2.1 Online Marketplace

An online marketplace is an online virtual place where buyers and sellers exchange product information and make transactions with the use of internet technologies, including information systems, payment technologies, social applications, visual simulation and other new digital technologies [19], [33], [51]. By bringing together numerous buyers and sellers and by managing transactions automatically, online marketplaces offer more opportunities for both of the transaction parties because of the network effects [31], [55]. Thus, online marketplaces are in essence as the endorsing institutions or transaction intermediaries between buyers and sellers who might have no interactions before [8], [51]. By offering transaction information and establishing reliable institutional context, online marketplaces facilitate transactions between the unseen sellers and buyers [11], [25]. Today, online marketplace has become the most efficient and important model for e-commerce transactions [33]-[34], and the volume of transactions in online marketplaces is explosively expanding [8], [28]. New marketplaces are emerging at an unprecedented rate by exploiting new digital technologies (e.g., the mobile online marketplaces, the social network based marketplaces, and the apps) or by meeting new market needs (e.g., Uber and Didi for transportation services, Airbnb for accommodation services). With the popularity and success of online marketplaces, the academic interest in this area is continually growing [2], [8], [15], [22], [28], [33]-[34], [42]-[43], [47], [51].

One of the main research interests in this area is to study buyer purchase decision inside of an online marketplace, where buyers are especially vulnerable to fraud and other opportunistic behaviors of sellers [28], [34], [52]. Trust, is then suggested in literature, the most fundamental social mechanism that helps diminish complexity and vulnerability that a buyer feels in an online marketplace [20], [34], and thus, as the key facilitator of online transactions [12], [27], [32]. Therefore, studies of online purchase decision are often centered on the concept of trust, trying to uncover the trust-building mechanisms [8], [43], [51] or identify the boundaries of the effects of trust on transaction activity [17], [22]. For instance, Pavlou and Gefen [51] studied the effects of three IT-enabled institutional mechanisms established by online marketplaces on the institution-based trust. Lu et al. [43] confirmed the positive roles of two set of marketplace-related attributes, i.e., IT-enabled institutional mechanisms and social attributes afforded by social technologies, in engendering trust at the institutional level. Although these studies shed light on trust and purchase decision in online marketplaces, they cannot tell us how online buyers make a choice when facing a variety of different online marketplaces. Solutions to this problem become particularly pertinent considering the rising number of online marketplaces in this digital economy (e.g., more and more marketplace apps in smart cellphones) and the fierce competition among them. This calls special attention to buyer decisions towards marketplaces rather than those inside of a marketplace. These two types of decisions are further explained in next section.

2.2 Purchase Decisions in an Online Marketplace

The decision making process is a little more complex when buying from online marketplaces than from the online storefronts or e-merchants. Unlike in online storefronts where buyers make transactions directly with one identifiable Internet merchants, in online marketplaces buyers have to deal with both the marketplace and the sellers resided in that marketplace. Therefore, there at least two decisions (shown as Figure 1) faced by buyers when trying to make a purchase from online marketplaces: (1) from which marketplace to make a purchase and (2) whether to transact with the sellers in a specific marketplace if the first decision has been made. As stated in last section, while the second type of decision (Decision 2 in Figure 1) has been extensively explored in literature [8], [43], [47], [51], [67], the first type of decision (Decision 1) is rarely examined. However, the intense competitions among online marketplaces make the first type of decision a significant and pertinent problem today. Only if the marketplace sponsors such as Amazon and Alibaba know buyer preferences in choosing online marketplaces can they take the right countermeasures to improve and sustain their marketplaces [16], [63]. This study then tries to fill the aforementioned knowledge gap by focusing on this untapped type decision in literature with the use of conjoint analysis approach.

Next we turn to the discussion of decision attributes for conjoint analysis method.

Figure 1:  Two types of purchase decisions in online marketplaces 

3.1 Logistics Service Quality Theory and Related Attribute

Logistics service quality theory roots in service quality research, which has been a priority theme in both marketing and logistics research since the mid of 1980s [54], [60]. Logistics service is viewed an indispensable part of e-commerce. For a process point of view, e-commerce service consists of two dimensions: e-service and logistics service [41]. E-service facilitates online transactions by offering services to buyers with the use internet technologies, such as, searching for product information, placing orders, paying and reviewing. Once online transactions have been made, logistics service is immediately needed to deliver products to buyers. Logistics service can be completed via either the self-managed logistics system or third-party logistics (3PL) system [59]. Therefore, the quality of logistics service offered by online marketplaces is an important part of the online shopping experiences perceived by customers. Logistics service quality (LSQ) has been recognized as an important attribute for enhancing customer satisfaction [49] and customer loyalty [6], [41] in e-commerce. Recent industrial reports showed that logistics service related factors, including delivery speed, reputation and delivery cost, are the major attributes considered by buyers when purchasing online [9]. In general, online marketplace sponsors include LSQ as one of the key indicators for post-purchase evaluation.

Current research generally treats LSQ as a multi-dimensional construct [4], [48]. For instance, Mentzer et al. [48] developed a scale of LSQ that has 9 dimensions including information quality and order quality. Kang and Kim [30] proposed a model of LSQ that covers three dimensions: service outcome quality, service delivery quality and service capability quality. For the purpose of simplification, this study takes the overall logistics service quality (LSQ) as one decision attribute for conjoint analysis. Therefore, one attribute springs out relying from LSQT, namely, LSQ, defined as “the extent to which the logistics systems offered by an online marketplace satisfy the demand of buyers who receive logistics service” [49].

3.2 Technology Adoption Model and Related Attribute

E-services in online marketplace are delivered via the IT infrastructures and new digital technologies such as visual simulation and streaming technologies [19] provided by marketplace sponsors. These digital technologies are integrated into and offered via a website or app. Therefore, platform characteristics [67], such as usability, ease of use, interactivity and social presence, have been suggested as key determinants of online purchase decision, satisfaction, and repurchase intention in e-commerce [1], [10], [26], [33], [42], [61].There are quite a few different theoretical perspectives that have contributed a long list of determinants related to website characteristics. Among them, technology acceptance model offers us a more parsimonious way to understand information technology adoption [13]-[14], [26], [67]. An e-commerce website, app or platform is an integration of information and digital technologies in essence, and thereby, online purchase decision should be explained in part by the technology acceptance model [20]. Based upon TAM, two technological attributes of online marketplace, namely perceived usefulness (PU) and perceived ease of use (PEOU), are generally recognized to have influences on buyer decision of online transactions [20], [26], [37], [67].

Thus, according to TAM, there are two determinants that we hypothesize that online buyers assess in determining which marketplace for online transactions:

Usefulness: the utility offered by the information and digital technologies in an online marketplace.

Ease of use (EOU): the cognitive effort needed to learn and to utilize the information and digital technologies in an online marketplace.

3.3 Institutional Theory and Related Attribute

E-commerce activities in online marketplaces are often perceived to be less legally protected and to be at greater risk because information about sellers and products cannot be fully assessed [34], [38], [51]. These negative perceptions such as risk and uncertainty become the major barriers to prevent buyers from engaging online transactions in online marketplaces [34], [52]. In this situation, online marketplaces, as the intermediary between unseen sellers and buyers, have to establish institutional mechanisms to foster buyers’ trust and reduce their risk to enhance transaction activity [8], [22], [35]. These IT-enabled institutional mechanisms (e.g., feedback mechanisms, third-party escrow services, and credit card guarantees) provide a certain degree of guarantee and protection for transactions by restricting the ability of a seller to engage in opportunistic behavior [7]-[8], [43], [51]. This institutional perspective [70] has been widely applied to online marketplace research [22], [43], [51]. If these institutional mechanisms are felt to be available and effective in facilitating and securing online transactions [8], [35], the overall transaction environment in online marketplaces will be perceived to be appropriate, well ordered, and favorable for doing personal business [43].

Therefore, prior work generally examined the impacts of the effectiveness of these institutional structures on the institution-based trust and purchase decision [17], [43], [51]. Drawing from institutional mechanism theory, there are four determinants that we hypothesize buyers assess in determining which marketplace for online transactions:

Effectiveness of feedback mechanisms (EFM): the extent to which the feedback mechanism in an online marketplace is able to provide accurate and reliable information about the sellers. Feedback mechanisms are buyer-driven reputation mechanisms that accumulate and disseminate information and electronic word-of-mouth (eWOM) (e.g., reviews, opinions, ratings, and recommendations) about sellers’ past trading behavior and performance [7], [22], [35]. They allow buyers to assess the entire community of sellers in a marketplace by providing an overview of the marketplace and its sellers’ past behavior. Thus, effective feedback mechanisms should therefore act as an informal buyer-driven certification system for sellers [7], [35], [51].

Effectiveness of escrow services (EES): the extent to which the escrow services are able to guarantee that their transactions in a marketplace will be fulfilled in accordance with their expectations [22], [43]. Escrow services is the most adopted e-commerce payment mechanism in which a trusted third party authorizes payment only after the buyer has received and approved the purchased products. They protect buyers from the threats that arise from the opportunistic behaviors of sellers (e.g., offering false product information, fraud and reducing the promised quality of product offerings) by adding a level of control to the transaction [52].

Effectiveness of information security (EIS): the extent to which the security mechanisms in an online marketplace are able to protect his/her private information from risk of improper or illegal use [43]. Security concerns are one of the major barriers that inhibit online purchase behaviors [38], [65]. Online marketplace sponsors then must invest heavily to reinforce the security mechanisms and technologies to protect buyer private information and privacy.

Effectiveness of dispute resolution (EDR): the extent to which online dispute resolution services are able to redress buyers’ disputes or complaints with sellers in accordance with their expectations [43]. Disputes about online transactions are growing explosively. These disputes must be well resolved so that online buyers and sellers can have full confidence in e-commerce [18], [43], [64]. Online Dispute resolution (ODR), the dispute resolution method supported by new IT and digital technologies, are thought to be more efficient, more cost effective and more flexible way to address complaints against sellers and protect the benefits of buyers after the transaction have been made.

So far, we have identified seven attributes relevant to buyer decision towards online marketplace (Decision 1) by relying upon prior work. These seven attributes are summarized in Table 1. In this study, we are interested in understanding the relative weight that buyers give to each attribute in making a decision on which marketplace to purchase from. We believe the conjoint methodology is the corresponding approach to achieve this objective.

Table 1:  Attributes for purchase decision towards marketplaces 

4.1 Instrument Development

To make sure that the attributes and levels reflected a real-life situation, we employed a pretest of our instrument with five academics in e-commerce field and ten buyers who were experienced in purchasing from the real online marketplaces including Taobao and JD in China. Based on their feedbacks, minor revisions were made to the wording and the levels of these attributes, which are finalized in the form above (Table 1). The pretest allowed us to ensure that the instrument contained no ambiguous attributes and the task was appropriate for online buyers. Further, these two steps also ensured the face and content validity of the profiles of online marketplace [58].

Next, we used the orthogonal factorial design to reduce the number of profiles used in the conjoint analysis [23]. The data quality would severely degrade if each respondent is required to assess every possible combination of the attributes and their levels. Recent research indicated that limiting the number of tasks to 20 does not cause a degradation of data [45]. Utilizing an orthogonal factorial algorithm, we determined that we needed a set of 18 profiles for each respondent to evaluate. After reading each profile, the subjects rated the likelihood of engaging online transactions with these marketplaces (the likelihood of decision was our dependent variable). The likelihood was rated from 0 to 100, where 0 represent that the subject would not to choose the option and 100 represent that the subject would definitely consider this option. An example of the profiles is illustrated in Appendix A.

4.2 Sample

Master of Business Administration (MBA) and senior undergraduate students from a Chinese university were recruited for this study. Subjects are required to have past purchase experiences from online marketplaces and are able to make evaluation based upon the combination of the attributes. A research invitation that contained an introductive message of this study and a link for a web survey was sent via Wechat, the most used social network application in China. Each subject was offered a gift and the chance to get several raffle prizes to stimulate more valid responses. The web survey opened for two weeks in August, 2016. In the end, 173 valid responses were obtained after removing 81 invalid responses, resulting in a valid rate of 68%.

For the conjoint approach, a minimum sample size of 75 is usually recommended. However, there is no clear required minimum sample size [45]. The profile of the respondents is shown in Table 2. All subjects have involved online purchase from online marketplaces before and 74% of them have at least two years of online purchase experience. Therefore, the subjects are experienced and qualified to rate after assessing the profiles.

As the survey was conducted, most of the respondents (98.8%) were aged between 20 and 35. Even though the subjects are mainly students, they without a doubt are the major group purchase consumers in China [9]. The use of students sample for e-commerce research was also advocated in previous studies [21], [66] as they have the opportunity to use the Internet for communication and commercial transactions. Therefore, the data could represent the major part of Internet group purchase customers in China.

Table 2: Sample statistics (n = 173) 

6.1 Findings

The results reveal some fundamental notions about how online buyers make marketplace choices. First, in general, the most relative important attribute considered by online buyers when making a marketplace choice is the effectiveness of dispute resolution. Dispute resolution mechanisms, as a kind of after-sale protection mechanisms, are the last but probably the most important procedure to protect the interests of buyers from the malicious behaviors of sellers. When the payment has been done, if buyers find any hidden problems related to the transactions (e.g., false description, counterfeit and shoddy products, post-contractual shirking), they tend to rely on online marketplaces (i.e. the middle man or the intermediary) to resolve their disputes and complaints with online sellers in an effectively, fair and impartial manner. However, if online marketplaces, as the endorsed transactional institution, fail to protect the legitimate rights and interests of buyers at this step, trust in these marketplaces will be deteriorated. Once online buyers don’t trust an online marketplace any more, they will give up online purchase or leave for others marketplaces [43]. Thus, our results disclose that online buyers are concerned a little more on the after-sale protection mechanism (i.e. dispute resolution mechanisms) than the pre-sale institutional mechanisms, including the feedback mechanisms, escrow service and information security service. Though these pre-sale institutional mechanisms have been well explored in prior research [22], [51], the post-sale institutional mechanisms have not drawn enough academic interests. This study then suggests the post-sale institutional mechanisms deserve more attention in future.

Second, only one attribute was considered minimally important for online marketplace choice: the logistics service quality (7.9 percent likelihood). Existing research generally views logistics service as a vital part of the e-commerce service and as one critical factor for e-commerce success [40], [41]. However, our results indicate that the logistics service quality is not an important attribute considered by buyers for online marketplace choice. Two plausible reasons are offered as below to clarify the discrepancy. First, unlike traditional survey studies in which respondents rate an item without relating to others, subjects in conjoint approach have to make trade-offs among the attributes and rate based upon the whole profile. This allows conjoint approach studies offer a level of insight not normally available from traditional questionnaire studies [58]. Second, choosing an online marketplace is only the first step to complete an online transaction. Though the logistics service quality is an important part of e-service [41], it might not be an important attribute considered by buyers when making a decision on marketplace choices. In the minds of online buyers, logistics services might be more associated with the logistics service provider than the marketplace, and thereby, the quality of logistics services should not be an important attribute of the marketplace.

Third, our results reveal that when considering marketplace choice, male buyers weigh the attributes significantly different from the female buyers. While female buyers give nearly equal relative weight to each attribute, male buyers weigh some attributes more importantly than others. Male buyers are more concerned with the dispute resolution services and the information security mechanisms, but attribute minimally importance to the logistics service quality. Our results suggest that male buyers are more risk-taking, and thereby, they would like to have more confidence in the information provided by the third parties, are more tolerant to deficiencies of the pre-sale mechanisms (e.g., escrow services, feedback mechanisms and logistics services), but are concerned with the post-sale mechanisms (i.e., dispute resolution mechanisms). On the contrary, female buyers are more inclined to assess every aspect of the information, indicating that they are more risk-averse when making online transactions. These findings are in accordance with what were observed in existing work [29], [36], [56], which suggested that male buyers have more positive perceptions towards e-commerce and are more tolerant to the perceived risk and uncertainty, whereas female buyers have more negative feelings about e-commerce and are conservative to make transactions.

Finally, three interesting patterns emerge when the results are viewed on a theory-by-theory basis. First, LSQT-based attribute is not considered especially relevant for marketplace choice. One plausible interpretation may be that with the maturation of the e-commerce, online buyers are capable of distinguishing the logistics service from the service and then can ascribe the quality of logistics service to the logistics provider rather than the marketplace sponsor. Second, the TAM-based attributes and the institutional mechanism theory based attributes are considered more relevant for marketplace choice, suggesting that both technology adoption model [13]-[14], [26], [67] and institutional mechanism theory [7]-[8], [43], [51] are two appropriate theoretical lenses for understanding the decision towards the marketplace. Third, no one theoretical lens dominates the marketplace choice decision when considering all three theories in together, suggesting that the phenomenon of marketplace choice should be best viewed from a contingency perspective. Rather, it is best explained by utilizing certain attributes from several different theories for different genders. For male buyers, the appropriateness of marketplace choice mainly varies based upon the effectiveness of dispute resolution mechanisms, the usefulness and the effectiveness of information security mechanisms. For male buyers, the appropriateness mainly depends upon the effectiveness of dispute resolution mechanisms, the usefulness and the effectiveness of feedbacks mechanisms.

6.2 Implications for Research

This study has a number of implications for future research. First, we have applied the conjoint approach to study buyer decision in e-commerce research. Despite the extensive applications of this method in marketing research, conjoint analysis has been rarely used in the field of information systems [58] and e-commerce. Followed Schwarz et al. [58], we also believe that the conjoint methodology is both practically relevant (as these results are easily digestible to practitioner audiences) and academically rigorous (as the attributes are theoretically based and have implications for how we theorize about decisions on marketplace choice). Thus, we encourage further application of the methodology in understanding many aspects decisions in e-commerce, online marketplace and online community research.

Next, theoretically, the results suggest a number of implications about the relative relevance of theories in understanding decisions of online buyers. The strongest theories are the institutional mechanism theory [7]-[8], [43], [51] and the technology adoption model [13]-[14], [26], [67] while the weakest is the logistics service quality theory [54], [60]. This suggests that online buyers are more concerned with the institutional mechanisms established afforded by IT and digital technologies to safeguard online transactions [8], [43] and the utility of technologies offered by the marketplace [67], but less concerned with the logistics service quality. From another perspective, this reflects the maturity of e-commerce at present. Online buyers are experienced and have the abilities to distinguish the responsibilities of each party (e.g., marketplace sponsors, logistics service providers, online sellers) in e-commerce.

Third, in this work, we offer a more comprehensive model of purchase decision in online marketplace by dividing the decision process into two steps: decision over the marketplace choice (decision 1), and purchase decision in a selected marketplace (decision 2). Then we focus on the first type of decision (marketplace choice) by outlining 7 attributes derived from three theories that we hypothesized to influence a buyer’s decision at the marketplace level. In doing this, we not only have examined the impacts of the marketplace-related attributes on decisions of online buyers, but also have contributed a few important theory-based factors that can be further explored in future research.

Finally, we further explore the impacts of gender on the decision of buyers when considering marketplace choice. The results reveal a significant structural difference of the relative importance of attributes between male and female buyers: while male buyers think that some attributes are more important than others, female buyers consider that each attribute is equally important. Thus, this study confirms the impacts of gender on buyers’ decisions suggested in prior research [36], [56]. With the new development of e-commerce towards social commerce and mobile commerce, the role of gender on decision of online buyers in these new areas should draw more research interests. Thus, we here call more attention the effects of gender in in future e-commerce studies.

6.3 Implications for Practice

The results also offer a few implications for the marketplace sponsors. First, institutional mechanisms, as the trust-building tools, have got a tremendous amount of investments from the marketplace sponsors [22]. However, our results suggest that the marketplace sponsors probably ought to set different investing priorities for different institutional mechanisms. In contrast to the pre-sale mechanisms such as the feedback mechanisms and the escrow services, the post-sale mechanisms (i.e. the dispute resolution mechanisms) are relatively more important for marketplace choice of online buyers. Thus, to gain a competitive advantage, online marketplace sponsors should put more investment in enhancing the post-sale mechanisms like the dispute resolution rather than equally investing on each institutional mechanism.

Second, our results disclose that the logistics service quality is the least important attribute for online byers to marketplace choice. However, this does not simply mean that logistics services are not important for the overall of the online purchasing experience of buyers. It might be because of that online buyers have not directly related the logistics services with the marketplace. This means that online marketplace sponsors should be very careful in interpreting and applying this finding to their logistics strategy. If they plan to build the self-owned logistics systems, the marketplace sponsor should by various means to help online buyers to relate the logistics services with the marketplace and take the logistics services as one key attribute of marketplace for commercial propaganda. And if they plan to rely on the third parity logistics systems, the marketplace sponsor should let the buyers be clearly aware of the situation, as well as enhance the controls and monitoring over the behaviors of the logistics service providers.

Finally, the marketplace sponsors should pay attention to the differences between male and female buyers, and should adjust their strategies in institutional mechanism investment. For instance, for the products (or subpages) that target on male buyers, the marketplace sponsors should pay more attention to the institutional mechanisms like the dispute resolution mechanisms and information security mechanisms, whereas for the products that target on female buyers (or subpages) that target on female buyers, they should focus on providing a full set of institutional mechanisms.

6.4 Limitations

Like any empirical research, this work has limitations. First, we selected 7 theoretically derived attributes that were hypothesized to impact buyer’ marketplace choices. Although these attributes are grounded from three theories for marketplace research, we recognize that other latent attributes influence the marketplace choice and could indicate different weightings of these (and other) attributes. Second, although we have considered a set of three theories that are relevant to marketplace choices, we cannot include all other important theories in this single study, e.g., social presence theory [21], [42]-[43] and social capital theory [3], [5]. We encourage future research to study marketplace choice by exploring other important theoretical perspectives. Third, MBA and undergraduate student samples were used in this study. While the use of student samples for e-commerce research was advocated in previous studies [21], [66], the sample may not represent actual buyers of online marketplaces. Students are more educated and more familiar with the use of new technologies like e-commerce. Therefore, their perceptions on online marketplace might be different from those of which are less educated and less familiar with new technologies, such as aged people and rural residents. Fourth, the attributes are measured as perceptions in this study. Although it is each buyer’s own perception of marketplace that determines his/her transaction decision making, it should be noted that the perception is a fuzzy concept and some perception is more complex and difficult to be measured than others. For example, usefulness of marketplace in this study could be a very fuzzy concept that the respondents have troubles to measure since online marketplace is inherently a complex tool. Thus, the findings of this research might be confounded. This limitation might be overcome by dividing usefulness into several more concrete sub-attributes such as facilitating conditions, hedonic motivation [67] or availability. Finally, caution should be taken when making generalizations from a single study, particularly when this study was conducted mainly in Chinese culture background. The effects of the marketplace-related attributes may be different in another broader institutional, societal, and cultural environment. Therefore, the results might not be simply generalized to other cultural contexts.