eWOM effectiveness: An investigation of the interaction
effects of eWOM emotionality, relationship closeness and
sender’s expertise on consumers’ purchase intention
Business Administration
Universiteit van Amsterdam
Teng Xiao
10256377
Supervisor: Afred Zerres
Statement of Originality
This document is written by Student Teng Xiao who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Purpose – Based on social influence theory, the purpose of this paper is to use positive electronic word-of-mouth (eWOM) settings to examine the interaction effects of eWOM
emotionality, relationship closeness, and sender’s expertise on consumers’ purchase intention.
Design/methodology/approach – Data were collected from online experiments. A 2 (eWOM emotionality: high vs. low) x 2 (relationship closeness: close vs. distant) x 2 (sender’s
expertise: high vs. low) experimental design was created, and finally 222 qualified samples
were analyzed to verify the hypotheses of this study.
Findings - The results confirm the significant interactions between relationship closeness and eWOM emotionality, and sender’s expertise and eWOM emotionality on consumers’
purchase intention. The study suggests that eWOM exerts higher effectiveness for close
relationship other than distant relationship only when eWOM is emotional, whereas the
effectiveness is stronger for experts rather than novices only when eWOM is rational. The
implications for researchers, practitioners are discussed.
Originality/value – Previous research exploring the interaction effects of eWOM emotionality, relationship closeness, and sender’s expertise is rare. This study aims to fill this
gap.
1. Introduction
The proliferation of the internet brings us the era of information explosion. Consumers are
presented with tons of advertisements in the new forms of media, such as online magazines,
blogs, and social media. Facebook, as one of the most typical model of the social media, has
been attracting two million active advertisers and the number is still dramatically growing
(facebook, 2015). Previously, advertisers generally refer to merchants who advertise in the
public medium to promote product sales. However, with the advent of the internet, the
definition of advertisers has been broaden to everyone. The transparent and speedy
transmission of information on the internet can make any personal reviews go wild all of a
sudden. Therefore, to generate more buzz, consumers are encouraged to share their comments
among their online social contacts, and this type of behavior is called electronic
word-of-mouth (eWOM).
By and large, eWOM extends the traditional word-of-mouth (WOM) to the online context. It
comprises vast amounts of consumer information on opinions and recommendations on
vendors/products from experienced consumers, which is expected to play an important role in
directing consumers’ purchase decision. Actually, many industry research has shown that
when making purchase decision, consumers is most likely to rely more on online reviews
posted by known or even unknown consumers more than traditional media (e.g. print
advertising, TV commercials) (Nielsen, 2010). Reported by ChannelAdvisor (2010), 91% of
respondents mentioned that they seek advice from online reviews, blogs, and other
user-generated content before buying new goods/service, and 46% of them indicated that their
buying behaviors are then impacted in the way they to purchase. Remarkably, predicted by
Thus, it is acknowledged that the more effective eWOM is, the more likely consumer’
purchase behavior will be influenced. Previous literatures have examined many factors
related to eWOM effectiveness. Cheng and Zhou (2010) summarized them into five
dimensions: the information source, ties between information source and receiver, receiver,
information, and other factors. The information source refers to source expertise and
trustworthiness. The higher level of expertise and trustworthiness implies higher eWOM
effectiveness (Brown, Broderick, and Lee, 2007). Interestingly, since advice from friends and
families indicate high trustworthiness and that from firms implies high expertise, both should
significantly affect people’s purchase intention. Nevertheless, as indicated by Schiffman and
Kanuk (1997), people tend to reply more on the feedback sought from interpersonal
communications (e.g. word-of-mouth) than the information received from organizational
sources (e.g. advertising campaign), because as for the former, the sender of the information
is generally perceived as altruistic. Thus, the tie between information source and receiver is
also perceived as a critical factor on the effectiveness of eWOM (Banal and Voyer, 2000), as
different ties imply different motivation behind. A closer tie between information sender and
receiver is generally considered more reliable, thus exerting more influences. Except tie
strength, another construct emerged in the research of relationship between senders and
receivers is homophily. It refers to the extent to which the receiver is congruent on certain
attributes, such as age, gender, and education etc. with the sender. It was found that
homophonous sources will be regarded as more reliable than heterogeneous ones (Brown and
Reingen, 1987). In addition, academics also pointed out that characteristics of receivers will
also influence the effect of eWOM. Banal and Voyer (2000) suggested that recipients’
expertise proved more negative impact on word of mouth. Besides, the demographics of
receivers, their propensity to trust, and their self-efficacy will also have impacts on eWOM
affect. The quality of eWOM (i.e. high quality vs. low quality), the valence of eWOM (i.e.
positive, negative, and neutral), the side of the information (i.e. one-sided vs. two-sided), and
the consistency of the information (i.e. consistency vs. inconsistency) are all relevant aspects.
Apart from information and information carriers, there are also many situation factors, such
as product category, website platforms which can also influence consumers’ perception of
eWOM (Senecal and Nantel, 2004).
In this study, I am interested in three factors that are from three dimensions (i.e. information
source, relationships between information source and receiver, information) discussed above.
That is, the sender’s expertise, which is part of information source; the relationship closeness,
which is involved with ties between senders and receivers; and eWOM emotionality, which is
one of features of the information. eWOM emotionality refers to the extent of the emotional
appeals showing in a message. Previous research categorizes message appeal into rational
appeals (e.g. “It saves me more time.”) and emotional appeals (e.g. “I don’t like it.”) (Kotler
and Keller, 2008), and two kinds of appeals would lead to different brand attitudes and
consumer behaviors (Park and Lee, 2009; Paul, Wu, and Wang, 2011; Petty and Cacioppo,
1984). On top of eWOM emotionality, the source of eWOM also affects consumer decision
making. eWOM from rapports is perceived more reliable whereas that from experts is
considered more referential. Since both types of source indicate high credibility, the effect of
eWOM from these sources on purchase intention is larger than that from anonymous sources
(Brown and Reingen, 1987; Steffes and Burgee, 2009), but which source has larger effects
remains unclear. Furthermore, because as mentioned before, eWOM emotionality and
eWOM source have been proved to be antecedents for eWOM effectiveness, it is highly
likely that there are interaction effects between them. However, little research has shed light
size of each targeted factors (i.e. eWOM emotionality, relationship closeness, and sender’s
expertise) as well as the interaction effects on consumers’ purchase intention.
The contributions of this research are twofold. From the theoretical perspective, this study
aims to enrich the marketing research of the impacts of eWOM-related factors on consumer
purchase behavior. This will help academics to further broaden and deepen their research on
more complex relationship among relevant factors that influence eWOM effectiveness on
people’s purchase intention. From the managerial perspective, my findings can be of use to
leverage eWOM to stimulate consumer’s purchasing behavior. Understanding relationship
closeness, sender’s expertise, and eWOM emotionality variables that affect consumers’
purchase behaviors could help marketers to identify influential individuals in the online
context and to effectively incorporate eWOM promotion as part of integrated marketing
communication strategy. Besides, since the findings show that the interactions of eWOM
emotionality and information source lead to different degrees of effectiveness, marketers can
make use of the results by stimulating the most effective eWOM and controlling the least
effective eWOM.
The paper is organized as follows. First, the theoretical background of eWOM, relationship
closeness, and sender’s expertise is introduced. Then, the sample, data and methodology are
described. Next, the results of the data are analyzed. Finally, it is concluded with a discussion
of implications for research and practice.
2. Literature Review
Four streams of literature can be found in this section. The first three section involve previous
research on the three main concepts (eWOM and eWOM emotionality, sender’s expertise,
theory which supports the arguments of my hypotheses. Finally, hypotheses and conceptual
framework are introduced.
eWOM and eWOM emotionality
WOM is defined as the informal and personal suggestions about the products, service, and
social issues (East et al., 2007). Research of WOM focuses on its impact as both an input of
consumer decision-making (Bloch, Sherrell, and Ridgway, 1986; Feick and Price, 1987) and
an outcome of the purchase process (Holmes and Lett, 1977; Richins, 1983). Consumers
generally believe that WOM credibility is higher than commercial advertisement.
Traditional WOM normally refers to offline face-to-face communication. However, with the
proliferation of Internet, more and more offline communication is moving to online. Thus, to
meet this change, various online platforms have been invented for online communication.
This brings to the advent of eWOM era. Hennig-Thurau, Qwinnerm Walsh, and Gremler
(2004, p.39) defined electronic world of mouth as “any positive or negative statement made
by potential, actual or former customers about a product or company, which is made available
to a multitude of people and institutions via the Internet”. Given the conceptual closeness of
traditional WOM and eWOM, consumer benefits that have been identified in the literature as
being relevant for traditional WOM also can be expected to be of relevance for eWOM.
However, on top of that, eWOM brings even more convenience to consumers due to the
unique features that can only be realized through Internet (Cheung and Thadani, 2012). First
and foremost, it diffuses at an unprecedentedly speed and scale with low costs. The easy
transmission of eWOM enables it to spread far further than traditional WOM (Chatterjee,
2001). Compared with WOM which typically occurs in private dialogs, eWOM is exposed
under open platforms which anyone can read it at anytime. Various platforms involve email,
traditional word of mouth only takes place in the face-to-face communication. This
contributes to its second attribute: more accessible and persistent. With eWOM, people are
enabled to trace relevant reviews that was created at anytime anywhere. That means they can
further compare old and newly-generated reviews of the same product, which helps them to
confirm the credibility of an advertised product. Last but not the least, eWOM protects
privacy. On the contrary to WOM in which opinions are exchanged among people who are
known to each other, eWOM allows people to comfortably share their opinions without
revealing their real identities, which enlarges the tie between the information sender and
receiver in the traditional WOM.
Since eWOM is a judgement of an object, it can be either positive or negative. Though
considerable literatures (Wright, 1974; Arndt, 1967; Lee, Park, and Han, 2008) suggest that
negative eWOM also has strong influence on consumers buying behavior, in this study, I
focus on positive eWOM because the purpose here is to research stimuli that increases
consumer purchase intention.
Even both eWOM are positive on the same topic, they can result in different degrees of
persuasive effects on receivers. Why? One of the reasons may due to the fact that each
eWOM is articulated with different levels of emotionality. This premise is validated by
Kotler and Keller (2008) who strongly supported that the type of message can be classified as
with emotional and rational dichotomy.” Emotional message involves with subjective and
abstract reviews based on personal feelings of a product. For instance, “I don’t like to drink
with this wine glass.” On the contrary, rational message refers to objective and concrete
reviews based on the specific facts of a product. For example, “This wine glass is easy to be
broken.” This classification is similar to the concept of two-dimensional consumers attitudes
maintained by Voss, Spangenberg, and Grohmann (2003). They proposed that consumers’
sensations and experience of the products while utilitarian dimensions resulted from
functions and usability performed by products.
Many empirical studies show that rational and objective reviews are much more persuasive
than emotional and subjective ones (Petty, Cacioppo, and Schumann, 1983; Wilson, Lindsey,
and Schooler, 2000; Petty, 2013). According to them, rational reviews are perceived to be
more informative than emotional reviews. Also, the quality of a rational review is generally
considered higher because of higher objectiveness which is used to assess information quality
(Park, Lee, and Han, 2007). Since it is almost impossible to objectively define a message that
contains only emotional appeals or only rational appeals, in this research, rational and
emotional appeal fall apart onto the extreme of the spectrum of eWOM emotionality, and I
measure the extent of eWOM emotionality.
With regard to the causal relationship between emotionality and purchase intention, Wu and
Wang (2011) confirmed that consumer purchase intention would be higher when receiving
rational appeal than emotional appeal, under high product involvement. Based on elaboration
likelihood model (ELM), consumers with high involvement are inclined to process
information through the central route, which means that they are likely to think rationally by
carefully examining product relevant information. According to the cognitive fit theory, when
the information type and consumer information processing route are consistent, the consumer
feels the cognitive fit and thus better towards the targeted object (Vessey and Galletts, 1991).
This indicates that consumers who actively search online reviews are mostly likely following
the central information processing route, thus accepting eWOM with low emotionality. Given
positive eWOM would stimulate consumer purchase intention, the higher likelihood of
receiver accept eWOM the greater chance they will purchase the product/service. Voss et al.
(2003) elaborated the mechanism by dividing the involvement into affective involvement and
hedonic attitudes that is a result of emotion attachment, and cognitive involvement predicts
utilitarian attitudes which arises when personal relevance is based in logic. Both dimensions
of attitudes will affect consumers purchase intention.
Relationship Closeness
According to Berscheid, Snyder, and Omoto (1989), the closest relationships are defined
between romantic partners, friends, and family members. The categorization comes from the
research carried out among colleague students at the University of Minnesota. Students were
asked to report one of the relationships they believed to be the closest of all their
interpersonal relationships. 47% of them considered romantic relationship as their closest
relationship, followed by friendship which 36% nominated it as their closest relationship. The
third frequently mentioned type was family relationship while that was only raised by a small
portion of people (14%).
Thus, in this study, in the light of findings from Berscheid etc. (1989), we define that “close
relationships” encompass romantic, friend, and family relationships. Other than these three,
all other type of relationships fall into the category - “distant relationships”, such as work
relationships, strangers.
Within the closest relationships, the closeness however can been further conceptualized in
three dimensions: strength, frequency, and diversity (Berscheid, Snyder, Omoto, 1989).
Strength is assesses by the degree to which participants believe that they are influenced by
the counterpart. The relationship closeness would have a causal effects on their daily
behaviors, plans, goals. Frequency, refers to the time that individuals spend together with
their partners. Research shows (Attridge, 2013) that the more time people interact with each
closeness – diversity – is based on the idea that the wider range of different activity domains
in which partners engage in together, the closer they will be.
Since relationship closeness influences people’s behavior, it is reasonable to assume it as one
of the crucial factors that influences people’s purchase behavior. Namely, the relationship
between information sender and the receiver might determine the receiver’s purchase
intention. This assumption actually has some theoretical substantiation. Brown and Reingen
(1987) hypothesized that generally strong relationship is perceived as more influential than
weak relationship in decision-making. In addition, research in the field of referral
effectiveness shows that the closer relationship between the sender and the receiver, the
greater the impacts of the sender’s WOM on the receiver’s purchase decision (Bansal and
Voyer, 2000; De Bruyn and Lilien, 2008). The work from Schiffman and Wisenblit (2014)
presents instrumental insights into the underlying mechanism, as they argue that close
relationships would imply high perceived source trustworthiness which determines how much
receiver’s purchase intention would be affected by sender’s opinions.
Sender’s expertise
Sender’s expertise, refers to the knowledge that the sender has in a particular field, especially
related to the commented product (Gotlie and Sarel, 1991). People with high expertise,
namely experts, are defined as the ones who possess professional competency and skills. In
current research, senders with low expertise refer to normal consumers who did not obtain
relevant knowledge.
Compared to novices, experts should be believed as more likely to persuade people by
providing accurate, compelling, and valid assertions. The persuasion occurs via two different
mechanisms. When receiver’s motivation and ability to process the forthcoming message is
confidence in their own perceptions (Tormala, Briñol, and Petty, 2006). Nevertheless, when
receiver’s motivation and ability is moderate, sender’s expertise influences by determining
the extent to which receivers will carefully scrutinize the persuasive appeals (Debono and
Harnish, 1988; Heesacker, Petty, and Cacioppo, 1983; Tobin and Raymundo, 2009). Given
the fact that receiver generally perceive sender to be independent from marketers, the
manipulative intent of eWOM sender is therefore less detected and salient as compared to
marketer-dominated sources (e.g. endorsers, salesperson) (Bansal and Voyer, 2000; Smith et
al., 2005). Under this condition, the expertise of sender serves as a peripheral or heuristic cue
for receivers when assessing the message.
In the consumer behavior literature, considerable research has shown that expertise is a key
dimension under the construct: source credibility (Dholakia and Sternthal, 1977). The higher
level of perceived expertise signals a higher credibility of the sender, so consumers are likely
to use experts’ judgements to examine the validity of product evaluation (Friestad and
Wright, 1994; Maheswaranet et al., 1992), which in the end influences consumers’ purchase
intention. This concern is echoed by Gilly et al. (1998), who advocates the contention that the
sender’s expertise positively affects receiver’s purchase intention.
Social influence theory
Social influence theory holds that one of the most significant determinants of an individual’s
behavior is the influence of other people (Burnkrant and Cousineau, 1975). Considering I am
arguing that eWOM from other people will influence consumer purchase intention, this
theory can help provide theory substantiation of my argument.
Particularly, Duetsch and Gerard (1955) classified two types of social influence:
informational social influence and normative social influence. Information social influence
messages, whereas normative social influence posits that people’s attitudes and beliefs
change due to shared social norms.
Under this mechanism, Kelman (1961) further distinguished three different processes:
internalization, identification and compliance. Internalization, which normally would occur
under informational social influence, refers to the fact that individuals accept the influence
because they believe the information substantively. According to McGuire (1969),
internalization are most likely to exert when the information sender is perceived as credible,
such as an expert or very knowledgeable person. Other than internationalization,
identification and compliance are processes that when people follow normative social
influence. Identification holds that people desire to be identified and visible by complying
with others’ views. The compliance is resulted from people’s expectation of receiving
rewards or avoid punishments from others. This schema works prominently when
information senders are socially proximal with receivers (Cocanongher and Bruce, 1971). In
other words, close relationships are more likely to exert normative social influence than
distant relationships.
Hypotheses and conceptual framework
Previous research largely investigated the main effects of message appeal (Kotler and Keller,
2008), or some effects of source trustworthiness and source expertise (Hovland and Weiss,
1951; Giffin, 1967). It would be interesting to combine the three factors together and examine
their influences in consumer purchase intention. Based on social influence theory,
relationship closeness may trigger normative social influence, while sender’s expertise could
evoke informational social influence. Under different types of social influence, there are
likely to be interactions effects between different types of sources and different types of
emotionality, and an interaction between expertise and eWOM emotionality. Focusing on
this, I developed following hypotheses as explained in the following sections.
Relationship closeness and eWOM emotionality
When eWOM emotionality is high, the message is subjective and based on personal feelings.
Because such comment is primarily affective and sensory, it triggers normative social
influence under which people’s need to be liked rather than right is more likely to be evoked.
In compliance or identification process, close relationship exerts much more influences to
consumers than distant relationship, because consumers hopes to achieve a favorable reaction
from the former by aligning their opinions.
H1 (a): With high eWOM emotionality, close relationship, as compared to distant relationship, results in higher level of purchase intention.
When eWOM emotionality is low, the message is objective and based on facts. With this type
of message, people are more likely to experience informational social influence, since they
tend to use the information provided by others as an evidence about the reality of the object
under consideration. Internalization process is then applied. However, because relationship
closeness is not an influential factor that will manipulate the process, close relationship and
distant relationship do not differ on the persuasion effectiveness scale under this condition.
Some may argue that close relationship can be perceived as a credible source, which would
be a key variable to induce people’s behavior in the internalization process (Kelman, 1961).
But some also advocate that distant relationship may possess capabilities that one’s close
social circle does not have and this raises the significance of their comments (Steffes and
Burgee, 2009). Both arguments seem convincing, and since they offset each other, it is
believed that close relationship and distant relationship would exert similar effectiveness on
H1 (b): With low eWOM emotionality, close relationship, as compared to distant relationship, results in same level of purchase intention.
Sender’s expertise and eWOM emotionality
If eWOM emotionality is low, which means the review focuses more on concrete and
tangible attributes, people have a tendency to evaluate it cognitively. Internalization process
is then applied. Under this process, eWOM senders who are with high expertise are exerting
more influence as they are perceived more credible than normal consumers.
H2 (a): With low eWOM emotionality, high expertise, as compared to low expertise, results in higher level of purchase intention.
If eWOM emotionality is high, which means the review conveys more emotional judgment,
people have a tendency to evaluate it affectively. Under normative social influence, however,
experts and novices have similar influence, as consumers do not feel the need to please
experts in order to be rewarded.
H2 (b): With high eWOM emotionality, high expertise, as compared to low expertise, results in same level of purchase intention.
Moreover, combining the constructs of relationship closeness and sender’s expertise
generates the following four types of recommenders: (1) close knowledgeable recommender,
(2) close unknowledgeable recommender, (3) distant knowledgeable recommender, (4)
distant unknowledgeable recommender. Since the relationship closeness and sender’s
expertise positively influence consumers’ purchase intention, eWOM effectiveness from a
close and knowledgeable person would be considered as strong whereas that from a distant
and unknowledgeable person would be regarded as weak. On top of that, because the level of
emotionality negatively relates to purchase intention, for distant and knowledgeable person
and close and unknowledgeable person, a highly emotional eWOM would have weak
influence while a highly rational eWOM would have strong effect. The prediction is
presented in Table 1.
Table 1. The Type of Recommender Relationship Closeness Close Distant Se nde r's E xp er tis e
Knowledgeable (1) eWOM effectiveness of High emotionality: Strong Low emotionality: Strong
(3) eWOM effectiveness of High emotionality: Weak Low emotionality: Strong
Unknowledgeable (2) eWOM effectiveness of High emotionality: Weak Low emotionality: Strong
(4) eWOM effectiveness of High emotionality: Weak Low emotionality: Weak
3. Research Methodology
Research design
Experimental methodology, particularly controlled experiment, was used to test the
hypotheses as it gives more control and its high internal validity ensures the precision of
low) x 2 (relationship closeness: close vs. distant) x 2 (eWOM emotionality: high vs. low)
between-subjects factorial design was used for hypotheses test. Rather than within-subjects
factorial design, between-subjects factorial design eliminates the carryover effects in the
research. So there were eight treatments in total as below. Participants were randomly
assigned to one of the eight treatments, and they were not aware of other treatments.
Experimental Weibo page design
Since their emergence, social network sites (SNS) such as Facebook, Twitter, Pinterest have
attracted millions of users. Those SNSs are with various technological affordances,
supporting a wide range of interests and practices. People can not only use the sites to
establish connections with others, but also receive all kinds of information from continuously
personal updates in their established social networks composed of friends, colleagues and
other acquaintances. Since this study is to research the influence of personal recommendation
on others, the characteristics of SNS make it a desired medium. However, because it is almost
impossible to access participants’ account to manipulate and control the experimental
settings, fictitious pages were designed by using Weibo as the template. Weibo, the Chinese
version of Twitter, now is the most popular SNS in China. It has 198 million monthly active
users (Paul, 2015), most of whom have integrated it into their daily lives. I used Weibo
instead of Twitter as the research medium, because in this study participants are mainly
In the experiment, each participant was given two fictitious Weibo pages to read: the search
result page with sender’s tweet, and the sender’s personal homepage. The search result page
(e.g. see figure 1) often comes up when people want to look for eWOM of certain objects.
With the key words searched, all related eWOM show up. The sender’s personal homepage
(e.g. see figure 2) displays his/her name, self-description, (authorized) title, and number of
followers/ following/ tweets. Besides, their latest two tweets and their latest updated
articles/album are shown to participants in the same page.
Figure 1. Search result page showing sender’s tweet
Participants and the object
222 hundred participants took part in the experiment. Considering the budget of operation
costs and constrain of time, this study used convenience sampling. It is a continuous process
to find participants for a survey, until the sample size requirement is attained (Schiffman and
Wisenblit, 2014). The sampling is appropriate because young adults are the major active
users of SNSs to spread eWOM which represent the population that the current study attemps
to target (Wang, Jackson, Gaskin, and Wang, 2014).
iPhone 6S was chosen to be the experimental product in the study. There are four reasons for
this choice: (1) electronic products are frequently purchased online; (2) among them, mobile
phone is the one that customers use eWOM most to make purchase decision for (Fan and
Miao, 2012); (3) it is one of most popular stuff that college students want to have (Buster,
2015); (4) since iPhone 6s is newly launched, at this stage, people tend to rely on external
information/ reviews to help evaluate the product.
Pre-experiment questionnaire
For participants who enter into treatments that involve “close relationship”, they were asked
to fill a questionnaire before reading the scenario. In the questionnaire, they were required to
list five people from their close relationships, and rate those five people’s expertise with
regard to iPhone 6S. In other words, participants would nominate the most knowledgeable
person and or the least knowledgeable one among the five. These questions helped
manipulate relationship closeness and sender’s expertise in the experiment.
Relationship closeness. For close relationship, the construct was manipulated by using one of the five people whom were nominated in the pre-experiment questionnaire. However,
since it is almost impossible to obtain all the details of every participants’ close relationship
beforehand, a mock-up profile picture and name with mosaic effect were created, and
participants were asked to image this account as from his/her close relationship. The mosaic
effect is added in order to avoid the distraction from a mock-up account on participants’
judgement. In contrast, for distant relationship, a stranger with fictitious name was used. No
mosaic effect is added as contrary to the close relationship condition. Since under this
condition, it is assumed that the participants did not know this person before, the name and
profile picture, unlike the manipulation in close relationship, can be clearly shown to
participants. Besides, for close relationship, “mutual-following” sign is shown in the search
result page, whereas distant relationship is not. Based on Berscheid et al (1989), in this study,
relationship closeness is assessed from three dimensions: strength, frequency, and diversity.
Thus, three items which measure three dimensions, with 5-point Likert scale were adopted in
the experiment.
Sender’s expertise. Since senders are divided into close and distant ones, the manipulation of their expertise should also be classified into two ways. For close relationship, high and low
expertise were manipulated by using the nominated ones in the pre-experiment questionnaire.
For distant relationship, high expertise was manipulated by displaying their authorized
qualifications/ titles (e.g. obtaining the Diploma in Professional Cookery), whereas low
expertise would not have any authorization. For example, a professional cook is more likely
to be considered as an expert than a normal consumers as for cuisine recommendations.
Besides, the expertise is also manipulated by displaying different amount of professional
languages and industry-related tweets. The more expertise the recommenders is, the more
addition, the number of their followers and the times that their tweets being forwarded are
also indication of their expertise. It is implied that the more followers they have and more
tweets being forwarded, the more expertise they are perceived with (Liao, Wagner, Pirolli,
and Fu, 2012). In this experiment, Roobina’s (1990) 5-item, 5-point semantic differential
scale was used to evaluate sender’s perceived expertise.
eWOM emotionality. The study classified eWOM appeal into high emotional and low emotional. The tweets in high emotionality may communicate strong personal attitudes and
feelings, whereas those in low emotionality may emphasize on price/quality ratio, function,
and benefit. Based on Wu and Wang (2011), in this study, ten items with 7-point Likert scale
were developed to measure eWOM emotionality characteristics.
Dependent variables
Purchase intention. Measurement of this construct was carried out by using two questions, 7-point Likert scale, ranging from extremely unlikely (1) to extremely likely (5). The scale
items were chosen based on Park, Lee, and Han (2007) research, and corresponding questions
are “How likely are you going to purchase this product?” and “How likely are you going to
recommend this product to others?”
Procedures
Participants will be invited to join the experiment conducted on Qualtrics. They are randomly
assigned to one of the treatments. If they are entering the treatments of “close relationship”,
they are requested to fill out the pre-experiment questionnaires. If not (i.e. entering the
treatments of “distant relationship”), they will be led to the scenario directly. There are eight
treatments in total, so eight scenarios as well. Following the scenario, there are questions that
information regarding the respondents, such as age, gender, and user experience with SNS.
Each experiment takes about 5-10 minutes to complete.
Pretest
A pretest was conducted on eight participants to examine the manipulation of the treatment
and if there are other potential problems in the research design. Participants entered the
pretest in the exactly same procedure as it would have been shown in the actual experiments.
Feedback and suggestions were taken from three aspects (i.e. readability, construct validity,
measurement validity) afterwards. Overall, participants perceived that the design is clear and
valid to measure the proposed concepts. Major amendments include rephrasing questions and
adjusting the order of constructs that are to be measured. Measures for purchase intention are
moved from the last bulk to the first one, since it is believed that people’s purchase decision
may be biased by the questions for the other constructs if it is asked in the end of the survey.
As purchase intention is the most important construct in this study, being placed in the first
bulk can also improve the accuracy of the results.
4. Results
Data cleaning
Firstly, the frequency check was conducted to identify errors in the data entry. The frequency
distribution shows how many times each of the scores of the items are in the data set. Since
there is no rare frequency showing in the frequency table, it is believed that no error is in the
data entry. Then, to handle the missing data in the data entry, listwise deletion was adopted.
This means I only examined cases without any missing data in any variable. No missing data
was found in the data set. Outliers are cases that are extremely inconsistent with the rest of
these variables, and if the Z-scores that exceeded 3 in absolute value were determined as
outliers (Sincich, 1996). By using this approach, no outlier was identified, meaning 222 valid
data points in total.
I included four counter-indicative items, so they are phrased in such a way that an agreement
with the item represents a low level of the construct being measured. This enables me to
detect acquiescence bias (i.e. the yea-saying bias) which means the tendency to respond in an
indiscriminately positive or negative way. Therefore, the four counter-indicative items (i.e.
rStrength2, rStrengths4, rEmotionality2, and rEmotionality4) need to be firstly recoded before analyzing. After reversing all scores, four new variables (i.e. Strength2, Strengths4,
Emotionality2, and Emotionality4) were added.
Sample Profile
Among 222 respondents, the percentage of female is slightly higher than that of male (i.e.
59% vs. 41%). The average age is 25 years old, with the minimum 18 years old and the
maximum 35 years old. This meets our expectation, since Weibo is mostly popular among
Chinese young adults. 87.8 percent of them has more than one year experience with this SNS,
and averagely they log into it one to few times per week. This tells that in general
respondents are all experienced with the platform, which helps to reduce the effect of the
SNS itself, since people tend to behave more naturally in a context that they are familiar with.
Reliability and Validity Tests
To test the construct validity, a principal component analysis (PCA) was conducted using
SPSS. The Kaiser–Meyer–Olkin measure verified the sampling adequacy for the analysis,
KMO = .859. Bartlett’s test of sphericity χ² (190) = 3656.429, p < .001, indicated that correlations between items were sufficiently large for PCA. An initial analysis was run to
obtain eigenvalues for each component in the data. Four components had eigenvalues over
Kaiser’s criterion of 1. In agreement with Kaiser's criterion, examination of the scree plot
revealed a levelling off after the fourth factor. Consequently, four factors were yielded and
rotated with a varimax with Kaiser normalization rotation. The items that cluster on the same
factors suggest that factor 1 represents relationship closeness, factor 2 perceived sender’s
expertise, factor 3 stands for purchase intention, and factor 4 describes ewom emotionality (see Appendix 2 for complete factor analysis results). Since factor analysis indicated that the
item measuring attitudes towards eWOM was loaded together with the other two items
measuring purchase intention (eigen-value = 2.365, Cronbach’s a = 0.861), the three items
were averaged to compose a purchase intention score. Except that, all items are loaded on the
factors as expected. Moreover, because all the scores are above 0.608, it indicates a good
construct validity (Cook and Campbell, 1979).
Reliability is measured by using Cronbach’s alpha coefficient (see Table 1 diagonal). As a
rule of thumb, academics regard a construct to have adequate inter-item reliability if
Cronbach’s alpha coefficient exceeds 0.70 (Leary, 2007). Therefore, three constructs indicate
high reliability (i.e. purchase intention: 0.845, relationship closeness: 0.964, sender’s
expertise: 0.930), and one construct has acceptable reliability (i.e. ewom emotionality:
0.704). The corrected item-total correlations show that all the items have a good correlation
with the total score of the scale (i.e. all above .30). Also, none of the items would
substantially affect reliability if they were deleted.
Correlation analysis was used to check if there is association between identified variables
(see Table 1). To compute the scale means, TOTExperience and TOTFamilarity was created
by computing the mean of each construct. As shown in Table 1, SNS experience and product
considered as covariates. However, both relationship closeness and sender’s expertise are
significantly associated with purchase intention, which meets my expectation.
Table 1. Means, Standard Deviations, Correlations
Variables M SD 1 2 3 4 5 6 7 8 1. Gender 1.590 0.493 - 2. Age 25.140 2.935 -0.256** - 3. Product familarity 3.323 0.960 0.032 0.016 - 4. SNS experience 3.232 0.730 0.134* -0.068 0.869** - 5. Relationship closeness 2.854 1.970 0.037 0.015 0.038 -0.002 (0.964) 6. eWOM emotionality 3.839 1.193 -0.017 0.138* -0.06 -0.025 -0.007 (0.704) 7. Sender's expertise 3.341 0.942 -0.034 -0.122 0.098 0.042 -0.013 -0.167* (0.930) 8. Purchase intention 3.697 0.795 -0.096 -0.028 0.026 0.055 0.235** -0.039 0.319** (0.861) **Correlation is significant at the 0.01 level (2-tailed).
*Correlation is significant at the 0.05 level (2-tailed).
Manipulation Checks
To verify the manipulation of the independent variables, the independent-samples T test was
performed. For each independent variable, the results show that the associated population
means are significantly different between two groups (see Table 2). Therefore, it can be
concluded that three variables (i.e. relationship closeness, eWOM emotionality, and sender’s
expertise) were all successfully manipulated.
Table 2. Manipulation Checks
Independent Variable Groups N Mean Std.
Dev T-statistics Relationship closeness Close 112 3.936 0.756 t= 4.720
eWOM emotionality High emotionality 114 3.807 0.708 t-= 2.118
Low emotionality 108 3.582 0.866 p< 0.05*
Sender's expertise High expertise 109 4.132 0.704 t= 9.473
Low expertise 113 3.278 0.639 p< 0.01** *p<0.05, **p<0.01
Hypotheses Tests
Main effects. The three-way ANOVA was conducted to determine if there is an interaction effect between three independent variables on purchase intention, as well as the main effects
of the independent variables. All tests of hypotheses were run at a 5% level of significance.
Table 3 summarizes the results of the ANOVA test. The results suggest that all main effects
are significant. Particularly, there was a significant main effect of eWOM emotionality on
purchase intention, F(1, 214) = 12.974, p < .01, η² = .057, indicating a low effect size. Besides, there was a significant main effect of relationship closeness on purchase intention,
F(1, 214) = 54.133, p < .01, η² = .202, indicating a high effect size. Also, there was a significant main effect of sender’s expertise on purchase intention, F(1, 214) = 180.108, p < .01, η² = .457, indicating a high effect size.
Table 3. ANOVA Results
Independent Variable df Mean Square F Sig. η² eWOM emotionality 1 2.97 12.974 .000** .057 Relationship closeness 1 12.393 54.133 .000** .202 Sender's expertise 1 41.234 180.108 .000** .457
eWOM emotionality * Relationship closeness 1 6.34 27.691 .000** .115
eWOM emotionality * Sender's expertise 1 17.211 75.178 .000** .26 Relationship closeness * Sender's expertise 1 1.195 5.221 .023* .024 eWOM emotionality * Relationship closeness *
Sender's expertise 1 10.944 47.804 .000** .183
Total 222 Dependent variable: Purchase intention
*p<0.05, **p<0.01
eWOM emotionality and relationship closeness. The table presents that there are significant two-way interactions between eWOM emotionality and relationship closeness, eWOM
emotionality and sender’s expertise, and also a significant three-way interaction of three
variables. To further explore the interaction effects, a simple main effect analysis was
adopted. For eWOM emotionality and relationship closeness, the results revealed that under
the condition of high eWOM emotionality, purchase intention is significantly higher
(F(1,214)=81.723, p<.01) for eWOM comes from a close relationship (N=58, Mean=4.193,
SD=0.665) than that from a distant relationship (N=56, Mean=3.407, SD=0.502). Hence
hypothesis 1 (a) is supported. On the other hand, under the condition of low eWOM
emotionality, the difference between close relationship and distant relationship is not
significant (F(1,214)=2.140, p=n.s.). Therefore, hypothesis 1 (b) is also supported. The
relationship is shown in Figure 1.
eWOM emotionality and sender’s expertise. In terms of eWOM emotionality and sender’s expertise, under the condition of low emotionality, purchase intention is significantly higher
(F(1,214)=237.909, p<.01) for eWOM comes from sender’s with high expertise (N=55,
Mean=4.280, SD=0.482) than whom with low expertise (N=53, Mean=3.650, SD=0.501).
Thus, hypothesis 2 (a) is supported. On the other hand, under the condition of high
emotionality, purchase intention is significantly higher (F(1,214)=11.577, p<.01) for eWOM
comes from sender’s with high expertise (N=54, Mean=3.982, SD=0.852) than whom with
low expertise (N=60, Mean=3.650, SD=0.504). In this sense, hypothesis 2 (b) is not
eWOM emotionality and recommender type. Moreover, as proposed earlier, grouping relationship closeness and sender’s expertise produces four types of recommenders (i.e. (1)
close knowledgeable recommender, (2) close unknowledgeable recommender, (3) distant
knowledgeable recommender, (4) distant unknowledgeable recommender). To test the
effectiveness of these four types of recommenders, a factorial ANOVA was conducted (see
Table 5). The results revealed a significant interaction between the type of recommender and
eWOM emotionality (F(1,214)=48.924, p<.01, η² = .407). Table 5 shows the descriptive statistics of eWOM effectiveness for four types of recommender under the condition of high
emotionality and low emotionality. As predicted, eWOM from a close and knowledgeable
person has the strongest influence (N=56, Mean=4.429, SD=0.55) whereas that from a distant
Surprisingly, as the table presents, except the treatment of eWOM from a close and
unknowledgeable person, other treatments indicate that eWOM emotionality is negatively
associated with people’s purchase intention.
The relationship shown in Figure 4 helps us to further explain the results. As it indicates,
when eWOM is emotional, compared to closed and unknowledgeable people, those who are
distant and knowledgeable people’s recommendation would lead to a higher purchase
intention (t=0.506, p<0.01). By contrast, when eWOM is rational, a close and
unknowledgeable person would exert much higher persuasiveness than a distant and
knowledgeable person (t=1.286, p<0.01).
Table 5. ANOVA Results for Recommender Type and eWOM Emotionality Independent Variable Df Mean Square F Sig. η²
eWOM Emotionality 1 2.97 12.974 .000** 0.057 Recommender Type 3 18.567 81.099 .000** 0.532 eWOM Emotionality* Recommender Type 3 11.201 48.924 .000** 0.407 Error 214 0.229 Total 222
Dependent variable: Purchase intention *p<0.05, **p<0.01
Table 6. The Type of Recommender
Mean (N,SD) Relationship Closeness
Close Distant Se nde r' s E x p ert
ise Knowledgeable High emotionality: 4.66 (28, 0.51) (1) eWOM effectiveness of
Low emotionality: 4.20 (28, 0.50) (3) eWOM effectiveness of High emotionality: 3.76 (30, 0.47) Low emotionality: 3.08 (26, 0.51) Unknowledgeable (2) eWOM effectiveness of High emotionality: 3.25 (26, 0.44) Low emotionality: 4.36 (27, 0.46) (4) eWOM effectiveness of High emotionality: 3.54 (30, 0.52) Low emotionality: 2.64 (27, 0.40)
5. Conclusion and Discussion
This study investigated the interaction effect of eWOM emotionality, relationship closeness,
and sender’s expertise on people’s purchase intention. Given the fact that the levels of
relationship closeness and sender’s expertise signal different credibility for eWOM receivers,
the two factors are the focal examination of the experiment. Besides, according to the social
influence theory, the emotionality of eWOM can manipulate the way people process the
information, and the magnitude of the effect then depends on the relationship closeness and
sender’s expertise, so the interaction effects of different levels of eWOM emotionality on the
two other factors were assessed in the experiment.
The results of the experiment demonstrate that all three targeted independent variables have
significant impacts on consumers’ purchase intention, with sender’s expertise (η² = .457) was found to have the strongest effect. This provides a strong evidence that consumers’ purchase
decisions are influenced by the expertise of people who comment on the products, and
eWOM from an expert tends to be more influential that that from a rapport.
Furthermore, the study shows that though generally relationship closeness affects consumers’
any significant effects. Conversely, when eWOM is in high emotionality, the relationship
closeness significantly influences the people’s purchase intention, with a closer relationship
implying a higher purchase intention. The results are the same as hypothesized with the
support of social influence theory. Specifically, when eWOM is emotional, consumers are
more subject to normative social influence as exerted by a close relationship in order to make
their values and behaviors more congruent to their close friends, families or partners.
However, this influence is discounted when eWOM is rational. As the nature of rational
eWOM is objective and attributes-focused, informational influence is more likely to be
applied than normative influence. Further, since both close relationship and distant
relationship can provide the same amount of information, their effect on consumers’ purchase
decisions would not differ significantly, provided with a rational eWOM.
In addition, the current research also suggests that a higher eWOM emotionality will result in
a high purchase intention, despite of the extent of sender’s expertise. This is contradictory to
our hypothesis, since under the treatment of high eWOM emotionality, eWOM from an
expert should have yielded the same results as from a novice. A probable interpretation of
that may due to the fact that the emotionality of eWOM from an expert may be reduced even
it was manipulated to show high emotionality. Thus, participants would have the tendency to
exert the internalization process rather than the identification or compliance process, which
means participants will objectively scrutinize eWOM other than basing their decisions on
subjective feelings. In this sense, senders who are knowledgeable and skilled would definitely
be more persuasive than novices. To confirm our speculation, an independent T test was run
to check the mean differences of eWOM emotionality between experts and novice (see Table
X). The results indicate that there was a significant difference between the two groups, with
eWOM from the high expertise group implying a lower emotionality. Through further
did not support hypothesis 2(b), the problem comes from the manipulation of eWOM
emotionality, instead of the argumentation based on social influence theory.
Table X. Mean Comparison
Variable Levels N Mean Std. Dev T-statistics Sender's expertise High expertise 109 3.6651 1.29154 t= -2.143 p< 0.05
Low expertise 113 4.0066 1.06824 Dependent variable: eWOM emotionality TOT
Also, the test on eWOM effectiveness of four types of recommenders points out that close
and knowledgeable recommender infers the highest persuasiveness, while a distant and
unknowledgeable one shows the lowest persuasiveness. However, inconsistent with my
prediction, only treatment of eWOM from a close and unknowledgeable person shows the
negative effect of eWOM emotionality, all the rest treatments show the reverse results. That
is, the higher level of eWOM emotionality infers a higher purchase intention. This could be
interpreted with the ELM model. Previously, I assumed that people who search eWOM are
with high involvement, so they will process the information via central route. Under this
route, it is likely that they will generate a great amount of cognition and elaborate the
message with careful and thoughtful consideration. Thus, a rational message which is
supposed to provide more cognitive arguments should be more persuasive than an emotional
message. However, the empirical results show that the level of eWOM emotionality enhances
its effectiveness. This implies that eWOM searchers may be likely to exert the peripheral
route. Under this route, people tend to judge by emotions (Petty and Cacioppo, 1996). As
argued by Vessey and Galletts (1991), when the information-processing route is consistent
with the message appeal, consumers will have a cognitive fit and better attitude towards the
product. Therefore, since the eWOM emotionality in this study is found positively related to
peripheral route. More emotional words appearing in eWOM will enhance the effect, which
leads to higher purchase intention.
Research Implication
This study makes several theoretical contributions. Firstly, this study is mainly built on social
influence theory to investigate the interaction effects of eWOM emotionality, relationship
closeness, and sender’s expertise. Previous research on eWOM effectiveness has examined
the main effects of each factor. However, this is the initial study that examined the interaction
of the targeted three factors. As evidenced by the two two-way interactions between eWOM
emotionality and relationship closeness and between eWOM emotionality and sender’s
expertise, eWOM emotionality has different impacts on the level of relationship closeness
and sender’s expertise on consumer purchase intention, which is due to the different
mechanism explained by social influence theory. The normative social influence illustrated
reason of the high effectiveness of emotional eWOM exerted by rapports, and the
informational social influence explained the persuasiveness of experts.
Secondly, this research investigated the influence of eWOM emotionality on consumer
purchase intention, which was rarely studied before. As one of the few studies, Wu and
Wang’s (2011) research advocates for the usefulness of rational appeals in the message.
eWOM emotionality was perceived to be negatively related with the quality of the message
as a message contains more emotional appeals would be considered as less objective and
informative. Since the quality of the message is closely linked to people’s purchase decision
(Doh and Hwang, 2009; Cheung, Lee, and Rabjohn, 2008), this study examined the direct
relationship between eWOM emotionality and purchase intention. Unexpectedly, the findings
show that the level of emotionality will help increase purchase intention. This may due to the
eWOM. Therefore, it is important to bear in mind that consumers tend to use simple
inference to judge the information, even though their actively opinion seeking behaviors
make them appear to have high involvement.
Thirdly, this study provides a new insight towards the research on the effect of relationship
on eWOM effectiveness. Previous studies mostly agreed with the theory coined by
Granovetter (1973) and Brown and Reingen (1987), who advocated the “strength of strong
ties”. It is believed that strong ties generally exert more influence that the weak tie.
Nevertheless, in the current research, it needs to be pointed out that the strength of the strong
tie is under the condition of an emotional eWOM. In other words, if eWOM is articulated in a
more rational way, the strength of the strong ties would be discounted. The findings support
that people react to eWOM with different cognitive process on the level of eWOM
emotionality.
Fourthly, in terms of the influence of sender’s expertise, the interaction with eWOM
emotionality assessed in this paper shed new light on purchase intention. Prior research
demonstrated that the higher expertise implies higher credibility, which helps stimulates
purchase intention (Doh and Hwang, 2009; Cheung and Thadani, 2012). In this study, the
effectiveness of this factor on purchase intention is subject to the level of eWOM
emotionality. The expertise of sender would exert its most effects when eWOM is rational
instead of emotional. Though the results of the treatment of sender’s expertise*high eWOM
emotionality are not the same as hypothesized, the reasons are attributed to the manipulation
of eWOM emotionality in the experiment rather than the theory. Thus, in general, the
proposed interaction effect between reiviwer’s expertise and eWOM emotionality is
Fifthly, this study summarized eWOM effectiveness of four types recommenders – close and
knowledgeable recommender, close and unknowledgeable recommender, distant and
knowledgeable recommender, and distant and unknowledgeable recommender, under
different levels of eWOM emotionality. This attempt not only offers new insights for the
recommender type research but also provides empirical guidance for interested practitioners
as discussion in the next section.
Managerial Implication
Since eWOM implies that manufacturers are no longer in exclusive control of information
flows (Lee, Cheung, and Sia, 2006), it is imperative that they respond with tactics to monitor
or influence the content. Therefore, the findings of this study provide practical implications
on what factors they can manipulate to leverage the positive eWOM and how they can help
with promotion. The implications are discussed based on four variables in this study:
relationship closeness, sender’s expertise, eWOM emotionality, and the recommender types.
Relationship closeness. Supported by the results, close friends, families or couples will exert stronger influence than strangers when they recommend products online. Thus, firms should
encourage consumers to share their eWOM among the social platforms. Additional benefits
provided by the firm, such as discount voucher, membership etc. can help stimulate the
behavior. Compared with traditional WOM, the disseminate of eWOM is easier to guaranteed
and tracked. For example, any product review shared on Weibo or Facebook can be
immediately detected by firms, so firms are able to respond quickly to minimize the
occurrence of negative eWOM and amplify the positive eWOM. Social networks allow
consumers to share and locate simultaneously, which saves much money and efforts for firms
to advertise. More importantly, this kind of consumer-generated advertisements appear more
Sender’s expertise. For firms that choose experts as endorsers, it is recommended that those experts spread the word in a rational way. Experts are closely associated with professionality
and authority, so rational eWOM is more fit with the characteristics so that it exerts more
influences. Furthermore, since the level of expertise of the sender increases consumers’
purchase intention, it is also suggested to manipulate the profile page in the social network in
order to help audience to perceive the endorser’s product expertise. For example, sections
like the self-description, the skill authentication, the number of followers, the recent updates
can all help lessen audience skepticism towards the endorser’s expertise.
eWOM emotionality. The results for all the eight treatments in this research indicate that higher eWOM emotionality implies higher purchase intention. Thus, even under the least
favored situation that eWOM is read from an unknowledgeable stranger, a rise of its
emotional appeals showing in eWOM would still help. Therefore, to achieve that, firms can
pre-offer more emotional key words that can be used in the feedback for senders, such as
“satisfied”, “chill”. In that way, with the available key words, there is a higher chance that
consumers will pick up the emotional words directly from the list. EWOM emotionality is
then guaranteed.
Recommender types. As the findings suggest, close and knowledgeable recommenders who communicate in an emotional way will result in the highest purchase intention. Thus, firstly,
firms should always encourage the behavior of sharing eWOM among consumers’ online
social circle. Moreover, the content of eWOM is preferred to contain more professional
words. This can be done by asking certain questions that can implicitly guide consumers to
generate more cognitive thinking. For example, questions like: How many times are you
using our products? Which function brings you the most benefits? How and why?
One of the limitations of this study relates to the relationship closeness manipulation. This
study only chose two ends (i.e. the closest person vs. stranger) in the relationship continuum
to research. However, many scholars hold that in fact there are many levels in between which
may influence people’s behavior differently (Marsden and Campbell, 1984; Mittal, Huppertz,
and Khare, 2008; Petróczi, Nepusz, and Bazsó, 2007). Future research can include more
levels in order to obtain comprehensive understandings of the influence of this factor. Besides,
social network site as a newly developed medium involves more complicated relationship
structure than traditional offline social connection. For example, people can hear a eWOM
from friend’s friend but whom they do not have any contact before. This relationship is
obviously stronger than the typical stranger that we are referring to. Therefore, it suggests
that future study may focus on relationship structure as an antecedent.
Another limitation to this study is the manipulation of emotionality. Due to the fact that there
is no clear definition for a pure emotional and a pure rational appeal, the manipulation of this
factor in this study can be some biased. For example, for the treatment involves high
expertise, high eWOM emotionality appears to be hard to reach. One the one hand, experts
need to use more professional terms to indicate their expertise. On the other hand, these
professional terms will reduce the emotionality of eWOM. Thus, future research should be
aware the effect from other factors when examining eWOM emotionality, as other factors
may indirectly enhance or reduce the level of eWOM emotionality even when the content
stays the same.
In addition, while this paper focuses on the interaction effect of expertise and eWOM
emotionality, only sender’s expertise is considered. In fact, previous studies also revealed that
reader’s expertise may also affect consumer behaviors (Johnson and Russo, 1984; Park and
Kim, 2009; Punj and Staelin, 1983). In future, studies may be conducted on examining the
Besides, this study chose to study eWOM on the social network site, but it is only one of the
online social platforms. There are also other forms such as blogs and forums which are
popular for consumers to find the discussion of the products. Different forms have their own
characteristics and target groups. For example, blogs tend to be used more often with experts
and forums are more likely to be surfed by fans. In this way, consumers’ purchase intention
may be affected by the form as well. Future study could compare the effectiveness of eWOM
on different forms.
Finally, due to the limit of the available time in this experiment, the research was conducted
in a static setting rather than an interactive one. It is noted that companies are willing to
involve in eWOM in order to build a good brand reputation, so it is realized that companies
will respond eWOM and interact with consumers. In the future, a more dynamic setting could
be taken into account to investigate the effectiveness of the interaction in eWOM
conversation.
6. References
Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The Relationship Closeness Inventory: Assessing the closeness of interpersonal relationships. Journal of personality and Social Psychology, 57(5), 792.
Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The Relationship Closeness Inventory: Assessing the closeness of interpersonal relationships. Journal of personality and Social Psychology, 57(5), 792.
Bhatnagar, A., & Ghose, S. (2004a). A latent class segmentation analysis of e-shoppers. Journal of Business Research, 57(7), 758-767.
Bhatnagar, A., & Ghose, S. (2004b). Segmenting Consumers Based on the Benefits and Risks of Internet Shopping. Journal of Business Research, 57(12), 1352–1360.
Bloch, P. H., Sherrell, D. L., & Ridgway, N. M. (1986). Consumer search: An extended framework. Journal of consumer research, 119-126.