Facebook C2C social commerce: A study of online ...

Decision Support Systems 83 (2016) 57–69

Contents lists available at ScienceDirect

Decision Support Systems journal homepage: www.elsevier.com/locate/dss

Facebook C2C social commerce: A study of online impulse buying Jengchung Victor Chen a,1, Bo-chiuan Su b,⁎, Andree E. Widjaja a,c,1 a b c

Institute of International Management, National Cheng Kung University, No. 1, University Road, Tainan City, Taiwan Department of Information Management, National Dong Hwa University, No. 1 Sec. 2, University Road, Shoufeng, Hualien, Taiwan Department of Information Systems, Faculty of Computer Science, Pelita Harapan University, 1100 M.H. Thamrin Boulevard, Lippo Village, Tangerang 15811, Indonesia

a r t i c l e

i n f o

Article history: Received 16 August 2014 Received in revised form 26 December 2015 Accepted 26 December 2015 Available online 6 January 2016 Keywords: Online impulse buying Facebook group Information quality Facebook “likes” Latent state–trait theory Observational learning

a b s t r a c t Facebook users are increasingly using the site to conduct commercial activities, by posting advertisements in groups and then buying or selling items from each other. This type of group is called as a C2C Facebook “buy and sell” group in the current work. Drawing from latent state–trait theory, heuristic information processing, and observational learning, we conducted an online field experiment to empirically investigate the effect of the information quality of the advertisement, the trait of the impulsiveness, and the number of “likes” it receives on consumers' urge to buy impulsively. The findings and implications of our study are discussed in the paper. © 2015 Elsevier B.V. All rights reserved.

1. Introduction While impulse buying has been extensively studied by scholars for many years [1,2], few studies have examined this issue in an online context [3]. Despite this, a number of recent studies have examined how various factors related to information systems artifacts influence online impulse buying within an e-commerce context, such as online store beliefs [4], system design [5], website quality [3,6], website attributes [7], website atmospheric cues [8], or website ease of navigation [9]. However, to the best of our knowledge, there are still very few studies, which have empirically investigated online impulse buying in a social commerce context, and this work thus addresses this gap in the literature. Social networking sites (SNSs), such as Facebook, are increasingly utilized to conduct commercial activities, a phenomenon known as social commerce [10]. One way to conduct such commercial activities is by using Facebook's group function, and since such activities will most likely involve one user (as a seller) and another user (as a consumer), the group is considered as conducting the consumer-to-consumer (C2C) commercial activities. In this study, we call these groups as C2C Facebook “buy and sell” groups. This study thus aims to investigate online impulse buying in C2C Facebook “buy and sell” groups. Prior studies have demonstrated a strong Abbreviations: C2C, Consumer-to-consumer; SC, Social commerce; UBI, Urge to buy impulsively; IQ, Information quality; LST, Latent state–trait theory; D&M, DeLone and McLean; IS, Information systems; OL, Observational learning. ⁎ Corresponding author. Tel.: +886 3 863 3109 (office); fax: +886 3 8633109. E-mail addresses: [email protected] (J.V. Chen), [email protected] (B. Su), [email protected] (A.E. Widjaja). 1 Tel.: +886 6 275 7575 × 53561 (office); fax: +886 6 275 1175.

http://dx.doi.org/10.1016/j.dss.2015.12.008 0167-9236/© 2015 Elsevier B.V. All rights reserved.

positive relationship between e-commerce website quality attributes and online impulse buying [4,6,7,11]. Nevertheless, this type of group is unique compared to ordinary e-commerce websites. As the commercial activities on Facebook groups are based on user-generated content, the most critical website quality attribute to be considered is the information quality of the advertisements posted by the sellers. The quality of information will be perceived and processed by the users (consumers) as an environmental cue, based on which they will then decide whether or not to make an impulse purchase. Moreover, one of the key features of Facebook is its use of “likes,” and thus the number of “likes” a post receives might also be considered by the users (consumers) in this context. This study examined the following research questions: “Do information quality and number of ‘likes’ affect consumers' urge to buy impulsively?” Also, “Does the trait of impulsiveness have interaction effect on the relationships that information quality and number of “likes” have with consumers' urge to buy impulsively?” The main goal of this study is therefore to contribute to the literature by empirically investigating the effect of information quality, the trait of impulsiveness, and number of “likes,” as well as their interaction effects, on consumers' urge to buy impulsively in a social commerce (C2C Facebook “buy and sell” group) context. 2. Theoretical framework and literature review 2.1. C2C Facebook “buy and sell” groups Facebook groups are now widely used to conduct C2C commercial activities, with users offering either new and secondhand items or services for sale via advertisements they post on the site. Users

58

J.V. Chen et al. / Decision Support Systems 83 (2016) 57–69

(consumers) who see such posts and are interested in making a purchase can then contact the other user (seller) directly to make a buying arrangement, such as seeing or testing the product, or making a payment and then receiving the item. The “Tainan market: buy and sell” Facebook group is an example of this type of group that operates in Taiwan (https://www.facebook.com/groups/TainanMarket.BuyandSell/), and a screenshot of that group is shown in Fig. 1.

2.2. Urge to buy impulsively and latent state–trait theory Beatty and Ferrell [12] defined impulse buying as “a sudden and immediate purchase with no pre-shopping intentions either to buy the specific product category or to fulfill a specific buying task” (p. 170). As it is difficult to measure actual impulsive behavior, consumers' urge to buy impulsively (UBI) was used in this study as a surrogate measure

Fig. 1. Screenshot of the “Tainan market: buy and sell” Facebook group.


of their impulse buying behavior. In support of this, Luo [13] argued that observing actual impulsive behavior in a controlled setting is problematic, and so consumers' UBI has been used as a proxy for impulsivity in numerous studies of online impulse buying [6,7,11,14]. Eysenck [15] stated that it is important to examine both individuals' intrinsic traits and their current state of mind to better understand their behaviors. Researchers have studied how a person's state of mind can be affected by retail environments, such as a store layout or website's characteristics, in order to trigger more impulse buying [1,6,11]. Meanwhile, other studies have seen impulsiveness as an impulse buying trait, which has also been found to positively influence intention to shop online [1,6, 16]. Finally, following Eysenck [15], it has been argued that latent state– trait theory can be used to explain online impulse buying in an ecommerce context [6], and thus it is adopted in the current study. The latent state–trait (LST) theory is a psychology theory developed by Steyer, Schmitt, and Eid [17], which claims that human behaviors are dependent on environmental cues or situations (states), individual factors (traits), and the interplay (interactions) between these two determinants. Previous research utilized LST theory as a theoretical framework to explain impulse buying within an e-commerce context [6]. Drawing on Wells, Parboteeah, and Valacich [6], the current work incorporates LST theory by operationalizing “state” as the degree of information quality in a related post, and the number of “likes” it receives. Meanwhile, impulsiveness is operationalized as the individual's “trait.” Fig. 2 depicts the research framework used in this study. 2.3. Information quality construct E-commerce researchers have identified the critical role of information quality based on the ideas presented in DeLone and McLean's (D&M) IS success model [18–20] and tested it from the perspective of online customers [21,22]. IQ is important as it can positively influence consumers' online purchase decisions [23,24]. Many of the past studies related to IQ have included dimensions such as “relevance,” “accuracy,” “understandability,” “completeness,” “format,” and “currency” [19,20, 22,25,26]. Meanwhile, IQ can be categorized into three categories: intrinsic quality, contextual quality, and representational quality [26,27].

59

Intrinsic quality refers to the “accuracy” of information that is given. Moreover, this information needs to be defined relative to the receiver of the information, and this refers to contextual quality. The IQ dimensions that belong to this category are “relevance,” “completeness,” and “currency” [27]. We argue that “easy to understand” (e.g., understandability) can be categorized into contextual quality since whether the information is seen as easy or difficult to understand depends on the knowledge that each individual already has about the focal subject. Finally, the representational dimension refers to the effectiveness of information presentation, and the IQ dimension of this category is “format.” There are four types of information which can be posted on Facebook: text, picture, audio, and video. Based on our empirical observations, text and picture are the most frequently used in this context. The picture IQ dimensions are somewhat different from the IQ construct conceptualized in IS research. Therefore, since the majority of ecommerce studies have associated picture IQ with product presentation and its specific design factors [28–30], the scope of this study is limited to the textual IQ rather than picture IQ. Besides, an earlier study found that the use of text-based media led to more impulse buying [31], and thus the existing literature indicates that it is appropriate to focus on using textual IQ in this study. 2.4. The significance of investigating information quality in a social commerce and impulse buying context While IQ has become one of the fundamental e-commerce design principles [32], the situational differences of IQ and its role in Social Commerce (SC) context are mostly unexplored. We, therefore, argue that there are three key reasons why IQ deserves further investigation in an SC context. First, unlike e-commerce, which might be primarily driven by experienced online marketers (companies), SC is entirely focused on user-generated content (UGC) [33,34]. As UGC has economic value [33], researchers have agreed that SC users are important as active contents creators [34], who play an important role in driving the transactions seen on online market places [35]. The contents created by the users (sellers) can be in the form of text-based product information [33,36]. This textual information serves as the external informational

Fig. 2. Research model.

60


cues that support the other users (consumers) in their buying behaviors [33], and this is the main source of the information upon which their purchase decisions are based [37,38]. We argue that the diversity of the users (sellers) with regard to their knowledge about IQ might be an issue in an SC context. IQ is often explored in studies of e-commerce websites [19,21], and it is thus likely that the majority of experienced online marketers have some knowledge about how to create messages that contain high-quality information. By contrast, the quality of the information created by users (sellers) might be varied, as they are likely to use many different ways to present textual information. This is because such users (sellers) are often ordinary people with diverse demographic characteristics and backgrounds (e.g., age, culture, education, and personality). These differences might directly influence users' understandings about IQ, and it is also likely that some users (sellers) might not be aware of IQ in this context, or of its significance. Second, IQ is a very important factor in C2C Facebook “buy and sell” groups due to the fact that such groups have very limited functions that only allow users (sellers) to post advertising information online [39]. For example, users cannot alter the environmental design of the site, such as changing the website designs, navigation structure, or color scheme. This situation is very different from the situation seen on ordinary e-commerce websites, for which online marketers are able to manipulate consumers' sense of vision by the use of store designs or functions that aim to stimulate impulse buying [9]. Besides, unlike most of the e-commerce websites, the group also does not have online reputation systems as well as third party assurance systems for reducing seller uncertainty. It is thus reasonable to argue that IQ is one of the most important environmental cues in a C2C Facebook “buy and sell” group context and is a key consideration both for the sellers to advertise their products, and for the potential consumers to make their purchase decision. Third, since the IQ construct has been conceptualized and investigated primarily in an e-commerce context, little is known whether the existing construct would have similar effects in an SC context. It is possible that the users (consumers) might react differently when seeing advertisements with varying degrees of IQ posted by the other users (sellers) on a Facebook group than is the case on ordinary ecommerce websites. For instance, as any users can practically sell the product on the group, it is likely that the other users (consumers) might feel uncertain about the seller and/or the product (see Dimoka, Hong, and Pavlou [40] for a review regarding seller and product uncertainty in the online markets). To date, little research has examined IQ in an SC context. For instance, Huang and Benyoucef [38] proposed a set of SC design principles in which IQ was included as one of the common design features. Their study inferred that IQ also plays an essential role in SC, and thus it should not be overlooked by SC designers. However, it should be noted that this earlier work did not empirically investigate IQ. A recent study found that customer perceived IQ and the effectiveness of the information content both have positive effects on social shopping intention [41]. While this study demonstrated the role of IQ in general, it did not look into the dimensions of IQ in a detailed manner. Besides, the study focused on Malaysian online social media groups and the issue of regular buying, rather than impulse buying. Due to the few prior studies on IQ in an SC context, and their limited scope, we contend that further research is necessary, thus motivating the current work. In addition, we argue that IQ is a more important factor in impulse buying than in a regular buying (planned purchase) context. Unlike regular buying, in which all sources of information are available, in an impulse buying situation the only available external information is that presented in the immediate environment [42]. As a result, Lee and Kacen [42] posited, “it is likely that information available inside a store will have a greater overall impact on an impulse purchase than a planned purchase” (p. 266). Similarly, the textual IQ generated by the users (sellers) would be the most significant available external information for the other users (consumers) when the latter engage in impulse

buying on a C2C Facebook “buy and sell” group. This makes IQ a more important driver in an impulse buying context than in a regular buying one. 2.5. The theoretical basis of information quality affecting impulse buying 2.5.1. Cognitive aspect of impulse buying Impulse buying is a type of rapid, less deliberate, on-the-spot decision making, and unplanned purchase [42–46], which occurs as a result of “minimal” information processing [4,47]. Although the literature might seem to indicate that impulse buying implies lack of rationality or alternative evaluation, this is not necessarily true, as it still involves a logical decision [48]. In fact, some researchers contend that impulse purchases are a result of a cognitive deliberation process [44,49], and so not necessarily irrational [50]. In a similar vein, others argue that impulse buying is not free from cognitive information processing [2,4,5]. A number of researchers have acknowledged the important role of cognitive components in impulse buying [2,11,51–54]. According to Peter and Olson [43], cognition is inevitably required to process and interpret the information that is available. When interacting with online websites, consumers will process product-related information presented using their cognitive reactions [11], and these are thus posited as an important component of the impulse buying process [55]. After all, given the fact that the affective component cannot make decisions [43], actual impulse purchases will be made only if cognitive forces are involved in the decision [56]. 2.5.2. Heuristic information processing and the role of information quality in impulse buying The information processing model suggests that consumers will process the stimuli available from the environment when making purchase decisions [43,48,57], including both regular and impulse purchase decisions. Research suggests that impulse buying is driven by a heuristic form of information processing [2,58,59], because the cognitive information processing that occurs in this context (rapid, simple, low effort, and less deliberate) is analogous to the concept of heuristics. Moreover, Peter and Olson [43] stated that impulse purchases represent a type of limited decision making, which is also synonymous with the notion of heuristics. Given the heuristic nature of impulse buying, and the support from the literature for this concept, we contend that how IQ influences impulse buying can be better explained by heuristic information processing (HIP) [60], with HIP being part of the heuristic–systematic model (HSM) of information processing proposed by Chaiken [61]. HIP is concerned with a limited processing mode that requires less cognitive effort [62] and that simplifies a consumer's decision-making process by using simple rules of thumb, such as the cues present in the available information [57,63]. The heuristic integration process in decision making is highly adaptive, may be constructed on the spot, and applied fairly automatically in response to the environment [43]. According to Chaiken, Liberman, and Eagly [62], when consumers process the information heuristically, they will focus on the subset of available information that enables them to use cognitive heuristics to make a decision. For example, consumers might focus on a heuristic cue such as message length to make a purchase decision [62]. As greater message length implies more strength [62], people might respond to a lengthy message more positively than to a shorter one. This argument is reinforced by a previous study which demonstrated that having more sufficient information increased people's willingness to process information heuristically, hence reducing their risk evaluations [63]. Another example of a heuristic cue is source expertise. As such, a message written by an expert (high source expertise) is considered to be more trusted than a message written by a non-expert (low source expertise) [62]. Applying the logic of the above discussion to the role of IQ in impulse buying, it is likely that consumers would focus on the dimensions of IQ and process them heuristically in a form of “IQ heuristic cue.” For instance, rich and lengthy text messages can improve the perception of


the product, thereby affecting advertisement effectiveness [64,65]. In view of this, “completeness” might imply information effectiveness, sufficiency, or strength. Consumers will also be concerned about the believability or reliability of the information source if the information provided is inaccurate [66] or irrelevant. Greater “accuracy” and “relevance” might imply that the message is more believable, reliable, and trustworthy. Late information has no added value [66], and so “currency” might imply value. A well-designed website could generate a more positive attitude toward the site [67], and hence a well-designed “format” might imply positive attitude or possibly a greater sense of professionalism. Meanwhile, users commonly see the characteristics of “simplicity” and “clarity” as representing an easy to understand design [68]. Therefore, “easy to understand” might imply simplicity as well as clarity. When consumers engage in impulse buying, they might heuristically perceive and process the aforementioned “IQ heuristic cues” as an “information signaling mechanism” (see Spence [69] and Dimoka, Hong, and Pavlou [40] for information signaling review). The effective, visible, clear, and credible product information signals (cues) in a form of textual product descriptions could help the consumers reducing their uncertainties regarding the product and facilitating their decision making [40, 70,71]. Research suggests that IQ is directly related to a perceived decrease in time and effort in decision making, thereby making the decision-making process more efficient [72,73]. Meanwhile, impulse buying is characterized by a rapid (decrease in time) and less effortful (decrease in effort) decision making. Based on these premises, it can be inferred that greater IQ would lead to more impulse buying. Following the logic, HIP, and information signaling mechanism, we argue that high IQ would make consumers' heuristic information processing stronger, so that it would be much more efficient when engaging in impulse buying, thereby further increasing their propensity to make impulse purchases. In addition, impulse buying is the result of an exposure to a stimulus [45]. Solomon [57] pointed out that the intensity of any stimulus is crucial with regard to whether it can be detected by the consumers' sensory system. Moreover, the level of intensity might imply the degree of quality. Prior studies have strongly demonstrated that high-quality environmental cues, such as a high-quality website, can be a salient stimulus to increase online impulse buying [6,11], and IQ itself is identified as a dimension of web quality [74]. Therefore, by referring to the aforementioned logic as well as the strong evidence presented in the literature, we posit the following hypothesis: H1. The advertisements posted on a C2C Facebook “buy and sell” group with high textual IQ (“relevance,” “accuracy,” “ease of understanding,” “completeness,” “format,” and “currency”) will stimulate stronger consumers' UBI than advertisements with low textual IQ.

61

The content of such OL information is similar to that seen with Facebook “likes.” Therefore, when consumers consider the number of “likes” when making an impulsive buying decision, it can be implied that they are involved in OL. When a post has a high number of “likes,” then an information cascade can occur, such that subsequent consumers will perceive the same positive effect signaled by that a high number of “likes,” thus triggering their UBI. The herding effect or herd behavior will occur when consumers see other people behaving in a certain way and then imitate these behaviors, possibly by ignoring their own information [79]. According to this logic, other consumers will follow as well as be influenced by the positive perceptions signaled by a high number of “likes,” and thus become engaged in a type of herd behavior. Based on the aforementioned arguments, the following hypothesis is proposed: H2. The advertisements with a high number of “likes” on C2C Facebook “buy and sell” group will trigger a stronger consumers' UBI than the advertisements with a low number of “likes.” 2.7. Impulsiveness and the interaction effects Wells, Parboteeah, and Valacich [6] noted that only considering the effect of website quality on consumers' UBI would provide a limited view of impulse buying. Consequently, in addition to website quality, individual traits such as a tendency to impulsiveness should also be considered in this context. According to Beatty and Ferrell [12], impulsiveness refers to “both the tendencies (1) to experience spontaneous and sudden urges to make on-the-spot purchases and (2) to act on these felt urges with little deliberation or evaluation of consequence” (p. 174). Consumers who have higher levels of impulsiveness have been found to be more likely to experience UBI in a traditional retail context [12]. Prior evidence also found that impulsiveness has a strong relationship with impulse buying intention [8,16] as well as consumers' UBI [6]. Youn and Faber [51] pointed out that individuals who rate high in impulsiveness are more sensitive and responsive to environmental cues, causing them to be more likely to engage in impulsive buying behavior. In line with LST theory, this study aims to explore the interaction effect of consumers' impulsiveness on the relationship between textual IQ and consumers' UBI. Moreover, this study also tries to explore the interaction effect of impulsiveness on the relationship between number of “likes” and consumers' UBI. We therefore postulate that, when interacting with advertising posts with varying textual IQ and number of “likes,” consumers' UBI will also be affected by their level of impulsiveness. The related hypotheses are stated as follows: H3. High impulsiveness consumers will experience a stronger UBI than low impulsiveness consumers.

2.6. Number of “likes” and the urge to buy impulsively The number of “likes” a post gets is one of the most important features on Facebook group. It can be assumed that when users click “like” on a post then they have positive attitudes, feelings, or perceptions toward it, based on an agreement with, recommendation of, or simply liking the content. The number of “likes” can thus be considered as a social environmental stimulus on a C2C Facebook “buy and sell” group. The link between the number of “likes” and consumers' UBI can be explained by observational learning [75] and the herding effect [76,77]. Zhang, Hu and Zhao [35] mentioned that online social interaction may also play an important role in online impulse buying. This type of interaction can be in the form of behavior-based social interaction, which is also known as observational learning (OL) [35,75]. OL is rooted in information cascade theory [78] and has been shown to influence purchase decisions [35,75]. According to Chen, Wang, and Xie [75], OL information contains “discrete signals expressed by the action of other consumers, but not the reasons behind the actions” (p. 240).

H4. There will be an interaction effect of consumers' impulsiveness on the relationship between textual IQ and consumers' UBI. H5. There will be an interaction effect of consumers' impulsiveness on the relationship between number of “likes” and consumers' UBI. 3. Research method 3.1. Experimental design An online field experiment was conducted to systematically study the effects of different factors of textual IQ as well as number of “likes” on consumers' UBI. We applied a full factorial mixed subjects design by manipulating the seven main constructs. To reduce excessive complexity in the experimental design, as well as the number of participants, we divided our experiment into two parts, part A and part B, as shown in Table 1 and Table 2, and used the same subjects for both parts.

62


Part A tested the textual IQ relevance, ease of understanding, and accuracy dimensions, employing a full factorial 2 × 2 × 2 between subjects design. Meanwhile, part B tested the textual IQ completeness, currency, and format dimensions along with number of “likes,” employing a full factorial 2 × 2 × 2 × 2 mixed subjects design. With regard to the participants who took part in parts A and B, we applied a within subjects design to reduce the effects of individual preferences. That is, the same participants were used in both parts of the experiment. We applied a counterbalancing design by changing the order of the two parts of the experiment in two groups – A first and then B (AB), or and B first and then A (BA) – to lessen the possibility of learning effects [80,81].

Table 2 Experimental design part B. Currency High

Low

Format Part B Completeness

High

Like

Low

Low High Low High

High

Low

High

Low

B1 (36)

B2 (35)

B3 (35)

B4 (35)

B5 (36)

B6 (32)

B7 (35)

B8 (33)

3.2. Product selection and factor manipulation We examined the various different kinds of products sold on twenty C2C Facebook “buy and sell” groups in Taiwan. We concluded that secondhand smartphones, particularly iPhone 5, were the most commonly sold items, and thus a secondhand iPhone 5 was used as the focal product in our experiment. We manipulated each factor based primarily on its operational definition, as shown in Table 3. To obtain more objective results, we examined numerous secondhand iPhone 5 advertisements posted on several C2C Facebook “buy and sell” groups in Taiwan with regard to the textual IQ and number of “likes.” After we manipulated the text IQ factors, the results were checked by three English speakers. As for the number of “likes,” we examined the average number for various Smartphone advertisements posted on four C2C Facebook “buy and sell” groups, and this was found to be three. As such, we decided to operationalize the high “likes” treatment with a figure of six, and low “like” treatment with a figure of zero. Instead of using actual Facebook accounts, we generated the number of “likes” by creating six different experimental Facebook accounts (all of which were newly created for this study, with no actual Facebook friends added to the accounts) in order to add the “likes,” as needed. These six experimental Facebook accounts had unusual names and were purposely created to avoid the possibility that the participants might notice their Facebook friends or their friends of friends clicking “like,” thereby isolating the influence of the number of “likes” from that of the people who clicked “like.” In other words, we controlled the effect of the participants' Facebook friend relationships in the high number of “likes” treatment. Due to the interrelated nature of the text IQ factors, when we manipulated these in part A, four factors in part B were kept constant. Likewise, when we manipulated the factors in part B, three factors in part A were kept constant. We also controlled other related information, such as picture IQ, seller Facebook account (experimental Facebook accounts were also used for this), and price, so that this information was the same across experimental manipulation conditions, and the differences were primarily in terms of textual IQ. 3.3. Participants and measurements Our target participants were Facebook users who were the members of a C2C Facebook “buy and sell” group in Taiwan. This subset of Facebook users was deemed more appropriate for this study, as they would be more familiar with the group setting compared to those Table 1 Experimental design part A. Accuracy High

Low

Ease of understanding Part A Relevance

High Low

High

Low

High

Low

A1 (36) A5 (36)

A2 (35) A6 (32)

A3 (35) A7 (35)

A4 (35) A8 (33)

who were not members of such groups. We used a country setting, Taiwan, because we need to consider the specific text information in the advertisement, such as price (in NT$) and pick up location (in Tainan or Kaohsiung). As Taiwan is becoming ever more internationalized (e.g., there are now many international students, foreign language teachers, international workers, and businessman living in Taiwan), more C2C Facebook “buy and sell” groups are being commonly used to conduct commercial activities. As a result, in Taiwan, there are numerous C2C Facebook “buy and sell” groups which have been actively used. For this reason, Taiwan was chosen as the setting for this study. The dependent variable was consumers' UBI, measured by three items adapted from previous studies [6,7]. To measure participants' impulsiveness, we adapted four items from Wells, Parboteeah, and Valacich [6]. Meanwhile, a series of manipulation check questions regarding text information quality were adapted from a series of past studies [25,26,83]. We used the manipulation check questions not as a survey instrument to conduct statistical analysis, but merely as a manipulation check to ensure our treatments were perceived differently by the participants. Following a previous study [6], all measurement items utilized a nine-point Likert-type scale anchored by 1 (strongly disagree) and 9 (strongly agree). 3.4. Experimental procedures We created an experimental website with a randomization Java Script algorithm code to randomly assign the participants into different experimental treatments, and thus the participants had an equal and independent chance of being placed in any of the treatments. The random assignment was only performed once, at the beginning of the experiment. Based on the results of this the participants were assigned to groups of the same size for both parts A and B of the experiment. For example, if one participant was randomly assigned into group 5 (A5) in part A, then that participant was also assigned into group 5 (B5) in part B. One random assignment was adequate, because our parts A and B focused on different factors. As this study employed an online field experiment, we created sixteen experimental Facebook groups that resembled C2C Facebook “buy and sell” groups (eight Facebook groups for each part of the experiment). We posted our manipulated advertisements using an experimental Facebook account on the sixteen experimental Facebook groups that we created. The experimental procedures consisted of five major steps, as follows. First, the participants were asked to provide their informed consent. Second, we introduced the C2C Facebook “buy and sell” groups and asked the participants to complete a demographic and impulsiveness scale survey. Third, and following the method used in a previous study [6], the participants were asked to read the instructions and a scenario consisting of a specific shopping task. The purpose of the scenario was to create a distinction between impulsive and normal buying [44]. Fourth, the participants were instructed to click the provided link and go to the experimental Facebook groups to see the advertisements. Finally, we asked the participants to complete the manipulation check and UBI survey items. Due to the within subjects design in the two


63

Table 3 Operational definitions of the variables. Experiment

Manipulated independent variables

Conceptual definition

Treatment

Variables operationalization

Supporting literature

Part A

Relevance

The degree to which information is relevant to users depending on the context or their situation.

Low

[25,26,82]

Ease of understanding

The degree to which information is easily understood by users.

Accuracy

The degree to which information is correct, unambiguous, meaningful, believable, and consistent. The degree to which all possible states relevant to users are represented in the stored information. The degree to which information is presented in a manner that is understandable to and interpretable by the user. The degree to which information is up to date.

The text information is less relevant/related/connected to the advertised product. The text information is highly relevant/related/connected to the advertised product. The text information uses abbreviations/jargon/specialist terms. The text information uses easy to understand words/terms. The text information is not that accurate/correct.

Part B

Completeness Format

Currency

“Likes”

Observational learning cue that is represented by the number of “likes” on a Facebook post.

parts of the experiment, and the mixed design in part B, each participant was required to answer the UBI survey three times. Fig. 3 presents a schematic of our experimental procedures. We conducted a pilot test prior to the main experiment. The pilot test was conducted by inviting 27 international graduate students who were members of a C2C Facebook “buy and sell” group in Taiwan. We asked them to participate in our pilot experiment to test the feasibility of the experimental procedures, validity of our experiment treatments, and the functionality of the experimental websites. We received valuable suggestions from the pilot test participants and improved our experimental design based on these. We found C2C Facebook “buy and sell” groups using the Facebook search function and keywords such as “buy and sell Taiwan,” “trade Taiwan,” and “secondhand devices Taiwan.” Twenty groups were found. We thus conducted the main experiment by inviting Facebook users from twenty C2C Facebook “buy and sell” groups, such as “Tainan Market Buy and Sell”; “Taipei: Buy, sell, trade”; “Taichung Swap Shop”; “Kaohsiung Buy & Trade”; and sixteen others. As of March 2015, these groups had from 704 to 23,927 members. To increase the exposure of our study invitation, we also posted the invitation on different Facebook groups, such as “Taiwan international student union,” “Foreign students in Taiwan,” and several other groups with a focus on Taiwan. It was assumed that the members of such groups might also be members of local C2C Facebook “buy and sell” groups. Additionally, we added three filtering questions in the demographic survey: “Are you currently a member of any buy and sell Facebook group in Taiwan?” “Are you one of the researchers' Facebook friends?” and “Have you ever participated in a similar experimental study on C2C Facebook buy and sell groups conducted by the researchers on Facebook before?” 4. Results and findings 4.1. Demographics and random assignment results Of the 316 responses received, a total of 277 usable samples were obtained. We had to drop 39 participants during the data cleaning process for the following reasons: the participants failed the manipulation check questions, were not members of any C2C Facebook “buy and sell” groups in Taiwan, had participated in the pilot experiment, and some outlier issues. Among the 277 participants, there were 142 (51.26%) male and 135 (48.74%) female respondents. Most of the participants were 25–34 years old (61.73%), and just under half (45.85%)

High Low

High Low High Low High Low High Low

High Low High

The text information is highly accurate/correct. The text information is less complete. The text information is more complete. The text information is not well formatted. The text information about the advertised product is well formatted/presented. The date the advertising was posted is less recent (Feb 6, 2015). The date the advertising was posted is more recent (February 13, 2015). 0 “likes” 6 “likes”

[83]

[25,26,82]

[25,26,82] [25,26,82]

[25,26,82]

[84–86]

reported that they often bought a product/service from C2C Facebook “buy and sell” groups. Table 4 shows the demographics data of the participants. After data cleaning, our random assignment of participants led to the sample sizes for each treatment condition shown in Table 1 and Table 2. The number of participants in the eight conditions in part A and part B of the experiment ranged from 32 to 36, and thus can be considered as relatively equal in terms of size. 4.2. Manipulation check, factor analysis, and control variables A manipulation check was performed to verify whether our treatments were perceived differently by the participants. We ran separate one way ANOVA analyses using SPSS 19 to compare the five manipulated factors based on manipulation check questions. The results of the Levene's tests in all five ANOVA analyses were insignificant (p N 0.05), confirming the homogeneity of variances assumption. Meanwhile, with regard to currency and number of “likes,” the participants accurately perceived our treatments. Factor analysis was then conducted to assess the measurement quality of our measurement items (impulsiveness and consumers' UBI). For the dependent variable, we used the average of all (three) consumers' UBI responses. We found that all the measurement items met the validity requirements. The factor loadings and Cronbach's alpha values were all above 0.7, indicating our measurement items had high reliability and validity [87]. Before proceeding to the main and interaction effect data analysis, we tested the possibility that other variables might influence the dependent variable. These control variables were: Facebook friend relationship with the researchers, gender [88], age [89], familiarity with the smart phone technologies, and familiarity with Apple products (e.g., the iPhone 5). For the dependent variable, as noted above, we used the average of all (three) consumers' UBI responses. The results of the ANOVA tests indicated that the control variables did not have any significant effects on the dependent variable. 4.3. Main effects, interaction effects, and findings We ran a generalized linear model analysis of variance (GLM ANOVA) using SPSS 19 to compare whether there were any significant differences among the experimental groups. Since the impulsiveness construct was a continuous variable, we dichotomized impulsiveness into two categories (high and low) using the mean splitting method [90]. As shown in Table 5, the results of the Levene's tests for

64


Table 4 Descriptive analysis results of the demographic results. Demographics

Category

Frequency (N = 277)

(%)

Gender

Male Female 25–34 years old 18–24 years old 35–44 years old 45–54 years old 55–64 years old Under 18 years old Graduate student Administrative Professional PhD student Sales Teacher Undergraduate student Engineer Other Military/Government Senior management Researcher $500–$1000 Less than $500 $1001–$1500 $1501–$2500 More than $2500 South Asian North Asian North American European African Central American Russian Pacific Islander Middle Eastern New Zealander Several times a year Several times a week Several times a month Once a week Once a month Every day Buy a product/service Look at the advertisements Click “like” on advertisements Sell a product/service Comment on advertisements Recommend advertisements to others (e.g., Facebook friends)

142 135 171 48 40 13 3 2 76 42 32 24 24 24 18 12 10 7 5 3 83 77 48 42 27 162 57 18 18 6 6 6 2 1 1 94 63 40 30 26 24 127 79 33 15 14 9

51.26 48.74 61.73 17.33 14.44 4.69 1.08 0.72 27.44 15.16 11.55 8.66 8.66 8.66 6.50 4.33 3.61 2.53 1.81 1.08 29.96 27.80 17.33 15.16 9.75 58.48 20.58 6.50 6.50 2.17 2.17 2.17 0.72 0.36 0.36 33.94 22.74 14.44 10.83 9.39 8.66 45.85 28.52 11.91 5.42 5.05 3.25

Age

Occupation

Income

Nationality

Frequency of visiting a Facebook C2C FB “buy and sell” group

Most frequent activities on C2C FB “buy and sell” group

homogeneity of variances were insignificant in both parts of the experiment, confirming that the equality of variances assumption was met. Table 5 shows that all of our independent variables (main effects) were significant based on the results of the GLM ANOVA tests (F N 4 and p b 0.05). Meanwhile, GLM ANOVA results also showed that there were significant interaction effects of format × currency (F = 7.656, p = 0.006) and completeness × format × impulsiveness (F = 5.232, p = 0.023). Nevertheless, like × impulsiveness was not significant since its F b 4 and p N 0.05 (F = 2.562, p = 0.110). Based on the results, we concluded that H1, H2, and H3 were supported, H4 was partly supported, and H5 was not supported. In addition, we visualize the significant interaction effect of completeness × format × impulsiveness in Fig. 4 and Fig. 5. 5. Discussion and implications In this section, we summarize and discuss five key findings as well as important insights of this study. First, we provide empirical evidence that the six text IQ dimensions (relevance, ease of understanding, accuracy, completeness, format, and currency) play a major role in

stimulating consumers' UBI. Prior studies acknowledged the significance of IQ in an e-commerce context [19,21,91], but these works did not explore the individual IQ dimensions of their relative importance. The main effect results indicate that the six text IQ dimensions had significant and positive effects on consumers' UBI. When seeing the advertisements posted on the C2C Facebook “buy and sell” group, those consumers who were exposed to a high level of textual IQ would tend to have higher UBI compared to those who were exposed to a low level of textual IQ. In light of HIP theory, this means that consumers processed the six IQ dimensions heuristically, thereby increasing their UBI. Our findings thus suggest that in order to increase consumers' UBI, textual information in the advertisements posted on a C2C Facebook “buy and sell” group should be highly relevant, easy to understand, accurate, complete, well formatted, and current. Second, our results indicate that impulsiveness positively affects consumers' UBI. This means that highly impulsive consumers would have higher UBI than less impulsive consumers. Prior impulse buying studies have agreed upon the vital role of individual differences with regard to impulse buying, such as the trait of impulsiveness [1,44,92]. Consistent with most previous studies, our results support the significant effect of consumers' impulsiveness on their UBI within a C2C Facebook “buy and sell” group context. Third, the number of “likes” represents a positive indicator for consumers, and the main effect results show that a high number of these could increase consumers' UBI. However, we found no empirical evidence regarding the interaction effect of consumers' impulsiveness on the relationship between the number of “likes” and consumers' UBI. This insignificant result can possibly be explained by Social Distance Theory (SDT), which concerns the terms, grades, and degree of intimacy of personal and social relations [93]. SDT is consistent with the consumer behavior literature which suggests that reference groups with different social roles, type of contact, and strength of social ties would have different impacts on consumers' purchase decisions [43,48]. Research suggests that impulsive consumers are affected by the presence of peer cohesiveness, which then influences their impulse buying decisions [94]. For example, Luo [94] found that impulse buying could be increased when impulsive consumers shop together with their peers. Likewise, James, Ching, and Luong [95] demonstrated that the presence of peers could increase the likelihood of impulse buying. Therefore, based on SDT and prior evidence abovementioned, it seems rational to argue that the presence of others (in a form of “likes”) with a closer social distance or stronger social ties might possibly encourage impulsive consumers to buy more impulsively. Nevertheless, in this study, we operationalized and controlled the status of the person who clicked “like” on the advertising posts, and this person was fictitious and thus not known to the participants (unknown). Consequently, this unknown person who clicked “like” might not really encourage the impulsive consumers in the current study to increase their UBI. This is possibly because the unknown person might be perceived by the impulsive consumers as having greater social distance. Hence, with regard to the insignificant result, we would argue that these impulsive consumers might be more influenced by “who” has clicked on “like” rather than the “number” of “likes” a post has received. That is, if members of an important reference group have “liked” a post, then this might encourage such consumers to shop more impulsively. However, although this conjecture sounds reasonable, it warrants a further empirical research with regard to investigating the effect of “who” clicked “like” aforementioned. Fourth, there is a significant interaction effect between format and currency, which jointly interact to increase consumers' UBI. Our findings here suggest that consumers tend to show higher UBI when the information they receive is well formatted and more current. The wellformatted information that is included in an advertisement might provide some “heuristic cues” to the consumers regarding “a positive attitude toward the advertisement” or “the sense of professionalism” of the seller. Well-formatted textual information might also be easier to read. Meanwhile, when consumers see that an advertisement has


65

Fig. 3. Experimental procedure.

recently been posted then this stimulates greater UBI compared to less recent post, as other consumers might have not yet seen or shown interest in the focal items. This implies that the combination of high format and high currency is critical to increasing consumers' UBI in this context. Finally, there are significant interaction effects among consumers' impulsiveness, completeness, format, and consumers' UBI. Our results demonstrate that “completeness” and “format” are the only textual IQ dimensions which significantly interact with consumers' impulsiveness. Specifically, we found that when “format” is low and “completeness” is high, high impulsiveness consumers are more affected by high “completeness” of textual information, and thus their UBI is higher than that seen for low impulsiveness consumers (see Fig. 4). Meanwhile, when both format and completeness are high, high impulsiveness consumers are not greatly influenced by high “completeness.”By contrast, low impulsiveness consumers are significantly affected and having a more significant increase in UBI than high impulsive consumers (see Fig. 5) in this context, and this finding is of particular interest. Consumers like to be well-informed about the product in an advertisement, and if an advertisement has more complete textual information then it needs to be properly formatted. Both high and low impulsiveness consumers seem to be influenced by both format and completeness. However, our findings suggest that low impulsiveness consumers' UBI can be more significantly raised by increasing both “completeness” and “format.” In general, the findings of this work shed light on the importance of the textual IQ dimensions of “completeness” and “format” for both types of consumers, and particularly with regard to how to increase impulse buying among low impulsiveness consumers. Overall our results are consistent with LST theory, which

suggest that the interaction between a person's state of mind and personality traits would jointly affect their UBI [96]. 5.1. Theoretical contributions This study contributes to both theory and online impulse buying literature in four ways. First, we empirically investigated the relationship between six different dimensions of textual IQ and consumers' UBI with regard to C2C Facebook “buy and sell” groups. This study, therefore contributes to the literature by providing empirical evidence that “relevance,” “ease of understanding,” “accuracy,” “completeness,” “format,” and “currency” are the salient IQ dimensions affecting consumers' UBI. Despite the fact that IQ has been argued as one of the key factors in an online environment [97], few online impulse buying studies have investigated these dimensions in detail. To the best of our knowledge, our study is one of the very few to comprehensively investigate the textual IQ dimensions in an SC context. Second, our study expands the applicability of LST theory by using it to explain the phenomenon of online impulse buying. There are a large number of impulse buying studies which apply the stimulus–response–organism (SOR) framework as their theoretical foundation [1], and the current work is one of the few that applies LST theory. Consistent with Wells, Parboteeah, and Valacich [6], the results of the current work show that LST theory can better explain online consumers' impulse buying phenomenon by taking the interaction effect of the consumers' state of mind and the personality traits into consideration. Third, our study enhances the view of impulse buying as a heuristic form of cognitive information processing, based on the theoretical

66


Table 5 Main and interaction effects. Dependent variable: urge to buy impulsively Experiment

Manipulated independent variables

Levene's stats (Sig.)

Treatment

N

Mean

Std. Dev

F

p (Sig.)

Part A

Relevance

1.508 (0.102)

Low High Low High Low High Low High Low High Low High Low High Low High Low High LFORM × LCUR LFORM × HCUR HFORM × LCUR HFORM × HCUR LLIKE × LIMP LLIKE × HIMP HLIKE × LIMP HLIKE × HIMP LCOMP × LFORM × LIMP LCOMP × LFORM × HIMP LCOMP × HFORM × LIMP LCOMP × HFORM × HIMP HCOMP × LFORM × LIMP HCOMP × LFORM × HIMP HCOMP × HFORM × LIMP HCOMP × HFORM × HIMP

136 141 135 142 138 139 158 119 272 282 270 284 276 278 277 277 316 238 136 134 140 144 158 119 158 119 84 46 88 54 72 68 72 70

2.95 3.91 3.13 3.73 3.17 3.69 3.03 3.83 3.14 3.68 3.21 3.61 3.16 3.66 3.16 3.66 2.97 3.85 3.16 3.26 3.15 4.07 2.83 3.48 3.09 4.22 2.68 3.31 2.68 3.89 2.80 4.04 3.71 4.17

1.786 1.894 1.895 1.897 1.858 1.911 1.823 1.875 1.730 1.821 1.760 1.823 1.733 1.811 1.683 1.894 1.679 1.772 1.759 1.765 1.711 1.752 1.565 1.692 1.780 1.810 1.555 1.685 1.624 1.674 1.608 1.892 1.772 1.708

18.198

0.000

7.316

0.007

5.276

0.022

12.590

0.000

13.160

0.000

7.466

0.006

11.697

0.001

11.405

0.001

35.496

0.000

7.656

0.006

2.562

0.110

5.232

0.023

Ease of understanding Accuracy Impulsiveness Part B

Completeness

1.003 (0.465)

Format Currency Like Impulsiveness Format × currency

Like × impulsiveness

Completeness × Format × Impulsiveness

perspective of HIP. Prior studies might have put too much emphasize on the SOR framework to explain the role of environmental cues in impulse buying in a more general fashion [1,2]. As a result, they might have overlooked the underlying cognitive information processing

mechanisms that occur when consumers process the focal stimulus or environmental cues while engaging in impulse buying. This study highlights the applicability of HIP in explaining the role of cognitive processes in online impulse buying. While a heuristic process has been argued

Fig. 4. Interaction effect of impulsiveness in the low format treatment.

Fig. 5. Interaction effect of impulsiveness in the high format treatment.


as the driver of impulse buying [2,58,59], few past studies of impulse buying incorporated the notion of HIP. By contrast, and based on HIP, the current study suggests that when engaging in online impulse buying consumers will heuristically process the environmental cues, especially product-related textual information cues with varying degree of quality, and will use those cues as information signaling mechanism. Lastly, we extend the notion of observational learning and the herding effect to an online impulse buying context. This study also provides empirical evidence to the online impulse buying literature that behavioral-based social interactions, such as observational learning and the herding effect are influential in increasing consumer's UBI through the effects of the number of “likes.” 5.2. Practical and managerial implications This study offers valuable insights to people who would like to sell products/services via a C2C Facebook “buy and sell” group. Our results highlight the importance of the IQ dimensions which can increase consumers' UBI. In particular, when posting advertisements on such a group, users (sellers) should focus on the following dimensions: “completeness,” “format,” and “currency,” as in our results these had the biggest effects on increasing consumers' UBI. Although this study focuses on C2C commercial activities, the findings may also be generalized to B2C commercial activities on Facebook. For instance, companies that set up a store on Facebook should make more efforts to provide high textual IQ, as this is likely to increase impulse purchases. In addition, the findings of this study could be used as a basis for a set of valuable design guidelines and principles for use on SNS, and especially Facebook, with regard to enhancing C2C Facebook “buy and sell” groups so that they are more effective and efficient. A recent news report stated that Facebook plans to fully support users' commercial transactions by providing a “Buy” button [98], a move that might greatly increase SC activities on Facebook. Consequently, as more companies and individual users are now engaging in various SC activities, and with even more expected to do so in the future, how to enhance impulse buying in an SC context is an issue of growing importance. 5.3. Limitations and future research As with any empirical studies, the current work is not without its limitations. There are three key limitations that may be addressed by future research. First, nationality and culture might be an issue in this context, and although we did not find any significant differences in consumers' UBI with regard to different nationalities, Asian participants dominated our sample. Future studies may thus replicate our approach and target a more “balanced” sample to overcome this limitation. Second, we did not measure consumers' actual impulse buying behavior but instead incorporated consumers' UBI as a proxy measure. Consumers' UBI measure has been widely accepted in previous impulse buying studies [6,7,11]. However, by only using consumers' UBI, we could not know for sure that the participants would actually buy the focal product when they reported higher UBI. It is thus suggested that future research may use actual impulse buying behavior in addition to consumers' UBI to get more convincing results. Lastly, we used experimental Facebook accounts to generate the number of “likes.” We did this to distinguish the effect between participants' friends clicking “like” and the number of “likes” itself. Although it might be possible that none of the participants' friends would have clicked “like” on the advertising posts, this condition might not accurately represent the real situation. Moreover, as mentioned in the discussion part, the interaction effect of like and impulsiveness on consumers' UBI was not significant, and this might possibly be caused by the effect of “who” clicked “likes.” Therefore, future empirical studies may replicate our study and then carefully consider the effect of who has clicked “likes”

67

in addition to the number of “likes” a post has received on consumers' UBI.

5.4. Conclusion The popularity of SC is growing rapidly, but little is known about consumers' impulse buying in this context, particularly with regard to C2C Facebook “buy and sell” groups. Our study empirically investigated the effect of environmental cues (textual IQ and number of “likes”) and individual differences (impulsiveness) on consumers' UBI. Based on the findings of this work, high textual IQ and a high number of “likes” could generally increase consumers' UBI. We found interaction effect between “format” and “currency,” and also interaction effect among “completeness,” “format,” and the trait of impulsiveness influencing consumers' UBI. However, we found no interaction effect between the number of “likes” and the trait of impulsiveness influencing consumers' UBI. Our results indicate that the textual IQ of the advertisement post along with the number of “likes” are important factors, as these can both increase consumers' UBI within the context of a C2C Facebook “buy and sell” group.

References [1] C. Amos, G.R. Holmes, W.C. Kenson, A meta-analysis of consumer impulse buying, J. Retail. Consum. Serv. 21 (2014) 86–97. [2] S.H. Xiao, M. Nicholson, A multidisplicinary cognitive behavioural framework of impulse buying: a systematic review of literature, International Journal of Management Reviews 15 (2013) 333–356. [3] C.A. Turkyilmaz, S. Erdem, A. Uslu, The Effects of Personality Traits and Website Quality on Online Impulse Buying, International Conference on Strategic Innovative Marketing (IC-SIM), Elsevier, Madrid, Spain, 2015 98–105. [4] T. Verhagen, W.v. Dolen, The influence of online store beliefs on consumer online impulse buying: a model and empirical validation, Information & Management 48 (2011) 320–327. [5] K.N.S.M. Khalifa, System design effects on online impulse buying, Internet Research 22 (2012) 396–425. [6] J.D. Wells, V. Parboteeah, J.S. Valacich, Online impulse buying: understanding the interplay between consumer impulsiveness and website quality, Journal of the Association for Information Systems 12 (2011) 32–56. [7] Y. Liu, H. Li, F. Hu, Websites attributes in urging online impulse purchase: an empirical investigation on consumer perceptions, Decision Support Systems 55 (2013) 829–837. [8] A. Floh, M. Madlberger, The role of atmospheric cues in online impulse-buying behavior, Electronic Commerce Research and Applications 12 (2013) 425–439. [9] S.-W. Lin, L.Y.-S. Lo, Evoking online consumer impulse buying through virtual layout schemese, Behavior and Information Technology (2015) http://dx.doi.org/10.1080/ 0144929X.2015.1056546. [10] T.-P. Liang, E. Turban, Introduction to the special issue social commerce: a research framework for social commerce, International Journal of Electronic Commerce 16 (2012) 5–14. [11] D.V. Parboteeah, J.S. Valacich, J.D. Wells, The influence of website characteristics on a consumer's urge to buy impulsively, Information Systems Research 20 (2009) 60–78. [12] S.E. Beatty, M.E. Ferrell, Impulse buying: modeling its precursors, Journal of Retailing 74 (1998) 169–191. [13] X. Luo, How does shopping with others influence impulsive purchasing? Journal of Consumer Psychology 11 (2005) 51–87. [14] I. Phau, C. Lo, Profiling fashion innovators: a study of self-concept, impulse buying and internet purchase intent, Journal of Fashion Marketing and Management 8 (2004) 399–411. [15] H.J. Eysenck, Cicero and the state–trait theory of anxiety: another case of delayed recognition, American Psychologist 38 (1983) 114–115. [16] X. Zhang, R. Prybutok, C.E. Koh, The role of impulsiveness in a TAM-based online purchasing behavior model, Information Resources Management Journal 19 (2006) 54–68. [17] R. Steyer, M. Schmitt, M. Eid, Latent state–trait theory and research in personality and individual differences, European Journal of Personality 13 (1999) 389–408. [18] W.H. DeLone, E.R. McLeann, Information systems success: the quest for the dependent variable, Information Systems Research 3 (1992) 60–95. [19] W.H. DeLone, E.R. McLeann, Measuring e-commerce success: applying the DeLone & McLean information systems success model, International Journal of Electronic Commerce 9 (2004) 31–47. [20] S. Petter, W.H. DeLone, E.R. McLeann, The past, present, and future of “IS success”, Journal of the Association for Information Systems 13 (2012) 341–362. [21] Y.-S. Wang, Assessing e-commerce systems success: a respecification and validation of DeLone and McLean model of IS success, Information Systems Journal 18 (2008) 529–557.

68


[22] S. Petter, W.H. DeLone, E.R. McLeann, Information systems success: the quest for the independent variables, Journal of Management Information Systems 29 (2013) 7–61. [23] W.H. DeLone, E.R. McLeann, The delone and McLean model of information systems success: a ten-year update, Journal of Management Information Systems 19 (2003) 9–30. [24] J. Aldás-Manzano, R. Currás-Pérez, S. Sanz-Blas, Online information quality as determinant of perceived risk reduction in e-commerce: an application to apparel virtual stores, International Journal of Internet Marketing and Advertising 6 (2011) 352–372. [25] N. Au, E.W.T. Ngai, T.C.E. Cheng, Extending the understanding of the end user information systems satisfaction formation: an equitable needs fulfillment approach, Management Information Systems Quarterly 32 (2008) 43–66. [26] R.R. Nelson, P.A. Todd, B.H. Wixom, Antecedents of information and system quality: an empirical examination within the context of data warehousing, Journal of Management Information Systems 21 (2005) 199–235. [27] R.Y. Wang, D.M. Strong, Beyond accuracy: what data quality means to data consumers, Journal of Management Information Systems 12 (1996) 5–34. [28] Z. Jiang, I. Benbasat, The effects of presentation formats and task complexity on online consumers' product understanding, Management Information Systems Quarterly 31 (2007) 475–500. [29] P. Baggett, Understanding Visual and Verbal Messages in Knowledge Acquisition from Text and Pictures, Elsevier Science Publishers, Amsterdam, 1989. [30] J. Miyoung, C. Jiyoung, Effects of picture presentations on costomers' behavioral intention on the web, Journal of Travel & Tourism Marketing 17 (2008) 193–204. [31] T. Adelaar, S. Chang, K.M. Lancendorfer, B. Lee, M. Morimoto, Effects of media formats on emotions and impulse buying intent, Journal of Information Technology 18 (2003) 247–266. [32] A. Molla, P.S. Licker, E-commerce systems success: an attempt to extend and respecify the DeLone and McLean model of IS success, Journal of Electronic Commerce Research 2 (2001) 1–11. [33] N. Hajli, J. Sims, Social commerce: the transfer of power from sellers to buyers, Technology Forecasting and Social Change 94 (2015) 350–358. [34] N. Hajli, Social commerce constructs and consumer's intention to buy, International Journal of Information Management 35 (2015) 183–191. [35] K.Z.K. Zhang, B. Hu, S.J. Zhao, How Online Social Interactions Affect Consumers' Impulse Purchase on Group Shopping Websites, 18th Pacific Asia Conference on Information Systems (PACIS 2014), Chengdu, China, 2014. [36] N. Hajli, A research framework for social commerce adoption, Information Management Computer Security 21 (2013) 144–154. [37] P. Do-Hyung, L. Jumin, H. Ingoo, The effect of online consumer reviews on consumer purchase intention: the moderating role of involvement, International Journal of Electronic Commerce 11 (2007) 125–148. [38] Z. Huang, M. Benyoucef, From e-commerce to social commerce: a close look at design features, Electronic Commerce Research and Applications 12 (2013) 246–259. [39] Facebook, What Are Facebook groups? Facebook Inc., 2015 [40] A. Dimoka, Y. Hong, P.A. Pavlou, On product uncertainty in online markets: theory and evidence, Management Information Systems Quarterly 36 (2012) 395–426. [41] E. Daliri, S. Rezaei, W. Khairuzzaman, W. Ismail, Online social shopping: the impact of attitude, costumer information quality, effectiveness of information content and perceived social presence, International Journal of Bussiness Environment 6 (2014) 426–450. [42] J.A. Lee, J.J. Kacen, Cultural influences on consumer satisfaction with impulse and planned purchase decisions, Journal of Bussiness Research 61 (2008) 265–272. [43] J.P. Peter, J.C. Olson, Consumer Behavior & Marketing Strategy, McGraw-Hill/Irwin, New York, NY, 2010. [44] D.W. Rook, R.J. Fisher, Normative influences on impulse buying, Journal of Consumer Research 22 (1995) 305–313. [45] F. Piron, Defining impulse purchasing, Advances in Consumer Research 18 (1991) 509–514. [46] D.W. Rook, M. Gardner, In the mood: impulse buying's affective antecedents, Research Consumer Behavior 6 (1993) 1–28. [47] C.J. Cobb, W.D. Hoyer, Planned versus impulse purchase behavior, Journal of Retailing 62 (1986) 67–81. [48] D.I. Hawkins, D.L. Mothersbaugh, Consumer Behavior: Building Marketing Strategy, McGraw-Hill/Irwin, New York, NY, 2010. [49] J.U. McNeal, An Introduction to Consumer Behavior, John Wiley & Sons, Inc., New York, NY, 1973. [50] G. Bayley, C. Nancarrow, Impulse purchasing: a qualitative exploration of the phenomenon, Qualitative Market Research: An International Journal 1 (1998) 99–114. [51] S. Youn, R.J. Faber, Impulse buying: its relation to personality traits and cues, Advances in Consumer Research 27 (2000) 179–185. [52] J.E. Burroughs, Product symbolism, self-meaning, and holistic matching: the role of information processing in impulsive buying, Advances in Consumer Research 23 (1996) 463–469. [53] Y. Chen, L. Zhang, Influential factors for online impulse buying in China: a model and its empirical analysis, International Management Review 11 (2015) 57–69. [54] U.M. Dholakia, Temptation and resistance: an integrated model of consumption impulse formation and enactment, Psychology & Marketing 17 (2000) 955–982. [55] P.W. Weinberg, W. Gottwald, Impulsive consumer buying as a result of emotions, Journal Bussiness Research 10 (1982) 43–57. [56] S.A. Jeffrey, R. Hodge, Factors influence impulse buying during an online purchase, Electronic Commerce Research 7 (2007) 367–379. [57] M.R. Solomon, Consumer Behavior: Buying, Having, and Being, Ninth ed. Prentice Hall, Upper Saddle River, NJ, 2011. [58] J.E. Michael, J.S. William, A. Pandit, Marketing, Tata McGraw-Hill, New Delhi, 2010.

[59] B. Verplanken, A. Sato, The psychology of impulse buying: an integrative selfregulation approach, Journal of Consumer Policy 34 (2011) 197–210. [60] S. Chaiken, Y. Trope, Dual-Process Theories in Social Psychology, Guilford Press, New York, 1999. [61] S. Chaiken, Heuristic versus systematic information processing and the use of source versus message cues in persuasion, Journal of Personality & Social Psychology 39 (1980) 752–766. [62] S. Chaiken, A. Liberman, A.H. Eagly, Heuristic and Systematic Information Processing Within and Beyond the Persuasion Context, in: J.S. Uleman, J.A. Bargh (Eds.), Unintended Thought, The Guilford Press, New York, NY 1989, pp. 212–252. [63] C.W. Trumbo, Heuristic–systematic information processing and risk judgment, Risk Analysis 19 (1999) 391–400. [64] H.A. Robinson, A. Wysocka, C. Hand, Internet advertising effectiveness, International Journal of Advertising 26 (2007) 527–541. [65] G. Baltas, Determinants of internet advertising effectiveness: an empirical study, International Journal of Market Research 45 (2003) 505–514. [66] P. Katerattnakul, K. Siau, Information quality in internet commerce design, in: M.G. Piattini, C. Calero, M. Genero (Eds.), Information and Database Quality, Springer 2002, pp. 45–56. [67] M.T. Elliott, P.S. Speck, Factors that affect attitude toward a retail website, Journal of Marketing Theory and Practice 13 (2005) 40–51. [68] B. Posetti, Clicking through endless seas: understanding user experience in the design of journalistic websites, in: A. Marcus (Ed.) Design, User Experience, and Usability: Interactive Experience Design, Springer International Publishing, Switzerland, pp. 82–93. [69] M. Spence, Time and communication in economic and social interaction, Quarterly Journal of Economics 87 (1973) 651–660. [70] J.E. Urbany, P.R. Dickson, W.L. Wilkie, Buyer uncertainty and information search, Journal of Consumer Research 16 (1989) 208–215. [71] A.R. Rao, K.B. Monroe, The effect of price, brand name, and store name on buyers' perceptions of product quality: an integrated review, Journal of Marketing Research 26 (1989) 351–357. [72] A.W. Gatian, Is user satisfaction a valid measure of system effectiveness? Information & Management 26 (1994) 119–131. [73] B.H. Wixom, H.J. Watson, An empirical investigation of the factors affecting data warehousing success, Management Information Systems Quarterly 25 (2001) 17–41. [74] A.M. Aladwani, P.C. Palvia, Developing and validating an instrument for measuring user-perceived web quality, Information & Management 39 (2002) 467–476. [75] Y. Chen, Q. Wang, J. Xie, Online social interactions: a natural experiment on word of mouth versus observational learning, Journal of Marketing Research 48 (2011) 238–254. [76] J. Zhang, P. Liu, Rational herding in microloan markets, Management Science 58 (2012) 892–912. [77] X. Li, L. Wu, Herding and social media word-of-mouth: evidence from groupon, Available at , DOI http://dx.doi.org/10.2139/ssrn.2264411 (2014) (SSRN: http:// ssrn.com/abstract=2264411). [78] S. Bikhcandani, S. Sharma, Herd behavior in financial markets, IMF Staff Papers 47 (2001) 279–310. [79] A. Banerjee, A simple model of herd behavior, Quarterly Journal of Economics 107 (1992) 797–817. [80] M. Shuttleworth, Counterbalanced measures design, https://explorable.com/ counterbalanced-measures-design2009. [81] P. Xue, L. Chen, R. Santhanam, Will video be the next generation of e-commerce product reviews? Presentation format and the role of product type, Decision Support Systems 73 (2015) 85–96. [82] Y. Wand, R.Y. Wang, Anchoring data quality in ontological dimensions, Communication of the ACM 39 (1996) 86–96. [83] V. McKinney, K. Yoon, F.M. Zahedi, The measurement of web-customer satisfaction: an expectation and disconfirmation approach, Information Systems Research 13 (2002) 296–315. [84] M. Deutsch, H.B. Gerard, A study of normative and informational social influences upon individual judgment, Journal of Abnorm Social Psychology 51 (1955) 629–636. [85] R.E. Burnkrant, A. Cousineau, Informational and normative social influence in buyer behavior, Journal of Consumer Research 2 (1975) 206–215. [86] M.K.O. Lee, N. Shi, C.M.K. Cheung, K.H. Lim, C.L. Sia, Consumer's decision to shop online: the moderating role of positive informational social influence, Information & Management 48 (2011) 185–191. [87] J.R. Hair, et al., Multivariate Data Analysis, Pearson Prentice Hall, New Jersey, 2010. [88] D.W. Rook, S.J. Hoch, Consuming impulses, Advances in Consumer Res. 12 (1985) 23–27. [89] J.J. Kacen, J.A. Lee, The influence of culture on consumer impulsive buying behavior, Journal of Consumer Psychology 12 (2002) 163–176. [90] J. Joseph, F. Hair, G.T.M. Hult, C. Ringle, M. Sarstedt, A Primer on Partial Least Squares Structural Equation Modelling (PLS-SEM), SAGE Publications, Inc, Thousand Oaks, CA, 2014. [91] D.J. Kim, D.L. Ferrin, H.R. Rao, A trust-based consumer decision-making model in electronic commerce: the role of trust, perceived risk, and their antecedents, Decision Support Systems 44 (2008) 544–564. [92] P. Sharma, B. Sivakumaran, R. Marshall, Impulse buying and variety seeking: a traitcorrelates perspective, Journal of Bussiness Research 63 (2010) 276–283. [93] C. Wark, J.F. Galliher, Emory bogardus and the origins of the social distance scale, The American Sociologist 38 (2007) 383–395. [94] X. Luo, Group dynamics of impulse buying: an extended social facilitation perspective, Advances in Consumer Research 31 (2004) 431.

J.V. Chen et al. / Decision Support Systems 83 (2016) 57–69 [95] C. James, G.S. Ching, T.-H. Luong, Impulse buying behavior of Vietnamese consumers in supermarket setting, International Journal of Research Studies in Management 3 (2014) 33–50. [96] J.D. Well, J.S. Valacich, T.J. Hess, What signal are you sending? How website quality influences perceptions of product quality and purchase intention, Management Information Systems Quarterly 35 (2011) 373–396. [97] F. Rahimnia, J.F. Hassanzadeh, The impact of website content dimension and e-trust on e-marketing effectiveness: the case of Iranian commercial saffron corporations, Information & Management 50 (2013) 240–247. [98] J. D'Onfro, Facebook is Diving Full-Force into Shopping, 2015 (Business Insider). Jengchung Victor Chen ([email protected]) is a professor and director at the Institute of International Management, College of Management, National Cheng Kung University, Taiwan. He has a PhD in CIS from the University of Hawaii, USA. He has published over 50 papers in refereed journals, including Information & Management, Decision Support Systems, European Journal of Information Systems, Computers in Human Behavior, Journal of Computer Information Systems, International Journal of Mobile Communications, Journal of Systems and Software, Industrial Management and Data Systems, and Computer Standards & Interfaces.

69

Bo-Chiuan Su ([email protected]) is an associate professor at the Department of Information Management, School of Management, National Dong Hwa University, Taiwan. He has a PhD in Business Administration with a specialization in information management from the University of Connecticut, USA. His research has appeared in International Journal of Electronic Commerce, Decision Support Systems, International Journal of Mobile Communications, International Journal of Business and Systems Research, International Journal of Management Theory and Practices, Journal of Information Management, Journal of e-Business, Information Management for Buddhist Libraries, Journal of International Management Studies, Journal of Global Business Management, Marketing Review, Electronic Commerce Studies, and others. Andree E. Widjaja ([email protected]) received his PhDat the Institute of International Management, National Cheng Kung University, Taiwan. He is also affiliated with the Faculty of Computer Science, Department of Information Systems, Pelita Harapan University, Indonesia. His research has appeared in International Journal of Human-Computer Interaction, Information Systems Frontier, Enterprise Information Systems, and several international conference proceedings. He had received the award for the best track paper in Information Technology of Decision Sciences Institute (DSI) 43rd Annual Meeting in 2012.