Sie sind auf Seite 1von 11

Journal of Consumer Behaviour, J. Consumer Behav.

11: 94104 (2012)


Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/cb.385

Factors affecting consumers trust in online product reviews


PRADEEP RACHERLA1*, MUNIR MANDVIWALLA2 and DANIEL J. CONNOLLY3
1
West Texas A&M University, Canyon, TX , USA
2
Fox School of Business, Temple University, Philadelphia, PA , USA
3
Daniels College of Business, University of Denver, Denver, CO , USA
ABSTRACT
In this study, we apply the uncertainty reduction theory from communication to delineate the antecedents of consumers trust in online product
reviews. We test the competing effects of information content (argument quality) and social component (perceived background similarity) on
consumers trust in reviews. We also hypothesize that the strength of the effect is moderated by consumers involvement. To test the hypotheses,
we adopted a 2  2  2 repeated measures experimental design. The results show that both argument quality and perceived similarity contribute
to increased trust but in varying degrees. The results provide mixed support to the moderating role of involvement. While argument quality
played an important role in the high involvement mode, perceived similarity explained more variance in the low involvement mode. We discuss
the implications of these results for both research and practice.
Copyright 2012 John Wiley & Sons, Ltd.

INTRODUCTION
Product review systems (PMS; e.g., Amazon.com, TripAdvisor.
com), a form of online word-of-mouth (eWOM), have become
one of the most popular information sources for modern
consumers. PRS are less personal but more ubiquitous WOM
platforms wherein consumers can post reviews about products
and services. These reviews are widely accessible to other
consumers but are disseminated only when other consumers
consult these reviews.
Online WOM has spurred increasing amounts of research
in the recent past (Ba and Pavlou, 2002; Dellarocas, 2003;
Dholakia et al., 2004; Chevalier and Mayzlin, 2006). However, one of the relatively underexplored areas of research
is the trust aspect of online reviews. Two problems are
generally cited in the case of PRS. Chevalier and Mayzlin
(2006) note that marketers have numerous incentives to
encourage promotional chat and therefore can (and do)
disguise their promotions as consumer recommendations. This
problem of genuineness is compounded by the fact that c2c
interactions in PRS occur in a computer-mediated environment
that lacks both nonverbal cues as well as social and contextual
cues (Einwiller, 2003). Consequently, consumers have to deal
with two different uncertainties. The rst is to infer the
characteristics of the product that is being considered. This task
is more difcult in the case of hedonic/experiential services
that are intangible and difcult to quantify in terms of features
and functionalities. Consumers also have to deal with the
uncertainty regarding the integrity and intentions of consumers
who provide reviews. Therefore, it is important to understand
how consumers manage to alleviate uncertainty and get the
information they want and are willing to trust. The question

*Correspondence to: Pradeep Racherla, West Texas A&M University, 2403


Russell Long Blvd, Canyon, TX 79016, USA.
E-mail: pracherla@mail.wtamu.edu
Statement of exclusive submission: The authors hereby assure that this
manuscript neither has been published in any journal nor is currently under
consideration for inclusion in any other journal, and that it will not be
submitted elsewhere until JCB has completed its review process.

Copyright 2012 John Wiley & Sons, Ltd.

that arises is why do consumers trust online reviews? The


primary objectives of the study are the following:
(1) To explicate the factors that drive consumers to trust
online reviews.
(2) To understand the moderating inuence of involvement
in consumers evaluation of online reviews.
To this end, we apply the uncertainty reduction theory
(URT) (Berger and Calabrese, 1975) to explore the antecedents
of consumers trust in online reviews. The basic premise of this
study is that online reviews are essentially c2c communication
episodes that have both informational as well as social
components. While informational content is the primary driver
of goal-directed behavior (in this case, product purchase), PRS
are also used as forums for social exchange, drawing people to
promote purchases, and regulate user behavior. Applying this
theory to online interactions, Ramirez et al. (2002) suggested
that in reduced cue and goal-directed environments, individuals seek any available information to alleviate uncertainty.
However, in the case of online reviews, the potential means
of interaction and information gathering are relatively
constrained. Consequently, whatever few social cues that are
available also become highly salient and important in forming
attitudes and determining behavior (Lea et al., 2001).
This paper is structured as follows: First, we briey
discuss the role of uncertainty and trust in online environments.
We then draw upon URT to delineate the constructs of our
research model. In the Methodology section, we present the
study context, research methods, and data collection strategies.
Subsequently, we present the results and discuss their
implications for both research and practice. We end the paper
with study limitations and future research directions.
LITERATURE REVIEW
Despite a decade since the emergence of e-commerce, the
uncertainty of online environments still makes many consumers
reluctant to engage in online exchanges. For customers
accustomed to face-to-face contact, the uncertainty and

Consumer trust in online product reviews


ambiguity inherent in virtual environments are likely to constrain interactions and transactions. Trust, by denition, mitigates such constraints. Previous studies emphasize that trust is
the central construct that sustains both interpersonal and economic relationships and that it is the ambient current of
e-commerce (McKnight, 2001). This stream of research has
primarily focused on various aspects of business-to-consumer
trust. However, recent research reveals that eWOM also plays
an important role in alleviating consumer uncertainty and
perceived risk pertaining to online transactions. The underlying
assumption of the trust literature is that humans have a fundamental need to reduce uncertainty, and understanding how trust
is built and maintained in an online environment is important to
explain the dynamics of eWOM.
URT AND TRUST IN PRODUCT REVIEWS
To explain how consumers alleviate uncertainty regarding
online reviews, we apply the principles of URT (Berger
and Calabrese, 1975). One of the aspects covered by URTrelated studies (Kellermann and Berger, 1984; Berger and
Gudykunst, 1991) is the communication between perfect
strangers, that is, interactions between individuals with no
previous interaction history. The fundamental assumption
of URT is that when strangers meet, their primary concern
is of uncertainty reduction or increasing predictability about
the behavior of both themselves and others in the interaction
(Berger and Calabrese, 1975: p.100). Uncertainty inuences
the level of liking, intimacy, and, consequently, the trust
between the communicators. Based on the background of
communicators and the context, initial interactions typically
tend to go through at least three distinct phases: entry phase,
personal phase, and attitude assessment phase.
(1) Entry Phase: Communication tends to be somewhat
structured, and the focus is more on the content and
quality of the verbal communication. At this stage, the
importance of nonverbal afnitive behavior is quite less,
as the individuals are still trying to assess the basic rules
of the communication episode.
(2) Personal Phase: As the uncertainty reduces over the period
of the communication, individuals focus more on selfdisclosure, that is, low intimacy demographic information.
For instance, Taylor and Altman (1966) had both college
students and navy recruits sort various conversational
topics on an intimacy continuum; they found that topics
falling into categories such as sociodemographics and
hobbies and interests were exchanged in the initial stages
of the conversation, and more intimate topics such as
political afliations and religion appeared much later in
the conversation.
(3) Attitude Assessment Phase: During this stage, individuals
begin to focus on personal attitudes and similarities/
dissimilarities in background information. In this regard,
most of the psychological theories emphasize that similarity is an antecedent to liking and continuance of communication (Newcomb, 1961). Perceived similarity not only
reduces uncertainty but also determines whether or not
the partners begin to trust each other.
Copyright 2012 John Wiley & Sons, Ltd.

95

The underlying assumption of this study is that online


reviews are essentially asynchronous c2c communication
episodes with both social and informational goals. One of
the main differentiating factors of PRS is that the interactions
here have neither a previous history nor an expectancy of
future interaction. Therefore, to reduce uncertainty, consumers
employ both active and passive strategies to evaluate information. Active strategies involve evaluating the contents of a
message to determine source expertise and bias (Buda and
Zhang, 2000). In the case of online reviews, this could mean
scanning other reviews of the same product. Passive strategies
include social observation wherein uncertainty is reduced by
witnessing the behavior of others and drawing conclusions.
In online environments, even though the availability of
traditional nonverbal and contextual cues is limited, other cues
or proxies for such information are used by information
seekers. As a result, alternate bases for comparison such as
an inclusive mindset or shared background might come into
play (including shared sociodemographic characteristics)
(Kollock and Smith, 1996).
The basic tenets of URT are also supported by theories
related to consumer information processing and search
behavior. The dual-process theory (Deutsch and Gerard,
1955) posits that two distinct sources of inuence affect the
persuasiveness of information: informational inuence and
normative inuence. Informational inuence is based on the
receivers judgment of the relevant content of the message.
Normative inuence refers to the inuence on individuals to
conform to the norms/expectations of others that are implicit
or explicit in the choices of a reference group. Under normative
inuence, individuals evaluate communication based not so
much on the content of the information itself but other cues
such as similarity in opinions, tastes, and preferences. In
addition, the elaboration likelihood model (ELM; Petty and
Cacioppo, 1986) introduces an important determinant of
information inuence: motivation of the receiver. Studies
based on ELM suggest that consumers, depending on their
motivation to process information, either consider productrelated information (central route) or focus on less cognitively
taxing cues (peripheral route). When the motivation to process
information is high, the product-related information in the
WOM is likely to be carefully considered with the rest of the
available information (Areni et al., 2000). This central route
becomes more relevant in the case of experiential products.
In the case of such products, WOM serves as the proxy for
ones own experience. Therefore, consumers carefully evaluate all the product and experience-related information. On the
other hand, consumers with low motivation to process
information are likely to seek some heuristic cues to minimize
their cognitive effort. In the case of online reviews, this could
mean looking for similarities in sociodemographic background
as a signal of t between the recommenders tastes/opinions.
The conceptual model in this study is composed of four
key constructs (see Figure 1). We hypothesize that both the
informational content (ICON) and the social component
(SIMILARITY) induce consumers trust (TRUST) in online
reviews. We further hypothesize that purchase involvement
has a moderating inuence on the relationship between
review characteristics and induced trust.
J. Consumer Behav. 11: 94104 (2012)
DOI: 10.1002/cb

96

P. Racherla et al.
Information Content
(ICON)

H1

Trust in the Review


(TRUST)
Social Component
(SIMILARITY)

H2 a&b

H4
H3

Purchase Involvement
(INVOLVE)

Figure 1. Conceptual model and hypotheses

CONSTRUCTS AND HYPOTHESES


Information content in the review (argument quality)
Uncertainty reduction theoryrelated theories suggest that in
initial interactions, a greater amount of verbal communication
between the communication partners alleviates uncertainty.
Further, the length and depth of discussion in the initial stages
also reduce uncertainty and enhance the trust between the
communicating parties. In the case of online reviews, the
extent of information available in a review helps the customers
assess the attributes of the product as well as the source of
information, and this, to an extent, reduces uncertainty. Following this logic, it can be assumed that consumers, when faced
with numerous product reviews, tend to be inuenced by how
a reviewer presents and frames the information in a review.
We term the presentation/framing of a message as argument
quality (Petty and Cacioppo, 1986). Information sources that
make strong claims (claims that are relevant, objective, and
veriable) tend to be more persuasive and are perceived more
credible. The persuasive strength of argument quality in a
message source has been established in numerous contexts.
For instance, Kempf and Palan (2005) found that argument
quality of face-to-face WOM communication tends to have a
positive inuence on subsequent brand attitudes and purchase
intentions. Similarly, Dillard and Shens (2005) study on
online healthcare communication found that strong arguments
containing concrete facts and sound evidence have a greater
likelihood of inhibiting uncertainty as opposed to messages
lacking in evidence and reasoning.
A product review is essentially an argument made by a
reviewer to either encourage or dissuade consumers from
buying a particular product or service. The manner in which
the reviewer argues for or against the product increases the
credibility and trust perceptions. Therefore, reviews with
better argument quality tend to be more trustworthy than
reviews with weak argument quality.
H1: Reviews with high argument quality have higher
TRUST scores than reviews with low argument quality.

Social component of the review (perceived background


similarity)
The uncertainty reduction theory suggests that communication
partners, especially in the initial rounds of interaction, resort to
uncertainty reduction strategies involving exchange and
collection of sociodemographic information that allows
Copyright 2012 John Wiley & Sons, Ltd.

them to predict each others attitude and behavior. WOM communication between individuals with a shared background is
much more efcient and effective than between individuals
from disparate backgrounds (e.g., Gilly et al., 1998). In this
sense, knowing the identity of the source of information helps
consumers nd similar others.
One of the key differences between traditional WOM and
eWOM is the strength of ties between consumers. In the case
of traditional WOM, research has shown that greater tie
strength leads to greater persuasive effect. However, in the
age of the Internet, consumers now have the option to go
beyond the limited set of sources and access information
from numerous other sources that can be considered as weak
ties (Granovetter, 1973). Weak ties play an important role in
social learning and are essential to the diffusion of ideas
(Ellison and Fudenberg, 1995). Previous research has identied two possible benets of accessing information from
weak ties: (i) increased diversity of information and (ii)
higher quality of product information since consumers can
access people with greater expertise not available within
immediate social circles.
However, given the anonymous nature of online communication and the potential for deception, consumers nd it
difcult to assess the quality of information provided in online
reviews as well as the motives of the information provider.
Therefore, consumers may look for social cues that establish
background similarity or dissimilarity. Yet, the ndings
regarding the importance of perceived similarity have been
equivocal and varied according to the context of the study
and the stage of the consumer decision-making process.
Recent studies have shown that perceived similarity is a
factor that drives trust in eWOM, especially in the evaluation
and purchase stages. For instance, Forman et al. (2008) found
that when reviewers shared their demographic information in
Amazon.com, it spurred identity granting behavior from
customers with similar characteristics and eventually had a
positive impact on sales. Perceived similarities between the
sender and recipient will serve as cues for the latter that the
product or service may be of interest to him and tailored to
his demographic prole and needs. This reinforces the degree
of trust and condence in the message.
H2a: Reviews with high perceived similarity between the
reviewer and the consumer have greater TRUST scores
than reviews with low perceived similarity.
On the other hand, ambiguity or dissimilarity may also lead
to effective online interactions (Joinson, 2001; Lea, Spears and
de Groot, 2001). Lack of sociodemographic information
makes communication partners create an idealized image of
each other. This leads to a greater focus on the informational
aspects of the message. Seasoned points of view and novelty
of information can outweigh the benets of social similarity,
allowing dissimilar sources to play a more important role as
information brokers (Burt, 2005), especially during the awareness and information search stages. It is possible that when the
task at hand is more difcult and requires more information,
consumers tend to give credence to socially dissimilar sources
with greater knowledge and expertise (for instance, experiential products such as hotels and restaurants present greater
J. Consumer Behav. 11: 94104 (2012)
DOI: 10.1002/cb

Consumer trust in online product reviews


knowledge and choice uncertainty as opposed to regular
products). Therefore, we consider an alternate hypothesis:
H2b: Reviews with low perceived similarity between the
reviewer and consumer have greater TRUST scores than
reviews with high perceived similarity.
Trust in a review
Mayer and Davis (1999: p.712) dene trust as the willingness
of a party to be vulnerable to the actions of another party based
on the expectation that the other party will perform a particular
action important to the trustor, irrespective of the ability to
monitor or control that other party. Trust and uncertainty
form two ends of the same continuum in the sense that the
higher the uncertainty, the lower the trust and vice versa. The
fundamental outcome of uncertainty reduction mechanisms is
enhanced trust. Further, the notion of willingness in the
above denition reects a trustors perception that the other
party is competent, open, and reliable and that there is a
complementarity of intention and expectation between the
two parties. The level of trustworthiness of a source is directly
related to how individuals perceive and respond to information
(Grewal et al., 1994). In the case of online product reviews,
consumers do not have complete information about reviewers
actions, thoughts, or motives, thereby creating a problem of
information asymmetry. To take an optimal decision, consumers have to resolve this asymmetry.
Therefore, we adopt decision trust (McKnight et al.,
2003) as our dependent variable. It is dened as the extent
to which a consumer is willing to depend on something or
someone in a given situation with a feeling of relative
security, even though negative consequences are possible.
In the case of PRS, consumers providing the reviews are
completely unknown entities. Even so, consumers are willing
to use various heuristics to assess the trustworthiness of a
source and take a decision based on what can be a potentially
misleading or false review. Therefore, the concept of
decision trust can be aptly applied in the context of this
study. We hypothesize that both information content and
the social component contribute to decision trust in
reviews..
Purchase involvement
Involvement has been dened as a subjective psychological
state, reecting the importance and personal relevance of an
object or event (Johnson and Eagly, 1990). Product and
purchase involvement are closely related since purchase
involvement reects the relevance and importance of a
purchase to a buyer. Extending this concept to leisure and
travel, researchers have looked at activity involvement as
an important construct to explain the extent to which consumers show greater sensitivity toward the subtleties of activity
attributes. Some of the important dimensions of activity
involvement include interest, pleasure, perceived probability
and consequence of risk of purchase, and self-expression
(Havitz and Dimanche, 1999; Laurent and Kapferer, 1985).
Research suggests that highly involved consumers use more
criteria for evaluation, accept fewer alternatives, process
information in detail (Chaiken, 1980), and form attitudes that
Copyright 2012 John Wiley & Sons, Ltd.

97

are more resistant to change (Celsi and Olson, 1988). The


degree of involvement has an impact on information processing, decision making, and response to external/productrelated stimuli. In the case of high involvement activities,
the probability and consequences of risk, both monetary
and nonmonetary, are higher in comparison to low involvement activities. Therefore, highly involved consumers tend
to rely on more cues and require more information than
consumers in a low involvement mode. We expect that
involvement moderates the relationship between review
characteristics and attitude toward the review.
H3: The positive effect of argument quality on TRUST
will be greater for subjects in high involvement mode.
H4: The positive effect of perceived similarity on TRUST
will be greater for subjects in high involvement mode.

METHODOLOGY
Study context and design
In this study, we focused on tourism and hospitality services.
The element of risk in buying these services is much higher
when compared to regular products due to the lack of try
before you buy or return in the case of quality being below
expectations features (Buhalis, 2003). Further, the intangible
and experiential nature of these services enhances the knowledge and choice uncertainty of consumers. Therefore, the
appeal of reference group evaluation is even higher during
the decision-making process.
To identify the appropriate stimuli for the purchase scenarios, we rst conducted focus groups with about 20 undergraduate students to identify various dimensions of involvement
with a trip scenario. Based on this information, we created six
trip scenarios that were rated by 70 students on Zaichkowskys
(1994) Personal Involvement Inventory scale. Through the
pretests, we identied two trip scenarios for high and low
involvement modes. In the high involvement scenario, subjects
were asked to imagine that they were traveling to a foreign
destination with their closest friends and were given the responsibility to research and book an expensive hotel room to stay in
for a week. In the low involvement scenario, subjects were
asked to book an inexpensive hotel room for a trip with their
parents to visit a distant relative. The two scenarios formed
the basis for the experimental stimuli.

Experimental design
We adopted a repeated-measures experimental design since
consumers typically read more than one review associated
with a product/service. The two independent constructs and
one moderating construct with two levels each led to a
2  2 (within-subjects)  2 (between-subjects) full factorial
experimental design. We created four reviews each for high
and low involvement purchase scenarios. The experimental
manipulations (four reviews for each involvement mode)
were designed as follows:
J. Consumer Behav. 11: 94104 (2012)
DOI: 10.1002/cb

98

P. Racherla et al.

Information content manipulation


Information content (ICON) has two dimensions: valence
(negative or positive), which is implicit in all the reviews
but not considered in this study, and argument quality, which
was explicitly measured. Since incorporating both positive
and negative valence was not feasible given the study design
constraints, we used only negative reviews as stimuli. This
decision was based on the notion of negativity bias
(Mizerski, 1982), which suggests that when consumers are
neutral, negative reviews tend to become more salient than
positive reviews. Applying this concept to online reviews,
Ba and Pavlou (2002) found that in the presence of large
number of positive reviews, negative reviews have a greater
inuence on trusting intentions (b = 0.856) than positive
reviews (b = 0.541). Further, it is unlikely that marketers
provide negative reviews about their own products and
services. Therefore, using only negative reviews, to an extent,
reduces consumers skepticism as to whether the reviews are
written by actual consumers or part of the promotional chat
provided by marketers. This may encourage consumers to
focus primarily on the quality of the review.
The argument quality of the reviews was manipulated
using Toulmins (1958) model of argumentation as applied
by Kim and Benbasat (2006). According to Toulmin, a
trust-assuring communication should have well-organized
structure and content. A trust-assuring argument should have
the following components: (i) a claim which is dened as an
assertion or conclusion put forward for general acceptance;
(ii) data, which are dened as evidence presented in the
argument to support the claim or the basis for the claim;
and (iii) backing, the portion of the argument that explains
why the data and claim should be accepted. For instance,
Figure 2 shows a comparison of two typical reviews from a
popular travel-related PRS. As can be seen, review 1 has all
the components. The reviewer presents enough evidence/
argument to support the claim that the hotel is overpriced.
Further, the argument is backed up by the assertion that
compared to the standards set by the Sheraton brand, this
particular location is not up to the mark and therefore not worth
the price. This indicates the reviewers experience about the

brand and shows his/her expertise in making the claims. In


comparison, review 2 only makes a claim and offers no
evidentiary support.
To extract the argumentation dimensions, we examined
about 50 randomly selected reviews and created a reference
sheet based on the various aspects of the reviews. Based on
this information, we created several reviews with varying
aspects of argument quality that were rigorously pretested
during the manipulation checks. One factor that cannot be
controlled in this context is the social categorization process.
The content of the review itself can form a strong basis for
the subjects to draw inferences. The manner in which the
review is written, the issues detailed in the review, the tone,
etc., can be sufcient cues for a subject to indulge in a social
categorization process, even in the absence of any personal
identiers (real name, location, email address, photo, etc.).
This can be considered as a potential and yet interesting
confounding effect in this study.
Social component manipulation
The social component of the review was manipulated using
sociodemographic information (photo, real name/screen name,
place of origin, and a short bio). These variables were drawn
from Marxs (2004) study, which found that people use 11
different types of information to assess the identity of a source
in online interpersonal communication. The most important
of these, as rated by the study subjects, included individual
identication (the who question), shared identication (demographics and lifestyle), geographical location (the where
from question), and photos (what does he/she look like
question). Real name of the source, for instance, was perceived
as much more reliable identity information than a screen name
(even though the real name provided can itself be false in many
cases). Keeping these issues in mind, the stimuli were manipulated with two conditions: the presence/absence of a photo, real
name, and age, place of origin and bio-graphical sketch or
absence of this information with only a screen name. To ensure
that the names do not reveal the gender, we matched them with
an established neutral names list. The geographical location of
the reviewer was chosen as the one in which the students

Figure 2. Toulmins (1958) argumentation model applied to online reviews

Copyright 2012 John Wiley & Sons, Ltd.

J. Consumer Behav. 11: 94104 (2012)


DOI: 10.1002/cb

Consumer trust in online product reviews


resided. Figure 3 shows one of the manipulations (review with
High ICON and SIMILARITY).
Manipulation checks
We conducted a pilot study with 86 undergraduate students
to test the efcacy of the manipulations. The measures were
adopted from previous studies in marketing and e-commerce
and modied to suit the context. All responses were measured
on a seven-point Likert scale. Argument quality measures were
adopted from Rainss (2007) study on healthcare communication that tested the effect of argument quality and anonymity in
a computer-mediated environment. The subjects were asked to
rate the reviews on six attributes: compelling, well-supported,
contained specic facts, contained detailed information, listed
concrete examples, and did not include detailed information
(the last item was reverse coded in the analysis). Perceived
similarity measures were derived from previous studies in traditional WOM and virtual communities (Gilly et al., 1998; Ma
and Agarwal, 2007). Subjects were asked to rate their agreement on the following aspects: outlook in life, tastes and preferences, social background, and review writing style. Trust
was measured on McKnight et al.s (2003) decision trust scale
that measured consumer intentions to purchase a product/service based on information provided in the reviews. All the
scales tested for high reliability above 0.85. The mean ratings
of the manipulations are shown in Table 1.

DATA COLLECTION AND RESULTS


In the nal study, 283 undergraduate students from a large
university in Northeastern United States participated for
extra credit (150 and 133 students randomly assigned to high
and low involvement modes, respectively). The students
were asked to imagine that they were going on a trip (for a
description of the trip scenarios, please see the text above).
In the scenario, students were given the pretrip responsibility
to search the Internet for a hotel that ts the trip objectives
and budget. Before booking a room in the hotel, the students
were asked to read the four reviews about the hotel written
by different travelers. The students were given a booklet
with the four reviews and were asked to assign trust scores
to each review. The reviews were placed in random order
and tested for sequential bias and order effects. The characteristics of the subjects were as follows: (i) sex: male (38%)
and female (62%) and (ii) age: 1823 years (84%), 2427 years
(12%), and 2832 years (4%). The responses from subjects
in the 2832 age group were excluded from the analysis
since they did not represent the general age category of the
target population.
Analysis of variance
We applied repeated-measures linear mixed models to test
the four hypotheses. The results are shown in Table 2 (all
table rows in bold indicate signicant effects).1
1
The HIGH level of each construct was specied as the reference category,
and therefore, the estimates in the table correspond to the LOW level of
the main factors.

Copyright 2012 John Wiley & Sons, Ltd.

99

The results show that the main effect of ICON is signicant


(p < 0.01; F = 32.58) conrming our hypothesis H1 that higher
argument quality in a review has a signicant effect on trust.
Similarly, the results show that the main effect of SIMILARITY on TRUST is highly signicant (p < 0.01; F = 69.37).
This supports hypothesis H2a and disconrms H2b. However,
this effect should be interpreted in the context of the interaction
term ICON * SIMILARITY that was found be signicant
(p < 0.01; F = 7.63). It shows that there is a signicant interaction between argument quality and the perceived similarity.
Therefore, reviews that have a combination of high ICON and
SIMILARITY have signicantly higher trust scores compared
to reviews with low ICON and SIMILARITY. The results show
mixed support for the moderating inuence of involvement.
The interaction term ICON * INVOLVE (p > 0.05; F = 0.02)
was not signicant, but SIMILARITY * INVOLVE (p < 0.05;
F = 4.01) was found to be signicant. These results disconrm
our hypothesis H3 but support H4. We further tested the difference in trust scores of the main manipulations (the four reviews
in two different involvement modes). The results are shown in
Table 3 and Figure 4.
We also performed a post-hoc analysis based on Scheffes
test of multiple comparisons (see Table 4). This test is a
method of multiple comparisons that is used for making not
only pair-wise comparisons but also all possible linear
combinations of group means. Sheffes test revealed the
following: (i) JORDAN (with high ICON and SIMILARITY)
was associated with signicantly higher trust scores in both
high and low involvement modes than SAM, MMJOY, and
LUCKY; (ii) MMJOY (with high ICON and low SIMILARITY) had signicantly higher mean score than LUCKY; (iii)
SAM (with low ICON and high SIMILARITY) had signicantly higher trust score than LUCKY; and (iv) there is
no signicant difference in the mean trust scores of SAM
and MMJOY.

DISCUSSION AND IMPLICATIONS


Discussion
Previous research in both conventional and eWOM (Duhan
et al., 1997; Park et al., 2007) suggests that reviewer characteristics, apart from the quality and informativeness of the
message, have a signicant impact on consumers evaluation
of WOM communication. Our results support these assertions.
The main effect of argument quality was found to be
signicant, indicating that the reviews with higher argument
quality are more trusted by consumers. This also indicates
that in the absence of nonverbal cues, consumers adapt their
relational behavior to the verbal and linguistic cues that are
available in a computer-mediated environment.
Surprisingly, the size of the effect was much smaller when
compared to perceived similarity. Perceived similarity contributed to greater increase in trust scores. Even in an online
environment, consumers categorize themselves in relation to
a reference group of which the WOM generator is also a part.
If consumers identify with this reference group, there is a
higher likelihood of trust and a favorable attitude toward
products/services. Two possible explanations can be put
J. Consumer Behav. 11: 94104 (2012)
DOI: 10.1002/cb

100

P. Racherla et al.

Figure 3. Example of experimental stimulus for High ICON and IDENTITY condition
Table 1. Mean ratings of the manipulated variables in pretests
Review
SAM
JORDAN
LUCKY
MMJOY

ICON

IDENTITY

4.62
6.15
4.47
6.06

5.63
5.84
2.74
3.29

forward for this result: (i) Since the subjects were evaluating
tourism/hospitality services, the expectation is of experiencing
hedonic rather than functional benets. These purchases have a
value expressive motive (i.e., when consumers feel that a
Copyright 2012 John Wiley & Sons, Ltd.

purchase says something about themselves in terms of their


values, tastes, and preferences). (ii) The greater importance of
perceived similarity could also be the result of the stage of
purchase process that the subjects were placed in during the
study. Since the subjects were told that they already completed
the information search and evaluation stages and were asked to
directly consider purchase, perceived similarity may have
played a greater role than argument quality.
At the same time, this result should be viewed in the
context of the interaction term ICON * SIMILARITY. Previous research suggests that there is a signicant interaction
between perceived similarity of the message source and the
J. Consumer Behav. 11: 94104 (2012)
DOI: 10.1002/cb

Consumer trust in online product reviews

101

Table 2. Type III tests of xed effectsa


Estimate
Intercept
ICON
SIMILARITY
ICON * SIMILARITY
INVOLVE
ICON * INVOLVE
SIMILARITY * INVOLVE

Std. Error

Numerator df

Denominator df

Sig.

0.11
0.13
0.13
0.17
0.14
0.14
0.13

1.00
1.00
1.00
1.00
1.00
1.00
1.00

355.01
834.61
670.91
1088.47
355.01
834.61
560.32

8371.84
32.58
69.37
7.63
1.08
0.02
4.01

0.00
0.00
0.00
0.01
0.30
0.90
0.05

5.23
0.16
1.17
0.47
0.02
0.02
0.25

Dependent variable: TRUST

Table 3. Estimated marginal mean trust scores in two involvement


conditions
Involvement
mode

Reviewer

Mean

Std.
error

Lower
bound

Upper
bound

High

JORDAN
LUCKY
SAM
MMJOY
JORDAN
LUCKY
SAM
MMJOY

5.580
4.181
4.967
4.763
5.226
4.394
4.907
4.949

0.105
0.133
0.103
0.122
0.115
0.145
0.113
0.133

5.374
3.919
4.764
4.523
5.000
4.108
4.684
4.686

5.787
4.442
5.171
5.003
5.452
4.680
5.129
5.211

Low

Figure 4. Estimated marginal means of TRUST in the two involvement


modes

quality of the message. For instance, found signicant


interaction between similarity and expertise, with the message having low similarity and low in expertise showing the
least inuence on the students. Such an effect is reected in
the results of this study, which showed that the trust scores
associated with JORDAN (high ICON and SIMILARITY)
were signicantly higher trust across both the involvement
conditions.
Two alternative explanations can be put forward to explain
the main effects. When consumers read and compare reviews,
the amount of information in a review denitely matters.
Increased availability of reasons and information increases
decision makers condence (Tversky and Kahneman, 1974).
In the case of online reviews, argument quality and perceived
similarity increases consumers accuracy of prediction and
diagnosis regarding a product. However, even in the absence
of these attributes, the mere increase in the availability of
information may serve as powerful peripheral cues that can
increase decision makers condence. In this case, greater
number of words and more sociodemographic information in
the high ICON and SIMILARITY manipulations may have
inuenced trust perceptions. Further, the order in which the
reviews were laid out for the subjects can lead to varying trust
perceptions. For instance, if the low ICON and SIMILARITY
review immediately follows the high ICON and SIMILARITY
review, the subjects can automatically assign a low score to the
former. In this study, we controlled for this problem by
randomizing the placement of reviews as well as giving clear
instructions to the subjects to assess each review independently
on its merits. Subsequent tests did not reveal any signicant
differences in the mean scores.
Interestingly, the results on the moderating role of
involvement showed mixed support. While information

Table 4. Scheffes test for multiple comparisons of the review format


(I) Reviewer
JORDAN
LUCKY
SAM
MMJOY

(J) Reviewer

Mean difference (I J)

Std. error

Sig.

Lower bound

Upper bound

LUCKY
SAM
MMJOY
JORDAN
SAM
MMJOY
JORDAN
LUCKY
MMJOY
JORDAN
LUCKY
SAM

1.077*
0.436*
0.474*
1.077*
0.641*
0.604*
0.436*
0.641*
0.038
0.474*
0.604*
0.038

0.111
0.086
0.100
0.111
0.097
0.102
0.086
0.097
0.100
0.100
0.102
0.100

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.706
0.000
0.000
0.706

0.858
0.266
0.277
1.296
0.833
0.805
0.606
0.449
0.159
0.671
0.402
0.234

1.296
0.606
0.671
0.858
0.449
0.402
0.266
0.833
0.234
0.277
0.805
0.159

Copyright 2012 John Wiley & Sons, Ltd.

J. Consumer Behav. 11: 94104 (2012)


DOI: 10.1002/cb

102

P. Racherla et al.

content was seen as a more important attribute in a high


involvement mode, the social component played a signicant
role in the low involvement mode. This result can be
explained by ELM, which states that consumers tend to
cognitively elaborate on a message in the high involvement
mode (Petty and Cacioppo, 1986) and look for peripheral
cues in low involvement situations. In the high involvement
mode (central route), the subjects carefully considered the
issues pertinent to the message. On the other hand, in the
low elaboration (peripheral route) mode, recipients use
simple cues. In this study, subjects in the low involvement
mode gave importance to peripheral cues such as the
demographic information of the reviewer. The mixed support
for this hypothesis could also be the artifact of using an
experimental setting with students. Since the subjects role
played their decision making task in a controlled setting,
there were no monetary or opportunity costs associated with
a suboptimal decision.
Two interesting results stood out in this study. First, the
consumers perceived an anonymous review with high
argument quality equally trustworthy as the review with
low argument quality but greater perceived similarity.
Second, the mean TRUST scores for all the reviews in both
the involvement conditions were above their respective
midpoint (4.00), indicating that all the reviews were
perceived to be fairly trustworthy. This outcome suggests
that a causal attribution could be at play. Attribution theory
suggests that causal analysis is inherent in an individuals
need to understand events. It is more pronounced when a
consumer is faced with a negative WOM message, a situation
in which a consumer wonders why another consumer would
communicate negative information about the product
(Laczniak et al., 2001). This leads to causal attribution from
the message receiver. In this study, perhaps the consumers
inferred that since the reviewer is going through the trouble
to provide a review, the information may be genuine to
some extent.
Implications
The study ndings suggest various strategies that can be
effectively employed to enhance consumers trust in online
product reviews.
Review sites need to devise metrics that can determine
argument quality and create lters that will allow the best
among the lot to be presented to the consumer. This is also
important since consumers generally face an overwhelming
number of reviews for any given product. PRS can use
methods such as text mining and document indexing tools
to sort reviews based on their content and structure, and then
enable the site with dynamic capabilities to provide the
necessary information to consumers on demand.
The ndings with respect to the social component have
implications for managers. Websites need to evolve mechanisms that match consumers with reviewers based on sociodemographic characteristics. These sites must gradually move
away from a mere information provision role to providing
social platforms that encourage information exchange between
socially similar consumers. As Yaniv (2004) suggests, simple
aggregation of information typically seen in recommender
Copyright 2012 John Wiley & Sons, Ltd.

systems creates conceptual difculties, as people are entitled


to varied tastes and preferences. A more promising strategy
would be to consider the personal match between the advice
seeker and the recommender, assuming that the greater the
similarity, the greater the impact and benet of the advice
received will be. Further, such information must be categorized
based on the extent of involvement that consumers face with
respect to each product.

LIMITATIONS AND FUTURE RESEARCH


DIRECTIONS
This study has certain limitations.
1. This study uses only negative reviews. As discussed
previously, the evidence on review valence is equivocal.
To the best of our knowledge, there are no studies that
reconcile these varying streams of thought. In real life
situations, consumers typically face numerous positive
and negative reviews associated with a given product.
This matter is further complicated by the recency bias
(Pain and Sharpley, 1989) in the sense that more recently
posted reviews are more easily accessible and tend to
have greater inuence although these reviews might not
be the best source of information. It is important for future
research to examine how consumers manage to weave
through the clutter and apply simple heuristics to achieve
the task at hand. Such research has important implications
for the design and management of PRS.
2. Another limitation is the operationalization of the social
component. We conceptualized the social component as
both reviewers identity information and perceptual
similarity based on sociodemographic characteristics of
the reviewer and the subjects used in the study. However,
previous research suggests that individual self-disclosure
about his/her background to a group and the groups
acceptance of this identity are analytically separate
processes that determine trust and credibility perceptions.
It can be assumed that consumers go through a similar
bind when faced with the identity information provided
by reviewers. Social cues may trigger not only perceptions
of similarity but also verication (verifying the genuineness
of the reviewer and consumption experience). Future
research should take other approaches to delineate this
important dichotomy.
SUMMARY OF RESULTS AND CONTRIBUTIONS
In this study, we apply the elements of URT to explore the
antecedents of consumers trust in online reviews. We
summarize our ndings as follows:
We nd that the argument quality is a signicant driver of
trust even in online reviews. In using argument quality, we
build upon some of the deciencies in previous studies
that typically used either the review valence (positive or
negative) (Ba and Pavlou, 2002) or review length (number
of words) (Mudambi and Schuff, 2010). We contend that

J. Consumer Behav. 11: 94104 (2012)


DOI: 10.1002/cb

Consumer trust in online product reviews


while consumers use these heuristics to assess online
reviews, they also read the reviews with the specic goal
of gaining information about a product or service. Therefore,
argument quality is of greater importance to customers.
We nd that the social component of online reviews is
equally important in building trust. The social dynamics
of the reviewerconsumer interaction have been generally
ignored in previous studies. Consequently, the predictor
variables explain very little variance in the dependent
variables (e.g., trust, sales, and satisfaction). Even though
online platforms are relatively anonymous and goal-directed,
most of the transactions/communications that occur are still
between two human beings. Therefore, it is important to
consider both the social and economic facets of consumerto-consumer interactions.
Our study focused on experiential products as opposed to
previous studies (e.g., Chevalier and Mayzlin, 2006) that
typically focused on products. Recent evidence suggests
that more and more consumers are relying on traveler
reviews more than commercially generated information.
More research is needed to understand how consumers
use online reviews when it comes to experiential products.

REFERENCES
Areni CS, Ferrell ME, Wilcox JB. 2000. The persuasive impact of
reported group opinions on individuals low vs. high in need for
cognition: Rationalization vs. biased elaboration? Psychology
and Marketing 17(10): 855875.
Ba S, Pavlou PA. 2002. Evidence of the Effect of Trust Building
Technology in Electronic Markets: Price Premiums and Buyer
Behavior. MIS Quarterly 26(3): 243268.
Berger CR, Calabrese RJ. 1975. Some explorations in initial
interaction and beyond: Toward a developmental theory of
interpersonal communication. Human Communication Research
1(2): 99112.
Berger CR, Gudykunst WB. 1991. Uncertainty and communication.
In Dervin B, Voight M (eds). Progress in communication
sciences. Ablex: Norwood, NJ; 2166.
Buda R, Zhang Y. 2000. Consumer product evaluation: the interactive
effect of message framing, presentation order, and source credibility. Journal of Product and Brand Management 9(4): 229242.
Buhalis D. 2003. eTourism: information technology for strategic
tourism management. Harlow, England: Financial Times Prentice
Hall.
Burt RS. 2005. Brokerage and closure: An introduction to social
capital. Oxford University Press: USA.
Celsi RL, Olson JC. 1988. The Role of Involvement in Attention
and Comprehension Processes. The Journal of Consumer Research
15(2): 210224.
Chaiken S. 1980. Heuristic versus systematic information processing
and the use of source versus message cues in persuasion. Journal
of Personality and Social Psychology 39(5): 752766.
Chevalier J, Mayzlin D. 2006. The effect of word of mouth online:
Online book reviews. Journal of Marketing Research 43(3):
345354.
Dellarocas C. 2003. The digitization of word of mouth: Promise
and challenges of online feedback mechanisms. Management
science 49(10): 14071424.
Deutsch M, Gerard HB. 1955. A study of normative and informational social inuences upon individual judgment. Journal of
Abnormal and Social Psychology 51(3): 629636.
Dholakia UM, Bagozzi RP, Pearo LK. 2004. A social inuence
model of consumer participation in network-and small-group-

Copyright 2012 John Wiley & Sons, Ltd.

103

based virtual communities. International Journal of Research


in Marketing 21(3): 241263.
Dillard JP, Shen L. 2005. On the Nature of Reactance and its Role
in Persuasive Health Communication. Communication Monographs
72(2): 144168.
Duhan DF, Johnson SD, Wilcox JB, Harrell GD. 1997. Inuences
on Consumer Use of Word-of-Mouth Recommendation Sources.
Journal of the Academy of Marketing Science 25(4): 283.
Einwiller S. 2003. When Reputation Engenders Trust: An Empirical
Investigation in Business-to-ConsumerElectronic Commerce.
Electronic Markets 13(3): 196209.
Ellison G, Fudenberg D. 1995. Word-of-mouth communication
and social learning. Quarterly Journal of Economics 110(1):
93125.
Forman C, Ghose A, Wiesenfeld B. 2008. Examining the Relationship Between Reviews and Sales: The Role of Reviewer
Identity Disclosure in Electronic Markets. Information Systems
Research 19(3): 291313.
Gilly MC, Graham JL, Wolnbarger MF, Yale LJ. 1998. A Dyadic
Study of InterpersonalInformation Search. Journal of the
Academy of Marketing Science 26(2): 83.
Granovetter MS. 1973. The strength of weak ties. The American
Journal of Sociology 78(6): 1360.
Grewal D, Gottlieb J, Marmorstein H. 1994. The moderating
effects of Framing and source credibility on the price-perceived
risk relationship. Journal of Consumer Research 21(3):
145153.
Havitz ME, Dimanche F. 1999. Leisure involvement revisited:
Drive properties and paradoxes. Journal of leisure research
31(2): 122149.
Johnson BT, Eagly AH. 1990. Involvement and persuasion: Types,
traditions, and the evidence. Psychological Bulletin 107(3):
375384.
Joinson AN. 2001. Knowing me, knowing you: Reciprocal selfdisclosure in Internet-based surveys. Cyberpsychology & Behavior 4(5): 587591.
Kellermann K, Berger CR. 1984. Affect and the acquisition of social
information: Sit back, relax, and tell me about yourself. Communication yearbook 8: 412445.
Kempf DAS, Palan KM. 2005. The effects of gender and argument
strength on the processing of word-of-mouth communication.
Academy of Marketing Studies Journal 10(1): 118.
Kim D, Benbasat I. 2006. The Effects of Trust-Assuring Arguments on Consumer Trust in Internet Stores: Application of
Toulmins Model of Argumentation. Information Systems
Research 17(3): 286.
Kollock P, Smith M. 1996. Managing the Virtual Commons:
Cooperation and Conict in Computer Communities. In S Herring
& J Benjamins (eds). Computer-Mediated Communication:
Linguistic, Social, and Cross-Cultural Perspectives. John Benjamins
Publishing Co: Amsterdam, Netherlands; 109128.
Laczniak RN, DeCarlo TE, Ramaswami SN. 2001. Consumers
responses to negative word-of-mouth communication: An
attribution theory perspective. Journal of Consumer Psychology
11(1): 5773.
Laurent G, Kapferer JN. 1985. Measuring consumer involvement
proles. Journal of Marketing Research 22(1): 4153.
Lea M, Spears R, de Groot D. 2001. Knowing Me, Knowing You:
Anonymity Effects on Social Identity Processes within Groups.
Personality and Social Psychology Bulletin 27(5): 526.
Ma M, Agarwal R. 2007. Through a Glass Darkly: Information
Technology Design, Identity Verication, and Knowledge
Contribution in Online Communities. Information Systems
Research 18(1): 4256.
Marx GT. 2004. Internet anonymity as a reection of broader issues
involving technology and society. Asia Pacic Review 11(1):
142166.
Mayer RC, Davis JH. 1999. The effect of the performance appraisal
system on trust for management: A eld quasi-experiment.
Journal of Applied Psychology 84(1): 123136.

J. Consumer Behav. 11: 94104 (2012)


DOI: 10.1002/cb

104

P. Racherla et al.

McKnight DH. 2001. What Trust Means in E-Commerce Customer


Relationships: An Interdisciplinary Conceptual Typology. International Journal of Electronic Commerce 6(2): 3559.
McKnight DH, Choudhury V, Kacmar C. 2003. Developing and
Validating Trust Measures for e-Commerce: An Integrative
Typology. Information Systems Research 13(3): 334359.
Mizerski RW. 1982. An Attribution Explanation of the Disproportionate Inuence of Unfavorable Information. The Journal of
Consumer Research 9(3): 301310.
Mudambi SM, Schuff D. 2010. What makes a helpful online
review? A study of customer reviews on Amazon. com. Mis
Quarterly 34(1): 185200.
Newcomb TM. 1961. The Acquaintance Process. Holt, Rinehart
and Winston: New York, NY.
Pain MD, Sharpley, CF. 1989. Varying the order in which positive
and negative information is presented: Effects on counselors
judgments of clients mental health. Journal of Counseling
Psychology 36(1): 37.
Park DH, Lee J, Han I. 2007. The Effect of On-Line Consumer
Reviews on Consumer Purchasing Intention: The Moderating
Role of Involvement. International Journal of Electronic
Commerce 11(4): 125148.

Copyright 2012 John Wiley & Sons, Ltd.

Petty RE, Cacioppo JT. 1986. Communication and Persuasion:


Central and Peripheral Routes to Attitude Change: SpringerVerlag.
Rains SA. 2007. The Impact of Anonymity on Perceptions of
Source Credibility and Inuence in Computer-Mediated Group
Communication: A Test of Two Competing Hypotheses.
Communication Research 34(1): 100.
Ramirez A, Walther JB, Burgoon JK, SunnafrankM. 2002. InformationSeeking Strategies, Uncertainty, and Computer-Mediated Communication. Toward a Conceptual Model. Human Communication
Research 28(2): 213228.
Taylor DA, Altman I. 1966. Intimacy-scaled stimuli for use in studies
of interpersonal relations. Psychological Reports 19(3): 729730.
Toulmin S. 1958. The Uses of Arguments. Cambridge University
Press: Cambridge, MA.
TverskyA, Kahneman D. 1974. Judgment under risk: Heuristics and
Biases. Science 185(4157): 11241131.
Yaniv I. 2004. The Benet of Additional Opinions. Current
Directions in Psychological Science 13(2): 7578.
Zaichkowsky JL. 1994. The Personal Involvement Inventory:
Reduction, Revision, and Application to Advertising. Journal
of Advertising 23(4): 159.

J. Consumer Behav. 11: 94104 (2012)


DOI: 10.1002/cb

Das könnte Ihnen auch gefallen