Вы находитесь на странице: 1из 10

An Empirical Study on Quality Uncertainty of Products and Social Commerce

Kyunghee Lee
KAIST Business School 85 HQEGIRO DONGDAEMOON-GU SEOUL 130-722 KOREA +82-10-9130-9327

Byungtae Lee
KAIST Business School 85 HQEGIRO DONGDAEMOON-GU SEOUL 130-722 KOREA +82-2-958-3629

nickend@business.kaist.ac.kr

btlee@business.kaist.ac.kr

ABSTRACT
With the advance of Social Network Service (SNS) like Facebook, Social Commerce (SC) such as Groupon now prospers, which provides daily deals at a highly discounted price by gathering buying power of consumers through SNS. From the perspective of quality-uncertainty, it is unusual to sell experience and credence goods/services on the internet as Groupon does. In traditional Ecommerce (EC) purchasing decisions rely on information provided after the actual use of products by other consumers, while in Groupon it heavily depends on opinion even before purchasing. For example, traditional sites use a third-party recommendation including feedback mechanism, while Groupon encourages consumers to post and share their preference on goods/services over SNS. Given this difference, focusing on the effect of SNS, we collect and analyze changes of sales for deals Groupon provided, using an econometric model that reflects our understanding of consumer behavior in the presence of different degrees of quality-uncertainty. The information from SNS is captured by using a function called Facebook Like that is a recommendation system in which suggestions are brought by ones friends, and it is a module that can be installed in any website. In this study, we demonstrate that the information from SNS positively affects sales for deals, which implies that SNS provides recommendation and encourages consumers to purchase by reducing encountered uncertainty. In addition, we also find that the effect of SNS is enlarged as the extent of the qualityuncertainty increases. This result means that under the presence of high degree of uncertainty, the information from SNS gives consumers a stronger belief in quality than information from a third-party. Besides, as many other studies proved, we also confirm that the internet turns experience goods into search goods by substituting in-store visits with virtual encounters.

1. INTRODUCTION
Recently, Groupon turn down Googles $6 Billion offer for acquisition1. Groupon is a deal-of-the-day website that is localized to major geographic markets worldwide. Launched in November 2008, Groupon serves more than 150 markets in North America and 100 markets in Europe, Asia and South America and has a cumulative 35 million registered users 2 . Groupon is now considering an initial public offering that would value the onlinecoupon company at as much as $25 billion3. Following Facebook and Twitter, Groupon is thought to be one of the most successful internet businesses nowadays. This is testimony to the emerging importance of Social Commerce (SC). Combined with a group-buying model, Groupon provides a dealof-the day that usually offers various kinds of local services, such as restaurants, spas and salons, at a highly discounted price, above 50%. There is a tipping point for every daily deal so that deals would be canceled if the point of sales is not reached. This reduces risk for retailers, who can treat the coupons as quantity discounts as well as sales promotion tools. Thus, they are likely to offer attractive deals to customers and Groupon provides an attractive marketing channel where an enormous number of customers look for good deals. Group-buying itself is not a new concept. In 1999, group-buying business models on the internet had been flourishing for a time. Bringing volume discounts to the internet, Dotcom companies like MobShop.com combined this with a dynamic pricing mechanism. They offered various sorts of products, ranging from dishes to cameras, with a pricing mechanism attracting more people to get high discounts. Of course, many researchers investigated the group-buying model with different perspectives. Among them, Kauffman discovered the effect of dynamic pricing mechanisms on features and timing of consumers purchase decision [28, 29]. Anand investigated the difference between posted-pricing and group-buying mechanisms and Chen developed an analytical model to point out sellers optimal price curve of the group buying mechanism [3]. However, success was not long-lasting and in less than 3 years most group-buying sites closed down. Kauffman pointed out several reasons that this model shrank, such

Keywords
Quality uncertainty of products, Social network, Social commerce, Groupon

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Conference10, Month 12, 2010, City, State, Country. Copyright 2010 ACM 1-58113-000-0/00/0010$10.00.

http://www.businessinsider.com/why-groupon-said-no-to-google2010-12
2 3

http://en.wikipedia.org/wiki/Groupon

http://www.businessinsider.com/why-groupon-is-worth-15-25billion-its-all-about-the-people-2011-3

as failing to achieve the critical mass, low barriers, and pricesensitive consumers [28, 29]. Considering the failure of the early group-buying model, we posit that one of differences between the old versions and Groupon is the variety of products Groupon provides. While the old version of group-buying sites sold products whose quality can be easily judged, Groupon sells goods and services for which quality and performance are difficult to evaluate before purchase. For instance, restaurants, hair-clinics, spas, and massage services are sold in Groupon rather than the computers and audio equipment that used to be sold in MobShop.com. This case might be treated as questionable because in e-commerce online retailers have always been concerned about how to reduce consumers perceived uncertainty of product quality. Due to the limitation of direct experience of goods/services, online consumers only have a limited choice of products whose prerevealed quality is hardly changed and clearly captured without using it or having direct experience. According to the search, experience, and credence products classification [30, 38, 39], these are distinguished as search products. If it is difficult to discern the quality of a product before using it, the product is classified as an experience product, whereas if you still have difficulty in evaluating the quality even after using it, it is classified as a credence product. In other words, search, experience and credence products are divided by the extent to which consumers feel difficulty in obtaining quality information before and after use. To expand the scope of products sold on the internet, recommendation systems including feedback and reviews were introduced for the purpose of dimming distinctions between experience and search products [25, 30] by providing information about actual experience of use. It enables online shoppers to gather more information on products and encourages them to purchase as they shop in the real world. Even though these mechanisms reveal some limitations, products sold on the internet are still restricted to search and a few experience type products [46]. Considering the issue about uncertainty, Jain suggested an interesting solution, source credibility, to reduce it [27]. In his paper, if a consumer faces uncertainty which experience goods contain, he/she more relies on where the information about the product comes from, not just on the information itself. As a method of obtaining credibility, social networks could be involved because people tend to easily believe in information that comes from a network to which he/she belong [43]. From this point of view, we focus on a fact that Groupon is called as SC. The reason why Groupon is regarded as SC is that Groupon substantially uses functions of SNS, especially Facebook. Facebook is a social networking service launched in Feb. 2004 and now has more than 600 million active users around the world4. This online-social network that consists of known friends around individuals allows him/her to create a personal profile, add other users as friends, and exchange messages, including automatic notifications when they update their profile. Through those activities, they can easily exchange opinions with their friends about topics that they are concerned with, and it may affect their way of thinking, which is like an off-line social network [17].

Therefore, one might gather that information generated and shared through Facebook is more likely to be accepted by a consumers friends and that this information could affect their purchasing decisions. As quality uncertainty of products can be reduced by getting information that comes from credible source [27], this would further influence purchases of products when information about the quality of the product cannot be acquired on the internet. In other words, the fact that certain information comes from SNS in which a network consists of his/her friends could compensate for consumers uncertainty about quality of products For those reasons, this paper explores the influence of information from SNS on consumers perceived uncertainty when they buy products whose quality information is hard to obtain. Using Groupon sales data, the effect of information on the sales of each deal is examined and in order to measure the degree of qualityuncertainty good/service contains, a classical scheme for classification along with quality uncertainty, called SEC classification (search, experience, and credence), is brought. The rest of this paper is organized as follows. The literature review is given to aid understanding about the theoretical foundation of this research. Next, an explanation of the empirical analysis follows, including the generation of hypotheses, regression models, method for categorization, and descriptive statistics of data. Then, we provide results from the analysis and present the discussion and implications afterwards.

2. Literature Review 2.1 Group-Buying


In 1999, the group-buying model on the internet was introduced by several pioneers, including Mobshop.com and Letsbuyit.com. They sold various kinds of products including cameras and computers with dynamic pricing mechanisms in which consumers could decrease the price of a product as others purchased more units. In each sale, there were several price-drop points and if the total number of products sold exceeded the points, then price would drop to the next level. As this model had diverse points of interests, such as pricing mechanisms, many researchers focused on this topic. With rigorous analysis of the pricing model, Anand investigated the difference between the posted-pricing model and the group-buying model. He found that the monopolists optimal group-buying schedule under varying conditions of heterogeneity in the demand regimes and compared its profits with those within the posted-pricing model [3]. Chen compared a fixed pricing mechanism with the group-buying auction (GBA) and found the sellers optimal price curve of the GBA [7]. In addition, it was also verified that Groupon-buying is likely to be more effective in a situation where there is larger low-valuation demand [8]. Kauffman, on the other hand, focused on consumers purchase behavior in dynamic pricing setting, or a group-buying model. He indicated a price-drop in a group-buying setting as uncertainty of future benefits, which results in consumers buying more at points before the price-drop occurs [28]. Even though this pricing model was considerably innovative on the internet and acceptable by both consumers and sellers, it also had some weaknesses mentioned by Kauffman [29], and it could not persist due to consumers indifference and so its business declined.

2.2 Search, Experience, Credence Products


4

http://en.wikipedia.org/wiki/Facebook

In the discipline of marketing, the quality of services has been well researched from various perspectives. Among them,

Parasuraman suggested one quite-well accepted conceptual model of service quality from intensive interviews: quality that a consumer perceives in a service is a function of the magnitude and direction of the gap between expected service and perceived service [42]. Along with other factors that cause changes in quality, quality varies to the extent to which the consumer has difficulty with judging goods/services performance, or expected quality, before or after purchase [38, 39]. By the Nelsons classification of search and experience products, each good/service can be classified as a search and experience good/service. Search attributes are a property which consumer can evaluate before use or purchase and experience attributes can only be judged after use or purchase. In addition, further research added to Nelsons classification a third category, credence properties, in which quality could not be determined even after use [11, 30]. For instance, only a few consumers have medical information sufficient to evaluate whether surgery, a credence-dominant service, is performed properly. Considering the lack of proper information about quality, as uncertainty is exposed to consumers, consumers may reduce the intention to buy as well as lessen actual purchases [19]. As consumers have more uncertainty about quality, they tend to rely on other information sources available, such as brand equity [32], in order to reduce the uncertainty. For example the demand for art performances is influenced by other sources of information such as reviews, reputation of the author, producer and cast which are the essential parameters by which to judge the quality of service [1]. Jain and Posavac also found that a source high in credibility can be employed to make experience claims more persuasive [27]. This result indicated that when consumers face a high level of uncertainty about quality, they heavily rely on the source of information, not just information itself.

works indicate that the uncertainty is an essential factor affecting consumers behavior on the internet. Particularly, Levin examined consumers preference of online/offline shopping along with search and experience goods/services [33]. He concluded that goods including search-dominant attributes are more appropriate when sold online while goods that have experience-dominant attributes are better sold offline. In addition, Hsieh investigated the effects of various relational bonds on consumers commitment across search-experience-credence goods/services on internet [24]. From the seller-side of view, Animesh indicated that bidding strategies of online advertisers differ in search, experience, and credence goods with empirical evidence [4]. Due to fact that experience or credence goods have a high degree of qualityuncertainty, consumers find buying these kinds of products inappropriate on the internet. To counter this problem of uncertainty, online retailers who want to sell various kinds of goods/services through online channels have tried to provide additional information related with reviews/experiences about products sold, which are called recommendation systems. In recommendation systems, online user reviews are an important source of information to consumers, substituting other forms of offline word-of-mouth communication about product quality [9]. Since information asymmetry definitely exists in the online environment, consumers need that kind of information to mitigate this problem. One point of thought is that a recommendation system is a tool for building the reputation by the seller in order to reduce consumers perceived uncertainty about products [16, 22]. As uncertainty increases, consumers tend to look for information to verify. In that sense, recommendation systems including feedback, rating, and WOM affect consumers intention to buy and actual sales [9, 12, 15], although there is still a conflict over whether the feedback system influences product sales positively or negatively [47]. It was also found that the presence of product reviews from other consumers and multimedia that enable consumers to interact with products before purchase have a greater effect on consumer search and purchase behavior for experience goods than for search goods [25]. These outcomes were predicted by Mcknight [37] in his paper indicating that the interaction provides the customer with evidence that the vendor has various positive attributes, thereby enhancing trusting beliefs. The systems enable online shoppers to gather more information on products and encourage them to purchase as they shop in the real world. Despite this, these mechanisms reveal some limitations on dimming the boundary of products sold on the internet, thus the types of products sold online are still restricted to search and a few experience type products [46]. It reveals that this additional information would be not enough to compensate for consumers uncertainty on the internet.

2.3 Trust in E-commerce and Recommendation Systems


Before examining how the quality uncertainty affects consumers behavior in EC, it is worth checking the fundamental theory of what trust is and how it is constructed, which is different across diverse disciplines including psychology, sociology, political science, and economics [34]. Trust is defined as the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the agility to monitor or control that other party [36]. Simply, trust is the willingness of an individual to behave in a manner that assumes another party will behave in accordance with expectations in a risky situation [13]. In this point of view, as trust declines, people are increasingly unwilling to take risks and demand greater protections against the probability of betrayal [44]. This situation also increases transaction costs because individuals have to engage in self-protective actions and prepare for the possibility of others opportunistic behavior [31]. Because the parties to a transaction are not in the same place, trust may be even more important in the virtual world, or online, than it is in the real world [23, 40, 44]. As Nohria and Eccles [41] pointed out, there is the lack of the entire human bandwidth which includes sight, hearing, smell, taste and touch, thus development of trust is far more difficult on the internet than the real world, although it is more important online. Combining the qualityuncertainty of different product-categories here, these previous

2.4 Social Network and SNS


A social networking service is an online service, platform, or site that focuses on building and reflecting of social networks or social relations among people5. Imitating off-line social network, SNS provides cheap and efficient ways to manage these relationships [14]. Facebook, especially, overlaps both off-line and on-line social networks [14] so it mitigates the distinction of the two networks. For this reason, it is assumable that Facebook reflects

http://en.wikipedia.org/wiki/Social_networking_service

characteristics of real world relations among people and obtains several virtues of them. When discussing how people obtaining benefits from networks, social capital is often used to refer to social networking. Social capital is regarded as the cumulative resources through the social network that consists of relationship of people [10]. Bourdieu and Wacquant [6] define social capital as the sum of the resources, actual or virtual, that accrue to an individual or a group by virtue of possessing a durable network of more or less institutionalized relationships of mutual acquaintance and recognition. While many studies focus on the link between social capital and positive social outcomes, Paxton investigated that social capital leads a person to rely on resources from other members of the networks to which one belongs [43]. These resources include useful information or personal relationships. If we expect that Facebook also has the above characteristics like most social networks, people would tend to rely on resources from networks to which they belong in Facebook. There is some evidence of a relationship between the use of Facebook and individuals social capital [17]. The fact that a group in Facebook consists of known people also supports that events happen in Facebook as it would happen in off-line social network. Based on this prior research and our guesses, it is assumed that in a certain situation the network in SNS such as Facebook would play an equivalent role as offline social networks. Then it could be expected that users of Facebook also draw on resources from the networks to which they belong. This feature may apply to a situation in which consumers face quality-uncertainty of products, and as additional information it may allow consumers to reduce uncertainty. We focus on this and assume that information from social network and SNS could generate a credible signal which can be used by consumers to compensate the exposed uncertainty.

which the Like button is attached. In other words, Facebook Like is an aggregated preference of users collected through SNS on topics or items. It is used not only within Facebook, but also distributed and posited globally no matter which platform or website it is installed. If a user likes items offered on a website, he/she just clicks the Like button as in Figure 1. Then, through the linked Facebook account, information of this item is posted in the user Wall, or board, with a remark of 00 like a link as long as it increases the total number of the Like as one unit. This action causes two different consequences, increasing the number of Likes, and posting it in the SNS page, so the results also should be interpreted differently. First, the increasing number of Likes are similar to a rating such as in recommendation system. A large number predicts that many users are interested in this topic or item, thus one can assume that if the number of Facebook Like is high, then he/she regards it as a sign of quality, or preference as a rating system does. Second, when you click the Like button, all friends who have connection with you in Facebook can see your posting about the item as in Figure 2. Then your friends can make their own opinions about the posting, or further spread this information by just clicking the Like button again, which means that this information you generate is distributed through your social network consisting of friends who have a relationship with you.

2.5 Facebook Like


In this paper, we use an installed-function called Facebook Like in order to capture the impact of information in SNS. Facebook Like is well defined in the Facebook developers page 6 : Facebook Like is currently designed for Web pages representing profiles of real-world things things like movies, sports teams, celebrities, and restaurants. By including Open Graph tags on ones web page, one can make ones webpage similar to a Facebook Page. Figure2. Facebook page after clicking Like button.

3. Empirical Analysis 3.1 Hypotheses


Building trust has been one of the most important features in Ecommerce because there is significant information asymmetry about product quality between sellers and buyers online [18, 23, 33, 37, 44]. Although there is some research indicating that various channels and information sources which the internet provides makes searching for information easier[11, 25, 30], quality-uncertainty still exists and it prevents consumers from purchasing products whose information about the quality is difficultly obtained in an online-shopping environment [33]. Thus, practically and academically, researchers have been working on resolving the unbalanced conditions using recommendation systems, such as feedback and rating systems [9, 12, 45]. As mentioned in the literature reviews, these systems influence consumers perceived usefulness, intention to buy and also the actual sales of online retailers [12, 15]. Since Facebook Like also has a function of rating products, it is reasonable to expect that Facebook Like, which is pre-purchased and aggregates preference of users collected through SNS on products, will also positively influence product sales of daily deals in Groupon as through other recommendation systems. H1: Users preference for products collected through SNS has a positive effect on sales

Figure1. Fackbook Like in Groupon page. Specifically, Facebook Like is a number that is constructed by the participation of users who are interested in topics or items to
6

http://developers.facebook.com/docs/reference/plugins/like/

Another perspective of understanding the influence of recommendation systems is that the impact can differ across various product categories. Especially since feedback systems exist to reducing quality-uncertainty of products, classification for search, experience, and credence goods/services are interesting topics in this area [2, 25, 46]. For example, consumers have more difficulty obtaining information about the quality of experienceproducts before use than search-products, and are more affected by the existence of reviews from other consumers/multimedia that enable consumers to interact with products [25]. Even though there have been many studies considering the effect on search and experience goods, however, just a small amount of research on the influence on credence goods and services was discussed on the internet. It might be caused by the fact that many sellers online do not attempt to sell credence goods or services because they contain too much uncertainty about the quality for consumers to purchase. On the other hand, creating and sharing information online now involves a huge variation of emerging SNS. A social network itself is not a new concept. Instead, it has always existed around people, and there are numerous studies about the importance of social networks in terms of trust and credibility. For instance, a social network builds trust within the network, and people tend to easily believe in information that comes from network he/she belongs to [43]. In this point of view, it is accepted that SNS, especially Facebook, works like an offline social network. Facebook is now regarded as a place for creating and sharing information, which leads users to get closer as they do in offline activities. Since individuals networks in SNS consist of friends they already know, information which is generated and shared through Facebook is likely to be thought of as more friendly, familiar, and credible by users, compared with information from other sources such as an internet portal or random online boards [17]. Specifically, Liu investigated whether social network information can be used to increase recommendation effectiveness by incorporating collaborative filtering [35]. As Grewal found, the influence of price on consumers perceptions of performance risk is greater when the credibility of the source is low [21]. Jain and Posavac also support the prediction that source credibility is more important for evaluating experience attributes than search attributes [27]. In same manner, we assume that information from SNS regarded as relatively credible might be a method for reducing consumers perceived

uncertainty about products so that it could lead consumers more likely to buy. Moreover, since uncertainty varies across search, experience, and credence products, it is hypothesized that Facebook Like will influence sales of products differently across the SEC classification and the effect will become larger as qualityuncertainty increases, as previous research suggests. Due to a fact that credence attributes are also characterized in the same way that search and experience products are characterized[11, 30], it is acceptable to expand previous predictions to the credence products. H2: The effect of users preference for products collected through SNS on sales is higher for experience goods than search goods H3: The effect of users preference for products collected through SNS on sales is higher for credence goods than experience goods H4: The effect of users preference for products collected through SNS on sales is higher for credence goods than search goods

3.2 Regression Models


These hypotheses are tested with the econometric model. First of all, sales of daily deal are used as a dependent variable in this model. In order to explain the effect of information from SNS, the number of Facebook Like in each deal and dummy variables, S and E, for distinguishing search, experience, and credence products respectively are used as independent variables. Moreover, interaction terms between Facebook Like and categories are given for verifying the difference of the effect across search, experience, and credence goods/services. There are also a number of control variables such as Price, Saved, Tipp, Pop, and Global. Price is product price of each deal and Saved indicates the amount of money obtained by the original price minus the offered price, or Price. Tipp is the minimum amount of sales that consumers can buy a product if total sales are above. Pop is the population of the city where the deal is provided. Finally, using Global, we distinguish companies that are selling products in multiple cities so that we can reduce the asymmetry of selling power between firms in this model. For example, Kodak was selling their products across more than 30 cities, which means that this company is relatively well known to customers and this fact may cause a difference in sales among given deals. In this case we use

Table 1. Collinearity Diagnostics (intercept adjusted) Eigen N 1 2 3 4 5 6 7 8 Value 2.302 1.669 1.312 0.969 0.827 0.510 0.409 0.003 Cond Index 1.000 1.174 1.325 1.541 1.668 2.125 2.371 29.920 price2 4.7E-06 0.173 0.007 1.4E-06 0.012 0.030 0.772 0.005 saved2 7.9E-05 0.167 0.013 0.004 0.062 0.100 0.652 0.003 Tipp 0.053 0.002 0.078 0.091 0.016 0.705 0.053 5.3E04 Proportion of Variation Pop 0.043 0.019 0.059 0.194 0.013 0.550 0.122 4.3E04 Global 5.3E04 0.036 0.101 0.370 0.475 0.004 0.012 0.001 FB 8.2E-04 2.4E-05 2.7E-04 3.8E-04 6.7E-05 2.8E-05 1.5E-06 0.998 inter_s 7.9E-04 2.0E-04 0.006 0.005 0.015 5.1E-04 1.8E-04 0.972 inter_e 7.1E-04 1.1E-05 1.1E-03 7.1E-05 2.2E-04 2.5E-06 2.0E-06 0.998

the Global variable to explain the relative large amount of sales. Equation (1) is for testing H1 only and (2) is brought to verify the remained hypothesis from H2 to H4.
(1) Salesi 1 1 Pr icei 2 Savedi 3Tippi 4 Popi 5Globali 1Facebooki
(2) Salesi 1 1 Pr icei 2 Savedi 3Tippi 4 Popi 5Globali 1Facebooki 2 ( Si Facebooki ) 3 ( Ei Facebooki )

One might be interested in the fact that there are only interaction terms between dummy variables and Facebook Like, and no intercept for the dummy. This is because we believe that categories themselves do not influence sales independently for several reasons. First, the firms familiarity with consumers that can cause a different level of sales across deals is well controlled by the Global variable. Also, firms working with Groupon are usually local and not famous so no firm has distinct sales power, or popularity, compared with other deals on Groupon. Finally, when it is run including the dummy for intercepts, the results also do not come out with any statistically significance for those variables (p=0.1842, p=0.8145 for Search and Experience, respectively). A diagnostic check for multi-collinearity among independent variables is also conducted. According to the result in Table 1, there is no significant collinearity between the variables, which is determined by observing the condition indices in third column having low scores less than 30. From the results of tests, therefore, this model is free from a multi-collinearity problem. In addition, there is an econometric issue about heteroscedascity, so White test is adopted for verifying it. Null hypothesis of the test implies homoscedascity, thus the high F value leads us to reject the null hypothesis and concludes that the model has a heteroscedascity problem. The result of the test comes out with Fstatistic (25.696) which is far exceeding the value at 5% significance level. Therefore, in this case the result of simple OLS estimate is going to be biased and instead GLS should be employed for handling this problem.

evaluated before purchase. In addition, services that have low scores at first, but high on the second scales are classified as experience dominated products while services having low score on both scales are distinguished as credence dominated products. At the 5% significance level, the midpoint 4 is used to determine whether the score is low or high (i.e. H 0 : M i 4, i={before, after} ). Table 2 shows the results of the classification. The results show that among 33 goods/services, 11 are determined as search, 20 as experience, and 2 as credence goods/services. While most goods/services Groupon provides are experience, which is in line with our presumption that Groupon sells products which have high quality-uncertainty, only two services, dental care and facial, are regarded as credence services. The validity of the pretest was established based on the fact that the respondents did not identify any service as being easy to evaluate prior to purchase, but difficult to evaluate after consumption [32]. Table 2. Categorization of products
Category Search Products Sports game Performance Admission Restaurant Food Dessert Caf Hotel Books Photo Subscription Fitness Yoga Sports lessons Education Activity Tour Occasion Salon Beauty clinics Spa and Massage Pub Wine Car services House services Dry cleaning Apparel Furniture Household items Other goods Facial Dentalcare Before use Mean Std. 4.51 1.61 4.88 1.55 4.37 1.45 4.37 1.29 4.42 1.30 4.37 1.23 4.49 1.49 4.74 1.18 5.23 1.34 4.30 1.06 4.60 1.29 3.35 1.41 2.81 1.37 2.77 1.17 2.77 1.29 3.35 1.43 4.07 1.42 3.56 1.42 2.79 1.25 2.79 1.12 3.16 4.12 3.47 3.58 2.84 3.31 4.14 3.81 4.12 4.17 2.79 3.00 1.09 1.33 1.26 1.35 1.09 1.30 1.44 1.20 1.18 1.10 1.06 1.29 After use Mean Std. 5.81 1.28 6.09 0.95 5.79 0.97 6.07 0.86 6.23 0.75 6.12 0.82 5.84 1.02 5.86 0.97 6.05 0.92 5.58 1.05 5.37 1.22 5.28 1.08 4.88 1.29 5.09 1.17 5.14 1.06 5.56 1.08 5.65 1.11 5.60 1.06 5.79 1.12 4.98 1.34 4.93 5.60 5.07 4.93 5.09 4.98 5.88 5.23 5.35 5.52 4.24 4.28 1.33 1.16 1.47 1.16 1.19 1.42 0.82 1.13 0.97 0.92 1.50 1.44

Experience

3.3 Categorization by Search, Experience, and Credence Goods/Services


Iacobucci [26] empirically explored goods/services according to search-experience-credence ratings. However, sample of goods/services used in his work are quite different than what is concerned in this research. Therefore, a pretest was conducted to determine whether consumers perceive differences in the search, experience, and credence characteristics of services, as suggested in the literatures [11, 24, 25, 32]. 43 KAIST business school students are asked to participate in this survey. First, participants are supplied with a short explanation about purchase decisions that described how some services can be easily evaluated before purchase, while others cannot even after use. Then participants are asked to evaluate their ability to judge the performance of each service before purchase using a sevenpoint scale ranging from Not at all to Very well. The total number of services is 33. Participants were then asked again to evaluate their ability to judge the performance of each service after using it on the same seven-point scale. Services that have high scores on both scales are regarded as search dominated products because their performance can be

Credence

3.4 Descriptive Statistics of Data


To analyze the hypotheses, data was obtained by crawling the Groupon site (www.groupon.com) with an automated crawler specifically designed for Groupon, and includes sales of daily deals across 124 cities in United States and Canada from Nov 2 to Dec 9, 2010. Data also contains various attributes about each daily deal. Details are presented in Table 3 and Table 4.

Table3. Definitions of Model Variables Variable Description Sales Amount of sales for each deal Price Price of goods/services Saved Original price minus offered price Tipp Sales point for a deal to be on Population Population of a city Facebook Number of Facebook Like for each deal Search Dummy variable for search goods/services Experience Dummy variable for experience goods/services Inter_s Interaction term between Facebook and Search Interaction term between Facebook and Exp. Inter_e Nofb Number of Facebook users in a city (Age:25~35) At first, Facebook Like data was not collected because of some technical problems, thus data collecting was conducted except for Facebook Like. Later, the numbers were captured manually by researchers, and through this work the validity of the data was established by checking the exact number on the web site with crawled results. Among 2673 samples except for data that have some missing observations or errors caused by crawler, 1378 samples are obtained excluding unattained information of Facebook Like associated with each deal. There were many samples that did not have Facebook Like information so we dropped out a large amount of data. For this reason, the number of cities that were covered in the data was also reduced from 124 to 53. Even though the number of samples decreases substantially, the remaining data is still enough to reflect the generality of the nature in the Groupon site. Table4. Descriptive Statistics Variable N MeanMedian Std Min Max
Sales Price Saved Tipp Population Facebook Nofb 1378 599 218 1270.03 0 1378 29 20 29.56 3 1378 55 24 141.02 3 1378 43 25 69.27 1 1378 510570 375744 514760.77 41058 1378 29 11 81.94 0 1378 290472 138740 316206.2 11120 17473 200 3150 1000 2833321 2000 1718500

for a search good, and that of Inter_e plus the Facebook coefficient (18.45 7.68 = 10.77) for an experience good. Because all the concerned coefficients are significantly differ from zero, supported by t-values (11.95, -3.74, and -3.81, respectively), it is shown that goods/services that belong to the credence category, are more increasingly affected by Facebook Like than search, and experience goods/services, which means that H3 and H4 are supported. Table5. Results of the GLS for Model (1) Variable Intercept Price Saved Tipp Pop Global Facebook R-Square Coefficient 124.46 -4.33 0.39 4.57 9.14E-05 71.53 10.77 0.6792 Std. 41.418 0.689 0.100 0.736 6.26E-05 57.186 2.033 Adj R-Sq Pr > |t| 0.0027 <.0001 0.0001 <.0001 0.1442 0.2112 <.0001 0.6778

4. Results 4.1 Results of the GLS estimation


According to Table 5, the result of the GLS estimate shows that the number of Facebook Like shows a significant and positive relationship with sales in goods/services (t=5.298, p<.0001). With this, H1 can be supported. Moreover, all the other control variables show a significant relationship with the dependent variable, sales, except for Pop and Global whose p-value is above .10. Verified by the fundamental theory of the economy, it is predictable that sales (demand) go down as the price increases. It is also confirmed that consumer tends to buy more when they are offered goods/services at a high-discount rate, which is justified as being the same as a lower price. The results, shown in Table 6, show that number of Facebook Like positively affects sales of goods/services across all predefined categories including search, experience, and credence (p<0.1, p<0.0001, p<0.05, respectively), and as suggested in the hypothesis, the influence of Facebook Like varies across different categories. The Coefficient of Facebook (18.45) indicates the impact of Facebook Like when a credence good is offered, that of Inter_s plus the coefficient of Facebook (18.45 - 7.82 = 10.63) is

However, H2 (that Facebook Like shows a stronger effect on experience than search goods/services) is not supported with the results. The related Wald test for deciding whether the coefficient of Inter_s and Inter_e are equal or not, indicates that these two coefficient are not significantly different in a statistical manner (F=0.0028). This outcome, however, can be predicted by the presence of an online environment that forces products which have the aspect of experience goods/services turning into search goods/services [11, 25, 30]. As previous research indicates, it might be caused by the fact that online, consumers can easily create and share product information that could not be obtained unless they actually use it. For example, they can share the experience that they obtain when they use products, and users of the internet easily access these reviews by searching. Through this procedure, one can reduce quality-uncertainty by consuming the information. Thus, our results can be explained in the same way. As the online environment blurs the distinction between search and experience products sold online, the effect of the information on sales of both search and experience products also should not vary. When the information cannot be obtained prior to use or even after use due to inherent features the product bears, consumer may look to a source as a cue. Consumers may verify experience and credence related information only when the source is credible [17, 43]. From the results, we confirm that under the presence of quality-uncertainty, the consumer also looks for additional information whose source is credible rather than relying only on information contained in the product message. Moreover, if the uncertainty increases, reliance on the source is also enlarged.

5. Conclusion
This research investigates the influence of information from SNS, such as Facebook, on consumers perceived quality-uncertainty of products in E-commerce. To verify this, we use 1378 samples of Groupon sales data obtained by crawling the Groupon website. An additional pretest was conducted to classify products according to the degree of quality uncertainty. Using classified product categories of search, experience, and credence products we conclude that information from SNS can

compensate for the quality-uncertainty that consumer faces and it can increase the likelihood of making a purchase. In addition, it is also supported that this influence varies across the degree of the quality-uncertainty which changes along with various products. In other words, as the difficulty of judging the quality of the product increases, the influence significantly get stronger. Moreover, our finding that the effects on search and experience are not significantly different is well supported by the fact that the internet blurs distinctions between experience and search goods by providing mechanisms that enable online shoppers to gather information on experience and search attributes [11, 25, 30]. Table6. Results of the GLS for Model (2) Variable Intercept Price Saved Tipp Pop Global Facebook Inter_s Inter_e R-Square Coefficient 127.53 -4.49 0.37 4.57 9.01E-05 74.03 18.45 -7.82 -7.68 0.6805 Std. 36.487 0.753 0.084 0.684 6.35E-05 63.868 1.544 2.089 2.016 Adj R-Sq F Value 0.002815 Pr > |t| 0.0005 <.0001 <.0001 <.0001 0.1561 0.2466 <.0001 0.0002 0.0001 0.6787 Pr > F 0.9577

a survey is conducted to capture individual conducts. The third limitation is the problem of convenience sampling when categorization is carried out. Future study should consider limitations discussed above. Furthermore, we have initiated additional research that will add more value to our findings in this research, by using time-series data of sales. Deals in Groupon have a special feature in that they sell products only for one day, thus actual bidding or purchasing timing can be observed. Using this, we would expect that according to the degree of qualityuncertainty, consumers actual bid-timing could be changed and this change also could be influenced by the source of information which is, in this case, SNS. Usually, sale periods are all different, so bid-timing is not observable in the real world case. With this data, however, we could derive the effect of information from SNS on actual bid-timing, and thus this research would provide a better understanding about consumers behavior under the presence of quality-uncertainty. This paper contributes to a better understanding of consumer behavior under the existence of quality-uncertainty. It also certifies that the more uncertainty there is, the more the consumer relies on the source, not the information itself. Using Groupon sales data, this research also analyzes the dynamics of Social Commerce empirically, and points out fundamental factors that impact sales. For managerial implications, the use of SNS for the purpose of marketing should be concerned with what characteristic of products the company sells. According to the results, SNS might be more efficient in a situation in which consumers face a high degree of uncertainty to verify the quality of products. Like the brand name, information from their friends also compensates the ambiguity by generating credible cues. Thus, a company that deals for products having inherent uncertainty should consider and manage its marketing strategy using SNS. SNS is one the most interesting issue on the internet nowadays; most people, however, have insufficient understanding of the implications on how and when it should be used. This paper shed lights on the effective use of SNS and on Social Commerce.

H0 :2 3 2 :coefficient of Inter_s 3:coefficient of Inter_e

From the previous studies, it is predicted that when it is difficult to obtain quality information before use or even after use, consumer may look to a source of information as a cue and verify experience information only when the source is credible [27]. This is because if a consumer could not distinguish the exact quality among different messages given, these are treated as equal and then consumers are likely to find other information that they can use to verify the quality of products, which is the source of information in this case. For instance, if he/she could not find information about the quality of products even after use, reviews do not contain any of useful information about products and are not different from information given before use. Then, he/she has to discover other cues to rely on. Because SNS consists of friends who have connections with users and who are relatively credible compared with unknowns, information that comes from SNS can be used by consumers to verify uncertain quality information. According to the results in this paper, we suggest that SNS is treated as a credible source of information and provides additional messages which are not contained in traditional information shared over the internet. Those additional messages can give consumers clues to help verify information when they are exposed to uncertainty. This study has several limitations. First, data does not cover all deals in Groupon during the period. Since there were some technical and observational errors, we eliminate a large amount of data and this affects the generality of our results. In addition, since we use aggregated preference, not individual, there is a little doubt about revealing the low-level behavior concluded in this paper. Although aggregated data is enough to represent the norm of consumer behavior conducted in Groupon, it would be better if

6. REFERENCES
[1] Abbe-Decarroux, F. 1994. "The perception of quality and the demand for services:: Empirical application to the performing arts," Journal of Economic Behavior & Organization (23:1), pp. 99-107. [2] Aggarwal, P. and R. Vaidyanathan. 2005. "Perceived effectiveness of recommendation agent routines: search vs. experience goods," International Journal of Internet Marketing and Advertising (2:1), pp. 38-55. [3] Anand, K.S. and R. Aron. 2003. "Group buying on the web: A comparison of price-discovery mechanisms," Management Science), pp. 1546-1562. [4] Animesh, A., V. Ramachandran, and S. Viswanathan. 2005. "Online Advertisers Bidding Strategies for Search, Experience, and Credence Goods: An Empirical Investigation," Citeseer. [5] Bennedsen, M., et al. 2007. "Inside the Family Firm: The Role of Families in Succession Decisions and Performance," Quarterly Journal of Economics (122:2), pp. 647-691. [6] Bourdieu, P. and L.J.D. Wacquant. 1992. An invitation to reflexive sociology. University of Chicago Press. [7] Chen, J., X. Chen, and X. Song. 2007. "Comparison of the group-buying auction and the fixed pricing mechanism," Decision Support Systems (43:2), pp. 445-459.

[8] Chen, J., et al. 2010. "Segmenting uncertain demand in group-buying auctions," Electronic Commerce Research and Applications (9:2), pp. 126-147. [9] Chevalier, J.A. and D. Mayzlin. 2006. "The Effect of Word of Mouth on Sales: Online Book Reviews," Journal of Marketing Research (JMR) (43:3), pp. 345-354. [10] Coleman, J.S. 1988. "Social capital in the creation of human capital," American Journal of Sociology), pp. 95-120. [11] Darby, M.R. and E. Karni. 1973. "Free competition and the optimal amount of fraud," Journal of Law and Economics (16:1), pp. 67-88. [12] Davis, A. and D. Khazanchi. 2008. "An empirical study of online word of mouth as a predictor for multi-product category e-Commerce sales," Electronic Markets (18:2), p. 130. [13] Deutsch, M. 1960. "The effect of motivational orientation upon trust and suspicion," Human Relations (13:2), pp. 123139. [14] Donath, J. and D. Boyd. 2004. "Public displays of connection," BT Technology Journal (22:4), pp. 71-82. [15] Duan, W., B. Gu, and A.B. Whinston. 2008. "The dynamics of online word-of-mouth and product sales - An empirical investigation of the movie industry," Journal of Retailing (84:2), pp. 233-242. [16] Eaton, D.H. 2005. "Valuing information: Evidence from guitar auctions on eBay," Journal of Applied Economics and Policy (24:1), pp. 1-19. [17] Ellison, N.B., C. Steinfield, and C. Lampe. 2007. "The benefits of Facebook "friends:" Social capital and college students' use of online social network sites," Journal of Computer Mediated Communication (12:4), pp. 1143-1168. [18] Gefen, D. 2000. "E-commerce: the role of familiarity and trust," Omega (28:6), pp. 725-737. [19] Girard, T., P. Korgaonkar, and R. Silverblatt. 2003. "Relationship of type of product, shopping orientations, and demographics with preference for shopping on the Internet," Journal of Business and Psychology (18:1), pp. 101-120. [20] Greene, H. 2008. "Econometric Analysis (6-th ed.)." Prentice-Hall, New Jersey. [21] Grewal, D., J. Gotlieb, and H. Marmorstein. 1994. "The Moderating Effects of Message Framing and Source Credibility on the Price-Perceived Risk Relationship," The Journal of Consumer Research (21:1), pp. 145-153. [22] Grund, C. and O. Guertler. 2008. "The effect of reputation on selling prices in auctions," Journal of Economics and Statistics (Jahrbuecher fuer Nationaloekonomie und Statistik) (228:4), pp. 345-356. [23] Hoffman, D.L., T.P. Novak, and M. Peralta. 1999. "Building consumer trust online," Communications of the ACM (42:4), pp. 80-85. [24] Hsieh, Y.C., H.C. Chiu, and M.Y. Chiang. 2005. "Maintaining a committed online customer: A study across search-experience-credence products," Journal of Retailing (81:1), pp. 75-82. [25] Huang, P., N.H. Lurie, and S. Mitra. 2009. "Searching for experience on the web: An empirical examination of consumer behavior for search and experience goods," Journal of Marketing (73:2), pp. 55-69. [26] Iacobucci, D. 1992. "An empirical examination of some basic tenets in services: goods-services continua," Advances in services marketing and management (1), pp. 23-52. [27] Jain, S.P. and S.S. Posavac. 2001. "Prepurchase Attribute Verifiability, Source Credibility, and Persuasion," Journal of

[28]

[29]

[30]

[31] [32]

[33]

[34]

[35]

[36]

[37]

[38] [39] [40]

[41]

[42]

[43]

[44] [45]

[46]

[47]

Consumer Psychology (Lawrence Erlbaum Associates) (11:3), pp. 169-180. Kauffman, R. and B. Wang. 2001. "New buyers' arrival under dynamic pricing market microstructure: The case of group-buying discounts on the Internet," Published by the IEEE Computer Society, p. 7034. Kauffman, R.J. and B. Wang. 2002. "Bid together, buy together: On the efficacy of group-buying business models in Internet-based selling." CRC Press, pp. 99-137. Klein, L.R. 1998. "Evaluating the potential of interactive media through a new lens: Search versus experience goods," Journal of Business Research (41:3), pp. 195-203. Kramer, R. and T. Tyler. 1996. "Whither trust," Trust in Organizations. Thousand Oaks: Sage Publications). Krishnan, B. and M. Hartline. 2001. "Brand equity: is it more important in services?," Journal of Services Marketing (15:5), pp. 328-342. Levin, A.M., I.P. Levin, and J.A. Weller. 2005. "A multiattribute analysis of preferences for online and offline shopping: Differences across products, consumers, and shopping stages," Journal of Electronic Commerce Research (6:4), pp. 281-290. Lewicki, R. and B.B. Bunker. 1996. "Developing and maintaining trust in work relationships," Trust in organizations: Frontiers of theory and research), p. 114. Liu, F. and H.J. Lee. 2010. "Use of social network information to enhance collaborative filtering performance," Expert Systems with Applications (37:7), pp. 4772-4778. Mayer, R.C., J.H. Davis, and F.D. Schoorman. 1995. "An integrative model of organizational trust," The Academy of Management Review (20:3), pp. 709-734. McKnight, D.H. and N.L. Chervany. 2001. "What trust means in e-commerce customer relationships: an interdisciplinary conceptual typology," International Journal of Electronic Commerce (6:2), pp. 35-59. Nelson, P. 1970. "Information and consumer behavior," The Journal of Political Economy), pp. 311-329. Nelson, P. 1974. "Advertising as information," The Journal of Political Economy (82:4), pp. 729-754. Ngai, E. and F. Wat. 2002. "A literature review and classification of electronic commerce research," Information & Management (39:5), pp. 415-429. Nohria, N. and R. Eccles. 2000. "Face-to-face: Making network organizations work," Technology, Organizations and Innovation: Towards' real virtuality'?), p. 1659. Parasuraman, A., V.A. Zeithaml, and L.L. Berry. 1985. "A conceptual model of service quality and its implications for future research," The Journal of Marketing (49:4), pp. 41-50. Paxton, P. 1999. "Is social capital declining in the United States? A multiple indicator assessment," American Journal of Sociology), pp. 88-127. Ratnasingham, P. 1998. "The importance of trust in electronic commerce," Internet research (8:4), pp. 313-321. Reinstein, D.A. and C.M. Snyder. 2005. "The influence of expert reviews on consumer demand for experience goods: A case study of movie critics," Journal of Industrial Economics (53:1), Mar, pp. 27-51. Weathers, D., S. Sharma, and S.L. Wood. 2007. "Effects of online communication practices on consumer perceptions of performance uncertainty for search and experience goods," Journal of Retailing (83:4), pp. 393-401. Yoo, B.-J., Y.B.J. and G.-W. Lee, L.G.W. 2007. "An Alternative Approach in Analyzing the Impacts of Online

Feedback System; A Bayesian Inference Model," Korea Society of Management Information Systems), pp. 395-400.

Вам также может понравиться