Information & Management 56 (2019) 236–248

Contents lists available at ScienceDirect

Information & Management

journal homepage: www.elsevier.com/locate/im

How do product recommendations affect impulse buying? An empirical T

study on WeChat social commerce
⁎
Yanhong Chena, Yaobin Lua, Bin Wangb, Zhao Pana,
a
School of Management, Huazhong University of Science and Technology, China
b
College of Business and Entrepreneurship, University of Texas Rio Grande Valley, Edinburg, TX 78539, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Social commerce is creating increasing opportunities for consumers to access product recommendations.
Product recommendations Evidence from practical and academic literature shows that product recommendations may lead to impulse
Signaling theory buying, yet relatively limited research exists on this topic. Based on the signaling theory, this research proposes a
Urge to buy impulsively model to examine how product recommendations on social media affect a user’s urge to buy impulsively. Our
Cognitive trust
results indicate that urge to buy impulsively is determined by affective trust in the recommender and affection
Affective trust
Product affection
toward the recommended product, which are influenced by both recommender-related signals (information
quality and similarity) and product-related signals (vicarious expression and aesthetic appeal).

1. Introduction buying, which is a sudden, powerful, and persistent urge to purchase

something immediately when driven by stimuli encountered within the
As a form of commerce facilitated by social media through the environment [89]. The impulse buying literature indicates that the
convergence of online and offline channels [81], social commerce has awareness effect of recommendations often inspires individuals to buy
undergone rapid changes and development in recent years. Social products that they otherwise would not have noticed [51,76,88]. Ac-
commerce focuses on consumer-driven commerce and social interac- cording to Stern’s four types of impulse buying [76], users’ buying
tion, which differentiate it from traditional e-commerce that focuses on behavior based on the recommendations of others can be viewed as
product-centered commerce with firm-provided information. In this suggestive impulse buying, which occurs when an individual sees a
research, we examine product recommendations on social media such product first and visualizes a need for it [84,88]. Lo et al. [51] also
as friends’ recommendations on social networking sites (SNS) and pointed out that peer recommendations that fulfill consumers’ ex-
bloggers’ recommendations on blogs. As one component of social pectations can be seen as motivational factors that evoke online impulse
commerce [26], product recommendations on social media can be buying. Therefore, we can reasonably infer that product recommenda-
much more influential compared with marketing messages created by tions generate or influence the needs of consumers and stimulate their
marketers. According to a 2016 McKinsey report, half of all online online impulse buying [51,66]. However, the extant literature on pro-
consumers rely on social media to obtain recommendations and make duct recommendations mainly focuses on planned purchase behavior,
their purchase decisions [60]. This trend is likely to strengthen as and few studies have examined impulse buying behavior [32]. For in-
consumers become more comfortable with social media and social stance, Hsu et al. [31] investigated the effect of bloggers’ re-
commerce. commendations on consumers’ online purchase intentions. Therefore, it
Product recommendations on social media affect both planned and is important for academics and practitioners to explore how product
impulse buying [31,74]. In the former case, social media users may recommendations on social media facilitate impulse buying.
actively seek out information on products that they intend to purchase, The current study investigates how product recommendations in
obtain other users’ recommendations and reviews, and make their de- WeChat social commerce affect consumers’ impulse buying behavior.
cision based on such information. In the latter case, users may en- As the most popular social media app in China, WeChat supports
counter recommendations for products or services that they did not WeChat store and WeChat payment that allow users to directly buy
plan to purchase while browsing social media or interacting with other products and services based on product recommendations made by
users. Product recommendations in this case can lead to impulse WeChat official accounts, thus providing a new marketing channel for

⁎
Corresponding author.
E-mail addresses: yachne@hust.edu.cn (Y. Chen), luyb@mail.hust.edu.cn (Y. Lu), bin.wang@utrgv.edu (B. Wang), victola.pz@gmail.com (Z. Pan).

https://doi.org/10.1016/j.im.2018.09.002
Received 29 June 2017; Received in revised form 22 July 2018; Accepted 4 September 2018
Available online 08 September 2018
0378-7206/ © 2018 Elsevier B.V. All rights reserved.
Y. Chen et al. Information & Management 56 (2019) 236–248

social commerce while promoting impulse purchases at the same time background and the relevant literature. Next, we present the research
[60,77]. WeChat official accounts provide an important way to obtain model and hypotheses, and these are followed by a discussion on the
information for users, therefore users browse official accounts mainly methodology and data collection. We conclude with our findings, im-
for getting information rather than searching or buying specific pro- plications, and limitations and suggestions for future research.
ducts [48]. That is to say, compared with a directed search, in which
the consumer has a specific purchase goal [36,61], consumer buying 2. WeChat social commerce and product recommendations
behavior in WeChat is associated with an exploratory search in which
consumers are less focused and no purchase is planned. Previous re- 2.1. WeChat social commerce
search shows that a consumer's exploratory search is often driven by
stimuli and often results in an impulse purchase [61]. Therefore, the Launched in January 2011 by Tencent, WeChat (or Weixin in
product recommendations may serve as the stimulus in WeChat that Chinese) is the largest social media app in China and is also widely used
leads to the consumer buying the recommended product impulsively. by Chinese populations outside of China. WeChat boasted 938 million
Moreover, when exposed to product recommendations by an official active monthly users in March 2017 [78], which accounted for more
account, followers often have little prior knowledge on the re- than half of the population in China and has rapidly become an integral
commended products. The official account possesses more information part of many Chinese people’s everyday lives [50]. WeChat official
than the followers do on the quality of the recommended product and accounts, a key functional module of WeChat, has become very popular.
may exaggerate the advantages of the product. In addition, social According to a survey reported by Technode [77], approximately 80%
commerce transactions are similar to e-commerce ones in that they are of WeChat users follow official accounts.
not conducted face to face, and this may increase the uncertainty of the WeChat official accounts are similar to Facebook pages and blogs
transaction [3]. where social media users can follow the accounts and receive pushed
The signaling theory provides a framework for understanding how articles. Followers can read information, opinions, reviews, or ideas
informational cues (i.e., signals) are used by one party to convey hidden posted by official accounts and interact with them by writing comments
or limited quality information to another party to facilitate a purchase on the articles. WeChat official account operators focus on pushing
or exchange. As past online impulse buying research has shown, con- content to build social connections and establish relationships with
sumers might process informational cues as an “information signaling their followers. Based on these social connections, WeChat official ac-
mechanism” to facilitate their decision-making [10]. Because a WeChat count operators can promote products or services through product re-
official account interacts with its followers mainly through its pushed commendations published to their followers, thus providing a new
posts, these posts may convey informational cues (i.e., signals) that channel for social commerce [77]. In the rest of the paper, we will use
establish relationships with account followers and support followers in “recommender” and “operator” interchangeably as we examine WeChat
their buying decisions. Therefore, we use the signaling theory to ex- official account operators that provide product or service re-
amine how WeChat official accounts use signals to differentiate them- commendations.
selves from other accounts and convey information about the quality of If the followers are interested in the recommended products, they
their recommended products to facilitate their followers' impulse can access the WeChat store and check out through the WeChat pay-
buying behavior. ment service. Followers can directly buy products and services without
This study makes the following contributions to the extant litera- leaving the WeChat app, which may easily stimulate users’ impulse
ture. First, our research examines how relationships developed between purchase desire when they are exposed to product recommendations
a WeChat official account and its followers in social media (i.e., the [19,79]. WeChat official accounts have been able to generate a sig-
social aspect) combined with product recommendations (i.e., the nificant amount of revenue. For example, one well-known account Rice-
commercial aspect) influence impulse buying. The extant impulse Cake Mama that focuses on child care reached 10 million followers and
buying literature mainly focuses on e-commerce and how impulse 80 million yuan in revenue in November 2016 [63].
buying may be influenced by online store characteristics such as web-
site quality [84], store belief [79], and website atmosphere [19]. Our 2.2. Product recommendations on WeChat official accounts
study examines the impact of informational cues present in person-to-
person interactions on impulse buying. The combination of social and There are two types of posts published by WeChat official accounts:
commerce factors allows us to empirically test our research model non-product-related posts and product recommendation posts, as illu-
unique to social commerce. Second, our study applies the signaling strated in Appendix A. The non-product-related posts are often the
theory to the understanding of how information cues trigger impulse reasons users follow the official account because they present valuable
buying in social commerce. Using the signaling theory, we investigate information. These posts often focus on specific topics, such as child-
how informational cues (i.e., signals) influence online buyers’ percep- care, beauty, or travel, in which the official account specializes. For
tions of trust, product attitude, and urge to buy impulsively example, a popular WeChat official account “Loujisiwei” publishes
[5,56,57,85]. Previous research has applied the stimulus–re- stories about historical misconceptions and comments on social events.
sponse–organism (SOR) framework to investigate the role of extrinsic In addition to pushing content based on specific topics, WeChat official
cues on impulse buying [9,66]. The current study is one of the few that accounts can also push content which designed to offer enjoyment and
applies the signaling theory. Third, our study highlights the pivotal role encourage user participation and facilitate interaction [54]. Unlike blog
of affect, especially the role of affective trust, in the impulse buying posts that focus on personal feelings, ideas, and opinions related to
process. Few studies have examined the role trust plays in impulsive specific events in everyday life [31,33], non-product-related posts by
buying [86]. An exception is Wu et al. [86], who pointed out that trust WeChat official accounts are original, informative, and specific, and the
belief in shopping websites is critical to online impulse buying. How- posts are sent from official accounts only once per day.
ever, Wu et al.’s [84] study does not distinguish between cognitive trust Product recommendations by WeChat official accounts are similar
and affective trust. The current research shows that affective trust is a to sponsored recommendations on blogs [34,52], which are articles
central construct for impulse buying in social commerce. Finally, the written by bloggers who receive payments or incentives from spon-
findings of this study provide valuable guidance for social commerce soring marketers to provide persuasive information about products or
recommenders on how to build trust and effectively present re- services on their personal blogs [52]. Lu et al. [52] pointed out that
commendation contents to facilitate consumers’ purchases of the re- sponsored recommendations posted by bloggers are a form of online
commended products. consumer review for readers when they evaluate products for purchase
In the remainder of this paper, we first introduce the research decisions. Product recommendation posts by WeChat official accounts

237
Y. Chen et al. Information & Management 56 (2019) 236–248

are also consumer-generated sponsored recommendation posts with 3.2. Signaling theory
direct purchase links. These recommendations contain detailed in-
formation about the recommended products such as product features, The signaling theory emerged from information economics studies
pictures, product experiences, and the advantages and benefits of the to explain scenarios with asymmetric information, i.e., when two par-
products. Such information is often presented as storytelling with spe- ties (e.g., sellers and buyers) have access to different information in a
cific usage experience. WeChat official accounts support different fonts, transaction [15,73,75]. According to the signaling theory, one party can
colors, layouts, and graphics in the recommendation posts. Content can invest in signals that reveal some relevant and meaningful information
also include audio and video. Moreover, an official account can publish to the other party, thus reducing uncertainty and facilitating a purchase
long articles, thus allowing content that is more comprehensive. or exchange [15].
In addition, product recommendations by WeChat official accounts Signals are informational cues sent out by one party to another to
are different from general product reviews and recommendations lead to desired outcomes [57]. We propose that official accounts may
posted on a product’s or seller’s website by peer consumers. Product use signals to influence followers’ perceptions of trust and product at-
recommendations by an official account target specific consumers and titude to induce impulse buying behavior. Because an official account
the recommender is often visible, whereas peer consumers on a pro- interacts with its followers mainly through its posts, it may convey
duct’s or seller’s website are usually anonymous or somewhat ambig- signals through its non-product-related and product recommendation
uous [6]. Most important of all, recommendations generated by peers posts. Li et al. [46] provided a typology for classifying Internet auction
on traditional websites are considered to be non-commercial con- features such as quality signals into product- and seller-related char-
versations because these recommendations are independent from a acteristics. Similarly, we categorize the signals conveyed by the posts as
marketer’s persuasive intent [74]. However, recommendations posted product- and recommender-related signals. The main objective of the
by WeChat official accounts and sponsored posts on blogs are com- official accounts is to establish relationships with their followers
mercial because they aim to review and promote products and services through non-product-related posts. Based on these relationships, the
[34,52]. Table 1 compares key differences among product re- official account recommends products to the followers. Therefore, an
commendations in WeChat official accounts, sponsored recommenda- official account may rely on informational cues in non-product-related
tions on blogs, and peer recommendations on traditional websites. posts (e.g., similarity) to stimulate trust and develop affective bonds
with its followers. The role of signals conveyed by non–product-related
posts can reflect the trustworthiness of the recommender. In contrast,
3. Theoretical framework and literature review signals conveyed by product recommendation posts (e.g., product de-
scription) help consumers evaluate the quality of the recommended
3.1. Impulse buying products [46]. Therefore, we propose that signals conveyed by non-
product-related posts are recommender-related signals and those con-
Rook [70] defined impulse buying as occurring “when a consumer veyed by product recommendation posts are product-related signals.
experiences a sudden, often powerful and persistent urge to buy something
immediately.” The process of impulse buying is characterized by a lack 3.2.1. Recommender-related signals conveyed by non–product-related posts
of cognitive deliberation and being dominated by emotions [1,9,79]. Signals conveyed by non-product-related posts are independent
During the impulse buying process, consumers often experience posi- from the recommended products and mainly serve the purpose of in-
tive affective reactions and reduced information processing and cog- creasing positive perceptions such as trust in the recommender [57].
nitive reactions [9]. For example, Parboteeah et al. [66] found that Previous research has shown that signals can enhance a customer’s trust
consumers are more likely to feel the urge to buy impulsively if they in transacting with the other party, thereby promoting purchase be-
perceive enjoyment when interacting with the shopping environment. havior [5,65]. For example, Benlian and Hess [5] used the signaling
In addition, cues from the environment such as website quality [84], theory to conceptualize the signaling role of IT features and their effects
store belief [79], and website atmosphere [19] influence impulse on trust toward other members and participation in online commu-
buying. In social commerce, consumers’ increasing exposure to in- nities. Similarly, trust can develop through signals conveyed by non-
formation and social interactions on social media makes the influences product-related posts, thus we consider information quality and simi-
of the purchase stimuli (e.g., social factors) stronger, thus leading to larity as recommender-related signals.
impulse buying [32,88]. First, participants follow the account mainly for valuable informa-
According to Luo [55], it is difficult to measure actual impulse tion [11]. Therefore, the quality of information provided by the official
buying in controlled settings because participants' responses or beha- account is important for the success of these accounts [72]. The in-
viors are often biased when they are asked to recall their impulse formation quality of the non-product-related posts may signal the un-
buying behavior. Previous research on impulse buying has used the observable trustworthiness of the recommender because trust arises
“urge to buy impulsively” as a surrogate measure for the actual impulse from knowledge accumulated based on the observed information
buying behavior [10,66,84]. The urge to buy impulsively is “a state of quality of these posts [39]. Second, similarity is an important cue upon
desire that is experienced upon encountering an object in the environment” which people build trust [58]. Non-product-related posts usually focus
[4]. Therefore, we also use the urge to buy impulsively as the surrogate on specific topics such as travel, baby, and beauty, which will leave the
of actual impulse buying in this study. followers with the impression that they share the same interest as the
official account owner [90]. Users usually follow an account because of

Table 1
Comparison of product recommendations on different platforms.
Characteristic WeChat official accounts Blogs (e.g., Weibo) Traditional websites (e.g., Taobao, JD)

Author/creator of recommendation contents Operators of the WeChat official accounts Bloggers Other consumers/users
Media richness Text, pictures, audio, and video Text, pictures, audio, and video Predominantly text based
(multimedia) (multimedia)
Self-presentation of the recommender High High Low
Visibility of the recommender Yes Yes No
Commercial conversation Yes Yes No

238
Y. Chen et al. Information & Management 56 (2019) 236–248

common interests or experience with the account owner. For example, 3.2.2. Product-related signals conveyed by product recommendation posts
new parents follow WeChat accounts that discuss how to raise a baby Product recommendation posts reveal direct information about the
because the account owners may also have babies and are interested in recommended product and mainly focus on conveying positive attri-
infant-related issues. In addition, these original non-product-related butes of the product. According to the signaling theory, sellers can use
posts are often presented as personal stories or experiences [69]. The signals to convey information about unobservable product quality to
operators of official accounts may reveal information about themselves buyers [15]. These signals or cues may be the brand name, price,
in non-product-related posts and their followers may further determine product description, or warranty [56,57,85]. In WeChat, the re-
their degree of similarity with the account owners by reading the commendation content is typically presented as a narrative or story
published articles. featuring the recommender’s own life experiences and attitudes and
Previous studies suggest that trust is important in users’ processing often includes pictures and videos [69]. The use of storytelling and
of product recommendations because of the high uncertainty and lack multimedia posts enables the recommender to directly reveal more
of face-to-face interactions in social media [26,74]. According to Smith vivid and visual information about the product and convey product
et al. [74], consumers must trust the recommender for recommenda- quality information. Thus, in this study, we examine the presentation
tions to be effective. Individuals who read product recommendations style of the recommendation content (vicarious expression and aes-
from credible persons can be easily persuaded to buy the product. In thetic appeal) as the signals that convey product and service quality
addition, the online impulse buying literature, however limited, shows information to the recipients. Vicarious expression refers to the degree
that trust is more important than risk. For example, Vonkeman et al. to which recommendation contents convey vivid experiences of a pro-
[80] examined the effects of perceived product risk and product af- duct that can be felt by the readers through stories and narratives
fection on online impulse buying and found that product risk does not [6,45]. Aesthetic appeal is the use of fonts, colors, layout, and graphics
substantially decrease the urge to buy impulsively. By contrast, Wu to present products in the product recommendation posts to attract
et al. [86] investigated the influences of the flow experience, tech- consumers [37]. These two variables have proven to be effective ele-
nology use features, and trust belief on online impulse buying and ments to present product information [6]. Moreover, these two vari-
found that trust has a positive impact on online impulse buying. Based ables are mood-relevant cues that can influence consumers’ emotions
on these literatures, we only consider trust in this study. [6,66], which is expected to play a crucial role in emotional and un-
In WeChat social commerce, because trust in the recommender is planned buying situations.
mainly interpersonal, we focus on two types of trust based on the in- We believe that presentation style sends a strong signal to con-
terpersonal trust literature: cognitive-based trust and affective-based sumers about product quality and that this informational cue will po-
trust [58]. Cognitive-based trust is based on a cognitive evaluation of an sitively influence consumers’ attitudes toward the product [17]. For
individual, group, or organization. It is “grounded in individual beliefs example, Wells et al. [85] identified the visual appeal of a website as a
about peer reliability and dependability, as well as competence” [83]. This potential signal with a strong effect on perceived product quality.
form of trust is rational and relies on objective information such as Previous studies have suggested that there are three components of
external information and experience to determine whether the other product attitudes: cognitive, affective, and conation [64,87]. The cog-
party in the relationship can be trusted [39]. Cognitive trust can cap- nitive component is based on the product information available to the
ture one’s confidence in another’s attributes under specific circum- customer (e.g., perceived product quality) [85], whereas the affective
stances [58], and it similarly depends on beliefs in the reliability and component represents the emotion or affective reaction of the consumer
dependence of others. In contrast, affective-based trust is “grounded in toward a product. According to the impulse buying literature, con-
reciprocal interpersonal care and concern or emotional bonds” [83]. This sumers’ impulse buying behavior is often characterized by high emo-
type of trust is emotionally based and is characterized by affective at- tional activation and this behavior mainly occurs when individuals
tachments from the basis for caring and benevolent actions [39]. Both experience strong positive emotions. Therefore, we choose product af-
cognitive trust and affective trust have received scholarly attention but fection as representing consumers' attitudes toward the recommended
they have not been extensively investigated in social media. It is likely product. Product affection here is defined as the extent to which the
that both cognitive trust and affective trust exist in the relationship product recommendation posts have the potential to trigger consumers’
between a WeChat official account and its followers [7]. positive affection toward the product [80,87].

Fig. 1. The Research Model.

239
Y. Chen et al. Information & Management 56 (2019) 236–248

4. Research model and hypothesis development 4.2. Product-related signals and product affection

Based on the above discussion, Fig. 1 summarizes our research 4.2.1. Vicarious expression and product affection
model on the impact of product recommendations on impulse buying Deighton et al. [16] pointed out that stories have an ability to
behavior in social commerce. Drawing on the signaling theory, we ex- generate empathy among users, thus creating a vicarious expression.
amine how recommender-related signals (i.e., information quality and According to the social learning theory, readers can learn an author’s
similarity) and product-related signals (i.e., vicarious expression and vivid feelings by reading a high level of vicarious expression, which
aesthetic appeal) affect a follower’s trust in recommender, affective ultimately influences his or her perception of pleasure through direct
attitude toward the recommended product, and ultimately the urge to emotional “contagion” [6,45].
buy impulsively. Prior research suggests that an author’s vivid expression of his/her
experiences can stimulate affective responses and facilitate a con-
sumer’s decision-making process [45]. Higher perceived vicarious ex-
4.1. Recommender-related signals and trust in recommender pression makes it easier for readers to imagine what it will be like to use
the product and generate similar feelings in their minds, subsequently
4.1.1. Information quality and cognitive trust in recommender leading to positive emotional effects [6]. Thus, given that product re-
Information quality refers to up-to-date, accurate, relevant, and commendations made by an official account often comprise stories and
original information provided by an official account to its followers examples, recommendations become specific, personal, and vivid [30],
[10,43,90]. Kim et al. [42] pointed out that a website’s information which could provide the followers with greater pleasure and elicit po-
quality is a signal that affects customers’ trust in the online store. When sitive emotions of the product. Based on the above, we hypothesize:
followers perceive that a WeChat official account presents accurate and
Hypothesis 3. Vicarious expression of the product recommendations is
timely information, they will think that the account will continue to
positively associated with product affection.
provide high-quality information and are more likely to have con-
fidence that the account is reliable and trustworthy [33,41]. Because
cognitive trust arises from accumulated knowledge and is based on 4.2.2. Aesthetic appeal and product affection
performance-relevant attributes, an official account that keeps pushing Aesthetic appeal refers to the presentation of product information
high-quality articles can build strong cognitive trust among its readers. through multiple visual elements and hedonic features [38], which
For example, Nicolaou and Mcknight [62] proposed that information jointly form a vivid and attractive information presentation, thereby
traits of accuracy, reliability, and correctness influence trusting beliefs enhancing users’ affective responses to the product [91]. Much research
of integrity and competence in a data exchange provider. Nicolaou et al. has shown that attractive and pleasurable presentation of product in-
[60] found that accurate and up-to-date information provided by a formation can enhance positive consumer responses toward the product
partner increases perceptions of competence trust. Thus, we have: [1]. For example, images in print advertisement [12] and visual pre-
sentations of a product [67] can generate an affective response toward
Hypothesis 1. Information quality is positively associated with
the product. Multimedia elements such as images, audio, videos, and
cognitive trust in recommender.
graphics presented in product recommendations are entertaining to
interested buyers and can provoke their internal excitement, stimula-
tion, or inspiration [12]. Thus, product recommendations displaying
4.1.2. Similarity and cognitive trust, affective trust in recommender charming visual cues (e.g., voice and graphics) are likely to enhance
Similarity arises from shared attributes such as demographic char- affective responses toward the recommended product.
acteristics, background, experience, and interests [2,80]. In this re-
Hypothesis 4. Aesthetic appeal of the product recommendations is
search, we focus on the perceived similarity between an official account
positively associated with product affection.
and its followers. Consistent with Zhang et al. [90], we define similarity
as the extent to which followers find a match with the interests or ex-
periences of the official account. Individuals tend to trust others who 4.3. Cognitive trust, affective trust, and product affection
are similar to them and have more confidence in a similar trustee. Prior
empirical studies show that similarity between individuals influences Previous research indicates that cognitive trust is the foundation of
trust development. For example, Mcallister [58] revealed that cultural affective trust because the latter is more likely to develop when a
or ethnical similarity between individuals affects interpersonal cogni- partner is seen to be reliable [39,44,58,92]. As cognitive reactions form
tive trust among managers and professionals in organizations. There- the basis for affective reactions, cognitive trust may influence affective
fore, similar interests or experiences may serve as social-based cues that trust [33]. A higher level of cognitive trust in the recommender will
followers use to reduce uncertainty and facilitate cognitive trust serve to reduce uncertainty and in turn encourage the readers to de-
building with a WeChat official account. Therefore, we have: velop emotional attachments to the recommender, thus leading to af-
fective trust. This leads to the following hypothesis:
Hypothesis 2a. Similarity is positively associated with cognitive trust
in recommender. Hypothesis 5. Cognitive trust is positively associated with affective
trust.
Similarity could result in reduced psychological distance [18]. Re-
search has shown that people with shared topics and similar interests The signaling theory suggests that trust in recommender has a sig-
may feel a closer bond with one another [2], which affects users’ per- nificant influence on consumers’ attitudes when limited information
ceptions of benevolence [53]. Therefore, similarity will in turn lead to about the recommended product is available [75]. When readers de-
enhanced affective trust [92]. Alnatour et al. [2] proposed that per- velop affective trust, they tend to build emotional bonds with the of-
ceived personality similarity affects users’ perceptions of the assistant’s ficial account and perceive it as being concerned about their best in-
benevolence. Johnson and Grayson [39] showed that a customer will terest [33]. Thus, when they read the recommendations from the
have an affective trust toward a service provider when s/he detects official account, they tend to consider the recommendations as acts of
elements of similarity with the provider. Therefore, we hypothesize: benevolence and the product recommended as important and relevant
[59]. Therefore, they may form positive affective responses toward the
Hypothesis 2b. Similarity is positively associated with affective trust in
recommended product. Further, if a credible recommender strongly
recommender.
recommends a product, receivers may substitute their own feelings with

240
Y. Chen et al. Information & Management 56 (2019) 236–248

the credible person’s recommendation. From this perspective, receivers above discussion, we hypothesize:
who trust the recommender will emotionally evaluate messages from
Hypothesis 7. Affective trust in recommender is positively associated
the recommender and develop positive affective responses toward that
with the urge to buy impulsively.
product. In a study of product review blogs, Huang [33] showed that
readers with higher affective trust toward a blog will have a positive Consumers in positive emotional states tend to reach a decision
attitude toward the products described in the blog. Therefore, we faster and more efficiently, which may lead to impulse buying [35].
propose that: Accordingly, when a consumer experiences a positive emotional re-
sponse toward a product, s/he would be more likely to buy the product
Hypothesis 6. Affective trust in recommender is positively associated
impulsively [80]. This leads to the following hypothesis:
with product affection.
Hypothesis 8. Product affection is positively associated with the urge
to buy impulsively.
4.4. Affective trust, product affection, and urge to buy impulsively

Previous research has considered positive affective reactions as a 4.5. Control variables
vital predictor of impulse buying [1,4]. For example, Parboteeah et al.
[66] confirmed that affective reactions (e.g., perceived enjoyment) to To examine the effectiveness of the research model, we included
the environment determine an individual’s urge to buy impulsively control variables that might affect impulse buying. Previous literature
online. Affective trust includes subjective affections such as inter- indicates that female impulsive buyers tend to buy products such as
personal concern and care. It is a positive affective relationship and jewelry and clothes that express their emotional and appearance feel-
emotional bond, and it results from positive emotions [26]. Thus, af- ings. By contrast, men are more impulsive when buying items related to
fective trust in recommender can stimulate consumers’ willingness to technology and sports equipment [14]. Age also influences impulsive
buy impulsively. buying, with younger consumers between the ages of 18 and 39 being
Further, Stern [76] suggested that if the buying decision process is more likely to buy impulsively [40]. Moreover, the inherent impul-
simpler, consumers are more likely to buy on impulse. The cognitive siveness of a consumer, defined as “both the tendencies (1) to experience
information processing that occurs in impulse buying is often simple, spontaneous and sudden urges to make on-the-spot purchases and (2) to act
involves low effort, and requires less deliberation [9], which is analo- on these felt urges with little deliberation or evaluation of consequence” [4],
gous to the concept of heuristics. According to the theory of heuristic can influence a consumer’s propensity to engage in impulse buying
information processing, heuristic information can simplify a consumer's [84]. Individuals with high impulsiveness are more likely to experience
decision-making process by drawing upon simple decision cues [8]. a stronger urge to buy impulsively. In addition, income and occupation
Following this logic, impulse buying is driven by a heuristic form of were also included as control variables.
information processing [10]. For example, Gwee and Chang [24] used
heuristic information processing to explain online impulse buying. They 5. Research methodology
found that scarcity and popularity claims foster heuristic information
processing and stimulate consumers’ desire for a product, which 5.1. Measurement development
eventually lead to impulse buying behaviors. Similarly, Chen et al. [10]
investigated information quality of advertisements in Facebook con- The survey questionnaire included two sections: the first section
sumer-to-consumer (C2C) “buy and sell” groups and revealed that in- comprised the respondent’s demographic information including gender,
formation quality as a heuristic cue can decrease time and effort in age, occupation, daily time spent on WeChat, the number of official
decision-making, thereby leading to more impulse buying behavior. accounts subscribed to, and the number of articles read every day; and
Given the heuristic nature of impulse buying and the relevant lit- the second section consisted of all constructs in the research model.
erature, we argue that how interpersonal trust influences impulse Most of the items for the constructs were adapted from previous re-
buying can be better explained by heuristic information processing. search to make them appropriate for this study (see Appendix B). A
McEvily et al. [59] pointed out that trust plays the role of a heuristic seven-point Likert scale ranging from “1-strongly disagree” to “7-
cue that reduces cognitive effort in decision-making. Trust in others strongly agree” was used to measure the items.
makes decision-making more efficient by simplifying the acquisition To confirm and ensure the validity of the measurement instrument,
and interpretation of information. When information is received from a back translation was done. We first translated the English version to
trusted source, the receiver is less likely to screen the information for Chinese and then the Chinese questionnaire was back translated to
validity [59]. Hence, trust in recommender can be seen as a heuristic English by another individual. The two English versions of the ques-
cue when followers read the recommendation posts [22,59]. When a tionnaire were compared to ensure their equivalence. We then con-
recommendation recipient perceives the source to be trustworthy, s/he ducted a pretest of the questionnaire to evaluate the face and content
may relinquish the thoughtful process of scrutinizing the message and validities. The participants in the pretest included two Internet mar-
instead unthinkingly accept the message. Thus, trust in the re- keting experts and five experienced users who regularly read articles
commender can reduce the amount of cognitive effort required and from WeChat official accounts. Except for a few minor modifications to
simplify the decision-making process, thus leading to increased impulse the wording, content, and structure of the questionnaire, they con-
buying. firmed that all items were appropriate and relevant to the corre-
The current study proposes that, for receivers of the recommenda- sponding constructs, which ensures content validity.
tion contents, cognitive trust in recommender will exert an indirect
influence on their urge to buy impulsively. As Parboteeah et al. [66] 5.2. Data collection and sample
pointed out, affective response is more important than cognitive re-
sponse in influencing the urge to buy impulsively. Users’ motivations We chose WeChat users in Mainland China as the research subjects.
for following a WeChat official account are primarily hedonic in nature, As mentioned earlier, WeChat is the most popular social media platform
thus followers may rely more on affective trust [74]. Further, cognitive in China and approximately 80% of WeChat users follow official ac-
trust is mainly built upon previously published non-product-related counts [77]. According to McKinsey [60], 31% of Chinese WeChat users
posts not directly related to the recommended product. Therefore, the initiated purchases on WeChat in 2016, doubling the percentage from
effects of cognitive trust in recommender on the urge to buy impulsively the previous year. Impulse-driven categories such as apparel and per-
will be fully mediated by affective trust in recommender. Based on the sonal care are the most popular on WeChat. These data show that

241
Y. Chen et al. Information & Management 56 (2019) 236–248

WeChat social commerce evolved quickly and became vitally important Table 3
in China, thus making China an ideal setting for investigating consumer Construct Reliability.
online impulse buying in social commerce. There are two types of Construct Item Standardized ρA Cronbach’s α CR AVE
WeChat official accounts: subscription accounts and service accounts Loadingsa
(for more details, see https://admin.wechat.com/). Service accounts
Information IQ1 0.812 0.813 0.811 0.868 0.637
are developed specifically for enterprises, whereas subscription ac-
quality IQ2 0.801
counts are available to all users. We mainly focus on subscription ac- IQ3 0.774
counts in this study. IQ4 0.806
The sample was randomly collected through the well-known online Similarity SI1 0.857 0.820 0.813 0.885 0.727
survey platform Sojump (www.sojump.com) in China. The survey hy- SI2 0.866
SI3 0.835
perlink was placed online, and only those who had WeChat experience
Vicarious VE1 0.870 0.864 0.860 0.913 0.780
were targeted for data gathering. We included screening questions to expression VE2 0.892
ensure that all respondents were WeChat users who followed more than VE3 0.888
one official account. Respondents were asked to recall their experience Aesthetic appeal AA1 0.815 0.887 0.884 0.918 0.743
AA2 0.883
with their most frequently read official account. Each respondent had
AA3 0.856
an opportunity to win monetary rewards such as a red packet after AA4 0.891
completing the survey. The data were collected in February 2017. A Product PA1 0.926 0.931 0.927 0.948 0.821
total of 251 respondents participated in this survey, and 18 were ex- affection PA2 0.923
cluded owing to incomplete or invalid answers, which resulted in a PA3 0.927
PA4 0.847
valid sample size of 233. Non-response bias was examined by com-
Cognitive trust CT1 0.842 0.814 0.812 0.883 0.726
paring the demographics of early respondents (those who responded CT2 0.865
within the first 2 weeks) versus late respondents (those who responded CT3 0.849
within the last 2 weeks). No significant differences were observed be- Affective trust AT1 0.747 0.833 0.830 0.878 0.621
tween the early and late respondents when analyzed using t-tests. AT2 0.771
AT3 0.842
Table 2 shows the demographic information of the subjects. Among Urge to buy UB1 0.900 0.880 0.874 0.923 0.798
them, 58.4% were male and more than 84.1% were aged below 30 impulsively UB2 0.879
years. The sample profile is consistent with findings from previous UB3 0.902
online surveys [77] that WeChat users are generally younger and there Impulsiveness IM1 0.881 0.938 0.937 0.953 0.841
IM2 0.925
are more men than women using the platform. More than 89.7% of the
IM3 0.934
respondents used WeChat for at least 1 h every day, and 96.1% of the IM4 0.927
respondents subscribed to more than five official accounts. Nearly half
a
of the respondents subscribed to 11 to 20 official accounts. Approxi- All standard loadings were significant at p < 0.001.
mately 60% of the respondents read more than five articles through
official accounts daily. factor explained 35.99% of the overall variance. No single factor
dominated the total variance, thus suggesting that common method
bias was unlikely to be a significant problem in our study. In addition,
5.3. Common method bias
we further assessed the method factor following the steps recommended
by Liang et al. [47]. A common method factor that was linked to all
Because our research data were collected from a single survey,
single-indicator constructs was included in the PLS model. The results
Harman’s single-factor test [68] was conducted to test common method
demonstrated that the loadings of the principal variables were all sig-
bias. Unrotated exploratory factor analysis indicated that the largest
nificant (p < 0.001), and none of the common method factor loadings
was significant. These results further indicated that common method
Table 2
bias was unlikely to be a serious concern in this study.
Sample Demographics.
Measure Item Frequency Percentage 6. Analysis and results
Gender Male 136 58.4%
Female 97 41.6% 6.1. Measurement model assessment
Age (years) < 18 2 0.9%
18–25 114 48.9% The measurement model was assessed to ensure item reliability,
26–30 80 34.3%
convergent validity, and discriminant validity. Item reliability was
31–40 34 14.6%
> 40 3 1.3% evaluated using Cronbach’s α , composite reliability (CR), average var-
Occupation Student 84 36.1% iance extracted (AVE), and Dijkstra-Henseler’s rho (ρA) [28]. As shown
Worker 134 57.5% in Table 3, the Cronbach’s α and CR of each construct were higher than
Unemployed/retired/ 15 6.4% the recommended value of 0.7, the AVEs of all the constructs were
housekeeper
above 0.6, and the values of ρA were more than 0.7 [28]. Hence, all
<1 37 10.3%
Hours spent daily on 1–2 50 21.5% constructs had good reliability [20].
WeChat 2–4 74 31.8% The convergent, discriminant and nomological validities were fur-
>4 72 30.9% ther examined for construct validity. As shown in Table 3, all AVE
Number of official accounts <5 9 3.9%
values were higher than the suggested benchmark level of 0.5 [25], the
subscribed to 6–10 46 19.7%
11–20 113 48.5% CRs were higher than the threshold of 0.70, and the standard loadings
21–30 31 13.3% of all items were higher than the desired threshold of 0.7 and significant
> 30 34 14.6% at 0.001, which indicated a good convergent validity.
Number of articles read per <5 93 39.9% Discriminant validity was evaluated in two steps. As shown in
day 6–10 98 42.1%
Table 4a, the Fornell and Larcker criterion indicates that the square
10–20 33 14.2%
> 20 9 3.9% roots of the AVEs (the bold numbers in the diagonal) were greater than
the interconstruct correlations [20]. As shown in Table 4b, the values of

242
Y. Chen et al. Information & Management 56 (2019) 236–248

Table 4a Table 5
Discriminant validity– Fornell-Larcker criterion. Model fit evaluation.
Construct SI VE AA CT AT PA UB IQ IM Discrepancy Value HI95 Conclusion

SI 0.853 SRMR 0.053 0.096 Supported

VE 0.198 0.883 dULS 5.276 27.203 Supported
AA 0.134 0.293 0.862 dG 1.067 29.457 Supported
CT 0.280 0.305 0.242 0.852
AT 0.291 0.281 0.479 0.509 0.788
PA 0.235 0.284 0.365 0.183 0.284 0.906
buy impulsively. R2 indicates the proportion of variance in a dependent
UB 0.126 0.144 0.227 0.142 0.162 0.268 0.893
IQ 0.270 0.253 0.255 0.413 0.526 0.159 0.032 0.798
variable explained by the exogenous variables, and R2 values higher
IM 0.154 0.023 0.028 0.021 0.013 0.140 0.258 0.001 0.917 than 0.20 indicate good explanatory power of the endogenous vari-
ables. The R2’s for cognitive trust, affective trust, product affection, and
Notes: SI, similarity: VE, vicarious expression; AA, aesthetic appeal; CT, cog- urge to buy impulsively were 0.465, 0.546, 0.435, and 0.439, respec-
nitive trust; AT, affective trust; PA, product affection; UB, urge to buy im- tively. These results show that the model has a good explanatory power
pulsively; IQ, information quality; IM, impulsiveness. on the dependent variables.

Table 4b
6.3. Overall model fit evaluation
Discriminant Validity– heterotrait–monotrait ratio of correlations.
Construct SI VE AA CT AT PA UB IQ We evaluated the goodness of fit of the research model by ex-
amining the standardized root mean square residual (SRMR), un-
VE 0.534
AA 0.428 0.621 weighted least squares discrepancy (dULS), and geodesic discrepancy dG
CT 0.648 0.665 0.579 values [28]. The lowerthe SRMR, dULS, and dG, the better the fit of the
AT 0.592 0.571 0.739 0.788 theoretical model [28]. Table 5 shows the results of the model fit. The
PA 0.561 0.594 0.665 0.492 0.550
SRMR value was below the recommended threshold of 0.08, and all
UB 0.425 0.433 0.541 0.449 0.430 0.570
IQ 0.627 0.604 0.600 0.787 0.708 0.460 0.213
discrepancies were below 95% of bootstrap quantile (HI95) [28], which
IM 0.458 0.168 0.182 0.166 0.116 0.400 0.556 0.033 suggested a good fit between the research model and the data.

Notes: SI, similarity; VE, vicarious expression; AA, aesthetic appeal; CT, cog- 6.4. Mediation analysis
nitive trust; AT, affective trust; PA, product affection; UB, urge to buy im-
pulsively; IQ, information quality; IM, impulsiveness.
To assess whether the indirect effects involved in the proposed
model were significant, we tested for mediation effects by applying the
the heterotrait–monotrait ratio of correlations (HTMT) were below the
steps recommended in Zhao et al. [91] using 5000 bootstrap samples at
threshold of 0.90 [28]. These results provided sufficient evidence of
the 95% confidence level. As shown in Table 6, all indirect effects were
discriminant validity for these constructs. Finally, nomological validity
significant at the 0.05 level, whereas the direct effects of cognitive trust
assesses whether the correlations between the constructs conform to the
on urge to buy impulsively was not significant, which suggested a full
theoretical framework [23]. The magnitude and significance of path
mediation of affective trust in the relationship between cognitive trust
coefficients in Section 6.2 confirmed that the model had nomological
and urge to buy impulsively. These analysis results supported H7.
validity.

6.5. Effect size

6.2. Structural model assessment
Effect size is used to identify the impact of an exogenous variable on
We used ADANCO 2.0.1 with bootstrapping to assess the structural an endogenous variable. According to Cohen [13], 0.02, 0.15, and 0.35
model [27]. The results of the structural model testing are shown in can be regarded as small, medium, and large, respectively, for the effect
Fig. 2. Most of the hypotheses were supported except for H6, which was size f2. The results in Table 7 show that the lowest value of the effect
represented as a dotted line. In addition, two control variables (age and size all exceeded the criterion of 0.02 except that affective trust in re-
impulsiveness) had significant positive effects on consumers’ urge to commender had a small effect with f2 = 0.018 on product affection.

Fig. 2. Research model testing results (*p < 0.05,**p < 0.01,***p < 0.001; ns: non-significant at the 0.05 level).

243
Y. Chen et al. Information & Management 56 (2019) 236–248

Table 6 recommendations with vivid stories and visual cues will increase fol-
Mediation analysis. lowers’ positive emotional responses and ultimately lead to the urge to
Relationship Direct effect Indirect effect buy impulsively.
Third, in support of H7, affective trust has a direct effect on the urge
Cognitive trust → Urge to buy impulsively −0.244 0.189* to buy impulsively. This suggests that, as followers form affective trust
Information quality → Affective trust 0.422** 0.228*
in the official account, they may be more likely to be affected by the
Similarity → Affective trust 0.171*** 0.226*
Vicarious expression → Urge to buy impulsively 0.064 0.164*
recommendation posts and develop an urge to buy the product re-
Aesthetic appeal → Urge to buy impulsively 0.277** 0.348* commended by the official account. Past research has recognized that
Similarity → Urge to buy impulsively 0.276*** 0.138* trust is one important determinant of online purchase behavior
[21,29,41]. The result of this study sheds light on trust as a direct de-
* p < 0.05.
terminant of unplanned, and even impulsive buying behavior.
** p < 0.01.
Fourth, affective trust has no significant effect on positive affective
*** p < 0.001.
reactions toward the product (p = 0.226), thus providing no support for
H6. As noted previously, there are three components of product atti-
Table 7
tude: cognitive, affective, and conation [64,87]. Although followers
Effect size analysis.
have affective trust in an official account, they may consider the pro-
Path Cohen’s f2 Effect size duct’s value and usefulness when reading product information in the
Hypothesized relationships
recommendation articles. That is, consumers may undergo cognitive
Information quality → Cognitive trust in 0.347 Medium responses instead of affective responses [67]. However, product affec-
recommender (H1) tion in our study only measures the affective reaction. It is worth noting
Similarity → Cognitive trust in recommender (H2a) 0.098 Weak- that a recent work by Hsu et al. [31] indicated that trust affects con-
medium
sumers’ attitudes toward products in blogs, but the effect is mediated by
Similarity → Affective trust in recommender (H2b) 0.080 Weak-
medium the perceived usefulness of the recommendations. This may suggest
Cognitive trust → Affective trust (H5) 0.560 Large that, although they have affective trust in the recommender, readers
Vicarious expression → Product affection (H3) 0.079 Weak- may first evaluate the recommendation and the product cognitively
medium when they see the product recommendation articles. For a better un-
Aesthetic appeal → Product affection (H4) 0.113 Weak-
derstanding of the relationship between affective trust and attitude
medium
Affective trust in recommender → Product affection 0.018 Very weak toward a product, future research can provide a comprehensive eva-
(H6) luation of consumers’ attitudes toward a product including both cog-
Affective trust in recommender → Urge to buy 0.056 Weak- nitive and affective components.
impulsively (H7) medium
Product affection → Urge to buy impulsively (H8) 0.086 Weak-
medium 7.2. Implications
Control variables
Impulsiveness → Urge to buy impulsively 0.227 Medium 7.2.1. Implications for theoretical research
Gender → Urge to buy impulsively 0.012 Very weak The current research enriches the existing literature on online im-
Age → Urge to buy impulsively 0.027 Weak
pulse buying in several ways. First, the current study applies the sig-
Occupation → Urge to buy impulsively 0.002 Very weak
Income → Urge to buy impulsively 0.007 Very weak naling theory to understand how extrinsic cues trigger impulse buying
in social commerce. Based on the empirical evidence offered in this
study, we conclude that recommender-related signals (information
7. Discussion and implications quality and similarity) are helpful in influencing trust in the re-
commender, and product-related signals (vicarious expression and
7.1. Discussion of findings aesthetic appeal) affect a reader’s positive attitude toward the re-
commended products, which in turn leads to impulse buying.
The current research examines how product recommendations af- Second, this study extends the previous research on impulsive
fects impulse buying in social commerce. Our empirical research based buying by examining this behavior in WeChat social commerce.
on a questionnaire survey of WeChat official account subscribers has Through building relationships with followers, WeChat official ac-
the following results. counts provide product recommendations to their followers. Our results
First, our results show that signals conveyed by non-product-related show that social-based recommendations have the ability to promote
posts such as information quality and similarity can evoke readers’ impulse buying. Previous research examining impulse buying in social
cognitive trust and in turn affect their affective trust toward the official commerce has been limited, and our study contributes to the under-
account. That is, if the official account provides up-to-date, accurate, standing of impulse buying in social commerce.
relevant, and original information, followers will cognitively trust the Additionally, few studies have examined the effect of trust on im-
official account. In addition, when followers detect elements of simi- pulse buying [86]. Trust has been shown to be a crucial issue in con-
larity with the operator of the official account, they will not only sumer buying behavior [26,43], but there has been little research on the
consider the operator of the official account to be reliable but also feel effect of trust in online impulse buying. Our empirical results show a
an emotional attachment to the account owner, which induces affective direct influence of affective trust on impulse buying. Moreover, our
trust. research employs a more comprehensive view to distinguish between
Second, signals conveyed by product recommendation posts such as cognitive and affective trust based on the research context. The results
vicarious expression and aesthetic appeal have significant impacts on highlight the role of trust in impulse buying and provide insights for
product affection. The storytelling and visual elements presented in the researchers in understanding impulse buying from the perspective of
recommendation contents can influence a consumer’s emotional re- interpersonal trust.
sponse to the product. Moreover, as expected, product affection sig-
nificantly influences the urge to buy impulsively. The result is con- 7.2.2. Practical implications
sistent with those from previous impulse buying studies which showed The results from this study provide some valuable insights to busi-
that a consumer’s positive affective response influences their impulse nesses engaged in social commerce, especially WeChat official account
buying behavior [49]. Therefore, the results indicate that providing operators. First, our results highlight the importance of trust

244
Y. Chen et al. Information & Management 56 (2019) 236–248

dimensions in increasing consumers’ urge to buy impulsively. Once with customers through content marketing in social media (e.g., mi-
consumers trust the recommender, their intention to purchase the so- croblog or WeChat) to enhance brand exposure, attract customers, and
cially recommended product is stimulated and they may buy the pro- increase sales.
duct impulsively [19]. Therefore, WeChat official account operators
should focus on improving their followers’ trust. Moreover, this study 7.3. Limitations and future research
offers insights into how to strengthen consumers’ trust in the re-
commender. The official account operators can signal their trust- This study has the following limitations and directions for future
worthiness by increasing information quality and similarity with the studies. The first limitation lies in the external validity. Our research
followers. First, the official account operators should continuously context is limited to WeChat, the best-known social media platform in
publish high-quality and informative articles. High-quality information China. Therefore, future studies can test our research model on other
is critical to attracting and maintaining trust relationships with fol- social media platforms such as Weibo or Facebook to cross-validate our
lowers. Further, the operators can post short biographical sketches on results. In addition, our research results might not be generalizable to
their background and interests as well as topical contents with messages other countries. Future studies may test our model in different cultures.
that resonate with their followers to highlight common interests. The Moreover, our research did not compare the product or service type in
posts can also include more personal information such as daily activities the research model, future studies may consider different product/ser-
to increase followers’ perceived similarity. vice scenario. Second, our research focuses on information quality and
Second, this study emphasizes the importance of storytelling and similarity revealed through non-product-related posts because fol-
aesthetic appeal when posting product recommendations. Personal and lowers subscribe to WeChat official accounts mainly to receive useful
vivid storytelling articles can enhance the reader’s positive attitude information contained in these posts. Product recommendation posts
toward the products. Storytelling articles can generate empathy and can also carry signals of information quality and similarity. Future re-
attract the reader’s attention more easily. Moreover, the findings search can examine how information quality and similarity revealed
highlight the importance of aesthetic appeal in creating affective re- through product-related posts affect the followers’ impulse buying be-
sponse. That is, when pushing product recommendations through an havior. Third, although the explanatory power of our research model is
official account, the owner should focus on how the product informa- acceptable, the explained variance of the urge to buy impulsively was
tion is delivered to create a vicarious expression and overall visual ef- only 43.9%, thus suggesting a need for future studies to identify more
fect that appeal to the readers. Our research provides guidelines for factors that help explain online impulse purchase activities. Factors
writing an effective message for product recommendations to stimulate related to market stimuli such as group buying, limited time offers [24],
readers to buy the recommended products. consumer characteristics such as product involvement [49], and other
Social commerce can be summarized in terms of two central activ- factors such as social influence may also affect impulse buying. In this
ities: putting e-commerce on social media platforms and/or adding and research, we only considered observable cues conveyed by the official
linking social media tools to e-commerce websites [81,89]. WeChat account’s posts that indirectly affect impulse buying through the lens of
social commerce falls under the former category. Therefore, although the signaling theory. Future studies can incorporate other theoretical
our study only focuses on WeChat social commerce in China, our perspectives to examine whether other factors may affect impulse
findings may be generalizable to social commerce in other countries. buying. Finally, our questionnaire survey used a self-reported scale to
Our results suggest that social commerce providers should focus more measure the variables, and this survey was conducted online. There-
on “relational” rather than “transactional” elements, as these “rela- fore, some bias might exist. Although common-method bias is not a
tional” elements (e.g., social trust) are likely to increase sales through serious issue in this study, future studies can use multiple methods or
consumers’ impulse buying behavior. In addition, the advent of WeChat collect data from multiple sources to cross-validate our findings.
social commerce has fostered content-driven social commerce [72] and
made it into something more accessible for both businesses and brands. Acknowledgements
Based on the results of our study, content-driven social commerce
through social media can start with high-quality contents and end in an This work is partially supported by grants from the National science
impulse purchase. Thus, social commerce providers can communicate Foundation of China (71332001, 71810107003 and 71602063).

Appendix A. Sample product recommendation articles published by a WeChat official account

245
Y. Chen et al. Information & Management 56 (2019) 236–248

Appendix B. Measurement items

Information quality (adapted from Nicolaou and Mcknight [62] and Kim and Park [43])
IQ1: The non-product-related information pushed by the official account is accurate.
IQ2: The non-product-related information pushed by the official account is relevant to my needs.
IQ3: The non-product-related information pushed by the official account is the latest.
IQ4: The non-product-related information pushed by the official account is original.
Similarity (adapted from Johnson and Grayson [39])
SI1. The operator of the official account is similar to me.
SI2.The operator of the official account has values similar to mine.
SI3. The operator of the official account has experience similar to mine.
Vicarious expression (adapted from Li et al. [45])
VE1: By reading product recommendation posts, I can feel what the author is trying to say about the recommended products and his/her usage
experience.
VE2: By reading product recommendation posts, I can imagine what the author is trying to say about the recommended products and his/her
usage experience.
VE3: By reading product recommendation posts, I can envision what the author is trying to say about the recommended products and his/her
usage experience.
Aesthetic appeal (adapted fromWatson et al. [82])
AA1: The graphics provided by the recommendation posts are attractive.
AA2: The product recommendation posts are visually pleasing.
AA3: The product recommendation posts are visually cheerful.
AA4: The product recommendation posts are visually interesting.
Affective trust in recommender (adapted from Johnson and Grayson [39] and Huang [33])
AT1: I enjoy reading articles in this official account.
AT2: I feel comfortable reading articles in this official account.
AT3: I feel articles in this official account are sincere.
Cognitive trust in recommender (adapted from Mcallister [58])
CT1: I think the operator of this official account is reliable.
CT2: This official account’s operator is very knowledgeable about his/her topics.
CT3: In general, I have confidence in the operator of this official account.
Product affection (adapted from Watson et al. [82])
PA1: I felt enthusiastic toward the recommended product.
PA2: I felt inspired by the recommended products.
PA3: I felt excited about the recommended products.
PA4: I felt interested in the recommended products.
Urge to buy impulsively (adapted from Wells et al. [84])
UB1: As I read the product recommendations in this official account, I had the urge to purchase items other than in addition to my specific
shopping goal.
UB2: As I read the product recommendations in this official account, I had a desire to buy items that did not pertain to my specific shopping goal.
UB3: As I read the product recommendations in this official account, I had the inclination to purchase items outside of my specific shopping goal.
Impulsiveness (adapted from Rook and Fisher [71])
IM1: “Just do it” describes the way I buy things.
IM2: I often buy things without thinking.
IM3: “I see it, I buy it” describes me.
IM4: “Buy now, think about it later” describes me.

References 453–482.
[8] S. Chaiken, Heuristic versus systematic information processing and the use of source
versus message cues in persuasion, J. Pers. Soc. Psychol. 39 (1980) 752–766.
[1] T. Adelaar, S. Chang, K.M. Lancendorfer, B. Lee, M. Morimoto, Effects of media [9] T.K. Chan, C.M. Cheung, Z.W. Lee, The state of online impulse-buying research: a
formats on emotions and impulse buying intent, J. Inf. Technol. 18 (2003) 247–266. literature analysis, Inf. Manage. 54 (2017) 204–217.
[2] S. Alnatour, I. Benbasat, R.T. Cenfetelli, The adoption of online shopping assistants: [10] J.V. Chen, B. Su, A.E. Widjaja, Facebook C2C social commerce: a study of online
perceived similarity as an antecedent to evaluative beliefs, J. Assoc. Inf. Syst. 12 impulse buying, Decis. Support Syst. 83 (2016) 57–69.
(2011) 2. [11] X. Cheng, S. Fu, G.-J. de Vreede, Understanding trust influencing factors in social
[3] Y. Bai, Z. Yao, Y.-F. Dou, Effect of social commerce factors on user purchase be- media communication: a qualitative study, Int. J. Inf. Manage. 37 (2017) 25–35.
havior: an empirical investigation from renren.com, Int. J. Inf. Manage. 35 (2015) [12] R.M.M.I. Chowdhury, G.D. Olsen, J.W. Pracejus, Affective responses to images in
538–550. print advertising: affect integration in a simultaneous presentation context, J.
[4] S.E. Beatty, M.E. Ferrell, Impulse buying: modeling its precursors, J. Retail. 74 Advert. 37 (2008) 7–18.
(1998) 161–167. [13] J. Cohen, Statistical Power Analysis for Behavioral Science, Lawrence Erlbaum
[5] A. Benlian, T. Hess, The signaling role of IT features in influencing trust and par- Associates, Hillsdale, NJ, 1988.
ticipation in online communities, Int. J. Electron. Commer. Stud. 15 (2011) 7–56. [14] A. Coley, B. Burgess, Gender differences in cognitive and affective impulse buying,
[6] A. Benlian, R. Titah, T. Hess, Differential effects of provider recommendations and J. Fashion Mark. Manage. 7 (2003) 282–295.
consumer reviews in E-Commerce transactions: an experimental study, J. Manage. [15] B.L. Connelly, S.T. Certo, R.D. Ireland, C.R. Reutzel, Signaling theory: a review and
Inform. Syst. 29 (2012) 237–272. assessment, J. Manage. 37 (2011) 39–67.
[7] F. Calefato, F. Lanubile, N. Novielli, The role of social media in affective trust [16] J. Deighton, D. Romer, J. Mcqueen, Using drama to persuade, J. Consum. Res. 16
building in customer—supplier relationships, Int. J. Inf. Manage. 15 (2015) (1989) 335–343.

246
Y. Chen et al. Information & Management 56 (2019) 236–248

[17] A. Dimoka, Y. Hong, P.A. Pavlou, On product uncertainty in online markets: theory buyers: trust in virtual communities and its effect on consumers’ purchase intention,
and evidence, MIS Q. 36 (2012) 395–426. Electron. Commer. Res. Appl. 9 (2010) 346–360.
[18] S.M. Edwards, J.K. Lee, C.L. Ferle, Does place matter when shopping online? [54] P. Luarn, Y. Lin, Y. Chiu, Influence of Facebook brand-page posts on online en-
Perceptions of similarity and familiarity as indicators of psychological distance, J gagement, Online Inf. Rev. 39 (2015) 505–519.
Interact. Advert. 10 (2009) 35–50. [55] X. Luo, How does shopping with others influence impulsive purchasing, J. Consum.
[19] A. Floh, M. Madlberger, The role of atmospheric cues in online impulse-buying Psychol. 15 (2005) 288–294.
behavior, Electron. Commer. Res. Appl. 12 (2013) 425–439. [56] T. Mavlanova, R. Benbunan-Fich, M. Koufaris, Signaling theory and information
[20] C. Fornell, D.F. Larcker, Structural equation models with unobservable variables asymmetry in online commerce, Inf. Manage. 49 (2012) 240–247.
and measurement error: algebra and statistics, J. Mark. Res. (1981) 382–388. [57] T. Mavlanova, R. Benbunan-Fich, G. Lang, The role of external and internal signals
[21] D. Gefen, E. Karahanna, D.W. Straub, Trust and TAM in online shopping: an in- in E-commerce, Decis. Support Syst. 87 (2016) 59–68.
tegrated model, MIS Q. 27 (2003) 51–90. [58] D.J. Mcallister, Affect- and cognition-based trust as foundations for interpersonal
[22] K. Giffin, The contribution of studies of source credibility to a theory of inter- cooperation in organizations, Acad. Manage. J. 38 (1995) 24–59.
personal trust in the communication process, Psychol. Bull. 68 (1967) 104–120. [59] B. McEvily, V. Perrone, A. Zaheer, Trust as an organizing principle, Organ. Sci. 14
[23] G. Goertz, Social Science Concepts: A User’s Guide, Princeton University Press, (2003) 91–103.
2006. [60] McKinsey & Company, McKinsey iConsumer China 2016 Survey: How Savvy, Social
[24] M.Y.T. Gwee, K.T.T. Chang, Effects of Persuasive Claims on Desirability and Shoppers Are Transforming e-commerce, (2016).
Impulse Purchase Behavior, Americas Conference on Information Systems, Chicago, [61] W.W. Moe, Buying, searching, or browsing: differentiating between online shoppers
Illinois, 2013. using in-store navigational clickstream, J. Consum. Psychol. 13 (2003) 29–39.
[25] J.F. Hair, R.E. Anderson, B.J. Babin, W.C. Black, Multivariate Data Analysis: a [62] A.I. Nicolaou, D.H. Mcknight, Perceived information quality in data exchanges:
Global Perspective, Pearson, Upper Saddle River, NJ, 2010. effects on risk, trust, and intention to use, Inf. Syst. Res. 17 (2006) 332–351.
[26] N. Hajli, J. Sims, A.H. Zadeh, M.-O. Richard, A social commerce investigation of the [63] Nicomama, http://www.nicomama.com/ (2016).
role of trust in a social networking site on purchase intentions, J. Bus. Res. 71 [64] T.M. Ostrom, The relationship between the affective, behavioral, and cognitive
(2017) 133–141. components of attitude, J. Exp. Soc. Psychol. 5 (1969) 12–30.
[27] J. Henseler, T. Dijkstra, "ADANCO 2.0.1",Composite Modeling, Kleve, available at: [65] C.X. Ou, K.C. Chan, Developing a competitive edge in electronic markets via in-
(2015) http://www.compositemodeling.com. stitutional and social based quality signaling mechanisms, Inf. Manage. 51 (2014)
[28] J. Henseler, G. Hubona, P.A. Ray, Using PLS path modeling in new technology 532–540.
research: updated guidelines, Ind. Manage. Data. Syst. 116 (2016) 2–20. [66] D.V. Parboteeah, J.S. Valacich, J.D. Wells, The influence of website characteristics
[29] K. Hsiao, J.C. Lin, X. Wang, H. Lu, H. Yu, Antecedents and consequences of trust in on a consumer’s urge to buy impulsively, Inf. Syst, Res. 20 (2009) 60–78.
online product recommendations an empirical study in social shopping, Online Inf. [67] J. Park, L. Stoel, S.J. Lennon, Cognitive, affective and conative responses to visual
Rev. 34 (2010) 935–953. simulation: the effects of rotation in online product presentation, J. Consum. Behav.
[30] K. Hsiao, H. Lu, W. Lan, The influence of the components of storytelling blogs on 7 (2008) 72–87.
readers’ travel intentions, Internet Res. 23 (2013) 160–182. [68] P.M. Podsakoff, S.B. MacKenzie, J.-Y. Lee, N.P. Podsakoff, Common method biases
[31] C. Hsu, J.C. Lin, H. Chiang, The effects of blogger recommendations on customers’ in behavioral research: a critical review of the literature and recommended re-
online shopping intentions, Internet Res. 23 (2013) 69–88. medies, J. Appl. Psychol. 88 (2003) 879.
[32] L.-T. Huang, Flow and social capital theory in online impulse buying, J. Bus. Res. 69 [69] J. Pulizzi, The rise of storytelling as the new marketing, Publ. Res. Q. 28 (2012)
(2016) 2277–2283. 116–123.
[33] L. Huang, Trust in product review blogs: the influence of self-disclosure and po- [70] D.W. Rook, The buying impulse, J. Consum. Res. 14 (1987) 189–199.
pularity, Behav. Inf. Technol. 34 (2015) 33–44. [71] D.W. Rook, R. Fisher, Normative influences on impulsive buying behavior, J.
[34] Y. Hwang, S.-H. Jeong, “This is a sponsored blog post, but all opinions are my own”: Consum. Res. 22 (1995) 305–313.
the effects of sponsorship disclosure on responses to sponsored blog posts, Comput. [72] J. Rowley, Understanding digital content marketing, J Mark. Manage. 24 (2010)
Hum. Behav. 62 (2016) 528–535. 517–540.
[35] A.M. Isen, The influence of positive affect on decision making and cognitive orga- [73] G.C.-C. Shen, Users’ adoption of mobile applications: product type and message
nization, Adv. Consum. Res. 11 (1984) 534–537. framing’s moderating effect, J. Bus. Res. 68 (2015) 2317–2321.
[36] C. Janiszewski, The influence of display characteristics on visual exploratory search [74] D. Smith, S. Menon, K. Sivakumar, Online peer and editorial recommendations,
behavior, J. Consum. Res. 25 (1998) 290–301. trust, and choice in virtual markets, J. Interact. Mark. 19 (2005) 15–37.
[37] Z. Jiang, I. Benbasat, Research note—investigating the influence of the functional [75] M. Spence, Job market signaling, Quart. J. Econ. 87 (1973) 355–374.
mechanisms of online product presentations, Inf. Syst, Res. 18 (2007) 454–470. [76] H. Stern, The significance of impulse buying today, J. Mark. 26 (1962) 59.
[38] Z. Jiang, I. Benbasat, Virtual product experience: effects of visual and functional [77] Technode, WeChat’s Impact: A Report on WeChat Platform Data, (2015) http://
control of products on perceived diagnosticity and flow in electronic shopping, J. technode.com/2015/02/10/wechat-impact-report/.
Manage. Inform. Syst. 21 (2004) 111–147. [78] Tencent, Tencent Announces 2017 First Quarter Results, Retrieved from: (2017)
[39] D. Johnson, K. Grayson, Cognitive and affective trust in service relationships, J. Bus. https://www.tencent.com/zh-cn/company.html.
Res. 58 (2005) 500–507. [79] T. Verhagen, W. van Dolen, The influence of online store beliefs on consumer online
[40] J.J. Kacen, J.A. Lee, The influence of culture on consumer impulsive buying be- impulse buying: a model and empirical application, Inf. Manage. 48 (2011)
havior, J. Consum. Psychol. 12 (2002) 163–176. 320–327.
[41] D.J. Kim, D.L. Ferrin, H.R. Rao, A trust-based consumer decision-making model in [80] C. Vonkeman, T. Verhagen, W. van Dolen, Role of local presence in online impulse
electronic commerce: the role of trust, perceived risk, and their antecedents, Decis. buying, Inf. Manage. 54 (8) (2017) 1038–1048.
Support Syst. 44 (2008) 544–564. [81] C. Wang, P. Zhang, The evolution of social commerce: the people, management,
[42] H.-W. Kim, Y. Xu, J. Koh, A comparison of online trust building factors between technology, and information dimensions, Commun. AIS 31 (2012) 5.
potential customers and repeat customers, J. Assoc. Inf. Syst. 5 (2004) 13. [82] D. Watson, L.A. Clark, A. Tellegen, Development and validation of brief measures of
[43] S. Kim, H. Park, Effects of various characteristics of social commerce (s-commerce) positive and negative affect: the PANAS scales, J. Pers. Soc. Psychol. 54 (1988)
on consumers’ trust and trust performance, Int. J. Inf. Manage. 33 (2013) 318–332. 1063–1070.
[44] R.J. Lewicki, E.C. Tomlinson, N. Gillespie, Models of interpersonal trust develop- [83] S.S. Webber, R.J. Klimoski, Client-project manager engagements, trust, and loyalty,
ment: theoretical approaches, empirical evidence, and future directions, J. Manage. J. Organ. Behav. 25 (2004) 997–1013.
32 (2006) 991–1022. [84] J.D. Wells, D.V. Parboteeah, J.S. Valacich, Online impulse buying: understanding
[45] M. Li, L. Huang, C. Tan, K. Wei, Helpfulness of online product reviews as seen by the interplay between consumer impulsiveness and website quality, J. Assoc. Inf.
consumers: source and content features, Int. J. Electron. Commer. Stud. 17 (2013) Syst. 12 (2011) 3.
101–136. [85] J.D. Wells, J.S. Valacich, T.J. Hess, What signal are you sending? How website
[46] S. Li, K. Srinivasan, B. Sun, Internet auction features as quality signals, J. Mark. 73 quality influences perceptions of product quality and purchase intentions, MIS Q.
(2009) 75–92. (2011) 373–396.
[47] H. Liang, N. Saraf, Q. Hu, Y. Xue, Assimilation of enterprise systems: the effect of [86] I.-L. Wu, K.-W. Chen, M.-L. Chiu, Defining key drivers of online impulse purchasing:
institutional pressures and the mediating role of top management, MIS Q. 31 (2007) a perspective of both impulse shoppers and system users, Int. J. Inf. Manage. 36
59–87. (2016) 284–296.
[48] X. Liang, Y. Yang, An experimental study of chinese tourists using a company- [87] P.C.S. Wu, Y. Wang, The influences of electronic word-of-mouth message appeal
Hosted wechat official account, Electron. Commer. Res. Appl. 27 (2018) 83–89. and message source credibility on brand attitude, Asia Pac. J. Mark. Logist. 23
[49] C. Liao, P.-L. To, Y.-C. Wong, P. Palvia, M.D. Kakhki, The impact of presentation (2013) 448–472.
mode and product type on online impulse buying decisions, J. Electron. Commer. [88] L. Xiang, X. Zheng, M.K.O. Lee, D. Zhao, Exploring consumers’ impulse buying
Res. 17 (2016) 153. behavior on social commerce platform: The role of parasocial interaction, Int. J. Inf.
[50] C.-H. Lien, Y. Cao, X. Zhou, Service quality, satisfaction, stickiness, and usage in- Manage. 36 (2016) 333–347.
tentions: An exploratory evaluation in the context of WeChat services, Comput. [89] M. Yadav, K. De Valck, T. Hennigthurau, D.L. Hoffman, M. Spann, Social commerce:
Hum. Behav. 68 (2017) 403–410. a contingency framework for assessing marketing potential, J. Interact. Mark. 27
[51] L.Y. Lo, S. Lin, L. Hsu, Motivation for online impulse buying: a two-factor theory (2013) 311–323.
perspective, Int. J. Inf. Manage. 36 (2016) 759–772. [90] K.Z.K. Zhang, S.J. Barnes, S.J. Zhao, H. Zhang, Can consumers be persuaded on
[52] L.-C. Lu, W.-P. Chang, H.-H. Chang, Consumer attitudes toward blogger’s sponsored brand microblogs? An empirical study, Inf. Manage. 55 (2018) 1–15.
recommendations and purchase intention: the effect of sponsorship type, product [91] P. Zhang, The affective response model: a theoretical framework of affective con-
type, and brand awareness, Comput. Hum. Behav. 34 (2014) 258–266. cepts and their relationships in the ICT context, MIS Q. 37 (2013) 247–274.
[53] Y. Lu, L. Zhao, B. Wang, From virtual community members to C2C e-commerce [92] A. Zur, C. Leckie, C.M. Webster, Cognitive and affective trust between Australian

247
Y. Chen et al. Information & Management 56 (2019) 236–248

exporters and their overseas buyers, Australas. Mark. J. 20 (2012) 73–79. Bin Wang is a Professor of Information Systems at the University of Texas Rio Grande
Valley. Her research interests include IT adoption, the performance of IT firms, electronic
commerce, mobile commerce, and social commerce. Her research has appeared in jour-
Yanhong Chen is a PhD student in Management Science & Information Systems at School nals such as Journal of Management Information Systems, Journal of Business Research,
of Management, Huazhong University of Science and Technology in China. Her research Information Systems Journal, Industrial Management & Data Systems, Information
focuses on electronic commerce, information management and social media. Technology and Management, Electronic Commerce Research and Applications, and
Computers in Human Behavior.
Yaobin Lu is a specially appointed Professor in Management Science & Information
Systems at School of Management, Huazhong University of Science and Technology in Zhao Pan is an associate professor of Management Science & Information Systems at the
China. His research interests include social commerce, mobile commerce, business mode, School of Management at Huazhong University of Science and Technology in China. He
electronic commerce, and related topics. He is the author of more than 50 publications in obtained his Ph.D. there in 2015. His research focuses on social commerce, social net-
leading international journals such as Journal of Management Information Systems, works, and social computing. His work has been published in Journal of Management
Decision Support Systems, Information Systems Journal, Information & Management, Information Systems, Computers in Human Behavior and other journals.
International Journal of Electronic Commerce, and Journal of Information Technology.

248