Beliefs about Usage, System and Institution on Intention to Use
Eva Chen1 and Ilka Weber2
1 John Molson School of Business, Concordia University eh_chen@jmsb.concordia.ca
2 Institute of Information Systems and Management (IISM), Universität Karlsruhe (TH) ilka.weber@iw.uni-karlsruhe.de
1 Introduction
Designing electronic marketshas become an important issue for electronic commerce.
Unlike traditional markets, electronic markets are supported by electronic media; they must be consciously designed since they are limited by the technical infrastructure. In essence, electronic markets are information systems that process and transport data and provide communication for agent interaction.
So far, there are many scientific approaches for analyzing and designing market institutions – nevertheless, a solid engineering practice for electronic markets is essential.
An understanding and deep knowledge of various research disciplines such as economics, computer science and jurisprudence is necessary as these disciplines are at least indirectly involved in the creation, design, evaluation and introduction of electronic markets (Roth 1999). So far, there is little knowledge on which institutions are suitable for certain situations or how the outcome of an electronic market should be measured and evaluated.
Furthermore, as Roth (1999) points out, the practical design of electronic markets has to deal with complexities, mainly of the economic environment itself, and the participants' strategic behavior. Dealing with such complexities requires more than simply attention to the institutional rules of a market. Additional methods and tools from other disciplines are needed to supplement traditional approaches. For example, experimental and computational economics are supplementary theories that help in understanding complexities and show how to deal with them.
Thus, more and more, an economist is regarded as an "engineer" (Roth 2002;
Varian 2002) who has extensive knowledge and a solid foundation in theory and methodology. Indeed, the design of market institutions shifts from a pure science to engineering - market engineering (Weinhardt et al. 2003). The purpose of market engineering is "to develop economically founded approaches and methods that support the designers in facing the difficulties associated with the design problem"(Neumann 2004).
While designing the institutional rules, the market engineer wants to achieve a certain effect and economic performance of the market. To automate the process of designing electronic markets in a systematic and structured manner tools are necessary (Neumann et al. 2005): These tools should close the gap between a structured design of electronic markets and the absence of methods and support the market engineer. So far, several auction platforms such as the Michigan Internet AuctionBot (Wurman et al. 1998), the Global Electronic Market (Reich and Ben-Shaul 1998), the Generic Negotiation Platform (Benyoucef et al. 2000) or the meet2trade platform (Weinhardt et al. 2005) have been developed as tools for designing and configuring auctions or even testing the designed auctions.
In this exploratory study, our purpose is to assess the perceptions of agents in an electronic market in order to understand the effects of the system characteristics (i.e.
reliability, quality, timeliness, etc) on institution and usage beliefs, as well as the influence on intention to use. We draw upon theories from two disciplines: Economics (Section 2) and Information systems (Section 3) to develop our research model (Section 4), and we test this model in a laboratory experiment using the meet2trade platform with ninety students in a Western European university (Section 5). The results demonstrate that perceptions of the institution, usability and system characteristics play a significant part in shaping the participants’ intention to use (Section 6). Based on our findings, we discuss the impact on electronic market design and suggest future work to build on this study (Section 7).
2 Electronic Market Systems and Auctions
The foundation for our research is the design of electronic markets, which has the potential to facilitate market activities by allowing for flexible mechanisms that adjust to the task and agents’ requirements. This study employs meet2trade as the platform to create an
auction mechanism governing the market interactions. The design of the auction system and the institutions are described below.
2.1 The auction system meet2trade
meet2trade is a generic, flexible trading platform facilitating an easy creation and automation of auction-based markets. The platform is flexible enough to host markets from a large variety of domains and to support various market mechanisms. Further, considering the user perspective, the platform is configurable: it meets the users' individual trading needs and supports users in selecting and configuring markets by adapting the client to his individual preferences for these tasks. Beside its flexibility and configurability the most powerful advantage is the facilitation of designing markets by a Market Modeling Language (MML). This language was developed to describe electronic market parameters and to enhance an easy development and creation of electronic auctions (Mäkiö and Weber 2004). A screenshot of an electronic auction created by meet2trade is shown in Figure 1.
Figure 1. Screenshot of Auction System Created by meet2trade
The innovativetrading concepts offered in this system - e. g. market configuration and platform flexibility - offer a starting point for a vast area of economic research. The meet2trade system delivers not only the platform to host these concepts but also provides a toolbox for their examination. The tools offered by meet2trade consist of the experimental system MES and an agent-based simulation environment AMASE. Basically, the intention of MES is to conduct economic experiments on electronic markets on the meet2trade system instead of deploying experimental standard software. Thus, MES and meet2trade
are strongly interweaved - MES is integrated in the underlying platform using components of meet2trade (Kolitz and Weinhardt 2006).
The workbench meet2trade follows a client-server architecture with a central server. The server provides the running platform for all available markets as well as the hosting of all data (e.g. user data, account data, product information, protocol data) and the data preparation. The clients connected to this central server display this data and provide an interface for submission of bids and for displaying relevant information.
2.2 Auction Mechanisms
In literature second-price auctions in a symmetric independent private values auction (SIPV)1 setting have been thoroughly analyzed and discussed. Moreover, literature on affirmative actions in auctions subsidizing a class of bidders is rarely. Often such classes accord to economic disadvantaged, less effective bidders. There are different forms of subsidizing such groups - advantages can be given in form of set-a-sides, discounts or bidding credits, or special payment terms (Rothkopf et al. 2003). Reasons for such policies stem from thoughts about non-economic aspects such as fairness, discrimination, populism etc. Examples of auctions with affirmative actions show, that bidding credits might pay for the seller and that auction revenue can be increased (c.f. to Corns and Schotter (1999) or Ayres and Cramton (1996)).
The interest in auctions with discounts brings up the desire to find explanations to the questions how affirmative actions such as discounts in auctions influence the bidding behavior and thus the market outcome as well as how institutional beliefs influence user perceptions and the intention to use auction systems Driven by these questions an experimental study was conducted analyzing the effect of institutional changes on bidding behavior as well as users’ perceptions of auction systems including beliefs about the information system as well as beliefs from the institution. Focusing on the institutional rules, a pure second-price auction (SPA) and a second-price auction with discount (DA) are employed and conducted in the laboratory experiment.
The DA is based on the second-price sealed-bid auction (Vickrey auction, (Vickrey 1961)), for short second-price auction, which is at the same time used as benchmark auction. In essence, the DA is a second-price sealed-bid auction augmented with a
1 The SIPV has been discussed in auction theory literature. For a more detailed description on this auction model confer for example to Wolfstetter (1999).
discount. The fundamental concept of that auction is, that exactly one seller is randomly selected to whom the discount is assigned. This bidder is called the designated bidder. In the DA the pricing policy rules the following: If the winning bidder is not the designated bidder, than the price to pay is the final price of the auction that is the second highest bid.
If the designated bidder wins the auction, then the payment is the discounted final price of the auction.
The goal of this experiment is to evaluate the bidders' strategic behavior and to measure the impact of that bidding behavior on the auction revenue in a controlled environment.
Additionally, the experiment aims at validating the beliefs of subjects about the institution and the system. From an individual perspective, the institution influences beliefs on uncertainty and risk associated to employing the auction system to transact in the market.
Therefore, the subjects’ perceptions towards the auction mechanism as well as the auction system are measured.
The experiment was conducted with the meet2trade workbench and the connected experimental system MES. The underlying institutional rules were configured and employed in meet2trade; the experimental sessions were configured, conducted and settled with MES.
3 Measuring System Perception
Information systems research has long developed a rich course of investigation into the factors and processes that intervene between technology investments and their economic return. Most studies are built on users’ perceptions of the system or behavior towards the system, which impacts on the system’s ultimate success in an organization. Our work also hinges on users’ perceptions as an indication for system success through improvement on design, but the environment differs as the unit of analysis is the individual interacting in a market environment rather than an organizational setting. In order to extract the salient perceptions for our work, we examine theories relating to technology acceptance, institution and system characteristics.
3.1 TAM
One of the predominant theories in IS research is the Technology Acceptance Model, TAM, which has been tested longitudinally (Venkatesh and Davis, 2000), in many different settings and with various technologies (Davis, 1989). The external variables
reflect system characteristics that influence the perceived ease of use (PEU), the user’s belief of the amount of effort needed to utilize the system, and both which affect the perceived usefulness (PU), the user’s belief of the degree to which using the system will enhance his or her performance. Furthermore, PU and PEU affect behavioral intention to use (IU) that serves to predict future system use.
While there have been critiques of TAM, such as the model’s inability to explain the external variables causing the fundamental beliefs on system usage (perceived usefulness and perceived ease of use) (Legris et al., 2003), TAM is shown to be appropriate in predicting acceptance even when users are given a prototype system for evaluation. Moreover, Davis and Venkatesh (2004) demonstrate that PU measured pre- prototype (i.e., users have no direct hands-on usage experience) is statistically powerful in predicting usage intentions six months after implementation. Figure 2 represents the version of TAM for prototype development (Davis and Venkatesh, 2004).
Figure 2. TAM, reproduced from Davis (Davis and Venkatesh, 2004).
Another point of caution when applying TAM is to consider the technology under investigation in order to extract all salient beliefs (i.e., beyond PU and PEU) that maybe specific to the system or the setting in which usage occurs (Legris et al., 2003).
3.2 Institution Theory
Given that the focus of this paper is on market systems and more specifically auctions, beliefs affecting IU in TAM must also reflect the very nature of auction systems. One particular factor involving auction systems is the mechanism presiding over the exchange
among participants. The mechanism constitutes the regulative institution that affects the method in which information is communicated and more notably the conditions for trade, e.g., market price (Smith, 1982). From an individual perspective, the institution influences beliefs on uncertainty and risk associated to employing the auction system to transact in the market (Ba and Pavlou, 2000; McKnight et al, 2002). Therefore, the evaluation of auction systems needs also include perceived measure of the institution shaping the exchange (Malone et al., 1987; Bakos, 1998).
However, from the perspective of organizational theory, institutions expand beyond the formal protocols to include cultural and normative beliefs on the social structure that regulates behavior (Scott, 2001). Building on both the formative and informative conceptualizations of institutions, Pavlou and Gefen (2004) capture the perceptions of the institutional structure to the construct of trust in the market. Furthermore in an e-commerce setting, Gefen, Karahanna and Straub (2003) related trust to TAM as an intervening variable from PEU to PU, as shown in Figure 2.
Figure 3. Integrating Trust to TAM, adapted from Gefen et al. (2003).
Moreover, the perception of trust in the market mechanism or the provider of the marketplace is affected by the perceived effectiveness of the feedback mechanism, which is essentially the system characteristics such as information accuracy and quality, as well as system reliability (Pavlou and Gefen, 2004).
3.3 System Characteristics
As a means of exploring the external variables suggested in TAM, perception of system characteristics, more specifically, information and system quality have been proposed to
influence beliefs on usage (DeLone and McLean, 1992). Later, Wixon and Todd (2005) decomposed information and system quality, and empirically related them to PU and PEU respectively. However, the relationship was established through information and system satisfaction, which were significantly linked, see Figure 3.
Based on this integrated model, the connection between system characteristics and PU, as well as PEU is unclear due to the fact that satisfaction is used as a moderating variable. On the other hand, Seddon and Kiew (1996) tested a direct relationship between information and system quality to PU. They found that system quality was a more significant predictor of PU than information quality. Nevertheless, there is little research to suggest the relationship among information and system quality, and belief from TAM from auction or market systems.
Figure 4. Integrating System Charateristics to TAM, adapted from Wixon and Todd (2005).
4 Research Model
The aim of this study is to assess users’ perceptions of an electronic market system using a second price sealed-bid auction mechanism, by means of expanding the TAM model to include system characteristics and the institutional-based belief towards the market mechanism. In essence, the independent variables are the perceived constructs on the attributes of electronic auction system, which affects perceptions on usage and the market mechanism that lastly influences the user’s behavioral intentions towards the electronic
auction system. In order to ascertain the goal of this study, the following research model is proposed, Figure 4, to illustrate all encompassing variables for consideration.
IU PU
PEU Perceived
trust System
timeliness
System reliability
Information quality
H1
H2a
H2b H2c
H3a H3b H3c
H4a H4b
H4c
H6b
H7c
H7b
H6a
H7a H5
Figure 5. Research Model.
4.1 Independent Variables
The independent variables reflect the main characteristics of the system in terms of information quality, which refers to the degree to which the system provides necessary information for the individual to interact in the market. System quality is decomposed into two separate constructs: system reliability that is the degree to which the electronic market is dependable and offers accurate data, and system timeliness, which refers to the degree which the system responses promptly to requests for information or action. Although, Wixon and Todd (2005) posit that these two variables capture of system quality, their study was unable to significantly demonstrate that system timeliness is indeed a dimension of system quality. Given that system reliability is dimension of system quality, a correlation can be inferred as Wixon and Todd showed a significant link between information quality and system quality.
H1: System reliability will be positively correlated to information quality.
Building on the work of DeLone and McLean (1992), Seddon and Kiew (1996), as well as Pavlou and Gefen (2004) that connect system characteristics to perceptions of usage and market mechanism (in the latter study), the following hypotheses are proposed:
H2a: System timeliness will have a positive direct effect on PU.
H2b: System timeliness will have a positive direct effect on perceived trust.
H2c: System timeliness will have a positive direct effect on PEU.
H3a: System reliability will have a positive direct effect on PU.
H3b: System reliability will have a positive direct effect on perceived trust.
H3c: System reliability will have a positive direct effect on PEU.
H4a: Information quality will have a positive direct effect on PU.
H4b: Information quality will have a positive direct effect on perceived trust.
H4c: Information quality will have a positive direct effect on PEU.
4.2 Intervening Variables
The intervening variables consist of the beliefs on usage from TAM (Davis et al., 1989) and trust in the institution (Gefen et al., 2003), which lead to the suggestion of the following hypotheses:
H5: PU will have a positive direct effect on IU.
H6a: Perceived trust will have a positive direct effect on IU.
H6b: Perceived trust will have a positive direct effect on PU.
H7a: PEU will have a positive direct effect on IU.
H7b: PEU will have a positive direct effect on PU.
H7c: PEU will have a positive direct effect on perceived trust.
4.3 Dependent Variables
The dependent variable is the behavioral intention to use (IU) an electronic auction system, which reflect future intentions towards the system and serves to predict usage. Therefore, it is important for system designers to understand the determinants of IU in order to induce participants to use the auction system.
5 Methodology
The research questions require an experimental design that allows us to compare the DA market institution to the second-price auction. In principle, both market institutions follow the rules of the underlying second-price auction mechanism and only differ in the existence of the discount or not in any other design parameter. The experiment follows between subjects design; it focuses on the isolated effect of levels of variables. The level of a treatment variable is only varied between single treatments and across subjects but not within one trial. The experimental design describes the nature and the number of treatments focused in the experiment2. The institutional rules vary according to two mechanisms:
- SPA: second-price auction
- DA: discount auction mechanism
Throughout the experiment only the institutional rules changed while keeping all other parameters on a constant level and leaving the environmental parameters unchanged. The setting SPA constitutes the benchmark case: auctions without discount, i.e. pure second- price auctions, are conducted and bidding behavior in these auctions is observed. The sessions of both settings are conducted separately and each subject participates only once in the experiment.
5.1 Conducting the experiment
The experiment was conducted at the experimental laboratory of the Institute of Information Systems and Management at Universität Karlsruhe (TH) from December 14th to December 16th, 2005. Participants were randomly selected from a database with more than 3,000 volunteers. All participants were undergraduate or graduate students mostly from the School of Economics and Business Engineering. Only a few subjects invited for participation have participated in a negotiation or auction experiment before; also only a few participants are experienced in negotiations or auctions. None of the subjects participated repeatedly.
The experiment was computerized and conducted with meet2trade and the meet2trade experimental system MES. The meet2trade market core was used to configure
2 Although the experimental design also focused on symmetric and asymmetric probability distribution of auction values, it is not the focus of this paper. Treatments examining this variable were statistically proven (by MANOVA) to have no influence on the factors in this study.
and to employ the institutional rules of the discount auction mechanism and the corresponding second-price auction; with the experimental system each session of the experiment was configured, conducted and settled.
All decisions of the participants as well as answers to questionnaires were entered in a computer terminal. In the laboratory, participants were randomly seated at one of 15 visually isolated cabins each equipped with a computer-terminal. The instructions were read aloud to all participants and each participant had to answer a quiz about the rules of the experiment and about the rules of the auction mechanism explained in the instructions.
All participants had to answer the questions correctly. Then, the first auction round started --there were no trial rounds. In each auction round five independent auctions were conducted at the same time by different groups of subjects. Recall that before the first round started the 15 participants were randomly assigned to one of the five groups. Each group consists of three subjects participating in the same auction. The assignment of participants to groups was fixed and did not change throughout the experiment. Before each auction round, participants were informed about their valuations for the object being auctioned in the current round as well as their actual experimental account on the computer screen.
In each auction round each bidder had to decide how much to bid for the object based on his induced valuation and to type the value of the bid in the bidding screen.3 By confirming this value the bid was submitted and entered in the experimental software. At the end of the auction, participants received a notification of the auction result displayed in the screen. Information about being the winning bidder, the name of the winning bidder (e.g., player 1 etc.), the final price of the auction, and the price to pay in case of being the winning bidder was indicated on the screen as well as on the experimental account.
In setting DA subjects were additionally informed about whether they were the designated bidder by displaying the information 'Discount: 20%' or whether they were a non-designated bidder by indicating the information 'Discount: no Discount' on the bidding-screen. Concerning the auction result, a participant being the designated bidder
3 In each round the 15 valuations assigned to the 15 subjects are randomly selected between [100,109] (10 valuations) and [146,150] (5 valuations). Each valuation is an integer number and each value out of the two intervals is assigned only once to the participants. In essence, each valuation of the 15 valuations is assigned exactly to one participant. Participants are informed, that the valuations are integer numbers and randomly drawn from the interval [100,150]. Information about the probability distribution function is not revealed to the participants.
and the winner in the auction was informed that the price to pay for the object was a discounted price.
Participants played the six consecutive auction rounds, each auction round limited to 2 minutes. After the six auction rounds, participants were asked to answer a screen- based questionnaire of 48 questions by entering the answers on the computer. The questionnaire comprised questions about the participants’ background, their behavior in conflict situations, their attitudes concerning auction systems as well as questions on the system and user interface design. A descriptive of their background is found in Table 1.
At the end of the experiment subjects were then called individually to be paid privately. The experimental session lasted about one hour. Table 2 summarizes the approximate duration of the different phases in an experimental session. Overall, 6 sessions with 15 participants in each session were performed. Three sessions involved the setting SPA and the rest for setting DA.
Participant characteristics Percentage
male 73.3 Gender
female 26.7
no experience 28.9
some experience 45.6
Auction Experience
experienced 25.5 collaborative 4.4 competitive 1.1 compromising 24.4 accommodating 25.6 Conflict
approach
avoiding 44.4 undergraduate 56.7 graduate 41.1 Education level
other 2.1 Information engineering
& management
11.1
Business engineering 48.9
Computer science (Informatics) Field of study 3.3
other 36.7 Table 1. Descriptive of Participants
Phases of Experimental Session Approximate Duration
Reading instructions 15 min
Questionnaire on instruction 14 questions in setting SPA 17 questions in setting DA
10 min
6 consecutive auction rounds 20 min
Questionnaire on subjects’ background, subjects’ behavior, and auction systems 48 questions
15 min
Payment of subjects 10 min
Total 1h 10 min
Table 2. Duration of phases in an experimental session
6 Results
The findings for this study are based on responses captured from ninety students participating in an auction experiment. Firstly, multivariate analysis of variance (MANOVA) is carried out to determine the effects of the treatments and discount in the experiment. Secondly, factor analysis is performed to determine the appropriateness of the measurement model. Thirdly, structure equation modeling (SEM) is conducted to assess the nomological network.
6.1 MANOVA
Before we can verify our research model, we need to determine effect of the treatments (SPA vs. DA) on our constructs (Lattin et al., 2003). Since our model encompasses seven constructs (system timeliness, system reliability, information quality, PU, perceived trust, PEU and IU), MANOVA allows us to measure the variance for both the main effect on the model, Table 3, and the interaction effect between factors, Table 4. Surprisingly, the results show that the institutional rules do not significantly affect the participants’
perceptions of the system characteristics (system timeliness, system reliability and information quality), beliefs on usage (PU and PEU) and institutional-based trust, and IU.
Effect Value F Hypothesis df Error df Sig.
Intercept Pillai's Trace .023 .275(a) 7.000 81.000 .962
Wilks' Lambda .977 .275(a) 7.000 81.000 .962
Hotelling's Trace .024 .275(a) 7.000 81.000 .962
Roy's Largest Root .024 .275(a) 7.000 81.000 .962
T Pillai's Trace .032 .388(a) 7.000 81.000 .907
Wilks' Lambda .968 .388(a) 7.000 81.000 .907
Hotelling's Trace .034 .388(a) 7.000 81.000 .907
Roy's Largest Root .034 .388(a) 7.000 81.000 .907
discount Pillai's Trace .057 .705(a) 7.000 81.000 .667
Wilks' Lambda .943 .705(a) 7.000 81.000 .667
Hotelling's Trace .061 .705(a) 7.000 81.000 .667
Roy's Largest Root .061 .705(a) 7.000 81.000 .667
T * discount Pillai's Trace .000 .(a) .000 .000 .
Wilks' Lambda 1.000 .(a) .000 84.000 .
Hotelling's Trace .000 .(a) .000 2.000 .
Roy's Largest Root .000 .000(a) 7.000 80.000 1.000
a Exact statistic; b Design: Intercept+Treatment+discount+Treatment * discount
Table 3. Multivariate Tests(b) for the Model
Variable Sum of squares df Mean Square F Sig.
Corrected Model PU 5.339(a) 2 2.670 .264 .769
PEU 9.165(b) 2 4.582 .598 .552
info_qlt 67.296(c) 2 33.648 1.764 .177
IU 16.530(d) 2 8.265 .757 .472
trust 17.030(e) 2 8.515 .887 .416
sys_rely 17.742(f) 2 8.871 .848 .432
sys_time 13.277(g) 2 6.639 2.281 .108
Intercept PU .580 1 .580 .057 .811
PEU 2.246 1 2.246 .293 .590
info_qlt 15.026 1 15.026 .788 .377
IU 4.664 1 4.664 .427 .515
trust 4.464 1 4.464 .465 .497
sys_rely 2.603 1 2.603 .249 .619
sys_time 3.441 1 3.441 1.183 .280
discount PU 3.714 1 3.714 .367 .546
PEU 9.126 1 9.126 1.191 .278
info_qlt 32.117 1 32.117 1.684 .198
IU 15.810 1 15.810 1.449 .232
trust 17.014 1 17.014 1.772 .187
sys_rely 3.742 1 3.742 .358 .551
sys_time 8.652 1 8.652 2.973 .088
Treatment PU 4.009 1 4.009 .396 .531
PEU 2.331 1 2.331 .304 .583
info_qlt 7.676 1 7.676 .402 .528
IU 1.039 1 1.039 .095 .758
trust 3.004 1 3.004 .313 .577
sys_rely 6.159 1 6.159 .589 .445
sys_time .370 1 .370 .127 .722
a R Squared = .006 (Adjusted R Squared = -.017);b R Squared = .014 (Adjusted R Squared = -.009); c R Squared = .039 (Adjusted R Squared = .017); d R Squared = .017 (Adjusted R Squared = -.005); e R Squared = .020 (Adjusted R Squared = -.003); f R Squared = .019 (Adjusted R Squared = -.003); g R Squared = .050 (Adjusted R Squared = .028)
Table 4. Tests of Between-Subjects Effects
6.2 Factor Analysis
Factor analysis serves to examine the validity of the constructs reflected by more than one item. Table 5 conveys the univariate statistics of the items and reliability values for each factor. The internal consistency is indicated by the Cronbach’s alpha, which is above 0.7 for all factors except for system timeliness. The Cronbach’s alpha for system timeliness maybe skewed due to the fact that this factor is measured by only two items (Lattin et al., 2003).
Mean Std. dev. Reliability
PU1 4.3667 1.5247
PU2 4.3333 1.3657
PU3 3.9889 1.5102
PU4 4.7111 1.4002
0.840
PEU1 5.8778 1.1595
PEU2 5.6333 1.3857
PEU3 5.8222 1.1859
0.907
Perceived trust 1 3.9556 1.6484 Perceived trust 2 4.5556 1.4696 Perceived trust 3 4.4111 1.7281 Perceived trust 4 4.2889 1.6776
0.776
Info_quality1 4.2889 1.7498
Info_quality2 4.0111 1.6860
Info_quality3 5.0667 1.5125
Info_quality4 4.9444 1.5533
Info_quality5 5.0556 1.3354
Info_quality6 4.5556 1.3748
0.829
Sys_reliability1 5.7333 1.1974 Sys_reliability2 5.4889 1.4162 Sys_reliability3 5.5889 1.1406 Sys_reliability4 5.3111 1.2238
0.822
Sys_timeliness1 5.0111 1.5397
Sys_timeliness2 4.9667 1.1163 0.664
IU1 3.9889 1.6860
IU2 4.1889 1.5857
IU3 4.1111 1.5466
IU4 4.4667 1.2649
0.842
Table 5. Univariate Statistics and Internal Consitancy
The convergent and discriminant validities are apparent in the rotated factor matrix, shown in Table 6, as related items load highly to similar factors and poorly to dissimilar factors. The items for PU, PEU and IU were adapted from Davis and Venkatesh (2004), those for perceived trust were from McKnight et al. (2002), and those for system characteristics were from Wixon and Todd (2005).
1 2 3 4 5 6 7
PU1 .732 .034 .134 -.025 .328 .060 .142
PU2 .654 .142 .111 .095 .081 .242 .047
PU3 .593 -.008 -.062 .202 .216 .232 .101
PU4 .739 -.048 .060 .038 .185 .232 .274
PEU1 .075 .095 .870 .152 -.056 .076 .122
PEU2 .076 .057 .795 .175 .147 .123 .092
PEU3 -.018 .083 .880 -.008 .026 .068 .217
Trust1 -.047 .229 .174 .097 .409 .533 .061
Trust2 .066 .369 .287 .094 .293 .616 .033
Trust3 .239 -.108 .018 .013 .030 .729 .267
Trust4 .224 .146 .068 -.058 -.047 .614 .072
Info_quality1 -.081 .244 .076 .463 .419 .147 .187
Info_quality2 -.153 .144 .011 .511 .358 .202 .258
Info_quality3 .100 .184 .269 .718 -.083 -.228 .182
Info_quality4 .083 -.003 .184 .700 .039 -.015 .182
Info_quality5 .276 .385 -.029 .572 .238 .139 -.002
Info_quality6 .133 .398 .071 .550 .307 .177 -.103
Sys_reliability1 .050 .600 .171 .345 -.001 .269 .020
Sys_ reliability 2 .126 .657 .046 .124 -.058 .293 -.116
Sys_ reliability 3 -.087 .824 .068 .045 .124 -.050 .150
Sys_ reliability 4 .018 .687 .156 .226 .044 .055 .376
Sys_timeliness1 .040 -.032 .292 .039 .036 .019 .681
Sys_ timeliness2 .174 .158 .034 .146 .173 .189 .610
IU1 .222 -.027 -.018 .111 .578 -.016 .161
IU2 .585 .026 .007 .136 .565 .079 .267
IU3 .476 -.049 .097 .022 .653 .167 .244
IU4 .451 .156 .166 .110 .505 .174 .014
Extraction Method: Unweighted Least Squares. ; Rotation Method: Equamax with Kaiser Normalization.
Table 6. Rotated Factor Matrix
6.3 Structural Equation Modeling
Based on the insignificant findings from the MANOVA, we combine the responses from the treatments to test the structural model. This was accomplished by the EQS software for estimating the path coefficients, correlation and variance explained using maximum likelihood (Bentler, 2004). The results of the modeling are reported in Figure 6 for all significant relationships. Although the fit indices are lower than the general recommendation, they remain in the acceptable levels for an exploratory study, meaning CFI and NNFI above 0.80, as well as RMSEA below 0.10 (Bollen, 1989).
The findings indicate that 70% of the variance of IU can be explained by PU (R2 = 22%) which is significantly influenced by system timeliness and perceived trust. While, perceived trust (R2 = 36%) is affected by system reliability and PEU (R2 = 10%), which is
influenced by information quality. The correlation between system reliability and information quality was found to be significant.
Figure 6. Research Model Results
7 Discussion ad Future Research
Based on a prototype of an electronic market system - the meet2trade system - this study outlines the antecedents that affect the intention to use such a system for trading in the marketplace. The participants were subjected to different institutions in order to examine their perceptions of the system in terms of characteristics, institution-based trust and usage as these impact their future intention towards such systems. Surprisingly, the results of this exploratory study reveal that objective manipulation of the institution has no effect on perception. In fact, IU is primarily affected by PU, which corresponds to Davis and Venkatesh (2004) findings that IU is mostly influenced by PU in prototype settings. This
implies that users express intentions to employ electronic market systems that give them a relative advantage in performing their trades.
In addition, perceived trust is an intervening variable between PU and PEU as shown by Gefen, Karahanna and Straub (2003). Thus, agents who perceived the system to be easy to use are likely to find it useful if they believe that the institution represented by the system is trustworthy. Moreover, there does not appear to have a direct relationship between PEU and PU stating the importance of the institution-based trust in shaping the perception of usefulness for such system.
The system characteristics are essential factors for market engineers as they form the beliefs on usage and institutions. Contrary to the literature, information quality affects only PEU and not PU. This maybe an artifact of the system under investigation, whereby the information provided affects the users’ view of the system complexity and not necessarily its usefulness. For example, the format in which the winning bids are shown influences the perception of system complexity (the mechanism used to generate these wins) but not on beliefs relating to enhancement of trade performance (PU). On the other hand, system timeliness affects PU, which is indicative of auction dynamics embedded into these market systems. The speed to which bids are processed by the system impacts the agents’ performance in the auction. Perceived trust is formed from the beliefs concerning system reliability. This finding reflects the importance of the dependability of the mechanism awarding the winners on the trust perceived by the users. Therefore, different system characteristics play different roles in formulating perceptions.
Furthermore, we showed that there is a difference between system reliability and timeliness, which would explain the need to separate these two factors rather than to lump them together as system quality.
In future work, we hope to compare perceptions between different groups of users by means of sampling techniques such as jackknifing and boot strapping, along with multi- group analysis in SEM.
Acknowledgement
We would like to thank Gregory Kersten, Christof Weinhardt and Dirk Neumann for their generous support and knowledgeable advice. This work has been partially funded by the Natural Science and Engineering Research Council, the Social Science and Humanities Research Council in Canada and the Alexander von Humboldt Foundation, Germany.
References
Ayres, I. and P. Cramton (1996): “Deficit Reduction through Diversity: How Affirmative Action at the FCC increased Auction Competition”, Stanford Law Review, (48), pp. 761- 815.
Ba, S., and P. A. Pavlou (2002): "Evidence of the Effect of Trust Building Technology in Electronic Markets: Price Premiums and Buyer Behavior," MIS Quarterly, 26, 243-268.
Bakos, Y. (1998): "The Emerging Role of Electronic Marketplaces on the Internet,"
Comm. ACM, 41, 35-42.
Bentler, P. M. (2004): Eqs 6 Structural Equation Program Manual. Multivariate Software Inc.
Benyoucef, M., R. Keller, S. Lamouroux, J. Robert, and V. Trussart (2000): “Towards a Generic E-Negotiation Platform”, Proceedings of the Sixth Conference on Re- Technologies for Information Systems, Zurich, Switzerland, pp. 95-109.
Bollen, K. A. (1989): Structural Equations with Latent Variables. New York, NY: John Wiley & Sons.
Corns, A. and A. Schotter (1999): “Can Affirmative Action be Cost Effective? An Experimental Examination of Price-Preference Auctions”, The American Economic Review, 89(1), pp. 291--295.
Davis, F.D. (1989): “Perceived Usefulness, Perceived Ease of Use and User Acceptance of Information Technology”, MIS Quarterly, 13 (3), pp. 319-340.
Davis, F. D., and V. Venkatesh (2004): "Toward Preprototype User Acceptance Testing of New Information Systems: Implications for Software Project Management," IEEE Transactions on Engineering Management, 51, 31-46.
DeLone, W., and E. McLean (1992): "Information Systems Success: The Quest for the Dependent Variable," Information Systems Research, 3, 60-95.
Gefen, D., E. Karahanna, and D. W. Straub (2003): "Trust and Tam in Online Shopping:
An Integrated Model," MIS Quarterly, 27, 51-90.
Kolitz, K. and C. Weinhardt (2006): “MES - Ein Experimentalsystem zur Untersuchung Elektronischer Märkte“, . Mareike Schoop, Christian Huemer, Michael Rebstock, Martin Bichler (editor) Service-Oriented Electronic Commerce - Proceedings zur Konferenz im Rahmen der MultikonferenzWirtschaftsinformatik 2006, Bonn, Gesellschaft für Informatik.
Lattin, J., J. Carrol, and P. Greeen(2003): Analysis Multivariate Data. Duxbury Press.
Legris, P., J. Ingham, and P. Collerette (2003): "Why Do People Use Information Technology?: A Critical Review of the Technology Acceptance Model," Information and Management, 40, 191-204.
Mäkiö, J. and I. Weber (2004): “Component-Based Specification and Composition of Market Structures”, M. Bichler (editor), Coordination and Agent Technology in Value Networks, pp. 127-137, Berlin, GITO.
Malone, T.W., J. Yates and R.I. Benjamin (1987): “Electronic Markets and Electronic Hierarchies: Effects of Information Technology on Market Structure and Corporate Strategies”, Communications of the ACM, 30 (6), pp. 484-497.
McKnight, D.H., V. Choudhury and C. Kacmar (2002): “Developing and Validating trust Measures for e-commerce: An Integrative Typology” Information Systems Research, 13 (3), pp. 334-359.
Neumann, D. (2004): “Market Engineering - A Structured Design Process for Electronic Markets”, PhD thesis, Universität Karlsruhe (TH), Karlsruhe, Germany.
Neumann, D., J. Mäkiö and C. Weinhardt (2005): “CAME – A Toolset for Configuring Electronic Markets”, Proccedings of 13th Europe Conference on Information Systems (ECIS 2005), Regensburg, Germany.
Pavlou, P. A., and D. Gefen (2004): "Building Effective Online Marketplaces with Institution-Based Trust," Information System Research, 15, 37-59.
Reich, B. and I. Ben-Shaul (1998): “A Componentized Architecture for Dynamic Electronic Markets”, SIGMOD Record, 27(4), pp. 40-47.
Roth, A. E. (1999): “Game Theory as a Tool for Market Design”, Technical report, Harvard University, Department of Economics and Graduate School of Business Administration.
Roth, A.E. (2002): “Game Theory, Experimentation, and Computation as Tools for Design Economics. Econometrica, 70, pp. 1341-1378.
Rothkopf, M. H., R. M. Harstad, and Y. Fu (2003): “Is Subsidizing Inefficient Bidders Actually Costly?”, Management Science, 49(1), pp.71-84.
Scott, R. W. (2001): Institutions and Organizations Sage Publications. Thousand Oaks London New Dehli: Sage Publications.
Seddon, P. (1997): "A Respecification and Extension of the Delone and Mclean Model of Is Success," Information System Research, 8, 240-252.
Seddon, P. B., and M. Y. Kiew (1996): "A Partial Test and Development of the Delone and Mclean Model of Is Success," Australian Journal of Information Systems, 4, 90-109.
Smith, V. (1982): "Microeconomic Systems as an Experimental Science", American Economic Review, 72(5), pp.923-955.
Varian, H. R. (2002): “When Economics Shifts from Science to Engineering”, The New York Times, August 29th, 2002.
Venkatesh, V., and F. D. Davis (2000): "A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies," Management Science, 46, 186- 204.
Vickrey, W. (1961): “Counterspeculation, Auctions and Competitive Sealed Tenders”, Journal of Finance, 16, pp. 8-37.
Weinhardt, C., C. Holtmann, and D. Neumann (2003): “Market Engineering”, Wirtschaftsinformatik, 45 (6), pp. 635-640.
Weinhardt, C., C. van Dinther, K. Kolitz, J. Mäkiö, and I. Weber (2005): “meet2trade: A Generic Electronic Trading Platform”, The 4th Workshop on e-Business (WEB 2005), December 10th.
Wixon, B.H. and P.A. Todd (2005): “A Theoretical Integration of User Satisfaction and Technology Acceptance”, Information Systems Research, 16 (1), pp. 85-102.
Wolfstetter, E. (1999): “Topics in Microeconomics: Industrial Organization, Auctions, and Incentives”, Press Syndicate of the University of Cambridge, Cambridge, United Kingdom.
Wurman, P. R., M. P. Wellman, and W. E. Walsh (1998): “The Michigan Internet AuctionBot: A Configurable Auction Server for Human and Software Agents”, K. P.
Sycara and M. Wooldridge (editors), Proceedings of the 2nd International Conference on Autonomous Agents (Agents'98), pp 301-308.
