactions, encourage honest behaviours and induce trust  6  between sellers and bidders. The feedback profile is the main source to which a buyer may refer for the seller's previous behaviour. It allows him to estimate whether the seller is trustworthy when making his bidding decisions. The following questions can be asked of the mechanism: Does the current format of the eBay feedback profile do a good job of helping buyers judge a seller's trustworthiness? Are trustworthy sellers easily distinguished from untrustworthy sellers with the eBay mechanism? Are there ways to improve the eBay mechanism's ability of distinguishing dishonest sellers from honest sellers using the currently available information gathered by eBay? Considering the shortcomings of the simple feedback aggregation mechanism as discussed in the section 1.4, we predict that alternative reputation mechanisms using time and/or price dimensions in the calculation and presentation of the reputation score, would work better in differentiate sellers of various reputation levels than the current eBay mechanism. In this research, we propose four alternative reputation mechanisms with different profile ID card formats. Our mechanisms include the time when the feedback is left, and/or the price of the related transaction in the reputation score calculation and presentation. The weighed calculation is in contrast to simply aggregating the number of positive, neutral and negative ratings like eBay. These four mechanisms are: (1) Timeweighted mechanism (TM); (2) Price-weighted mechanism (PM); (3) Price-timeweighted-double mechanism (PTDM); and (4) Price-time-weighted-single mechanism (PTSM). By performing laboratory experiments using an experimental website, we examine whether these proposed mechanisms do a better job than the eBay mechanism (EM) in terms of discriminating "dishonest" from "honest" sellers. We predict that the  7  proposed mechanisms help buyers formulate a more accurate assessment of a seller's trustworthiness. We seek answers to the following questions: 1) Do the proposed reputation mechanisms induce higher trust in honest sellers, and lower trust in dishonest sellers, as compared to the eBay mechanism? 2) Can the proposed reputation mechanisms distinguish types of behaviour that cannot be distinguished by the eBay mechanism? 3) Do the bidders feel more confident in their judgment concerning a seller's trustworthiness using the proposed reputation mechanisms? 4) Do the proposed mechanisms provide more helpful information about sellers' trustworthiness to bidders? 5) How will the proposed mechanisms affect bidders' maximum bids? 1.6  Organization of the thesis Section 2 provides background on previous research in online auctions, including the  introduction of information asymmetry in online auctions, a definition of and discussion about trust in online auctions, and a discussion of the relationship between uncertainty and bid amount in online auctions. The research framework and hypotheses are presented in section 3. Section 4 introduces the experiment design and procedure. In section 5 and 6, we present the results of the experiment and a data analysis discussion. Section 7 summarizes the limitations and future research.  8  2  Theoretical foundation and literature review  2.1  Information asymmetry between buyers and sellers in online auctions  Most online auction websites are auction listing sites. They usually do not process the auctioned goods nor handle payments and shipments; all fulfillment details are worked out by the buyers and sellers independently. Compared to auctions held in traditional auction houses, online auctions are more convenient, but involve more risk and uncertainty, because they involve interactions with strangers who they have had little or no interaction before. The standard procedure after an online auction ends is for the buyer to pay the seller in advance for a product or service to be delivered sometime in the future. Different from traditional auctions, online auction buyers cannot view the item while they are bidding, cannot take the item immediately when they win the auctions, and cannot benefit from a professional opinion about the auction items. The buyers must trust that the sellers will actually send the good in return for the payment of the bid amount. But how do they know the sellers are trustworthy? In fact, there have been a large number of cases of fraud reported by buyers in online auctions (Lucking-Reiley, 2000b). Sellers possess significantly more information regarding the quality of the goods and their own behaviours, as compared to the buyers (Mishra, Heide, and Cort 1998). Buyers face more uncertainty in online auction transactions (Kauffman and Wood, 2001), than in traditional auctions. First, there is uncertainty about product quality. Although online auction websites try hard to mitigate this problem by using technology to make it possible to post electronic images of items, provide text descriptions, and answer bidders' questions via email and phone call, it is almost impossible for bidders to physically inspect the goods before bidding online. Thus, buyers may not have easy access to  9  information regarding the true quality of the product, and therefore may be unable to judge product quality prior to purchase (Fung and Lee, 1999). This makes them vulnerable to additional risk from potentially incomplete or distorted information about the quality of products (Ba and Pavlou, 2002). Another uncertainty is seller identity. Online auction participants can easily remain anonymous or change their identities, so it is difficult to bind an identity to one participant. Most auction sites identify sellers and bidders by email addresses, which can be easily obtained from multiple sources without monetary cost. Without proper reputation mechanism, an opportunistic seller can very easily masquerade as an honest one, luring an unsuspecting buyer into a fraudulent transaction (Ba and Pavlou, 2002; Eaton, 2002). The difference between the information the two transacting parties possess is referred to as "information asymmetry". It can occur with respect to knowledge about product quality or knowledge about the transacting parties' behaviour. It often leads to various problems, including inappropriate decisions and outcomes, and unfair exchanges of value (Kauffman and Wood, 2000) which, in the case of online auctions, expose buyers to more risks associated with fraudulent transactions (Ba and Pavlou, 2002). Information asymmetry in online auctions may give rise to opportunistic behaviour such as the misrepresentation of product quality, which could lead to mistrust among participants or even market failure (Akerlof 1970). Therefore, opportunism could potentially erode the foundations of electronic markets and jeopardize the proliferation of the electronic economy. Following Ba and Pavlou's (2002) definition of opportunistic  10  behaviour in online auctions, it includes unjustifiable delays in product delivery, misrepresentation of product characteristics, receipt of payment without delivery of a product, and other forms of illegal activity and fraud. 2.2  Reputation mechanisms  In an effort to reduce the number of fraudulent transactions in online auctions, online reputation mechanisms have emerged to provide information on sellers' reputations, helping build trust among potential participants in online auctions (Ba and Pavlou, 2002). The mechanisms serve both as a source of information and as a potential source of sanctions (Kollock, 1999). As a source of information, they enable bidders to know how a seller has been evaluated by others with whom the seller has transacted previously, thereby aiding in a potential buyer's decision whether to bid in an auction, and the decision on how much he is going to bid. Also, the attribution of a negative reputation may work as a sanctioning mechanism to punish dishonest behaviour. This makes the owner of the reputation act in a more trustworthy manner (Keser, 2002). A good reputation mechanism must meet two challenges. Firstly, it must provide information that allows buyers to distinguish between trustworthy and untrustworthy sellers. Second, it must encourage sellers to behave honestly, and give them incentives to be trustworthy. (Resnik, Zeckhauser, & Friedman 2000; Resnick & Zeckhauser, 2001). eBay's reputation system is a "one-dimensional construct," in which the reputation scores are calculated by simply aggregating the number of positives and negatives. It does not comprehensively demonstrate a seller's previous behaviour. Malaga (2001) summarized this as a "no categorization" problem. Zacharia, Moukas & Maes (1999) put  11  forward the concept of a multidimensional and weighted reputation score, in which feedback is categorized according to certain dimensions, and weights are discounted for in the reputation score calculation according to the categories to which the feedback belongs. In this research, we propose several formats for reputation mechanisms by adding time and/or price dimensions into the calculation and presentation of reputation information, in an attempt to solve the problems with eBay mechanisms discussed in the literature, and try to determine if this will lead to better discernment between the trustworthy and untrustworthy sellers. 2.2.1 Time dimension As defined by Wilson (1985), reputation is a characteristic or attribute ascribed to one person. Operationally, it is usually represented as a prediction about likely future behaviour (Malaga, 2001). As a predictor of sellers' future behaviour, the larger the number of ratings used in the evaluation of reputation values, the better the predictability a reputation mechanism has. The reputation values are associated with human individuals, and humans change their behaviour over time (Zacharia, Moukas, & Maes, 2000). A seller may behave honestly in the initial stage, masquerading as a trustworthy seller, in order to build up a good reputation. After that, the seller may begin cheating in transactions, taking advantage of the existing "good reputation." It is desirable that the predicted reputation values be closer to the current behaviour of the individuals rather than their overall performance (Zacharia, Moukas, & Maes, 2000; Malaga, 2001). In other words, the more recent the reputation rating, the better it can  12  predict sellers' likely future behaviour. Thus, ratings should be discounted over time so that the most recent ratings have more weight in the evaluation of a user's reputation (Zacharia, Moukas, & Maes, 2000). The eBay mechanism, which lack time discounting, does not provide enough incentive for reputable sellers to continue to provide high quality products and services. As reported by Wood, Fan, and Tan (2002), sellers with high reputation scores are more likely to engage in opportunistic behaviour. Reputable sellers profit from their established reputation, and the current eBay reputation system does not provide sufficient penalty for such behaviour. 2.2.2 Price dimension The value of the transaction related to the feedback should be another dimension in the reputation mechanism. Dellarocas (2002) pointed out that simple feedback aggregation mechanisms such as eBay's fail to facilitate efficient transactions in settings in which each seller sells products of different qualities. Sellers can build a good reputation by honestly selling low value items and then use this reputation to help them "make a killing" by cheating on a few high value items. Also, as noted in Livingston (2004), current eBay seller profiles may not truly and comprehensively reflect a seller's trustworthiness. Sellers could build reputations by selling relatively low value items, and then cheat in auctions of more expensive goods. There is no information about the transaction price associated with a specific piece of feedback presented on the eBay JD card, or included in the reputation score calculation. The simple aggregated numerical ratings on the eBay JD card fail to convey the subtleties  13  of online auction interactions. The current eBay reputation mechanism may distort information about a seller's trustworthiness, and thus cannot help buyers to make accurate bidding decisions (Resnick et al. 2000). 2.3  Buyer's trust in online auctions According to Swan and Nolan (1985), trust is especially critical when two situational  factors are present in a transaction: uncertainty and incomplete product information, which leads to information asymmetries in online auctions. To promote trust and reduce opportunism in online auctions, credible signals should be provided to facilitate differentiation among sellers. The introduction of the reputation system increases both the level of buyer's trust and the level of seller's trustworthiness (Keser, 2002). Resnick and Zeckhauser (2001) reported that during their five-month eBay data collection period, 89.0% of all seller-buyer pairs conducted just one transaction, and 98.9% conducted no more than four. Therefore, most online auction participants are strangers to each other. It is difficult to build trust between strangers. Strangers do not have shared past histories or the prospect of future interactions, and they are not subject to a network of informed individuals who will punish bad and reward good behaviour by any of them. "The temptation to 'hit and run' outweighs the incentive to cooperate, since the future casts no shadow (Resnick, Zeckhauser, Friedman & Kuwabara 2000)." In online auctions, reputation mechanisms seek to build trust between participants who have never met by creating an expectation that other people will look back upon the reputation feedback, so as to restore the shadow of the future to each transaction. By sharing opinions about the seller after each transaction, a meaningful history of that seller will be constructed. Future buyers may base their bidding decisions on a sufficiently extensive  14  public history. If the buyers do behave this way, the seller's reputation will affect his/her future sales. Hence, he/she will seek to accumulate as many positive points and comments as possible, and to avoid negative feedback (Resnick, Zeckhauser, Friedman & Kuwabara 2000). The trust-building process is driven by the buyer's calculation that the costs to the seller of acting in an untrustworthy manner exceed the benefits of such actions. Good feedback will lead buyers to trust sellers, not only because good feedback provides a signal of trustworthiness to potential buyers, but because sellers also have incentives to guard their good feedback profiles (Ba & Pavlou 2002). 2.4  Uncertainty in online auctions and the bid amount  In this research, we study bid amounts in online auctions, under the condition of uncertainty related to seller identification and product quality. Lucking-Reiley, et al. (2000) reported that bid price in online auctions is affected by the market value of the item, the minimum bid, the number of negative comments, and the auction length. Kauffman and Wood (2001) extended the research on the bid price by analyzing the effects of uncertainty on purchases for e-Commerce, based on Nagel and Holden's (1995, p.75) economic value analysis (See Figure 2).  15  Positive differentiation value Reference value  Negative differentiation value t Total economic value 1  Figure 2 Nagel and Holden's economic value analysis  In Nagel and Holden's analysis (1995, p.75), buyers may be willing to pay an additional positive differentiation value because of a product's unique features. While, negative differentiation value is a reduction in value due to unique undesirables. Kauffman and Wood (2000) applied this analysis to online auctions (See Figure 3), in which seller and product uncertainty negatively affect the price of items. According to this model, higher uncertainty levels in online auctions lead to lower bids. Bids made by different bidders on an item with reference value may vary due to personal judgements on the level of uncertainty with the auction.  16  Positive differentiation value: • Convenience (++)  Negative differential value: • Seller uncertainty • Product uncertainty  Reference value  Figure 3 Kauffman and Wood's analysis of effects on purchases for traditional and e-Commerce markets  Using a reputation mechanism, bidders judge a seller's trustworthiness, thereby predicting the future behaviour of the seller, which comprises a large part of the uncertainty in the auction. An efficient reputation mechanism may help bidders accurately assess the level of this uncertainty, and allow them to make their bidding decisions accordingly.  17  3  Research framework  3.1  Independent variables - reputation mechanisms  We propose four reputation mechanisms. In these proposed mechanisms, as in the eBay mechanism, a seller's reputation score and positive feedback percentage is calculated according to the amount of positive, neutral and negative feedback from distinct buyers. Unlike in the eBay mechanism, weights for feedback time and/or transaction prices related to the feedback are added into the reputation calculation and presentation in our mechanisms. 3.1.1  Existing eBay mechanism  The eBay reputation mechanism encourages buyers to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment, after each transaction. To calculate a seller's reputation score, eBay adds up the numerical rating over time. Only ratings from unique buyers are used in computing the overall reputation presented as "feedback score". An "ID Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives," "neutrals" and "negatives." The numbers of ratings of each type given to the seller are presented in three time periods: the past sever days, the past month, and the past six months (See Figure 4).  18  Feedback Summary  .  582 positives,582 are from:unique users: 0 heijtrals  18 r.euat-Y&s 18 are from Lriquo users  ,•  —-  -  >  d}  9  • ID  - ^ ^ ^ - ^ ^ s ! - ^ : ^ - - -  C3TCl  'iw35 (  Summary of Most Recent Reviews Pas: 7 days  -See airfeedback^eviews,for ns35!  564)  <j,M«mbaislnct;25 Oct03 Location: Canada  Positive!  :2p;  Neutral :;;Negati»e;  Feedback Score Positive Feedback  0  Past.month-  Past6mo;  ;82  582;  0 3.1.8^.-. •  C '•'-•16:'.  554 97.0%  Figure 4 eBay mechanism  The seller in Figure 4 has received 582 positives and 18 negatives from unique users. His reputation score = 582 - 18 = 564 His positive feedback percentage = 582 / 600 = 97.0% 3.1.2  Time-weighted mechanism  With Time-weighted mechanism, buyers are encouraged to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment, after each transaction. To calculate a seller's reputation score, ratings are grouped according to the time they were left, and different weights are given to these groups. Ratings older than six months are discarded; ratings between one and six months ago are given a 0.2 weighting; between seven days and one month old, 0.3; within the last seven days, 0.5. A seller's reputation score is calculated according to the number of instances of positive and negative feedback. The positive feedback percentage is calculated by aggregating the time weighted positive percentage of the three groups. Only ratings from unique buyers are used in computing the overall reputation score. An "ID Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives,"  19  "neutrals" and "negatives." The number of feedback responses and the percentage of positives are presented for different time periods, as well as the weighted percentage of positives, using the same weights as above (See Figure 5). Feedback Summary 582  poi.twes.  —  582 are from -..rque users.  6 neutrals  v.-.^j^Sf |Q  —  —  —  C3fd  ns35 (564)  vJ.Membefjlnce: 25 Oct 03 Location: Canada  18 nQgatff8S:;18 are from unique users.-'  Summary of Most Recent Reviews  ;  S e e all feedbackTeyiewsfor. ns35. •  Past 7 c a y s  / days • 1 month age  .Positive-  20  62;  Neutral  ;0,  0  ^Negatives  '5;  Weight Positive Feedback  O.S 80.0%  13;-.'  1 month - 6 months a g o ; .503  Total 582-  0:  p,  U  18  03i/ 82.7%  Figure 5 Time-weighted mechanism  The seller in Figure 5 has received 582 positives and 18 negatives from unique users. His reputation score = 582 - 18 = 564 His positive feedback percentage = 80.0% x 0.5 + 82.7% x 0.3 + 100% x 0.2 = 84.8% 3.1.3  Price-weighted mechanism  Using the Price-weighted mechanism, buyers are encouraged to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment, after each transaction. To calculate a seller's reputation score, ratings are grouped according to the price of the related transaction, and different weights are applied to these groups. Ratings of transactions between $0 and $20 are given a 0.2 weighting; between $20 and $100, 0.3; and more than $100, 0.5. A seller's reputation score is calculated according to the number of positive and negative feedback responses. The positive feedback percentage is calculated by aggregating the total price-weighted positive percentage of the three groups. Only ratings from unique buyers are used in computing the overall reputation score. An 20  "JD Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives," "neutrals" and "negatives." The number of feedback responses and the percentage of positives in different price bands are presented, as well as the weighted percentage of positives, using the same weights as above (See Figure 6). Feedback Summary  < V.**-  |fj) CSTCi  582;positrWsV;582^ 6 neutrals  <j M.mborslnce:25 Oct03Locatlon: Canada.,  18;ne'gative's',18 are from unique jsers.  vlj  -Its35 (564V  • Summary of Most Recent Reviews • See ail feedback''re<iews',for. ns35.  TiansactionValiie » 3 • $23  $2C• J1UU  JIDD,^  Positive  ,/flS?/  ,121;:  .43;  Neutral  0  0  0  0.  ./..Oi.  18'  fj • , i|  sj ,-Negative-  total >582 fj;' '"'1.8:  Figure 6 Price-weighted mechanism  The seller in Figure 6 has received 582 positives and 18 negatives from unique users. His reputation score = 582 - 18 = 564 His positive feedback percentage = 100% x 0.2 + 100% x 0.3 + 70.5% x 0.5 = 85.2% 3.1.4  Price-time-weighted-double mechanism  With a Price-time-weighted-double mechanism, buyers are encouraged to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment, after each transaction. To calculate a seller's reputation score, different weights are applied to the ratings according to the time they were left and the related transaction price. With respect to time, ratings older than six months are discarded; ratings between one and six months old are given 0.2 weighting; between seven days and one month, 0.3; within the last seven days, 0.5. With respect to price, ratings for transactions between $0 and $20  21  are given a 0.2 weighting; between $20 and $100, 0.3; and more than $100, 0.5. The overall score is calculated by subtracting the number of negatives from the number of positives. The weighted percentage of positives is the average of time-weighted percentage of positives and price-weighted percentage of positives. Only ratings from unique buyers are used in computing the overall reputation score. An "ID Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives," "neutrals" and "negatives." The number of feedback instances and the percentage of positives are presented in different time bands and price bands, as well as the weighted percentage of positives, using the same weights as above (See Figure 7). Feedback Summary 582 fzsihioi. 582 are f-orn. i n quo users  <eb'"/ ID card  O.neutrals:  Mambvisince 25 Oct03 Location: Canada*  1^(564)  Summary of Most Recent Reyievys :.18-negatiyes^18 arefront unique users.  Time  .See'all feedback reviews for ns35.. Positive: Neutral Negative.  / days -1 jmonth;agq  1 month - 6 monthsvagb  $20$100  iiob  t20  •20-  .':E2;'  •500  . .418  •121  .43  b  0  0  0  0  pS;  &•  1  fs. ••Wi  Weight Positive Feedback  t r a nsa cli o n. Va I u e:  Past7 'days:  80:0%  627%  0.2  :50:2  100.0%:  Figure 7 Price-time-weighted-double  .84.8%  100 0%  v  'p)i''  Total  + '  .582.  .a  0 •'•  "•IB,  Wi ibo;o%; 7b.5%85:2%  mechanism  The seller in Figure 7 has received 582 positives and 18 negatives from unique users. His reputation score = 582-18 = 564 His positive feedback percentage: •  Time dimension = 80.0% x 0.5 + 82.7% x 0.3 + 100% x 0.2 = 84.8%  •  Price dimension = 100% x 0.2 + 100% x 0.3 + 70.5% x 0.5 = 85.2%  22  •  Overall positive feedback percentage = ( 84.8% + 85.2% ) / 2 = 85.0%  3.1.5 Price-time-weighted-single mechanism Using a price-time-weighted-single mechanism, buyers are encouraged to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment, after each transaction. To calculate a seller's reputation score, different weights are assigned to the ratings according to the time they were left and the transaction price. With respect to time, ratings older than six months are discarded; ratings between one and six months old are given a 0.2 weighting; between seven days and one month, 0.3; within the last seven days, 0.5. With respect to price, ratings for transactions between $0 and $20 are given a 0.2 weighting; between $20 and $100, 0.3; and more than $100, 0.5. The overall score is calculated by subtracting the number of negative feedback responses from the positive. The overall positive feedback percentage is calculated by weighting the percentage of each category by both its time and price weights, then summing them up. Only ratings from unique buyers are used in computing the overall reputation score. An "ID Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives," "neutrals" and "negatives." The number of feedback responses and the percentage of positives are presented in different time bands and price bands, as well as the weighted percentage of positives, using the same weights as above (See Figure 8).  23  Feedback Summary 582 positives. 582  are (torn un:qje users.  ID card  ns35 (564)  Member since: 25 Oct OSLocation: Canada  0 neutrals;  Summary of Most Recent Reviews Time Past 7 days  Transaction Value Weight  $3 • $20  0.2  0.5  14 (100.0%)  7 days -1/month ago  1 :month - 6months '"" -ago '•'  Total  0.3  i i (1C0 o%;  .360 (100.0%)  0.3  4 (100 0%)  12.000,0%)  '05(1:3 0%)  $100 + ,0.5  2 (26.6%)  6 (31.6%)  35(ido:o%)  520-$100 Positive Feedback  Figure 8 Price-time-weighted-single mechanism  The seller in Figure 8 has received 582 positives and 18 negatives from unique users. His reputation score = 582 - 18 = 564 His positive feedback percentage = 100% x 0.2 x 0.5 + 100.0% x 0.3 x 0.5 + 28.6% x 0.5 x 0.5 + 100% x 0.2 x 0.3 + 100.0% x 0.3 x 0.3 + 31.6% x 0.5 x 0.3 + 100% x 0.2 x 0.2 + 1005% x 0.3 x 0.2 + 100% x 0.5 x 0.2 = 71.9% 3.2  Dependent variables  3.2.1  Trust  We follow Ba and Pavlou's (2002) definition of trust in online auctions as "the subjective assessment of one party that another party will perform a particular transaction according to his or her confident expectations, in an environment characterized by uncertainty." This definition captures two important attributes of trust: first, the confident expectation encompasses the possibility of a (mutually) beneficial outcome, and second, the uncertain environment suggests that delegation of authority from one party to another may have adverse (harmful) effects on the entrusting party.  24  Trust in an online auction is credibility-based trust, which is the belief that the other party is honest, reliable, and competent. It originates from a subjective calculation of the costs and benefits of the other party's cheating, subject to the other party's reputation as it is perceived by a network of online auction participants (Ba, & Pavlou, 2002). 3.2.2  Confidence and helpfulness in bidding judgment  In situations characterized by uncertainty related to product quality and seller identification, bidders must judge a seller's trustworthiness, and thereafter make their bidding decisions, including whether or not they are going to bid, and how much they are going to bid. A reputation system can be an information source for the bidders upon which the bidders make decisions. We may evaluate the effectiveness of a reputation system by examining whether the reputation mechanism induces confidence in bidders when making bidding decisions, and whether the information provided by the reputation mechanism is helpful in decision making. We perform measures on bidders' confidence in their judgments about sellers' trustworthiness and the perceived helpfulness of the reputation system. 3.2.3  Bid amount  As noted in section 2.5, bids made by different bidders may vary because of their different personal judgments about seller uncertainty. A bidder may bid less than the reference value of an item in an online auction, since he is unsure whether the seller will honor the terms of the auction and successfully complete the transaction. Thus, the difference between the bid and reference price of the item reflects the bidder's perception concerning the uncertainty level of the auction, the seller's trustworthiness. We measure  25  the bid to compare the efficiency of different mechanisms in helping bidders to judge sellers' trustworthiness. 3.3  Hypothesis development  3.3.1 Bid amount According to Kauffman and Wood (2000), the higher the uncertainty in an online auction, the lower bidders will be willing to pay. The reputation mechanisms are there to help bidders accurately assess the uncertainty associated with the auctions. With information in time and/or price dimensions, we predict that the proposed reputation mechanisms may help bidders better assess the uncertainty associated with the auctions, thus bidding less in auctions with dishonest sellers, and more to honest sellers, as compared to the eBay mechanism. Hla  Compared to EM, under TM, bidders place lower value bids to sellers who  have negative feedback responses increasing over time.  Hlb  Compared to EM, under PM, bidders place lower value bids to sellers  who have negative feedback responses increasing along with the transaction value.  Hlc  Compared to EM, under PTDM, bidders place lower value bids to sellers  who have negative feedback responses increasing over time or along with the transaction value.  26  Hid  Compared to EM, under PTSM, bidders place lower value bids to sellers  who have negative feedback responses increasing over time or along with the transaction value.  3.3.2  Trust  As discussed in previous sections, the buyer's trust in sellers is critical in markets with information asymmetry, such as online auctions. A reputation system is a good way to promote buyer trust. With information on time and/or price dimensions, we expect reputation mechanisms to better help buyers to judge seller behaviour, thus inducing more buyer trust in honest sellers, and less trust in sellers behaving dishonestly. We propose the following hypotheses on buyer trust:  H2  Compared to the eBay mechanism (EM), the Time-weighted mechanism  (TM), Price-weighted mechanism (PM), Price-time-weighted-double mechanism (PTDM), and Price-time-weighted-single mechanism (PTSM) induce higher levels of trust in sellers who do not have negative feedback responses increasing over time or along with the transaction value.  H3a  Compared to EM, TM induces a lower level of trust in sellers who have  negative feedback responses increasing over time.  H3b  Compared to EM, PM induces a lower level of trust in sellers who have  negative feedback responses increasing along with the transaction value.  27  H3c  Compared to EM, PTDM induces a lower level of trust in sellers who have  negative feedback responses increasing over time or along with the transaction value.  H3d  Compared to EM, PTSM induces a lower level of trust in sellers who have  negative feedback responses increasing over time or along with the transaction value.  According to Resnick and Zeckhauser et al. (2000), a good reputation mechanism should be able to distinguish trustworthy sellers and non-trustworthy sellers. With time and/or price information provided, we hypothesize that the proposed reputation mechanisms perform better in distinguishing behaviour patterns. Here we are comparing the difference of trust levels between trustworthy seller and non-trustworthy sellers, as well as between non-trustworthy sellers with different dishonest behaviour patterns, across the five mechanisms. H4a-1 Compared to EM, TM induces more difference in buyer trust level, between sellers with completely positive feedback responses and sellers who have negative feedback responses increasing over time. H4a-2 Compared to EM, TM induces more difference in buyer trust level, between sellers who do not have feedback responses increasing over time or along with transaction value, and sellers who have feedback responses increasing over time.  28  H4b-1 Compared to EM, PM induces more difference in buyer trust level, between sellers with completely positive feedback responses and sellers who have negative feedback responses increasing along with the transaction value. H4b-2 Compared to EM, PM induces more difference in buyer trust levels, between sellers who do not have feedback responses increasing over time or transaction value, and sellers who have feedback responses increasing over time. H4c-1 Compared to EM, PTDM induces more difference in buyer trust levels, between sellers with completely positive feedback and sellers who have negative feedback responses increasing over time and along with the transaction value. H4c-2 Compared to EM, PTDM induces more difference in buyer trust levels, between sellers who do not have negative feedback responses increasing over time or along with the transaction value, and sellers who have negative feedback responses increasing over time and along with the transaction value.  H4d-1 Compared to EM, PTSM induces more difference in buyer trust levels, between sellers with completely positive feedback responses and sellers who have negative feedback responses increasing over time and along with the transaction value.  H4d-2 Compared to EM, PTSM induces more difference in buyer trust levels, between sellers who do not have negative feedback responses increasing over time or along with the transaction value, and sellers who have negative feedback responses increasing over time and along with the transaction value.  29  3.3.3  Confidence  We expect the mechanisms with time and/or price dimensions to outperform the eBay mechanism by inducing more confidence in judgments about seller trustworthiness, and mechanisms with time and price dimensions to outperform the mechanisms with only time or price dimension. H5a  Compared to EM, TM induces more buyer confidence in judgments about  seller trustworthiness. H5b  Compared to EM, PM induces more buyer confidence in judgments about  seller trustworthiness. H5c-1 Compared to EM, PTDM induces more buyer confidence in judgments about seller trustworthiness. H5c-2 Compared to EM, PTSM induces more buyer confidence in judgments about seller trustworthiness. H5d-1 Compared to TM, PTDM induces more buyer confidence in judgments about seller trustworthiness. H5d-2 Compared to TM, PTSM induces more buyer confidence in judgments about seller trustworthiness.  H5e-1 Compared to PM, PTDM induces more buyer confidence in judgments about seller trustworthiness.  30  H5e-2 Compared to PM, PTSM induces more buyer confidence in judgments about seller trustworthiness. 3.3.4  Helpfulness  In the series of hypotheses related to helpfulness, as in the confidence series, we expect that the mechanisms with time and/or price dimensions to be more helpful in judging the seller trustworthiness, as compared to the eBay mechanism; and that mechanisms with time and price dimensions are more helpful in judging seller trustworthiness, compared to mechanisms with only a time or price dimension. H6a  Compared to EM, bidders consider TM more helpful in judging seller  trustworthiness. H6b  Compared to EM, bidders consider PM more helpful in judging seller  trustworthiness. H6c-1 Compared to EM, bidders consider PTDM more helpful in judging seller trustworthiness.  H6c-2 Compared to EM, bidders consider PTSM more helpful in judging seller trustworthiness. H6d-1 Compared to TM, bidders consider PTDM more helpful in judging seller trustworthiness.  H6d-2 Compared to TM, bidders consider PTSM more helpful in judging seller trustworthiness.  31  H6e-1 Compared to PM, bidders consider PTDM more helpful in judging seller trustworthiness. H6e-2 Compared to PM, bidders consider PTSM more helpful in judging seller trustworthiness.  32  4  Experiment design  4.1 Overview We used a controlled laboratory experiment to test the proposed hypotheses. We recruited 150 university students and staff members to the experiments; 66 of them were males, 84 were females, with an average age of 23.6. On average, they had Internet experience of 7.3 years, spent 22.7 hours on Internet every week. 41% of the participants had bid online, and 29% had won online auctions. An online auction website for the experiments was built with an interface similar to that of eBay.ca (See Figure 9). Participants were invited to play an online auction game. In the experimental website, participants browsed and bid in six auctions, and also answered online questionnaires immediately after each bid and after the entire experiment.  - Ho  Edit  1  Wow  f svontas  Toots-  Hrip  home.I register I sign..in/out I services I site map Ihelp B r o w s e . [ S e a r c h .j  *>• BaCK IP list Of Items  Sell  |' MyeOay  j  Community-  Llit«d In •cittoowi . C o m o u f r.ft Elactronlc»Portabl« Audiofl.Vld«o>MP3 P l i « i n » t o a l « ' I P a d - . . 1  10GB A p p l e IPod with Extras in Original[ Box_  Item numben30540624a3i;;  • Y o u a r e s i g n e d In :  . W a t c h tliis Item ftrack it in My.eBay)  Sellor information  •••.Current bid:- • C$110Jill • •  ilo>eauetion (374 )  . - RoceBidV  Feedback ratingi 374 ' : PoalHv* hMdb*< fc V» / * » w-;v.-R.«fll*t«f«d Canada  ;  Time left:  27 h o u r s 18 m l n s 3 day l«Ung .•••.•••••.••-•/.•.•.•. er.d.».-13--^ufl-0').'O5 > iO^S'.EVl"'  History:.-  Bfi.sjj.JopiJUac.lv.r.evicyyB A s k seller a quflstion  4 b|08 i,c.(in.ivo.zesting bid)  :Vv\Vi9W-eeller'B oth6 :  High bidder .-pesasons  Si O :feuwiwlth'Confldci  ••.Location: ••.•..•.• -.Toronto  ••.4- ghipP' .S a"d-payment-details-. .••• n  Description  .:. Seller, a s s u m e s all.responsibility,for,listing;this-item:-v;ri*A  . This,has been my best friendfor the last year (it was purchased new at.a-.retailstore-on 09/t4/2003).-.lt"s m.excellentshape andworkslike a charm. It c o m a s m the.original-.packaging with all the accessories including earphones,.remote, carrying c a s e , FireWire cable; A C adapter.-  -.  \  FEATURES • H o l d s over 2.000. songs at near-CD quality.on 10GB hard dnva -.-  .  -  •.  Figure 9 Screen capture of the experiment website  33  A research design with one treatment of five levels was adopted. Participants were randomly assigned in a group using one of the five reputation mechanisms. The treatment in the experiments is the reputation mechanisms: EM, TM, PM, PTDM and PTSM. 4.2  Experiment settings  4.2.1  The game  In the experiments, participants were invited to join an online auction game in which they were given C$600 to bid in six different auctions, selling six different items, held by six different sellers. Participants competed against each other in the auctions within the groups. Their task was to decide how to allocate the C$600 by placing the maximum bid for each auction. They did not have to spend all C$600, and they had the opportunity to win the leftover funds after the study. If they won auctions, they would receive lottery tickets that qualified them to win their "pot" of unspent funds. The chances of winning the pot depended on whether they bid high enough to win auctions and receive lottery tickets. The reliability of each seller and the likelihood of successful completion of the transactions were pre-set. There was a probability, based on the reliability of the seller, that a specific transaction would not be completed successfully. The consequence of winning an unsuccessful transaction was that the winner would not receive the associated lottery tickets for winning that auction. Thus, participants needed to consider the trustworthiness of sellers very carefully before placing the bids, in order to maximize the probability of winning their leftover funds.  34  4.2.2 Seller profiles We designed six seller profiles for the experiments, using 600 pieces of feedback in total. All sellers registered on the same day, which was approximately six months before the date of the experiments. The seller's behaviour patterns differed from one another. One seller was honest for his "whole life" with all positive feedback. The other 5 sellers had both positive and negative feedback, and no neutral feedback. Some of them had built good reputations during the initial period after their registration, while some had negative feedback as time proceeded. Some cheated in the transactions with high value items, and behaved honestly in those with low value items. Table 1 summarizes the positive and negative feedback totals for the six sellers in the time dimension and price dimension. (See Table 1 and Table 2).  35  Number of feedback Sellers  Time dimension (in months)  Seller details  Price dimension (C$)  1  2  3  4  5  6  Total  $0-20  $20-100  $100+  Total  Pos.  102  103  98  100  97  100  600  418  121  61  600  Neg.  -  -  -  -  -  -  -  -  -  -  -  Pos.  99  100  95  97  94  97  582  406  117  59  582  Neg.  3  3  3  3  3  3  18  12  4  2  18  Pos.  102  103  98  100  97  82  582  406  117  59  582  Neg.  -  -  -  -  -  18  18  12  4  2  18  Honest  Random (representative eBay seller)  Time increase (dishonest)  Pos.  99  100  95  97  94  97  582  418  103  61  582  Neg.  3  3  3  3  3  3  18  -  18  -  18  Pos.  99  100  95  97  94  97  582  418  103  61  582  Price low low (dishonest)  This seller h a s perfectly positive feedback. T h i s seller h a s 18 p i e c e s of negative feedback evenly distributed in t h e most recent sixmonth period, falling into t h e 3 price bands. T h i s seller h a s 18 p i e c e s of negative feedback in the most recent month, but d o e s not h a v e a n y in the previous 5 months. The negative f e e d b a c k falls into the 3 price bands. T h i s seller h a s 18 p i e c e s of negative feedback evenly distributed in the most recent sixmonth period. A l l the negative feedback relates to transactions belonging to the median price band. T h i s seller h a s 18 p i e c e s of negative feedback  evenly  distributed  in  most  Price low high (dishonest)  month Neg.  3  3  3  3  3  3  18  -  -  period.  the 18  18  the  recent sixAll  negative  feedback to  relates  transactions  belonging  to  the  high price band. T h i s seller h a s 18 Pos.  102  103  98  100  97  82  582  418  121  43  582  negatives most  in  the  recent o n e  month, while d o e s not  Time price  have  any  previous  increase  months.  (dishonest) Neg.  -  -  -  -  -  18  18  -  -  18  18  negative  in 5  All  the  feedback  relates  to  transactions belong  to  price band.  Table 1 Seller's profiles  36  high  Sellers  Behaviour pattern changes according to Time  Transaction price  Honest  No  No  Random  No  No  Time increase  Yes  No  Price low low  No  Yes  Price low high  No  Yes  Price time increase  Yes  Yes  Table 2 Summary of seller's behaviour patterns  All participants in the five treatment groups bid in auctions held by these six sellers. Under different mechanisms, the seller's profiles were presented in different ways. Except for the honest seller, all other sellers had the same reputation scores under the eBay mechanism and also the scores follow the similar trends. Thus, these six sellers could not be differentiated using the eBay reputation score. Figure 10 shows the eBay scores of the six sellers during the most recent six months.  Reputation score  — « — Honest — a — Random -~&-"- T i m e i n c r e a s e —x—  P r i c e low low  — 3 K — P r i c e low h i g h — • — T i m e price  Figure 10 eBay reputation scores of the six sellers during the most recent six months  37  However, under the proposed mechanisms, including information in time and/or price dimensions and corresponding weighting added in the reputation score calculation, the six sellers can be distinguished (See Figure 11).  Positive percentage  105.0%  -•— • —  Honest Random Time i n c r e a s e  -X— —  Price low low Price low high  75.0%  70.0%  Mechanisms eBay  TM  PM  PTDM  PTSM  Figure 11 Positive percentages of the six sellers under five treatment mechanisms  38  4.2.2.1 Honest seller This seller behaved honestly and did not receive any negative feedback since being registered, and the feedback relates to transactions belonging to all three price bands (See Figure 12). We created this extreme positive seller profile for experimental purposes. It is not representative of the experienced honest online auction sellers, because even a very well-behaved seller has the chance to receive some negative feedback over time in the online auction business.  Number of feedback responses - time dimension 100 80 Number of feedback responses 40  • Positive  6 0  20 2  3  4  5  Time (month) Number of feedback responses - price dimension 450 400 350 Number of feedback reponses  ^ B Positive  2 Q Q  121  1 5 Q  • 61  100 50 $0-20  $20-100  $100+  Price ($C)  Figure 12 Behaviour pattern - Honest seller  39  4.2.2.2 Random seller This seller's behaviour does not change according to time or the price of the transactions. The seller has the negative feedback evenly distributed in the most recent six-month period, falling into the three price bands (See Figure 13). It is almost impossible to satisfy every buyer in transactions. Negative feedback is inevitable for frequent online auction sellers. Having a small amount of negative feedback does not mean the seller is doing business badly or cheating on purpose. Hence, we regard this random seller as a good representative of a normal experienced online auction seller.  Number of feedback responses - time dimension 120 100  100  fl*  9 7  +•  fl*  97  80 Number of feedback responses  • Positive 60 4  • Negatvie  Q  20 3 3  A  3;  4  Time (month)  Number of feedback r e s p o n s e s - price dimension 450 400 350 300 Number of 250 feedack 200 reponses 150 100 50  -406  I Positive I Negatvie  117 59 "3 $0-20  $20-100  —z $100+  Price ($C)  Figure 13 Behaviour pattern - Random seller  40  4.2.2.3 Time increase seller This seller's behaviour changes according to time. The seller has negative feedback in the most recent month, but does not have any in the previous five months. The negative feedback falls into the three price bands (See Figure 14).  Number of feedback responses - time dimension 120 100  J  f  l  - ^ 2 3 ^ ^  2  97  80 • Positiws  Number of feedback 60 responses  • Negatvie  4 Q  20 2  3  4  Time (month)  Number of feedback responses - price dimension 450  -406-  400 350 300 Number of 250 feedback 200 responses 150  FS Positive • Negatvie 117.  100 50  -59 12 $0-20  ~\ $20-100  2  $100+  Price ($C)  Figure 14 Behaviour pattern - Time increase seller  41  4.2.2.4 Price low low seller This seller's behaviour changes according to the price of the transactions. The seller has the negative feedback evenly distributed in the most recent six-month period. All the negative feedback relates to transactions belonging to the median price band (See Figure 15).  Number of feedback responses - time dimension 120 100  -SSL  100  80 Number of feedback 60 responses ^  • Positive • Negatvie  20 -A  sfas 3  3  l  4  Time (month)  Number of feedback responses - price dimension 450  .4.18-  400 350 300 Number of 250 feedback 200 responses 150  I Positive I Negatvie  103  100  61 18  50 $0-20  $20-100  $100+  Price ($C)  Figure 15 Behaviour pattern - Price low low seller  42  4.2.2.5 Price low high seller This seller's behaviour changes according to the price of the transactions. The seller has the negative feedback evenly distributed in the most recent six-month period. All the negative feedback relates to transactions belonging to the high price band (See Figure 16).  Number of feedback responses - time dimension 120 100 80 • Positive  Number of feedback 60 reponses ^  • Negative  20 13.11 A . 3  ••hili^VMIBIIIBM^IM  2  3  4  3  A  5  6  Time (month)  Number of feedback responses - price dimension 450  ,  44S-  400 350 300 Number of 250 feedback 200 responses 150  I Positive I Negatvie  103  100  6118  50 $0-20  $20-100  $100+  Price ($C)  Figure 16 Behaviour pattern - Price low high seller  43  4.2.2.6 Price time increase seller This seller's behaviour changes according to both the time and price of the transactions. The seller has negatives in the most recent month, but does not have any in the previous five months. All the negative feedback relates to transactions belonging to the high price band (See Figure 17).  Number of feedback responses - time dimension 120 100  .aa.  100  •~o~  80 Number of feedback 60 responses ^  • Positive • Negative  20  1$ 3  4  Time (month)  Number of feedback responses - price dimension 450  -4.18-  400 350 300 Number of 250 feedback 200 responses 150  1 Positive m Negatvie  121  100  43  50 $0-20  $20-100  18  $100+  Price ($C)  Figure 17 Behaviour pattern - Price time increase seller  44  4.2.3 Auctions and items We had six auctions in the experiments. All of them were seven-day listing auctions located in Vancouver, Canada, with current bids of about C$20, which started from C$1. The end time of all six auctions was in three-and-a-half days. According to a pre-study survey of commerce undergraduate students on their interest in online auction products, we chose the following items for the auction experiments: 1) Timex sports watch, 2) Dell flash 128M memory key, 3) Samsung DVD player, 4) DVD movie: Friends Collection, 5) RCA MP3 player, 6) Mountain Equipment sports bag. We provided detailed and professional item descriptions on the auction page, as well as retail reference prices of around C$100. Pictures of the items were also presented. Please refer to Appendix 3 for details of the auctions and items. 4.2.4 Proxy bidding Participants used the proxy bidding method, similar to the one used by eBay. A bidder enters the maximum amount that he is willing to bid, and the amount is kept confidential from other bidders. Then the system will bid for him as the auction proceeds, bidding only enough to outbid other bidders. If he is outbid, the system immediately increases his bid. This continues until someone exceeds his maximum bid, the auction ends, or he wins the auction. Proxy bidding provides convenience because bidders do not need to constantly monitor the auction. The winner of the auction pays the second highest proxy bid plus one minimum bid increment. In real online auctions, bidders may check back to the auction from time to time. If they are outbid in the auctions, they may re-submit a higher maximum bid as long as the  45  auction has not ended. In the experiments, in contrast, the participants had only one chance to enter their maximum bid for each auction. Once they had submitted their bids, they could not change them. They would not know the result of the auctions until the end of the entire study, which was two months afterward. Therefore, they did not know whether they won or lost in the auction right after they submitted their bids, nor even after the end of the experiments. Given this, they were encouraged to consider carefully the maximum amount they were willing to place in the auctions. 4.3  Experiment procedure  The experiments took place over the course of six weeks. There were 150 UBC students and faculty members who participated in the study. They were paid a guaranteed C$15 for participation of around 80 minutes in the experiments. Step 1: Training. After understanding the auction task, participants were trained how to bid in online auctions, and how to read sellers' reputation information, presented in one of the five reputation mechanisms. A practice exercise in bidding was given after the training, to make sure every participant fully understood how to bid in online auctions. Participants were encouraged to ask any questions they had about online auctions during the experiments. Step 2: Main task. Participants were asked to browse six auctions and bid in them. All six auctions were presented to the participants simultaneously, so that they could switch from auction to auction and make their decisions on their bids in each auction, before they actually placed the bids.  46  Step3: Questionnaire (See Appendix 3). After each auction, whether the participants bid or not, they were asked to fill out a questionnaire, asking about their experiences with the specific auction they had just completed. After all six auctions and the associated questionnaire following each auction was completed, participants were asked to complete another questionnaire about their feelings related to all six auctions. Step 4: Debrief. Participants were told that they would be informed if they won the lottery tickets and their pot of unspent funds.  47  5  Results  5.1  Data analysis overview  5.1.1  Reliability analysis  The data was checked for completeness before statistical tests were performed. All questionnaires were checked for missing data. Means for the dependent variables - trust, confidence, helpfulness, and maximum bid - were calculated according to the five treatment conditions. Next, we calculated the reliability of these measurements. The values of Cronbach Alpha are presented in Table 3. Results show that the Cronbach Alpha of all of our dependent variables is over 0.8, which suggests the measurements are valid. Dependent variables  Number of items  Number of cases  Cronbach Alpha  Trust  10  867  0.9735  Confidence  2  890  0.8413  Table 3 Reliability analysis  48  5.1.2  Factor analysis upon Trust and Confidence  Principle component extraction with Varimax (Kaiser Normalization) rotation was conducted upon items with dependent variable Trust and Confidence. Two factors were extracted, and they account for 82.0398% of the variance. Results are shown in Table 4 Table 5. Initial Eigenvalues Component  Total  % of Variance  Cumulative %  1  8.589396  71.5783  71.5783  2  1.255384  10.4615  82.0398  Extraction method: Principal component analysis Table 4 Results of factor analysis on Trust and Confidence: total variance explained  Component 1  Component 2  Trust item 1  0.899  0.212  Trust item 2  0.903  0.220  Trust item 3  0.870  0.210  Trust item 4  0.728  0.212  Trust item 5  0.914  0.229  Trust item 6  0.801  0.301  Trust item 7  0.922  0.231  Trust item 8  0.898  0.224  Trust item 9  0.886  0.227  Trust item 10  0.880  0.205  Confidence item 1  0.229  0.899  Confidence item 2  0.231  0.898  Table 5 Results of factor analysis on Trust and Confidence: rotated component matrix  Factor analysis showed that all Trust items are loaded into the first component, and both of the Confidence items are loaded into the second component. According to Hair, et al. (1998), loadings of +0.50 or greater are considered to be practically significant.  49  Therefore, we believe that the Trust and Confidence are actually two separate dependent variables. 5.1.3 Correlation between Trust and Confidence We did both Linear and Quadratic regressions to investigate the correlation between dependent variable Trust and Confidence (See Table 6).  Unstandardized Coefficients B  Std. Error  Standardized Coefficients Beta  (Constant) Confidence  1.3134 0.6731  0.2347 0.0416  (Constant) Confidence Confidence **2  2.4744 0.2165 0.0434  0.8304 0.3160 0.0298  Model Linear Quadratic  t  Sig.  Adjusted R square  0.4755  5.5954 16.1612  0.000 0.000  0.22523  0.1529 0.3254  2.9800 0.6850 1.4580  0.003 0.494 0.145  0.22621  a. Dependent variable: Trust b. Independent variable: Confidence Table 6 Linear and Quadratic regression statistics with Trust and Confidence  The R squares of the linear regression and quadratic regression were close to each other. However, only the square term of linear regression was statistically significant. Therefore, we may conclude that Trust and Confidence are strongly linear correlated to each other. However, we can not determine causality at this time, and further research need to be done on it. In the following sections, we discuss the results in detail of One way Analysis of Variance (ANOVA) on the variables to analyze the effects of the mechanisms according to the treatment conditions.  50  5.2  Maximum bid  According to the rules of the game, participants did not need to bid in all of the auctions. We had two ways to deal with the cases in which participants did not place bids, when we analyze the value of the bids. First, we eliminated all the "no-bid" cases in the bid value analysis. The second way is to set all the "no-bids" to be C$10. 5.2.1 Eliminate "no-bids" To eliminate the "no-bids", we ignored the auctions in which participants did not place bids in the analysis of the bid value. We compared the bids received by the completely Honest seller, Random seller, Time increase seller, Price low low seller, Price low high seller and Price time increase seller in pairs between the weighted mechanisms and the EM. A one way ANOVA was run to examine the significance of the mean difference between the time and/or price weighted mechanisms and eBay mechanism (See Table 7).  51  Sellers  EM Number of bids  Honest  Mean of bids Mean diff against E M Significance Number of bids  Random  Mean of bids Mean diff against E M Significance Number of bids  Time increase  Mean of bids Mean diff against E M Significance  Price low low  Number of bids Mean of bids Mean diff against E M Significance  Price low high  Number of bids Mean of bids Mean diff against E M Significance  Price time increase  Number of bids Mean of bids Mean diff against E M Significance  All sellers  Total number of bids  TM 24  PM 26  PTDM  PTSM  85.0000  28 87.3907  87.5000  25 97.2252  0.4171  1.9736  2.0829  11.8081  -  0.972  0.871  0.850  0.312  22  27  27  24  26  78.6115 8.4808  69.8441 - 17.2482  -  0.393  0.031  3.8627 0.684  16  12  24  15  12  59.5631 -  70.4167  85.3683  84.1740  10.8535  25.8052  24.6109  67.1333 7.5702  -  0.221  0.020  0.045  0.422  25 78.7612 -  24 78.8546 0.0934  24 64.0517 - 14.7095  22 73.8877 - 4.8735  25 85.0500 6.2888  -  0.992  0.136  0.593  0.557  24  27 91.5748 0.4260  25 72.4284 - 19.5724  21  26  82.5286 - 9.4723  80.7146 - 11.2862  -  0.977  0.098  0.478  0.360  14 63.5000 -  14 68.4293 4.9293  25 59.8112  13 72.0400  11 63.8191  3.6888  8.5400  0.3191  -  0.680  0.678  0.486  0.978  125  130  153  120  125  85.4171 -  87.0923 -  92.0008 -  -  -  -  -  25  83.2296 -  -  86.6850 0.4073 0.971  Table 7 Bids to sellers under different mechanisms  From the results of the experiments, we found that compared with EM, bidders under PM were willing to place lower value bids to the Random seller (sig. 0.031) and the Price low high seller (sig. 0.098), and placed higher value bids to the Time increase seller (sig. 0.020). Although mean differences exist in other comparisons between the proposed mechanisms and EM, none were significant with a p value less than 0.05. We cannot conclude that bidders place higher or lower bids under the proposed mechanisms, compared with EM. Thus,.none of the HI series of hypotheses are supported.  52  According to the rules of the experiment auction game, bidders were not required to bid in all the auctions held by the 6 sellers. They might bid to as many sellers as they liked. Some bidders bid to sellers who they believed would honour the terms of the auctions, and did not bid to those they did not consider to be trustworthy. As a result of this, we have a different number of bids made to the sellers, with different mechanisms. Generally, under all mechanisms, honest sellers (the Honest seller and the Random seller) received a larger number of bids, while dishonest sellers (the Time increase seller, the Price low low seller, the Price low high seller and the Price time increase seller) received a smaller number of bids. Therefore, the statistical power of the bids to different sellers under different mechanisms varied from each other, since the number of bids both within and between mechanisms varied by seller. For example, under PM, we had 25 bids to the Honest seller, however, only 11 bids to the Time price increase seller. 5.2.2  Set "no-bids" to C$10  The current bids of all the auctions in the experiments were C$20. For the participants who did not bid in particular auctions, we considered that they valued the items less than C$20. Therefore, to analyses the value of the bids, we set all the no-bids as C$10, which is the midpoint of the interval C$0-$20. We performed ANOVA upon the bids received by Honest seller, Random seller, Time increase seller, Price low low seller, Price low high seller and Price time increase seller, in pairs between the weighted mechanism and the EM (Table 8).  53  Mechanism  A  B  eBay  Honest Mean diff. (B-A)  Random  Time increase  Price Low low  Price Low high  Mean diff. (B-A)  Mean diff. (B-A)  Sig.  Mean diff. (B-A)  Sig.  Mean diff. (B-A)  Sig.  Sig.  Sig.  Price time increase Mean diff. (B-A)  Sig.  Time  0.2040  0.987  1.0637  0.922  -4.3707  0.611  -6.6847  0.511  3.0593  0.836  0.4665  0.961  Price  7.6847  0.536  -7.0990  0.448  33.0031  0.002  -18.9354  0.063  -19.0720  0.118  15.1999  0.077  2.1590  0.855  -0.0477  0.997  10.3653  0.343  -13.4703  0.179  -18.4932  0.172  1.0635  0.914  10.8422  0.386  8.3916  0.481  -4.8850  0.572  3.3412  0.767  -7.2261  0.57  -6.5975  0.463  Price-time double Price-time single  Table 8 Bids to seller under different mechanism ("No bids" are replaced with C$10)  We did not find difference with the value of the bids between mechanism in a significant level, except those between EM and PM, in auction held by Time increase seller (0.002), Price low low seller (0.063) and Price time increase seller (0.077). 5.3  Trust  We calculated the mean of trust in every seller, with each mechanism. The results are shown in Table 9.  Sellers Mechanisms Honest  Random  Time increase  Price low low  Price low high  Price time increase  EM  6.4259  5.0582  4.1250  5.2622  5.2715  4.1333  TM  6.5933  5.4900  3.8967  5.4333  5.5233  3.8559  PM  6.3656  5.1433  4.9367  4.8215  4.7093  4.7193  PTDM  6.3700  5.5237  4.0722  5.1633  5.0033  3.8833  PTSM  6.2733  5.4967  3.9367  5.4019  5.1537  3.6115  Table 9 Mean of trust in sellers 5.3.1  Trust in sellers  An ANOVA was run to compare the mean difference of trust in sellers. We made the comparison in pairs, between the eBay mechanism and the weighted mechanisms. The results of the ANOVA are shown in Table 10.  54  Sellers •Vfpphanicmc  Honest  Random  Time increase  Price low low  Price low high Mean Diff. (B-A)  Price time increase Mean Diff. Sig. (B-A)  B  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  TM  0.1674  0.193  0.4318  0.140  -0.2283  0.492  0.1711  0.499  0.2519  0.2830  -0.2774  0.422  PM  -0.0604  0.668  0.0851  0.769  0.8117  0.006  -0.4407  0.114  -0.5622  0.019  0.5859  0.057  PTDM  -0.0559  0.687  0.4655  0.085  -0.0528  0.863  -0.0989  0.702  -0.2681  0.279  -0.2500  0.440  PTSM  -0.1526  0.456  0.4384  0.113  -0.1883  0.533  0.1396  0.587  -0.1178  0.579  -0.5218  0.108  A  Sig.  EM  Table 10 ANOVA results of trust in sellers  There is no significant difference in trust in the Honest seller between the weighted mechanisms and the eBay mechanism. PTDM induces more trust in the Random seller at a marginally significant level (0.085). H2 is only partially supported. For trust in the Time increase seller, the Price low low seller, the Price low high seller, and the Price time increase seller, there is no significant difference observed between the proposed mechanisms and the eBay mechanism, with the exception of the trust in the Price low high seller (0.019) and the Price time increase seller (0.057) under PM. Therefore, Hypotheses H3a, H3b, H3c and H3d cannot be supported by the experiment data. 5.3.2  A N O V A results of trust between Random seller and other sellers  In addition to performing tests on the trust level, we also ran ANOVA on the difference in trust level between pairs of sellers, so as to examine the mechanisms' capability of distinguishing trustworthy and non-trustworthy sellers. We predict that the larger the difference in trust level between two sellers, the better they are distinguished by the mechanism. In this analysis, we set the Random seller, who is representative of  55  normal experienced online auction sellers, to be the baseline. We wished to determine the capability of the mechanisms to distinguish dishonest sellers with behaviour patterns changing according to time and/or price from a representative online auction seller with unchanging behaviour pattern - Random seller (See Table 11).  Pairs of sellers Mechanisms  Honest vs Random  Time increase vs Random  Price low low vs Random  Price low high vs Random  Price time increase vs Random Mean Diff. Sig. (B-A)  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  Mean Diff. (B-A)  Sig.  TM  -0.2446  0.371  -0.6937  0.073  -0.2490  0.263  -0.1651  0.570  -0.7586  0.059  PM  -0.1257  0.643  0.6929  0.018  -0.5142  0.041  -0.6325  0.050  0.4514  0.197  PTDM  -0.5016  0.050  -0.5519  0.090  -0.5527  0.033  -0.7188  0.012  -0.7649  0.029  PTSM  -0.5712  0.050  -0.6604  0.038  -0.2872  0.207  -0.5414  0.053  -0.9785  0.011  A  B  EM  Table 11 A N O V A results of difference in trust between Random seller and other sellers  From the results, we see that the Time increase seller and the Price time increase seller can be better distinguished from the Random seller using the TM with a significance level of 0.073 and 0.059 respectively. Also, the Price low low seller and the Price low high seller are distinguished from Random seller with a significance level of 0.041 and 0.050 respectively. Similarly, TPDM easily distinguishes the Price low low seller, the Price low high and the Time price increase seller from the Random seller (sig. 0.090, 0.033, 0.012 and 0.029 respectively). Also, TPSM can perfectly distinguish between nontrustworthy sellers with changing behaviour patterns (Time increase seller 0.038, Price low high seller 0.053, and Price time increase seller 0.011) and a seller with stable behaviour patterns (Random seller). As a result, hypothesis H4a-2 and H4c-2 are fully supported by the results; H4b-2 and H4d-2 are partially supported.  56  5.3.3 Tukey of trust between Random seller and other sellers In addition, we ran Tukey Post Hoc tests which provide more conservative results in pairwise comparison with trust difference between Random seller and other sellers (See Table 12).  Mechanism Pairs of Sellers  A EM  TM  PM Honest vs Random PTDM  B  EM  TM  PM Time increase vs Random  PTDM  Price low low vs Random  EM  Sig.  95% Confidence Interval Lower Bound  Upper Bound  0.2446  0.2444  0.855  -0.4306  0.9197  PM  0.1257  0.2444  0.986  -0.5495  0.8008  PTDM  0.5016  0.2444  0.247  -0.1736  1.1768  PTSM  0.5712  0.2444  0.139  -0.1040  1.2464  EM  -0.2446  0.2444  0.855  -0.9197  0.4306  PM  -0.1189  0.2423  0.988  -0.7883  0.5505  PTDM  0.2570  0.2423  0.826  -0.4124  0.9265  PTSM  0.3267  0.2423  0.662  -0.3428  0.9961  EM  -0.1257  0.2444  0.986  -0.8008  0.5495  TM  0.1189  0.2423  0.988  -0.5505  0.7883  PTDM  0.3759  0.2423  0.531  -0.2935  1.0454  PTSM  0.4456  0.2423  0.356  -0.2239  1.1150  EM  -0.5016  0.2444  0.247  -1.1768  0.1736  TM  -0.2570  0.2423  0.826  -0.9265  0.4124  PM  -0.3759  0.2423  0.531  -1.0454  0.2935  0.0696  0.2423  0.998  -0.5998  0.7391  EM  -0.5712  0.2444  0.139  -1.2464  0.1040  TM  -0.3267  0.2423  0.662  -0.9961  0.3428  PM  -0.4456  0.2423  0.356  -1.1150  0.2239  PTDM  -0.0696  0.2423  0.998  -0.7391  0.5998  TM  0.6937  0.3119  0.177  -0.1680  1.5555  PM  -0.6929  0.3119  0.178  -1.5547  0.1688  PTDM  0.5519  0.3119  0.396  -0.3099  1.4137  PTSM  0.6604  0.3119  0.218  -0.2014  1.5222  EM  -0.6937  0.3119  0.177  -1.5555  0.1680  PM  -1.3867  0.3065  0.000  -2.2335  -0.5399  PTDM  -0.1419  0.3065  0.990  -0.9886  0.7049  PTSM  -0.0333  0.3065  1.000  -0.8801  0.8135  EM  0.6929  0.3119  0.178  -0.1688  1.5547  TM  1.3867  0.3065  0.5399  2.2335  PTDM  1.2448  0.3065  0.3980  2.0916  PTSM  1.3533  0.3065  0.000 0.001 0.000  0.5065  2.2001  EM  -0.5519  0.3119  0.396  -1.4137  0.3099  TM  0.1419  0.3065  0.990  -0.7049  0.9886  PM  -1.2448  0.3065  0.001  -2.0916  -0.3980  PTSM PTSM  Std. Error  A-B  TM  PTSM PTSM  Mean Difference  0.1085  0.3065  0.997  -0.7383  0.9553  EM  -0.6604  0.3119  0.218  -1.5222  0.2014  TM  0.0333  0.3065  1.000  -0.8135  0.8801  PM  -1.3533  0.3065  0.000  -2.2001  -0.5065  PTDM  -0.1085  0.3065  0.997  -0.9553  0.7383  TM  0.2490  0.2344  0.825  -0.3985  0.8965  PM  0.5142  0.2344  0.188  -0.1333  1.1617  57  TM  PM  PTDM  PTSM  EM  TM  PTDM  0.5527  0.2344  0.133  -0.0948  1.2002  PTSM  0.2872  0.2344  0.737  -0.3603  0.9346  EM  -0.2490  0.2344  0.825  -0.8965  0.3985  PM  0.2652  0.2324  0.784  -0.3768  0.9072  PTDM  0.3037  0.2324  0.687  -0.3383  0.9457  PTSM  0.0381  0.2324  1.000  -0.6038  0.6801  EM  -0.5142  0.2344  0.188  -1.1617  0.1333  TM  -0.2652  0.2324  0.784  -0.9072  0.3768  PTDM  0.0385  0.2324  1.000  -0.6035  0.6805  PTSM  -0.2270  0.2324  0.865  -0.8690  0.4149  EM  -0.5527  0.2344  0.133  -1.2002  0.0948  TM  -0.3037  0.2324  0.687  -0.9457  0.3383  PM  -0.0385  0.2324  1.000  -0.6805  0.6035  PTSM  -0.2656  0.2324  0.783  -0.9075  0.3764  EM  -0.2872  0.2344  0.737  -0.9346  0.3603  TM  -0.0381  0.2324  1.000  -0.6801  0.6038  PM  0.2270  0.2324  0.865  -0.4149  0.8690  PTDM  0.2656  0.2324  0.783  -0.3764  0.9075  TM  0.1651  0.2609  0.969  -0.5557  0.8860  PM  0.6325  0.2609  0.115  -0.0883  1.3534  PTDM  0.7188  0.2609  0.051  -0.0020  1.4397  PTSM  0.5414  0.2609  0.237  -0.1794  1.2623  EM  -0.1651  0.2609  0.969  -0.8860  0.5557  PM  0.4674  0.2587  0.374  -0.2473  1.1821  PTDM  0.5537  0.2587  0.209  -0.1610  1.2684  PTSM PM Price low high vs Random PTDM  PTSM  Price time increase vs Random  EM  TM  PM  PTDM  0.3763  0.2587  0.594  -0.3384  1.0910  EM  -0.6325  0.2609  0.115  -1.3534  0.0883  TM  -0.4674  0.2587  0.374  -1.1821  0.2473  PTDM  0.0863  0.2587  0.997  -0.6284  0.8010  PTSM  -0.0911  0.2587  0.997  -0.8058  0.6236  EM  -0.7188  0.2609  0.051  -1.4397  0.0020  TM  -0.5537  0.2587  0.209  -1.2684  0.1610  PM  -0.0863  0.2587  0.997  -0.8010  0.6284  PTSM  -0.1774  0.2587  0.959  -0.8921  0.5373  EM  -0.5414  0.2609  0.237  -1.2623  0.1794  TM  -0.3763  0.2587  0.594  -1.0910  0.3384  PM  0.0911  0.2587  0.997  -0.6236  0.8058  PTDM  0.1774  0.2587  0.959  -0.5373  0.8921  TM  0.7586  0.3459  0.188  -0.1970  1.7142  PM  -0.4514  0.3459  0.688  -1.4070  0.5042  PTDM  0.7649  0.3459  0.181  -0.1907  1.7205  PTSM  0.9785  0.3488  0.045  0.0149  1.9422  EM  -0.7586  0.3459  0.188  -1.7142  0.1970  PM  -1.2100  0.3429  0.005  -2.1574  -0.2626  PTDM  0.0063  0.3429  1.000  -0.9411  0.9537  PTSM  0.2199  0.3459  0.969  -0.7356  1.1755  EM  0.4514  0.3459  0.688  -0.5042  1.4070  TM  1.2100  0.3429  0.005  0.2626  2.1574  PTDM  1.2163  0.3429  0.005  0.2689  2.1637  PTSM  1.4299  0.3459  0.001  0.4744  2.3855  EM  -0.7649  0.3459  0.181  -1.7205  0.1907  TM  -0.0063  0.3429  1.000  -0.9537  0.9411  PM  -1.2163  0.3429  0.005  -2.1637  -0.2689  0.2137  0.3459  0.972  -0.7419  1.1692  PTSM  58  PTSM EM TM PM PTDM  -0.9785  0.3488  0.045  -1.9422  -0.2199  0.3459  0.969  -1.1755  0.7356  -1.4299  0.3459  0.001  -2.3855  -0.4744  -0.2137  0.3459  0.972  -1.1692  0.7419  -0.0149  Table 12 Tukey results of difference in trust between Random seller and other sellers  We found fewer mean differences are significant at the 0.05 level from the Tukey results. Since the Tukey test is more conservative than the ANOVA we performed in section 5.2.2, we believe that the true results of the trust difference between Random seller and other sellers across mechanisms lie in between the results of the two tests. 5.4  Confidence We compared the mean of the confidence level between different groups based on  reputation mechanisms, and performed ANOVA on the data. See Table 13 for the results:  A  EM  TM  PM  B  B - A (Mean diff erence)  Significance  TM  0.1131  0.197  PM  -0.1464  0.089  PTDM  -0.0559  0.532  PTSM  0.787  0.365  PTDM  -0.169  0.053  PTSM  -0.0344  0.685  PTDM  0.0905  0.29  PTSM  0.2251  0.007  Table 13 ANOVA results of confidence in judgments  From the results, we see that compared with the EM, participants using the PM had less confidence in their judgment about the trustworthiness of the seller, at a marginally significant level of 0.089. Participants in the PTDM group were less confident in their judgement on seller trustworthiness than those in the TM group, with a marginally  59  significant level of 0.053. Participants using PTSM were more confident compared with those in the PM group, at a significance level of 0.007. No significant results were found in the rest of the pair comparisons. Thus, among all eight hypotheses about confidence, only H5e-2 is supported by the experiment data. We also ran ANOVA on the confidence in judgment across sellers. Please see Table 14 for the results. We predicted that with the information in time and/or price dimensions, people would be more confident in their judgment in relation to sellers whose behaviours changed according to the time and/or price of the transactions.  Sellers Mechanisms Honest A  B  Mean Diff. (B-A)  Sig.  Random Mean Diff. (B-A)  Sig.  Time increase Mean Diff. (B-A)  Sig.  Price low low Mean Diff. (B-A)  Sig.  Price low high Mean Diff. (B-A)  Price time increase Mean Diff. (B-A)  Sig.  Sig.  TM  0.21  0.195  0.23  0.210  0.11  0.669  0.29  0.176  0.05  0.776  -0.1  0.729  PM  -0.09  0.588  -0.29  0.150  0.06  0.788  -0.1  0.662  -0.3  0.131  -0.14  0.612  PTDM  -0.02  0.921  -0.11  0.622  -0.22  0.397  0.1  0.673  0.12  0.499  -0.17  0.548  PTSM  0.73  0.546  0.16  0.411  0.21  0.384  0.17  0.400  0.05  0.773  0.02  0.931  PTDM  -0.23  0.171  -0.34  0.097  -0.33  0.161  -0.19  0.428  0.07  0.702  -0.07  0.806  PTSM  -0.11  0.482  -0.07  0.724  0.1  0.659  -0.12  0.539  0  1  0.12  0.674  PTDM  0.07  0.671  0.18  0.387  -0.28  0.174  0.20  0.424  0.35  0.036  -0.03  0.891  PTSM  0.19  0.234  0.45  0.029  0.15  0.444  0.27  0.212  -0.07  0.079  0.16  0.559  EM  TM  PM  Table 14 A N O V A results of confidence i n judgments by sellers  To our surprise, most of the mean differences between mechanisms are not significant. The only significant results we have are those of the Price low low seller, between PTDM and PM (0.036), and between PTSM and PM (0.079). 60  5.5  Helpfulness We compared the mean of bidder perceived helpfulness for the reputation mechanisms  based on different conditions, and ran ANOVA to test if there were any differences between the groups (See Table 15).  A  EM  TM  PM  B  B - A (Mean difference)  significance  TM  0.2834  0.001  PM  0.0935  0.007  PTDM  0.3169  0.001  PTSM  0.1724  0.002  PTDM  0.0335  0.871  PTSM  -0.1110  0.635  PTDM  0.2235  0.065  PTSM  0.0790  0.741  Table 15 ANOVA results of helpfulness in decision making  All four proposed mechanisms - TM, PM, PTDM and PTSM - are found to be more helpful, positively and significantly, compared with the eBay mechanism, at the significance level of 0.001, 0.007, 0.001 and 0.002 respectively. Also, PTDM induces more perceived helpfulness than the PM, at a significance level of 0.065. Therefore, hypotheses H6a, H6b, H6c-1, H6c-2, and H6e-lare very well supported by the experiment data, while H6d-1, H6d-2, and H6e-2 cannot be concluded.  61  6  Discussion and conclusion  6.1  Discussions  6.1.1 Bids Based on the results of the bids placed in the experiment, we could not draw a conclusion about how the mechanisms affect bidders' the maximum amount bidders are willing to bid. This may be explained in a number of ways. First, people have different tastes related to the auction items, and they value the items differently, even with the reference retail price information provided. To decide their bids, people not only consider the uncertainty associated with the auctions or the sellers' trustworthiness; they also consider how much they like and need the items. It is difficult to control and measure people's personal evaluation of the items. Second, people have different attitudes towards risk-taking in the auctions. Some risk averse people place lower value bids even on items being sold by honest sellers, and some risk lovers place higher bids on items being sold by dishonest ones. Further, people have different ways of assessing the risks associated with the auctions. Even if they understand very well from the reputation mechanism that a seller may not be trustworthy from the reputation mechanism, some would still place high value bids in that auction. Third, according to the rules of the experiment game, participants did not need to bid in all the auctions. Many participants did not bid in particular auctions because they did not believe that the seller would honour the terms and complete the auctions successfully. Thus we could not have complete information about the bids to dishonest sellers, and we  62  cannot do the data analysis comprehensively. Further, because of the small number of bids to some sellers, the power was relatively low. Lastly, some participants played the game strategically. They estimated that most of the other participants would bid a lot to honest sellers, and little to dishonest ones. In order to win the auctions with as little money as possible, they wanted to avoid competition in auctions held by honest sellers, and wasting their money. These participants put quite a lot of money in the auctions held by sellers they did not consider to be trustworthy. Although they understood that it would be more likely to be "ripped off by these sellers, they still preferred to take the risk, in order to win the extra cash at the very end. These strategic bids have caused some noise in the experimental data, so that honest sellers received high value bids, while dishonest ones received low value bids. 6.1.2  Trust  The results discussed in section 5.2 demonstrate that both the eBay mechanism and the weighted mechanisms can very well distinguish the dishonest sellers and the Random seller from the Honest seller, by inducing different levels of trust in sellers. We cannot conclude that one outperforms another. The reason may be that the eBay mechanism has been doing a good job in discriminating dishonest sellers and the Random seller from the Honest sellers. In our experiments, the pre-designed Honest seller has perfectly positive feedback, and no negatives. It was reasonable for experiment participants to conclude that this was an excellent seller, regardless of the mechanism. That is why we could not find a difference between the eBay mechanism and weighted mechanisms, in terms of capability in distinguishing the perfectly honest sellers from other sellers.  63  We also found that weighted mechanisms induce more of a difference in trust between the Random seller and the dishonest sellers, compared to the eBay mechanism. This indicates that time and price weighted mechanisms do help to distinguish dishonest sellers with different behaviours. This provided evidence that time and price information does help judge seller's trustworthiness. 6.1.3  Confidence in judgment  According to the analysis in section 5.3, none of our hypotheses about confidence in the judgment of seller trustworthiness were supported, with the exception of H4e2. We may not conclude that the weighted mechanisms induce more confidence in judgment compared to the eBay mechanism. This may be caused by the complexity of the calculations and presentations used in the mechanisms. Some people in the experiments had difficulties in understanding the mechanism and the reputation score calculation, especially the price time weighted double and price time weighted single mechanisms, although most of them thought that they had fully understood them. It took more time to introduce the weighted mechanisms than the eBay mechanism to the experiment participants. This caused negative effects in people's confidence in their judgments. Further, the weighted mechanisms create more uncertainty for bidders by revealing information regarding the time and/or price dimensions of seller behaviours, factors which are invisible in the eBay mechanism. This uncertainty also has negative effects on bidder confidence in their judgments.  64  6.1.4  Helpfulness in decision making  From the results of the experiments, we found that people considered the Time weighted, Price weighted, Price time weighted double and Price time weighted single mechanism more helpful than the eBay mechanism when making bidding decisions. Contrary to our expectations, participants in Price-time-weighted-double mechanism and Price-time-weighted-single mechanism did not find the information provided by the mechanism more helpful than the Price-weighted mechanism or Time-weighted mechanism, at a significant level. This indicates that subjects did not find much difference in helpfulness in decision making between price-time two-dimension weighted mechanisms and price or time one-dimension weighted mechanisms. In H5d-1, H5d-2, H5e-1 and H5e-2, we predicted that the two-dimension weighted mechanisms would provide more information than the one-dimension weighted mechanisms. The more information about the seller's previous behaviours provided, the more a bidder might understand a seller's trustworthiness and therefore predict his future behaviour, hence the more helpful a bidder would find the mechanisms to be. However, the results are unexpected. One possible explanation is that the algorithms of the twodimension weighted mechanisms are too complicated to digest in the relatively short time given to subjects in the experiments. Most of the experiment participants did not have much experience in online auction bidding, nor were they familiar with the online auction environment, or reputation mechanisms. The rating calculations of the Time-weighted mechanism and the Price-weighted mechanism are straightforward; on the contrary, those of the Price-time-weighted-double mechanism and the Price-time-weighted-single mechanism are more difficult to understand. The additional information provided by the  65  additional dimension of Price-time-weighted-double mechanism and Price-timeweighted-single mechanism may have caused positive effects on bidder perceived helpfulness, while the complexity of the calculation algorithms may have caused negative effects on it. The outcome of the positive and negative effects confuses the participants about the true helpfulness of the information provided by the mechanisms. 6.2 Conclusion From the analysis and discussion above, we may conclude that the mechanisms with time and/or price dimensions may help online auction bidders to better distinguish sellers of different reputation levels. Compared with the eBay mechanism, the weighted mechanisms are considered to be more helpful to bidders in making their bidding decisions, including whether they are going to bid or not, and the value of their bids. We can not conclude whether the bidders using the weighted mechanisms are more or less confident with their bidding decision, compared with the eBay mechanism. Also, from the results, we may not say people tend to place higher or lower value bids using the weighted mechanisms.  66  7  Limitations and future research  7.1 Limitations Three limitations of this study must be highlighted. First, we recruited participants for the study on campus. Most were students who did not have much online bidding experience. Thus, they were neither familiar with the bidding method nor the way the mechanism works. Also, they were not very aware of the risk associated with online auctions. Although every participant was trained in how to bid online before they actually bid in the game, it is nearly impossible that they would have very good bidding sense and behaved in the same way as experienced online auction bidders. Second, participants were told that they were bidding on an experimental website, competing against other participants in the same game, in six fake auctions. They understood that they would not lose anything, except for a chance to win the lottery, if they got "ripped off by the seller. They might not have the same feelings as if they were bidding in real online auctions, using their own money. Thus, their attitude towards the risk associated with the online auctions in the game might be different from that in the real auctions. In order to simulate a genuine auction experience, and also encourage participants to bid in as many auctions as possible, we made the six auctions sell six different items. However, we did not control the auction item differences in this study. People had different tastes related to the six auction items. These different tastes would affect their  67  attitudes and bids in the auctions. And also, some items might be considered more risky than others to be bought from online auctions. 7.2  Future research In this research, we tried to eliminate the bidders' difference in preferences in auction  items, through an online auction game in which winning bidders would not actually receive the items, but a chance to win extra cash. However, from the feedback of some participants, they assumed that they were bidding in real online auctions, and considered their needs and interests in the items when they decided their bids. Future research may investigate the effects of the bidder's difference in preferences on the bids, or even eliminate the difference. Most of the participating bidders of this research were university students, who did not have much online bidding experience, which was a limitation of the study. Future research may involve some experienced bidders in online auctions. The results would be more generalizable. In the current research, we set the weights of the time dimension according to the recentness of the feedback for one to six months, seven days to one month, and within the last seven days to be 0.5, 0.3 and 0.2 respectively, and the weights of the price dimension according to the transaction value of high, medium and low as 0.5, 0.3 and 0.2 respectively. We did not verify these weights. Further research may investigate the optimum values of the weights.  68  Bibliography Akerlof, G.A. (1970). The market for 'lemons': Quality uncertainty and the market mechanism. Quarterly Journal of Economics 84 (3), pp. 488-500.  Ba, S., & Pavlou, P. (2002). Evidence of the effect of trust building technology in electronic markets: Price premiums and buyer behaviour. MIS Quarterly 26 (3), pp. 243-268. Beam, C , & Segev, A. (1998). Auctions on the Internet: A field study. Working paper, Haas School of Management, University of California - Berkeley. Cheung, C , & Lee, M. K. O (2000). Trust in Internet shopping: A proposed model and measurement instrument. In Proceedings of the Association for Information System 2000 Americas Conference. Long Beach, CA, pp. 681-689. Dellarocas, C. (2002). Goodwill hunting: an economically efficient online feedback mechanism for environments with variable product quality. 4 workshop on th  Agent-mediated Electronic commerce (AMEC IV), Bologna, Italy. Eaton, D. H. (2002). Valuing information: Evidence from guitar auctions on eBay, working paper, Murray State University. eBay Inc. (2003). 2003 Annual Report. Retrieved September 23, 2004, from http://investor.ebay.com/annual.cfm Gefen, D., Karahanna, E., & Straub, D. W. (2003). Trust and TAM in online shopping: An integrated model. MIS Quarterly, 21 (1), pp 51-90.  69  Hair, J.F., Anderson, R.E., Tatham, R.L., and Black, W.C. (1998) Multivariate Data Analysis,fifth edition Prentice-Hall, Inc.. Kane, M. (2002, April 10). Auction fraud tops FBI's Internet scam list. Retrieved September 23, 2004, from CNET News.com http://zdnet.com.eom/2100-1106-879660.html Kauffman, R., & Wood, C. (2000). Running up the bid: Modeling seller opportunism in Internet auctions. Proceedings of the AMCIS 2000 Conference, Long Beach CA. Kauffman, R., & Wood, C. (2001). What makes a buyer pay more for the same item in Internet auction? Presented at the Computer and Information Systems Technology (CIST) Workshop in Miami. Keser, C. (2002). Trust and reputation building in e-Commerce. IBM T.J. Watson Research Center, Yorktown Heights, NY, Centre Interuniversitaire de Recherche en Analyse des Organsisations (CIRANO), Montreal, July 23, 2002. Kollock, P. (1999). The production of trust in online markets, in: E.J. Lawler, M. Macy, S. Thyne, H.A. Walker (Eds.), Advances in Group Processes vol. 16, JAI Press, Greenwich, CT. Fung, R. & Lee, M. (1999). EC-Trust (trust in Electronic Commerce): Exploring the antecedent factors. Proceedings of the 5 Americas Conference on Information th  Systems, pp. 517-519.  70  Internet Fraud Complaint Center. (2003). IFCC 2002 Internet Fraud Report. Retrieved September 24, 2004, from http://www 1 .ifccfbi.gov/strategy/2002 IFCCReport.pdf Livingston, J. A. (2004). How valuable is a good reputation? A sample selection model of Internet auctions. Bentley College. Lucking-Reiley, D. (2000a). Vickrey Auctions in practice: From nineteenth-century philately to twenty-first-century E-Commerce. Journal of Economic Perspectives, 14 (3), pp. 183-192. Lucking-Reiley, D. (2000b). Auctions on the Internet: What's being auctioned and how Journal of Industrial Economics, 48 (3), pp. 227-252.  Lucking-Reiley, D., Bryan, D., Prasad, N., & Reeves, D. (2000). Pennies from eBay: The determinants of price in online auctions. Working paper, Vanderbilt University. Malaga, R. A. (2001). Web-based reputation management systems: Problems and suggested solutions. Electronic Commerce Research, 1 (4), pp. 403-417. Nagle, T. T., & Holden, R. K. (1995). The strategy and tactics of Pricing: A Guide to  Profitable Decision Making. Englewood Cliffs, NJ: Prentice Hall. National Consumers League. (2002a). Online Auctions Dominant Consumer Fraud.  Retrieved August 23, 2004, from http://www.nclnet.org/internetfraud02.htm  71  National Consumers League (2002b). Internet Fraud Statistics. Retrieved September 23, 2004, from http://www.fraud.org/2002intstats.htm Mishra, D. P., Heide, J. B., & Cort, S. G. (1998). Information Asymmetry and Levels of Agency Relationships. Journal of Marketing Research, 35 (3), pp. 277-295. Nagle, T. T., & Holden, R. K. (1995). The strategy and tactics of pricing: A guide to  profitable decision making. Englewood Cliffs, NJ: Prentice Hall. Resnick, P., Zeckhauser, R., Friedmen, E., & Kwabara, K. (2000). Reputation systems: Facilitating trust in Internet interactions. Communication of the ACM , 43 (12), pp. 45-48. Resnick, P., & Zeckhauser, R. (2001). Trust among strangers in Internet transactions: Empirical analysis of eBay's reputation system. Working paper. University of Michigan. Swan, J. E., & Nolan, J. J. (1985). Gaining customer trust: A conceptual guide for the salesperson. Journal of Personal Selling and Sales Management, 5 (2), pp. 39-48.  Wilson, R. (1985). Reputation in games and markets. In A. Roth (ed.), Game Theoretic Models of Bargaining. Cambridge: Cambridge University Press, pp. 27-62 Wood, C. A., Fan, M., & Tan, Y. (2002). An examination of the reputation systems in online auctions. Presented at the Workshop for Information Systems and Economics (WISE 2002), Barcelona, Spain, December 2002.  72  Zacharia, G., Moukas, A., & Maes, P. (2000). Collaborative reputation mechanisms for electronic marketplaces. Decision Support Systems, 29 (4), pp. 371-388.  73  Appendix 1 Experiment instructions for participants 1.  Instructions for the study i n £eneral  •  This is a study on online auctions. You are going participate in an online auction game, in which you will view some online auctions and have the opportunity to bid in them.  •  The study will take approximately 80 minutes.  •  You will be rewarded $15 for participating at the end of the study. In addition, you will have the opportunity to win extra cash based on your performance in the game. You will compete in a pool of ten subjects for a reward, each of whom has participated in the same study as you.  •  Please feel free to ask any questions you may have during the study.  2.  Instruction to Pre-study questionnaire  •  Please read and answer all questions based on your experiences and feelings.  •  After answering questions, please click on the "save responses" button at the end of the questionnaire and call the research assistant.  •  Please do not click on the "bid in auctions" button at the end of the questionnaire until told to do so.  74  3.  Instructions for the training section - online bidding  •  Auction page: three types of information can be found in an online auction: auction information, seller information, and detailed item descriptions. Auction information includes an image of the item, the current bid, how much time is left, bid history, the current high bidder and the location of the seller. Seller information details the seller's eBay user JD, their reputation rating, the time and place they registered, and provides a link to the seller's reputation profile. You can find more item images, item features and specifications in the detailed item description.  •  Placing a bid: If you are interested in an item and want to bid on it, you need to carefully look over the auction page for the aforementioned information, and then decide the maximum price you are willing to bid. The online auction website uses a helpful bidding system called "proxy bidding" to make bidding on auctions more convenient for buyers. When you place a bid, you enter the maximum amount you would be willing to pay for the item. Your maximum amount is kept confidential from other bidders and the seller. The proxy system compares your bid to those of the other bidders. The system places bids on your behalf, using only as much of your bid as is necessary to maintain your high bid position. The system will bid up to your maximum amount. You will be outbid if another bidder has a higher maximum. If no other bidder has a higher bid, you win the item, and you pay only the minimum needed to win the auction.  75  Example 1 Maximum bid You  $100  Bidder 1  $30  Bidder 2  $40  Result: Y o u win the auction with a bid at $41, the minimum amount required to win the auction. Example 2 Maximum bid You  $62  Bidder 1  $75  Bidder 2  $80  Result: Bidder 2 wins the auction with a bid of $76.  Given the mechanism used for proxy bidding, it is always in your interests to reveal the maximum you would be willing to pay, because if it is possible to win the auction for less, the proxy bid will do so. Therefore, we will use proxy bidding for all auctions in this experiment.  76  4.  Instructions for the training section - mechanism  4.1. eBay mechanism •  Seller's reputation:  In online auctions, you are engaged in transactions with strangers with whom you have had no interaction before. If you win an auction, you are required to pay before receiving or inspecting the items. Thus sellers could try to take advantage of you by shipping an inferior product, or even failing to ship anything at all. The feedback system of the auction website presents you with the opportunity to review comments left by others who have engaged in transactions with a particular seller. Before you bid on an item, it is V E R Y I M P O R T A N T to review the seller's reputation score and feedback details, so as to have some idea about the seller's reputation and trustworthiness. How does the reputation score work?  o After each transaction, the buyer is encouraged to leave a numerical rating of positive (+1), neutral (0) or negative (-1), plus a text comment. o To calculate a seller's reputation score, eBay adds up the numerical rating over time. Only ratings from unique buyers are used in computing the overall reputation score. o An "eBay LD Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives", "neutrals" and "negatives". You can find the number of ratings of each type given to this seller in three time periods: the past 7 days, the past month, and the past 6 months.  77  Example: Feedback Summary  ID card  380 positives. 380 are from unique users.  iloveauction (374)  *j Member since: 0 May 0 3 Location: Canada  || Summary of Most Recent Reviews  0 neutrals.  fj 6 negatives. 6 are from unique users.  Past 7 days  Past month  |]  Positive  10  72  fj  Neutral  0  0  See all feedback reviews for iloveauction.  Past 6 mo.  1  Feedback Score Positive Feedback  374 98.4%  Reputation score = 380 - 6 = 374 Positive feedback percentage = 380 / 386 = 98.4%  What does the reputation score mean?  The reputation score represents the "average" behaviour of the eBay seller over time. It is calculated as the average feedback rating left for the seller in the past. Because past behaviour is a good predictor of future behaviour, the reputation score helps to understand how a seller will behave in future transactions.  Your tasks:  1.  Explore the eBay feedback system.  2.  Practice placing bids online.  3.  Answer a questionnaire after the practice session.  4.  Ask questions about anything you don't understand.  78  .2. Time-weighted mechanism Seller's reputation: In online auctions, you have to engage in transactions with strangers with whom you have had no interaction before. If you win an auction, you are required to pay before receiving or inspecting the items. Thus sellers could try to take advantage of you by shipping an inferior product, or even failing to ship anything at all. The feedback system of the auction website presents you with the opportunity to review comments left by others who have engaged in transactions with this particular seller. Before you bid on an item, it is VERY IMPORTANT to review the seller's reputation score and feedback details, so as to have some idea about the seller's reputation and trustworthiness.  How does the reputation score work? o After each transaction, the buyer is encouraged to leave a numerical rating of positive (+1), neutral (0) and negative (-1), plus a text comment. o To calculate a seller's reputation score, eBay groups the ratings according to the time when they were left, and puts different weights on these groups. Ratings older than 6 months are discarded; ratings between 1 and 6 months ago are given a 20% weighting; between 7 days and 1 month ago, 30%; within the last 7 days, 50%. A seller's reputation score is calculated by aggregating the weighted scores of the three groups. Only ratings from unique buyers are used in computing the overall reputation score.  79  o  An "eBay LD Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives", "neutrals" and "negatives". You can find the number of feedback and the percentage of positives in different time periods, as well as the weighted percentage of positives, using the same weights as above.  Example: Feedback Summary 380 positives. 380 are from unique users  <Sf}V  lloveauction (374)  ID CelTCl  Member since: 0 May 03 Location: Canada  0 neutrals.  S u m m a r y of Most R e c e n t R e v i e w s  6 negatives. 6 are from unique users.  See all feedback reviews for iloveauction.  Past 7 days  7 days - 1 month ago  1 month - 6 months ago  Total  Positive  10  62  308  380  Neutral  0  0  0  0  Negative  1  4  B  Weight Positive Feedback  1  0.5  0.3  0.2  90.9%  98.4%  98.7%  94.7'/l  Reputation score = 380 - 6 = 374 Positive feedback percentage = 90.9% x 0.5 + 98.4% x 0.3 + 98.7% x 0.2 = 94.7%  What does the reputation score mean?  •  The reputation score represents the "average" behaviour of the eBay seller over time. It is calculated as the average feedback rating left for the seller in the past. Because past behaviour is a good predictor of future behaviour, the reputation score helps to understand how a seller will behave in future transactions.  80  •  More recent transactions are treated as more important in calculating the reputation score because the most recent behaviour of a seller (the last 7 days) is a better predictor of how the seller will behave today than is the behaviour of that seller in the more distant past.  Your tasks: 1. Explore the eBay feedback system. 2. Practice placing bids online. 3. Answer a questionnaire after the practice session. 4. Ask questions about anything you don't understand.  4.3. Price-weighted mechanism •  Seller's reputation: In online auctions, you have to engage in transactions with strangers with whom you have had no interaction before. If you win an auction, you are required to pay before receiving or inspecting the items. Thus sellers could try to take advantage of you by shipping an inferior product, or even failing to ship anything at all. The feedback system of the auction website presents you with the opportunity to review comments left by others who have engaged in transactions with this particular seller. Before you bid on an item, it is VERY IMPORTANT to review the seller's reputation score and feedback details, so as to have some idea about the seller's reputation and trustworthiness.  81  How does the reputation score work? o After each transaction, buyer is encouraged to leave a numerical rating of positive (+1), neutral (0) and negative (-1), plus a text comment. o To calculate a seller's reputation score, eBay groups the ratings according to the price of the related transaction, and puts different weights on these groups. Ratings of transactions between $0 and $20 are given a 20% weighting; between $20 and $100, 30%; more than $100, 50%. A seller's reputation score is calculated by aggregating the total weighted scores of the three groups. Only ratings from unique buyers are used in computing the overall reputation score. o An "eBay ID Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives", "neutrals" and "negatives". You can find the number of feedback and the percentage of positives in different price bands, as well as the weighted percentage of positives, using the same weights as above. Example: Feedback  Summary  3 8 0 positives. 3 8 0 are from unique users. 0 neutrals.  6 negatives. 6 are from unique users.  See all feedback reviews for iloveauction.  jepf ID card  ilovemictiou (374)  VI | M«mb«r since: 0 May 03 Location: Canada | S u m m a r y of Most R e c e n t R e v i e w s .i ={ Ti.ins.iction V a l u e a J j ID-$20 $20-1100 |  Positive  Total  160  380  »|  Neutral  0  0  0  »j  Negative  3  2  6  Weight  0.2  0.3  0.5  97.8%  98.8%  98.9%  Positive F e e d b a c k  132  $100 -  98.6%  Reputation score = 380-6 =374 Positive feedback percentage = 97.8% x 0.2 + 98.8% x 0.3 + 98.9% x 0.5 = 98.6%  82  What does the reputation score mean?  •  The reputation score represents the "average" behaviour of the eBay seller over time. It is calculated as the average feedback rating left for the seller in the past. Because past behaviour is a good predictor of future behaviour, the reputation score helps to understand how a seller will behave in future transactions.  •  Transactions with higher prices are treated as more important because they involve more value, and have more weight in determining the average behaviour of a seller on a per-dollar basis. For example, one 100-dollar transaction is, in some sense, as important as ten 10-dollar transactions.  Your tasks:  1. Explore the eBay feedback system. 2. Practice placing bids online. 3. Answer a questionnaire after the practice session. 4. Ask questions about anything you don't understand.  4.4. Price-and-time-weighted-double mechanism •  Seller's reputation:  In online auctions, you have to engage in transactions with strangers with whom you have had no interaction before. If you win an auction, you are required to pay before receiving or inspecting the items. Thus sellers could try to take advantage of you by shipping an inferior product, or even failing to ship anything at all. The feedback system of the auction website presents you with the opportunity to review comments left by others who have engaged in transactions with this particular seller. Before you  83  bid on an item, it is VERY IMPORTANT to review the seller's reputation score and feedback details, so that to have some idea about the seller's reputation and trustworthiness.  How does the reputation score work? o After each transaction, buyer is encouraged to leave a numerical rating of positive (+1), neutral (0) and negative (-1), plus a text comment. o To calculate a seller's reputation score, eBay puts different weights on the ratings according to the time when they were left and the related transaction price. With respect to time, ratings older than 6 months are discarded; ratings between 1 and 6 months ago are given 20% weighting; between 7 days and 1 month, 30%; within the last 7 days, 50%. With respect to price, ratings of transactions between $0 and $20 are given a 20% weighting; between $20 and $100, 30%; more than $100, 50%. The overall score is the average of the time-weighted score and the price weighted score. Likewise, the weighted percentage of positives is the average of time-weighted percentage of positives and price-weighted percentage of positives. Only ratings from unique buyers are used in computing the overall reputation score. o An "eBay LD Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives", "neutrals" and "negatives". You can find the number of feedback and the percentage of positives in different time bands and price bands, as well as the weighted percentage of positives, using the same weights as above.  84  Example: Feedback Summary 380 positives. 380 are from unique users.  eb'Y" ID card  iloveauction (374)  Member since: 0 May 03 Location: Canada 0 neutrals.  S u m m a r y of Most R e c e n t R e v i e w s Time  6 negatives. 6 are from unique users.  See all feedback reviews for iloveauction.  Transaction V a l u e  Past 7 days  7 days -1 month ago  1 month - 6 months ago  $0$20  $20$100  $100 +  Total  Positive  10  62  308  132  160  88  380  Neutral  0  0  0  0  0  0  0  Negative  1  1  4  3  2  1  6  Weight  0.5  0.3  0.2  0.2  0.3  0.5  90.9%  98.4%  98.7%  97.8%  98.8%  98.9%  Positive Feedback  94.7%  98.6%  Reputation score = 380 - 6 = 374 Positive feedback percentage = [ (90.9% x 0.5 + 98.4% x 0.3 + 98.7% x 0.2) + (97.8% x  0.2 + 98.8% x 0.3 + 98.9% x 0.5) ] / 2 = (94.7% + 98.6%) / 2 = 96.7%  What does the reputation score mean?  • The reputation score represents the "average" behaviour of the eBay seller over time. It is calculated as the average feedback rating left for the seller in the past. Because past behaviour is a good predictor of future behaviour, the reputation score helps to understand how a seller will behave in future transactions. • More recent transactions are treated as more important in calculating the reputation score because the most recent behaviour of a seller (the last 7 days) is a better predictor of how the seller will behave today than is the behaviour of that seller in the more distant past.  85  • Transactions with higher prices are treated as more important because they involve more value, and have more weight in determining the average behaviour of a seller on a per-dollar basis. For example, one 100-dollar transaction is, in some sense, as important as ten 10-dollar transactions. Your tasks: 1. Explore the eBay feedback system. 2. Practice placing bids online. 3. Answer a questionnaire after the practice session. 4. Ask questions about anything you don't understand. 4.5. Price-and-time-weighted single mechanism •  Seller's reputation: In online auctions, you have to engage in transactions with strangers with whom you have had no interaction before. If you win an auction, you are required to pay before receiving or inspecting the items. Thus sellers could try to take advantage of you by shipping an inferior product, or even failing to ship anything at all. The feedback system of the auction website presents you with the opportunity to review comments left by others who have engaged in transactions with this particular seller. Before you bid on an item, it is VERY IMPORTANT to review the seller's reputation score and feedback details, so that to have some idea about the seller's reputation and trustworthiness. How does the reputation score work? o  After each transaction, buyer is encouraged to leave a numerical rating of positive (+1), neutral (0) and negative (-1), plus a text comment.  86  o  To calculate a seller's reputation score, eBay puts different weights on the ratings according to the time when they were left and the transaction price. With respect to time, ratings older than 6 months are discarded; ratings between 1 and 6 months ago are given 20% weighting; between 7 days and 1 month, 30%; within the last 7 days, 50%. With respect to price, ratings of transactions between $0 and $20 are given a 20% weighting; between $20 and $100, 30%; more than $100, 50%. The overall score is calculated by weighting each score by both its time and price weights, then summing these up. The same weights are applied to the percentage of positives to calculate the overall percentage of positives. Only ratings from unique buyers are used in computing the overall reputation score.  o  An "JD Card" is used to show the summary of the reviews of a seller for the past six months, classified as "positives", "neutrals" and "negatives". You can find the number of feedback and the percentage of positives in different time bands and price bands, as well as the weighted percentage of positives, using the same weights as above.  Example: Feedback Summary 380 positives. 380 are from unique users, 0 neutrals  dfif | D CclTd  s V:  M e m  '  ) e , s  '  iloveauction  (374)  8 May03 Location: Canada  n c e :  S u m m a r y of Most R e c e n t R e v i e w s 6 negatives. 6 are from unique users.  See all feedback reviews for iloveauction.  Time „  j j  " Transaction Value  7 days -1 month ago  1 month - 6 months ago  Weight  0.5  0.3  0.2  0.2  6 (85.7%)  32 (97.0%)  94 (98.9%)  $20-$100  0.3  2 (100.0%)  24 (100.0%)  134 (98.5%)  $100 +  0.5  2 (100.0%)  6 (100.0%)  80 (98.8%)  $0-$20  °f f Feedback P  c  i , i V  1  Total  98.1%  Reputation score = 380 - 6 = 374  87  Positive feedback percentage = 85.7% x 0.2 x 0.5 + 100.0% x 0.3 x 0.5 + 100.0% x 0.5 x 0.5 + 97.0% x 0.2 x 0.3 + 100.0% x 0.3 x 0.3 + 100.0% x 0.5 x 0.3 + 98.9% x 0.2 x 0.2 + 98.5% x 0.3 x 0.2 + 98.8% x 0.5 x 0.2 = 98.1%  What does the reputation score mean? • The reputation score represents the "average" behaviour of the eBay seller over time. It is calculated as the average feedback rating left for the seller in the past. Because past behaviour is a good predictor of future behaviour, the reputation score helps to understand how a seller will behave in future transactions. • More recent transactions are treated as more important in calculating the reputation score because the most recent behaviour of a seller (the last 7 days) is a better predictor of how the seller will behave today than is the behaviour of that seller in the more distant past. • Transactions with higher prices are treated as more important because they involve more value, and have more weight in determining the average behaviour of a seller on a per-dollar basis. For example, one 100-dollar transaction is, in some sense, as important as ten 10-dollar transactions. Your tasks: 1. Explore the eBay feedback system. 2. Practice placing bids online. 3. Answer a questionnaire after the practice session. 4. Ask questions about anything you don't understand.  88  5. Instructions for the main task Now you are invited to participate in an online auction game, in which you are given $600 to bid in 6 different auctions, which are held by 6 sellers separately. In these auctions, your competitors are other participants of this study. Your task is to decide how to allocate the $600 by placing your maximum bid for each auction. You do not have to spend all $600, and you have the opportunity to win the leftover funds after the study.  If you win auctions, you can receive lottery tickets, which qualify you to win your "pot" of unspent funds. The number of tickets you receive depends on how many auctions you win.  Number of auctions won  Number of tickets for the lottery  0  0  1  1  2  3  3  6  4  10  5  15  6  21  Your pot is equal to your initial $600 less the total amount you bid in all auctions. Thus, if you bid $70 each in all six auctions, your pot is $600 - 6 x $70 = $180. Your chances of winning your pot, however, depend on whether you bid high enough to win auctions and receive lottery tickets. You should also be aware that not all eBay sellers are completely reliable. Sometimes sellers collect payment from auction winners but the auctioned item is either never delivered or delivered in such a way (different from expected, damaged,  89  used, late, etc.) that the buyer is not satisfied with the transaction. For each seller, we assess their reliability and determine the likelihood of successful completing of the transaction; for each auction you win, there is a probability based on the reliability of the seller that your transaction would not be completed successfully. In this experiment, the consequence of an unsuccessful transaction is that you would not receive the associated lottery tickets for winning that auction. Thus, you should consider the trustworthiness of sellers before placing your bids.  Your tasks: 1. Browse each auction and read the sellers' profiles. 2. Choose whether or not to bid in each auction; place your maximum bid. 3. Answer questionnaires. After you are done with each auction, whether you bid or not, you will be presented with a questionnaire. Please complete the questionnaire before you bid in any other auctions and carefully answer all questions based on your thoughts, beliefs, and impressions at the moment you made your bidding decision. Once you are done with a particular auction and have started answering the questionnaire, you will not be allowed to go back to the auction page or change your bid in that auction. 4. Feel free to ask questions during the process of bidding. Once you are done with all 6 auctions, please let the research assistant know. Dismissal •  Please make sure not to talk about the content of this study with other people.  •  We will contact you if you win the lottery.  90  Appendix 2  Seller's profiles presented in five mechanisms  Seller 1: Honest seller Feedback Summary  ID card  "600 positives:; 600 are from unique users:  •tommy gee ( 6 0 0 )  Member since: 25 Oct 03 Location: Canada:  0 neutrals.  Summary of Most Recent Reviews 0 negatives. 0 are frorr unique users.  Past;/ days  Sec all 'eedbick review? for tommy goe.  Positive  ;25:  Neutral.  0  Negative  Of;  Feedback Score < Positive Feedback  600 100.0%  Honest seller - eBay mechanism FeedbackiSummary. 600 positives:;600'are;frbmuniqueusers.  «*  ID card  : tommy ge e: (: 600)  Membel since: 25 Oct03 Location: C a n a d a ,  Summary of M o s t Recent Reviews  0 neutrals:  0 hegatives; 0 are from uhique:users:  Past 7 days  7 days -A. month ago  1 month:-6 months.ago  '-25;  75;  ;5&  C  .0  0  ..Negative;  0  -0-  .10;;:  Wei'jht  ;0:5  03  100.0%  100.0%  total  !  Positive;, Neutral  See all fee'lOack levisws for tommy nee.  Positive Feedback  100.0%  600f: 0 ;  ,0  100:0%  Honest seller - Time mechanism Feedback Summary 600 positives. 600 are from unique users. 0 neutrals.  llLw" (JO 'I  JQ Catd  toirunv gee ( 6 0 0 )  Membei since: 25 Oct03 Location: Canada:  Summary of Most Recent Reviews 6 negativesXOiare from:unique: users.  See al! feedback reviews for tommy oee.  Tiansaction Value $0 • 520  $20 • $100  Positive  '418,  '121  Neutral  0  0  Negative  0  '0-  Weiylit  02  Positive Feedback:  100.0%  :100.0%  Honest seller - Price mechanism  91  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Feedback Summary 600 positives: 600 are .from unique users.  ebY ID card  tommy gee ( 6 0 0 )  Membeisince: 25 Oct03 Location: Canada  0 neutrals.  Summary of Most Recent: Reviews Time  0 negstrves. 0 are from unique users  , , . P a c t 7  7 days • 1 mar.th  n  See ali feedback reviews far tommy gee.  Transaction Weight Value  riavc  1 mcnti - 6 mcnths  ago;  ago  0:5:  0.3  0.2  53 (100.0%)  347;(igo:g%)  J  $3-$20  0.2  1B (ICO.3%)  $20 - $1C0  0.3  •;(ip6:o%)  15 (100.0%)  1C1 ( O . 0 % )  $100*  .0.5  2 (103 0%)  7 ('CO.0%)  52 (X3.0%)  si  Total  iPositive Feedback  Honest seller - Price time single mechanism  92  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Seller 2: Random seller Feedback Summary  efaY" ID card  ,582;po'si)ives: 582>are from unique users..  gift deal (564)  Member since: 25 Oct03 Location: Canada  Summary of Most Recent Reviews  .0 neutrals.  18 niifptives. 18 are rom unique usury. f  See allfeedbacli'reviewgfqngiftdeal.  Past.7 days  Past month;  Positive  OA.  ;37:.  Neutral  0  •P  Negative  j.  Feedback Score "Positive Feedback  •Past 6 mo.. '582 '  0  564 970%  Random seller - eBay mechanism  Feedback Summary  «^|M"tDCard  l582;positive's;.582 are,from uhique'users.  ;  Oineutrals.  Summary of Most Recent Reviews  18 negatives. 18 are from unique users \ See all feedback'reviews for qittdeai.  giftdeal (5S4>  Member since. 2fl Oct 03 Location. Canada  Positive,  j Neutral |  Negative  Weitjhl,. Positive Feedback^  P a s t ? days'  7-days.-1 month ago  24'  'm.  0 ft"  0.5  0 y.2  1.month',-'6 mohths.ago, .is5  :  Tolal  'Mi':  0 15'  o:3  i0.2  97.3%  97:0%  D. . :  .W'  Random seller - Time mechanism  93  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Feedback Summary \582 positwes: 582are from uniquevusers:  e b Y ID card .v; .M«mbsr sines: 25  .0 neutrals..  siftdeal (564)  Oct03 Location: C a n a d a .  Summary of Most R e c e n t R e v i e w s 18 negatives 18 are from unque users.  Time  4  Past 7 days  ii  See all feedback reviews for aiftdeal.  | Positive.  Weight Positive,.. Feedback  $20,$100  S O  73  485'  :4OB'  117  . :*59:  0  ,P  0  0  b  0  '4:  2  .0:3.  0.5  v! .-Negative;.•  j  10120  24  *i Neutral:  :i j  Transaction V a l u e .  7 cays •• 1 1 rronth - B month ago months acs  '2-  los:  03  0.2  96.0%  37 3%  S7.0%  q ::2.-96.5%  97:1%  96.7%  Total •582  0 ;'"18.:  96:7% 96.8%  Random seller - Price time double mechanism  Feedback S u m m a r y 582. positives.':582 are from uniqueusers:  ebflDcard  aiftdeal (564):  i:Membetslnoa:25 Oct03;Location: Canada  0 neutrals.  i Summary of Most R e c e n t Reviews'; Time  18 negatives 18 are from unique users See all feedback reviews for cftceal.  Past? d s • •• Ttansactioii Value Weiyht  0.3  $100:*? 0.5  days -1 rro--;h aiifV  1 mont" - S montlis ago  0.3:  0.2  51 (98.1%)  338(96.8%)  5(83:3%)  15'(100.0%)  97(?7.0%)  2 (100.0%);  7(87.5%)  '50 (98:0%)  0.5  $0-$20 ,0.2 i7;(ibb:o%)_ $20- $100  ?  Total  J  Positive Feedback  Random seller - Price time single mechanism  94  Appendix 2  Seller's profiles presented infivemechanisms (Continued)  Seller 3: Time increase seller Feedback Summary  eb'Y" ID card  582; positives. 582 are from unique users:  fridayititef564)  Membei since': 2s;Octb3:LocatioroCanada, •;0 neutrals:  S u m m a r y of M o s t R e c e n t R e v i e w s Past 7 d a y s  18.negatives. 18 are from unique users. Positive  'Past: month  ~20'  •Past 6.mo:  82:'  '58?  Neutral'.  0  C  •  Negative  5  ,13>:  •18:  564-  Feedback Score-  97.0%'  Positrve-Feedbnck:  Time increase seller - eBay mechanism  Feedback Summary  1  •5K"^.6isfti^Ets^.Sp2''^jfrbrh.uhiqira)us'sr^'.  i  —  i Cttfi*/  —  :  ID Card  iridayiiite (564);  Member since: 25 Oct 03 Location: Canada., 0 neutrals.  S u m m a r y of M o s t R e c e n t R e v i e w s -?ast 7 days • -7:days-i.mohth/ag'o  18 negative;. 18 are from uniqje users. .Positive.:See all feedback reviews for'fridaynite. v  20  1 month -6'rnbnths;ago, Total  62  500  535  Neutral  0;  0  0  ,0.'-,  Negative;:  5-  vt'3-""-  0'  'fe-  Weight :Positive:Feedback..  0.5  0.3  .02  ,8o:o''.  8217%  •lOOiOS  Time increase seller - Time mechanism  Feedback Summary  e & Y ID card  ••582;positives: f^'are.from unique users:  fridnviure: (564).-  . Member since: 25 Oct03 Location: Canada 0 neutrals.  i S u m m a r y of Most R e c e n t R e v i e w s Transaction Value  18 negatives. 18 are from uniquo Lsers.  jq-,i20 • • Positive--  Seevall feedbackteviews for, fridaynite.  i !  Neutral .Negative,  ;  Weight  I  Positive Feedback  $20-510:1  $100 +  Total  ,406:, '  117;  ?S9  532;  0  •o  12  . .4'':.  0  0  %  '18;  0.2  03  ;'0;5.  97.1%  96.7%  96.7%  96.8'i  >'  Time increase seller - Price mechanism  95  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  F e e d b a c k Summary , 582;positives;>582 arefrom unique users:  ©blf  | Q CdFCi  fridaviute (564)  Membsrsinca: 25 Oct03 Location: Canada-  0 neutrals.  S u m m a r y of Most Recent Reviews Time  18 negatives. 18 are from un-q-s users.  See ali feedback reviews for fridaynite.  Transaction Value  7 days • 1 1 month - 3 month ago months ago  Pas! 7 cays Positive'  62  JO.-. $20  $20, $100  sen-; q  406/'  ' mi  0  ,0:  Neutral  0  0  Negative  f55::.'  "13;  *•  ;12i.'  0.5  =0:3;  0.2-  0.2  80.0%  82.7%  100.0%  Weight Positive Feedback  84.8%  $100: > •''  Total  ;59  v  ;  0  f0.3  0.5  97 1% 93.7%  98.7%  :.° -Sis-  Time increase seller - Price time double mechanism  Feedback S u m m a r y  ebV ID card  582 positives 582 arerom unique users. f  ;  fiida>iute (564)  i Membtf since: 25 Oct 03 Location: Canada  Oneutrals:  Summary of Most Recent Reviews  i8:negatives:»18,are from;unique,users:  ! j  Time Past 7 days  See ail feedback review? for fridaynite.  Transaction Value Weight $0 - $20  0.5  7 days - * month ago: C.3  1 mbnth-B months ago 0.2  02  14 (82.4%)  44 (83.0%)  348 (*CU.0%j  $20 - $100 jfJ:3  4.(80.0%)  '2 (80.0%;  •ioi (100,0%):  2(66:7%);  6 $5.7%]  $100+ 'TJ:5  Total  51 (100 Cl%)  Positive Feedback  Time increase seller - Price time single mechanism  96  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Seller 4: Price low low seller  Feedback Sum 582 pDsitives: 582 areifrorri:unique users: ;  «*'/" ID card  icvfire (564)  Membersincc: 25 Oct03 Locabon: Canada  0 neutrals.  -Summary of Most R e c e n t Reviews . Pas: 7 cays  ^^ega'twe^ ' -Positive;  :Seeali ,feedback, reviews., foricytlre':' :  Past 6;mo.  Past.month  :5B2;  24  Neutral  .0  0  .0  "Negative'  :-.f;  •'•3  IB.  Feedback Score Positive Feedback  564 97 0%.  Price low low seller - eBay mechanism Feedback Summary  ebY ID card  582 positive':. 582 are from;uhique:users.  Kvfire (564)  I Member since: 25 Oct 03 Location: Canada  0 neutrals.  I S u m m a r y of Most R e c e n t Reviews j  .18 negatives^ ;w  Positive'  Past 7 days  7 days -" month ago  24  ,73  •485  0  o  Neutral :See'3ll feedback feviews'for icvfire:  Negative  Weight  Ijmonth-6. months ago  r" 0.3  0.5  '582. .0  ;15:  Positive Feedback  Total  :ri8j  :  •0.2;  •97.6V,  97:3^  •9G.6%  Price low low seller - Time mechanism Feedback S u m m a r y 582 passives 582 a'e fiom unique users. 0.neutrals:  ®b¥"lDcard  icyfiief 564)  Member since: 25 Oct 03 Location: Canada.  Summary of Most Recent Reviews transaction Value  18 negatives. 18 are from unique users.  50 - 520 Positive'  See a'l feedback reviewsfcricvfire.  Neutral  $20 - 5100.  Positive Feedback  Total  '418'  -1G3  .,.:;S'l:,  532  0  0  ii  ; #  •18:::  p|  ,18L  Negative..  Weight  $103 +  D.2  .0.3  0.5  100.0e  85.1%  100.0'.  :  95.5%  Price low low seller - Price mechanism  97  Appendix 2 Seller's profiles presented in five mechanisms (Continued) F e e d b a c k Summary 582 positives. 582 arefror.unique users.  ebY" ID card  icyfire(564):  :M«mboisinca: 25 O c t 0 3 Location: Canada  0 neutrals.  Summary of Most Recent Reviews TiausactionValue  Tiine 18 negatives 18 are from urique users.  Past-7 days '•'•Positive -  See: all-feedback'''reviews foricyfire.  Neutral  7 days • 1 1 month • 6 monthago months ago  '••24.,  •':73.:.  ••485-  0  :0:  0  0.3  0.2  ^Negative  •  Weight  •.5>.  :. Positive -Feedback  56.0%  07.3%  97.0%  96.6%  100:0% C5."% '00.0% 95.5%  Price low low seller - Price time double mechanism  Feedback Summary 582 positivesr582 are.from uniqueusers.:  ebY ID card  icyfii'eiQ564)  Member since: 25 Oct 03 Location: Canada  0: neutrals:.  Summary of Most Recent Reviews Time  18negatives.::18 are from unique users. Past 7 days -See all feedback reviews for.'icyfire.  -•"•-.• Transaction Value Weight $0 • $20 C.2 $20• $100 0 3 $100 +  05  0.5  7 d  ^ s  1 n o n t h  ago 0.3  17(100.0%); .,51(100,0%) 5 (83.3%) 15 362%) 2 (100 0%)  7 (100.0%)  1 month • 6 rnontro ago; 0.2  Total  '350:(100.0%) '•• .83,(84:7%),52 (100.0%)  Positive Feedback  Price low low seller - Price time single mechanism  98  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Seller 5: Price low high seller  Feedback Summary 582 positives. 582 are from Lnique users,  ] ©tjY ID card si .  r^Mksiaii^*) • , .  :!::Mombei sine*.: 25 Oct03 Location: Canada.  0 neutrals!  4 Summary of Most Recent Reviews ':i  18 negatives 18 are from u m q j e u s e r s .  See all feedbabk'reviews for twihklestar.i  Past 7-days  Past month  . 24,  »97<  Neutral  0  0  Negative  1-j  |  Positive;  I |  Feedback Score Positive Feedback  Past 6 mo: '582 0  ••.•'3* ••  18  564 97:6%  Price low high seller - eBay mechanism Feedback Summary 58? pooitiviiv. 582 uru t-orn unique users.  ebY ID card  tiviirldestnr (564)  Membai since: 25 Oct03 Location: Canada  0 neutrals.  Summary of Most Recent Reviews :Past 7 days  18 negatves. 18 are from jr.iquo users. '| .Positive-, See'aii feedback reviews for.twinklestar.  7days-,1 month ago  1 month • B msntr.s ago  ~73'  24  ""•~4J85S/  Total f582;  Neutral  ;0  0  ;0 •'•  50;  Negative.  •iv  \2  15'  ' 18*  ;  Weight Positive Feedback:  0.5  03  0.2  96.0%  97.3%  97.0%  96.6%  Price low high seller - Time mechanism  Feedback Summary 582 positives. 582 are from unique users.  « * V ID card  t w u i k l e s t a r ( 564)  . M e m b t r s i n c e : 2 5 : O c t 0 3 Location: Canada -  0 neutrals.  S u m m a r y of Most R e c e n t Reviews Tiansaction Value  18 negatives. 18 are from unique users. "Positive:, . S'ee'all fe'edbackreviewsYortwinklestar.'-  10-120  $20 -$100  '•••^13];:  103  $103 + •'  Tola! •582  Neutral  0  0  0  0  •Negative  •Oi  pm  "'18;  :;'i8°v  77.2%  88.6%  Weight Positive Feedback  0.2  S6:3*  100.0%  100.0%  Price low high seller - Price mechanism  99  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Feedback Summary  e f e ¥ ID card  582 positives 582 are from unique users:  twiiiklestar (564)  j Member since: 25 Oct 03 Location: Canada  0 neutrals.  ! S u m m a r y o f Most Recent Reviews Tiriie  18jnegatiyes; 18 are from unique users-  Pa"sfc7 days  ;  Positive'.  See al feedback iewev.3 for tv/ir.klestar. !  ;24V-:  Neutral  0  Negative  Positive. Feedback  $0$20  $20:$100.:  .'•48S  ?418  0;  0  q  ;V'2t.  IS'  ''Oi  "A-:6!" --'  0.2  03  05  03-  96:0%  :97:3%.  Weiijht  Transaction Value  7 days• 1 1 month-6 month ago 'monthsia'gb  "0:2  97.0%  $100" Total .+.• '  w.  o  :P  18  •M'  .  A  0.5  '.96.6.%. 100:0% 100.0% 77:2% 88.6'V  Price low high seller - Price time double mechanism  FeedbackiSummary  r  582:positrves::.582 are:from unique users:, \ $3J)Y ID  Cdfd  , twiaklesfai^j 564)  Membersinca: 25.0ct 03 Location: Canada  0 neut'als.  Summary of Most Recent Reviews 18;negatjyes.18!are .frqrn^uriique^users.,  Time  ;  Rasti7'days , See ail feedback r'eviev^ fdr.twinklestar.  7 days -1 month ago  TiaiisactionValue  Weight  05;  $0 - $23 0.2 17 (100.0%) $20-'$100 0.3 5 (1C0 0%) $100+: 0.5 2.(56.7%)  03  51(100:0%) 15(100.0%) 7 (77 6%)  '1 month-6'months ago  0.2  350 (ioo.:%; 83 (100.0%) 52 (77.6%)  Positive Feedback  Price low high seller - Price time single mechanism  100  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Seller 6: Price time increase seller  F e e d b a c k' S u m m a r y  0  582 positives. 582 are from unique users.  p-~--; ' f  ID C3td  Member since: 25 Oct 03 Location: Canada  0 neutrals.  S u m m a r y of Most R e c e n t Reviews Past 7 days  18 negatives:: 18 are from unique: users.  'SeeaHfeedback reviews fdr:ns35:  .Positive:  , '20  Neutral  0  Negative  ''5'".  Feedback Score  Past month  Past 6:mo. 562;  62 ;  0  0 .,18.'-.  564  Positive Feedback  97.0%  Price time increase seller - eBay mechanism F e e d b a c k Summary 582 positives. 582 are from  cn.qLO  users.  0 neutrals.  ;«*Y ID card  iis35 ( 564)  ' " !-Member since: 25 Oct 03 Locatlon:Canada-  I Summary of Most R e c e n t Reviews »] v.i . . j Positr»e  18 negatives 18 are from unique users See aljfeedback reviewsfor ns35  -'Past? days:. 7 days -1 month ago ' " \ .' ' ' "".. v62:'; 20  ..j Neutral vi Negative .  Weight: Positive Feedback  1 month - 6 norths ago 'Total 600  :-'582t •'Oi"  0  0  0  %  Wx  03  05  0.3  80.0%  82.7%  Price time increae seller - Time mechanism  Feedback Summary  ID card  •582 positrves.582.are from uniqueusers.  11935(564)  ! Member since: 25:0ct03 Location:.Canada  0.neutrals.  Summary of Most Recent Reviews Transaction Value  18.negatives: 18 are from unique.users. si-  . 'See.'ali.fe"edba'ck;reviewsfor:ris35.  S3-$20  $20-.$100  $100 +-  Total  Positive:  v*B.\'  '-121i  42  : 582'  Neutral  $  0  0  b  0  18:  • is?;  ;  Negative  Weight Positive Feedback  0.2  -0.<3'  0.5  100.0%  100.0%  70.5V,  85:2%:  Price time increase seller - Price mechanism  101  Appendix 2  Seller's profiles presented in five mechanisms (Continued)  Fjeedb'acJc:SM.i^mai^,  r—~-  | gjjf |[J card  582:pqs^  us35(564)  j. Member since: 25 Oct03:Looation: Canada  .Oineutrals: 18 neyatryes. 18 are frcm unique users.  j Summary of Most Recent Reviews i Time  Past 7 7 days -1 ' days' month ago  See all'feedba'ckreviews for ns35:  Positive .  •62 •  •>5op:  '$0';j20  (20$100.  nop +  Total  • '3)1.8  121,  '43';"  532  0  b  iff  0  :'o  0  0  Negative  %  '#;;  Ql  0  Weight;  0.5  >°£<  02  80.0%.  82 7%  Neutrai  |j ?!  :20u  Transaction Value'-'  1 r.o"th - 6 •tenths age  Positive Feedback  •Ui  '•'P-2'  100.0%. 84.8'/,  /OS-;.-.  n.s  100.0% 100.0%. 70.5% 85.2%  Price time increase seller - Price time double mechanism  Feedback Summary  r—-—-  582 positives 582 are from uniqje users,  j  C h Y ID Cai*d  11555(564)  I Member since: 25:Oct03 Location: Canada  0:neutrals:  i Summary of Most Recent Reviews :i8:negatryesp8;are:frp'm:u^  Time  See ail feedback reviews (or ns35.  Past 7. days ••.,••,•• Tiansaction Value Weight  $0 - $20 $20 - $130  0.2  0:5  ' Y -Vmonth =™ ''' 0,3 ca  s  1 month - 6 months ago' ' 0.2 Total  14:(100.0%)  44 (103.0%)  350 (130.0%;  S&3 4 (100.0%)  12 (1C0.0%)  1C5  $100;+' 0.5  2 (29.6%)  6(31.6%)  (-.30 0%)  35 (100.0%)  Positive Feedback  Price time increase seller - Price time single mechanism  102  Appendix 3 Description of the items used in the experiments Item 1: Timex sports watch - listed by Honest seller home | register 1 sign in/out 1 services 1 site map | help. ®  e b ¥ <  > t  £^  4* Back to,lis| of items  | Browse j Search | Sell | MyaBay | Community |  Listed in category:  Jawaller/ Si Watches >Watcri«jf>Wriitwatcha;  Sports watch: Timex Rush Easy Set Alarm You are signed in  Current bid:  C$20.00  Race Bid > Time left  82 hours 44 mins 7-day listing  •  INDIGLO night-light  tommy gee ( 6 0 0 ) Feedback rating: 600 Positive f e e d b a c k : 100.044 Registered C a n a d a  Read feedback reviews A s k seller a question  History:  5 bids i c U-QQ starting bid)  View seller's other auctions  High bidder:  hawkfoot99  Location:  Vancouver  0 S^y with Confidanca  Shipping and payment details  Description E a s y set alarm  Seller information  Ends 3C--Au?-04 0 5 i 2 8 i i 0 EST  *  •  Item numb8r3387881292 W a t c h this item (track it in My eBay) j  • Water resistant to 50 meters •  QUICK-DATE easy set date window  •  1-12 hour alarm  •  1-59 minute alarm  a Rotating hour/minute rings for alarm setting •  Resin case with metal plate  •  Full arabtc numerals with markers  a Resin strap This auction is for a brand new watch in original box. Three colors are available. Now... Wear your performance on yourwristl  Reference price atwww.timex.com: $99.99 !  Seller assumes all responsibility for listing this item  Appendix 3  Description of the items used in the experiments (continued)  Item 2: Dell memory key - listed by Random seller home | register j sign in/out | services | site map | helfi  e h r ¥  |  Back tO list ° f ' t . r § m  Browse  | Search  Listed in category.  | Sell  | My eBay  ®  | Community |  Computer & Elact«jnics>Desktop PC Driues & Components>Flash Memory Drives  Dell 128M Flash Memory Key  Hem number 4434282992  j You are signed in  Watch this item (track it in My R R a y l i  Current bid:  Piece B i d >  ^ • u ^ R j  Time  Seller information  C$20.00  giftdeal (564 )  j  Feedback rating: 564  Positive feedback:  Registered C a n a d a  83 hours 7 mlns 7-day listing  Ends 3 Q - * . u g - 0 4 Q 5 i 3 I . l l t S T  History:  5 bjds, ( C $l.Q$ starting bldj  High bidder:  jimbolobo  Location:  Vancouver  *  97.0°*.  Read feedback reviews Ask seller a question View seller's other auctions 0 H Y , * ) t h Confidence.  Shiooinq and payment detail^  Description  Seller assumes all responsibility for listing this item.  The U S B Memory Key is another great innovation from Delll Just plug the Memory Key into your U S B port and experience the convenience of sharing or transporting highdensity files. This product is specifically designed with you in mind — it offers ease-of-use, great compatibility, convenience and Dell's industry-leading prices. One 128 M B Memory Key stores the same information as almost 69 traditional floppy drives. With such innovation, you can easily eliminate your data storage hassles and increase your computing freedom. Storage capacity: 128MB Depth: 0.6" Height: 2.8" Weight: 0 . 8 0 z Width: 1.1" Compliant standards: F C C , U L , I C E S , C S A , N O M , C E , G O S T , C-TICK, V C C I , MIC, BSMI Interface type: U S B Product highlights: •  Easy to use - Operates as a letter drive, just like a floppy (No driver needed after Windows® 98}  •  Great compatibility • Plugs into any U S B connection in a desktop or notebook computer  e Convenient - Attach to key-chain, drop in pocket or purse and always have it with you •  Share files easily - Share M P 3 s with friends, carry files between computer labs and home, or share important data and presentations with customers and coworkers  Current price at www.dell.ca is $99.99 !  D O L L  104  Appendix 3 Description of the items used in the experiments (continued) Item 3: Samsung DVD player - listed by Time increase seller home | register j sign in/out 1 services | site map I help Q Browse  #  Back lo list of items  | Search  | Sell  | My eBay  | Community"  Listed in category: Computer & E l t C t r o n i « > H o m a Audio 8. V i d t o H o m a V i d e O D V D Player  S a m s u n g DVD-S221 P l a y e r You are signed in  Watch this item (track it in My eBay) I  , Current bid:  C$20.00 P l a c e Bid >  Time left:  Seller information fritjaynite  (564)  Feedback ratingi 364 Positive f e e d b a c k i 97.0<tf> Registered C a n a d a  01 hours 27 mins 7-d«v kiting  Read feedback reviews  Ends 30-A.jg-i)4 0 4 i l l ! l 2 EST  A s k seller a,„guestton  History:  § bids (C $1.00 Starting bid)  View seller's other auctions  High bidder:  tybecker  Location:  Vancouver  0  Buy with Confidenca  4- Shipping and payment derails  Description  Seller assumes all responsibility for listing this item.  This is a brand new Samsung S221 DVD player.  Samsung has dramatically improved the performance and the function of DVD players by utilizing its unique technology. Samsung DVD-S221 DVD Player features 2x audio playback, a phantom surround processor used to deliver a rich simulated surround sound effect through just two speakers, advanced disc navigation functions, and dual mode optical pickup to ensure complete compatibility with D V D , DVD-R, Video C D , and both CD-R and C D - R W discs that are encoded with M P 3 music files. Its three-terminal video interface delivers the ultimate in DVD Video quality even when connected to the latest large-screen direct view and projection televisions. An advanced graphic user interface provides clear on-screen icons and menus, making it simple to control the S221 's advanced navigation features.  Composite, S-Video and Component Video terminals Dual Mode optical pickup compatible with C D - R / R W Media Integral M P 3 decoder plays back music files from CD-R/W Phnatom Surround Processor delivers two-speaker surround effects Advanced Navigational system with Motion Zoom Number of disc: 1 Surround sound: DTS, Dolby Digital Jog Dial Control: On unit Audio outputs: 5.1 channel, R C A , Optical, Coaxial Video outputs: Composite, S-Video, Component Headphone jack: Yes Remote control: Basic Dimension: 9.4" depth, 3.1" height, 16.9" width, 6.2lb weight Reference price at www.samsung.cum: $99 !  Appendix 3 Description of the items used in the experiments (continued) Item 4: DVD movie - listed by Price low low seller  home I register I siqn in/out I services I site map I halfJ ® |  •  Back Id list of items  DVD:  Browse  | Search  L i f t e d In c a t e g o r y :  | Sell  | My eBay  Entertelnment>DVDf  | Community |  81 M o u l e f  > D V D > C o m e d y  Friends Collection - Volumes 1-6 (6 Discs)  Item number,3355E89433  : You are signed In  Watch this i t e m (trar.lf i t i n M y nRsy) i  Current bid:  C$20.00  Time left:  B3 hours 52 mint  Seller information  ; Race Sid > 7-day luting History:  Feedback rating i  564  Positive feedback! 1 7 . 0 4 %  Registered  C a n a d a  Read feedback reviews  Ends S0-Aug-04 0613b! 1.4 EST  Ask  5 bids CC *1.00 starting bid)  View seller's other auctions  High bidder:  johnchrome  Location:  Vancouver  seller a question  O Buy with Confidence  •f Shipping and payment detail? Seller assumes all responsibility for listing this item.  The comedy series loved all over world, more popular in syndication than "Seinfeld," "Frasier" and "The Simpsons!" Now for the first time, the top episodes of the longrunning series as voted by fans and the series' creators. Loaded with special extras in a special 6 disc sell Episodes include: The Pilot, The One with Two Parts (Parts 1 and 2), The One with All the Poker, The One Where R o s s Finds Out, The One with the Prom Video, The One Where No One's Ready, The One with the Embryos, The One with Ross* Wedding (Parts 1 and 2), The One with All the Thanksgivings, The One Where Everybody Finds Out, The One with the Blackout, The One with the Candy Hearts. The One Where Ross and Rachel...You Know, The One with the Football, The One That Could Have Been (Parts 1 & 2 ) , The One with Chandler in a Box, The One-Hundredth, The One with All the Resolutions, The One Where R o s s Got High, The One with the Proposal (Parti &2) STARRING: Courtney Cox, David Schwimmer. Jennifer Aniston, Lisa Kudrow, Matt LeBlanc, Matthew Perry. SPECIAL FEATURES: e  Encoding: Region 1  •  Number of Discs: 6  •  Full Screen  e  English Dolby Digital Stereo  » Behind-The-Scenes Footage •  Scene A c c e s s  Fului eshop is sailing it at $98,991  106  Appendix 3  Description of the items used in the experiments (continued)  Item 5: MP3 player - listed by Price low high seller home | register | sign in/out \ services I site map I help & Browse  Back tO liSt Of items  RCA  | Search  | Sell  | My eBay  | Community  Listed in category; Consumer Electronics > MP3, Portable Audio > MP3 Pla  Lyra 128M Mp3 player RD1080  Item number:53/762ei16  i You are signed In  Watch this item (track it in My eBay) ;  Current bid:  Seller information  C$20.00 Place Bid >  |  fwinkiestar ( 5 6 4 ) F e e d b a c k rating:  564  Positive f e e d b a c k : Q7.0<**>  Time left:  Registered  82 hours41 mins 7 - d » y IlstV.g Ends 30-Aug-04 G 5 : 2 5 : i S EST  (c $t.oo  History:  5 bids  High bidder:  table1230  Location:  Vancouver  Read feedback reviews Ask  Parting bid)  C a n a d a  seller a question  View seller's other auctions 0  « ' t h Confidence  •f BMu$a#: and Q| ymtnt jftffltfj Description  Seller assumes all responsibility for listing this item.  Tiny, Mighty and P a c k e d with Power! This funky R C A Lyra player is all the rage among digital music enthusiasts. You wont want to be without this colorful and compact device - you wont have any reason to be since it is made small enough to fit in the palm of your hand (unit weighs only two ounces). The small size of this personal digital player does not minimize the power that is packed into it. R C A has packed this chic Lyra player with 128MB of memory and a digital F M tuner. If over two hours of audio enjoyment is not enough to satisfy your needs, you can use the External M M C (Multi-Media Card) slot for expandable memory. Awesome things come in small packages this Lyra Player is no exception. PRODUCT FEATURES: a Personal Digital Player with 128MB Built-in Memory a P l a y s Multiple mp3 Compression Rates a M p 3 / W M A 2 Compatible a 128MB Built-in Flash Memory a Digital F M Tuner with 10 Presets •  U S B Connection for Faster Downloads  a Upgradeable for Future Audio Compression Formats a A c c e s s o r i e s Included: X-Phone Stereo Headphones, Carry Case with Belt Clip, U S B Connection Cable, Music Management Software C D ROM,  2 " A A A " Batteries  a Product Dimensions: 2.3" H x 2.4" W x 1.2" D a Weight: 2.00 ozs The listing price at www.rca.com is $99.  107  Appendix 3 Description of the items used in the experiments (continued) Item 6: Sports bad - listed by Price time increase seller home | register | sign in/out | services | site map | hel£ CA  4*  B a c k tg |JSt Qf Items  | Browse  | Search  Listed in category)  | Sell  | My eBay  0  | Community |  Sports > Clothing and Accessories > Bags  Mountain E q u i p m e n t Co-op: A star d a y pack  Item nurnber:55780G2873  ; You are signed in  JH|  Watch this item (track it in My fiRay) !  Current bid:  Setter information  C$20.00  • ••• • •• • ••  Place Bid > Time left:  |  81 hours 17 mlns 7-day listing  ^H^HluL  End* 30-Aug-0«l 04i01il6  History:  5 bids (c *i.oo starting bid)  High bidder:  edinaBB  Location:  Vancouver  ns35 ( 5 6 4 l  .. •  •  •.  •• • ••.  •  . . • •. •  •  •  Feedback rating) 364 Positive f e e d b a c k : 97.9Va Registered C a n a d a  Read feedback reviews Ask seller a question View seller's other auctions D.Buy ^ C o n f i d e n c e  • * ShiDOina and navmant details  Description  Seller assumes all responsibility for listing this item.  The A-Star will tote your board(s), avy gear, and spare clothing in style and comfort, with enough capacity for luxurious day trips, carefully packed overnighters, or hut-tohut touring.  The harness on the front holds a snowboard or skis, and is armored for durability against metal edges by a Hypalon® patch and grippy rubberized straps. If desired, skis can be hung diagonally so they don?t gouge your ankles. A special lower loop lets you mount fat skis quickly and easily. The outermost pocket is contoured for a shovel blade. N B X I comes a stuff pocket ? it?ll hold your snow kit and features sleeves for your probe and shovel handle. It has its own drawcord closure and a compression strap that pulls the opening snugly beneath the shelter of the pack lid, yet you can open the pocket quickly without unclipping the lid. Both the blade pocket and the avy gear stuff pocket remain quickly accessible even when ycu have boards strapped on. The upper lid compartment features a wallet-sized security pocket with a keyclip. The underside of the lid has a zippered map pocket. In the main pack compartment is a mesh hydration pouch holster, positioned next to your back for optimum balance. A covered hose port at the top of the pack lets you route the drinking tube over either shoulder. PRODUCT •  FEATURES:  Smooth, snow-shedding 630-denier Superpack nylon,  e Bottom fabric is rugged 1050-demer nylon. e Comfy stretch-woven fabric surfaces on back pad, shoulder strap linings, and waistbelt linings. e Thermo-moulded, laminated foam backpad provides comfort while allowing full body flexibility e Backpad is tapered over hips to allow the waistbelt to wrap into a snug, comfortable fit. e Flexible framesheet and two aluminum stays transfer weight efficiently to the hips e Contoured padded shoulder straps, with stablizer straps and a quick-adjust sternum strap. a Contoured waistbelt, padded with dual-density foam e Additional open-cell foam lumbar pad for increased comfort. e A l l exposed zippers are weathertight YKK® Uretek Retail price at Mountain Equipment C o u p : C$99.99.  108  Appendix 4  Questionnaire  There were two parts in the questionnaire. They are presented to experiment participants in different stage of the study. Question 12 in part 1 and Question 10 in part 2 are open ended questions. All other questions are 7 point Likert-scale questions. Part 1; Questionnaire with each auction Subjects were required to complete this questionnaire right after they finish each auction, according to their feelings about the particular auction and the seller, whether they bid or not in the auction. 1. I think this seller is honest. (Gefen and Karahanna, 2003) (Trust) 2. I think this seller acts sincerely in dealing with buyers. (Cheung and Lee, 2000) (Trust) 3. I think this seller cares about buyers. (Gefen and Karahanna, 2003) (Trust) 4.  I think this seller is not opportunistic. (Gefen and Karahanna, 2003) (Trust)  5. I think this seller will provide good service. (Gefen and Karahanna, 2003) (Trust) 6. I think this seller will act predictably. (Gefen and Karahanna, 2003) (Trust) 7. I think this seller is trustworthy. (Gefen and Karahanna, 2003) (Trust) 8.  I believe that the seller will deliver the product I purchase according to the posted delivery terms and conditions. (Ba and Pavlou, 2002) (Trust)  9. I believe that the seller will deliver the product I purchase without unjustified delay. (Ba and Pavlou, 2002) (Trust)  109  Appendix 4  Questionnaire (continued)  10. I believe that this seller will deliver a product that matches the posted item description. (Ba and Pavlou, 2002) (Trust) 11. I am willing to bid on this auction. 12. Why are or why aren't you willing to bid? 13. I am confident that my judgment about this seller's trustworthiness is correct. (Confidence) 14. I believe that I can accurately assess the reliability of this seller. (Confidence) 15. The information provided by eBay ID Card is helpful to me in judging the seller's trustworthiness. (Helpfulness) 16. I understand the information contained on the ID Card. 17. The information on the ID Card is easy to understand. 18. The ID Card provided information not available in the comments. 19. The ID Card was useful, over and above the comments, in learning about the seller. Part 2; After auctions questionnaire  When the subjects were done with all 6 experimental auctions, they were asked to the following part of questionnaire according to their experiences with the 6 auctions. 1. I fully understand how to bid in auctions. (Understand) 2. I fully understand the risk associated with bidding in online auctions. (Understand) 3. I fully understand the experimental reputation mechanism. (Understand)  110  Appendix 4  Questionnaire (continued)  4. I fully understand the ID card. (Understand) 5. I have made my judgment about the sellers' trustworthiness based on: 2  3  eBay I D  4 Equal  card  5  6  7 Feedback comments  6. I read sellers' positive comments carefully. 7. I read sellers' negative comments carefully. 8. I fully understand the rules of the experiment and how I will be rewarded. (Understand) 9. This auction game is interesting. 10. Please describe your strategy to win this game and maximize your leftover fund.  Ill  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0091655/manifest

Comment

Related Items