Omega 35 (2007) 22 38

www.elsevier.com/locate/omega

Consumer trust in e-commerce in the United States,

Singapore and China
Thompson S.H. Teo , Jing Liu
Department of Decision Sciences, School of Business, National University of Singapore, 1 Business Link 117592, Singapore
Received 27 July 2004; accepted 4 February 2005
Available online 24 March 2005

Abstract
This study examines the antecedents and consequences of consumer trust in the United States, Singapore and China. The
results show that reputation and system assurance of an Internet vendor and consumers propensity to trust are positively
related to consumer trust. Consumers trust has a positive relationship with attitude and a negative relationship with perceived
risk. Implications of the results are discussed.
2005 Elsevier Ltd. All rights reserved.
Keywords: e-commerce; Trust; The United States; Singapore; China

1. Introduction
Trust plays an important role in many social and economic interactions involving uncertainty and dependency.
Since uncertainties exist in transactions over the Internet,
many researchers have stated that trust is a critical factor
inuencing the successful proliferation of e-commerce [1].
The concept of trust is crucial because it affects a number
of factors essential to online transactions, including security
and privacy. Moreover, although e-commerce brings benets to both vendors and customers, it also has limitations,
such as the physical separation between buyers and sellers,
and between buyers and merchandise. In order to reduce the
barriers, vendors must develop a trustworthy relationship to
foster customer loyalty.
Consumer trust is an important aspect of e-commerce, and
understanding its antecedents and consequences is a prime
concern for the following reasons. First, the antecedents of
Corresponding author. Tel.: +65 6874 3036;
fax: +65 6779 2621.
E-mail addresses: [email protected] (T.S.H. Teo),
[email protected] (J. Liu).

0305-0483/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.omega.2005.02.001

trust enable us to know the relative importance of factors

affecting trust. Understanding these factors would play an
important role in devising appropriate measures to facilitate
trust. Second, the consequences of trust would enable us to
better understand the importance of trust and its effect on
online buying behavior.
In the organizational trust literature, Mayer et al. [2] proposed a model incoporating both a trusting party (trustor)
and a party to be trusted (trustee). They discussed the
trustors perceptions about the trustees characteristics. In
the context of e-commerce, Jarvenpaa et al. [3] examined
whether customers perceptions of an Internet stores reputation and size affect their trust in the store. Meanwhile,
drawing on the theory of reasoned action, researchers have
also investigated the consequences that trust has on consumers attitudes, intentions, and behaviors [35]. In a similar vein, Bendoly et al. [6] examined the impact of channel
integration on consumers loyalty to a multi-channel rm.
The global nature of the Internet raises questions about
the trust effects across cultures as well. Although trust may
form in a variety of ways, whether and how trust is established depend on the cultural factors (e.g., societal norms,
values, etc.) that guide peoples behaviors and beliefs

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

(e.g., [7]). Based on Hofstedes [7] four dimensions of

national culture, Doney et al. [8] developed a framework
of cognitive trust-building processes which illustrated the
importance of culture in the development of trust. In a
similar vein, Jarvenpaa and Tractinsky [4] suggested that
culture might affect the antecedents of consumer trust in
an Internet store. Chai and Pavlou [9] developed a crosscultural e-commerce adoption model and empirically tested
it using data collected from the United States and China.
They found that trust directly inuences consumer attitude across cultures; i.e., trust is important for all cultures
studied.
Other research has proposed a typology of trust types
[10], developed a measurement scale for trust [11], emphasized the importance of system trust in business-toconsumer transactions [12], proposed trust model [1315],
integrated trust with the technology acceptance model
[16,17], examined antecedents of trust [18,19], and examined the importance of social presence on trust dimensions
[20].
Although trust has been examined for over 50 years,
most of the research on consumer trust focuses on consumers in English-speaking countries and newly industrialized countries. For example, previous research has examined countries such as the US [21] and Australia [3].
Other studies compared the formation of consumer trust
between two different countries, e.g., Yamagishi and Yamagishi [22] provided empirical evidence that Japanese citizens often report lower levels of trust compared with their
American counterparts. In a similar vein, Jarvenpaa and
Tractinsky [4] examined consumer trust in the US and Israel, while Gefen et al. [23] examined cultural diversity
and trust in IT adoption between the United States and
South Africa.
Sin et al. [24] reviewed recent cross-cultural consumer
studies and found that about half of these studies involve
only two cultures. However, studies that investigate only one
or two cultures may have limited values compared to studies done in several cultures. The reason is that studies done
in several cultures would give us a better and deeper understanding of the effects of culture on behavior [25]. Hence,
we empirically examine a model of consumer trust in ecommerce vendors using data collected from three countries: the United States, Singapore and China.

2. Research model and hypotheses

Our study extends previous research by examining
the antecedents of consumer trust in the context of ecommerce based on a dyad of trustorsthe consumers, and
trusteesthe e-commerce vendors. The consumers trusting beliefs affect their attitudes to the vendors and their risk
perception. Consumers attitudes to the vendor and their
perceived risk, in turn, affect their willingness to buy. The
research model and hypotheses are shown in Fig. 1.

23

2.1. Characteristics of trustees (e-commerce vendors)

2.1.1. Reputation and size
Research in traditional industrial buyerseller relationships revealed that buyers perceptions of sellers reputation
and size are the factors of trustworthiness [26]. This is because reputation and size provide assurances of the vendors
ability, benevolence, and integrity [4]. Jarvenpaa et al. [3]
asserted that customers perceptions of an Internet stores
reputation and size affect their trust in the store.
Reputation is dened as the extent to which buyers believe a seller is professionally competent or honest and
benevolent [26]. Researchers have recognized that a rms
reputation is a valuable intangible asset that requires a
long-term investment of resources, efforts, and attention to
customer relationships. Reputation is vulnerable because it
is harder to form a reputation than to lose it [27]. It is a
fragile strategic asset that can be easily tarnished or damaged if not carefully protected [28]. Consequently, a vendor
with good reputation is perceived to be reluctant to jeopardize their reputation by acting opportunistically [29]. In the
traditional marketing literature, reputation has been shown
to be positively related to the buyers trust in the seller [30].
In Internet shopping, perceived reputation of a vendor has
also been revealed to be signicantly related to consumers
trust in the vendor [35]. Therefore, we postulate that:
H1. The perceived reputation of an e-commerce vendor
is positively related to the level of consumers trust in
the vendor.
Doney and Cannon [26] dened a sellers size as its overall size and market share position. Since a large marketshare rm must serve a more diverse and heterogeneous set
of customers [31], a large overall size and market share suggests that the rm consistently delivers on its promises to its
consumers and many consumers tend to trust it. Otherwise,
it would not have been able to maintain its position in the
industry [26]. Large organizational size also indicates that
the rm is likely to possess expertise and necessary support
systems that encourage trust and loyalty [32]. Larger rms
also tend to have more well-developed Web sites to encourage transactions [33]. Finally, in an e-commerce environment, large size suggests that the vendor is able to assume
the risk of product failure or transit losses and to compensate buyers accordingly [3]. Hence, it follows that:
H2. The perceived size of an e-commerce vendor is
positively related to the level of consumers trust in the
vendor.

2.1.2. Multi-channel integration

The emergence of the Internet as a channel for commerce has made companies increasingly recognize the importance of having multiple interaction channels with consumers and increasing the integration level among all the

24

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

Characteristics of Trustees
(E-Commerce Vendors)

Perceived Reputation

Perceived Size

H1 (+)

H2 (+)
Multichannel Integration

Attitude
H3 (+)

System Assurance

H7 (+)

H6 (+)

H4 (+)
Consumer Trust

Characteristic of Trustors (Consumers)

H5 (+)

H9 ()

Willingness to Buy

H10 ()

H8 ()
Perceived Risk

Propensity to Trust

Fig. 1. Research model and hypotheses.

communication channels (e.g., telephone, fax, catalog, physical store, email, Internet Website, etc.). Multi-channel integration is fast becoming a serious source of competitive
advantage as consumers demand more and more exible access to products and services. Gulati and Garino [34] examined the spectrum of strategies to determine the best level of
integrating physical and virtual operations and discussed the
advantages and trade-offs involved in each choice from the
vendors perspective. Further, Daniel and Wilson [35] identied integration across channels to enable multi-channel
service as one of the key dynamic capabilities necessary for
e-business transformation.
It is important for vendors to manage customer interactions across different interaction channels using a common
set of information and processes, and leveraging information
learned on any channel to provide better services or more targeted offers on other channels. It is imperative that vendors
always give customers the choice of interacting with them in
the manner that is most comfortable for customers. Fully integrated interaction channels would enable vendors to obtain
a unied view of customers across all interaction channels
and deliver consistent and personalized service. This provides vendors with an effective way of creating a smoother
and more satisfying shopping experience for customers and
extending assisted service to online customers, which helps
to improve customers satisfaction and trust. Especially, to
vendors who have physical stores, the physical distribution
ensures that products are available at the right places at the
right time in the right quantities to satisfy customer demand

[36]. Hence, the physical presence will increase consumers

condence in a Web vendors products, no matter whether
the products are delivered online or ofine.
Typically, customers expect all touch points or interaction
channels to provide accurate and unbiased information, easy
navigation, and consistent touch [37]. To consumers, provision of different interaction channels would enable them to
communicate with the vendor through various means, which
in turn would bring them more convenience and increase
consumers condence in the vendor. For instance, if the integration level of the interaction channels of a vendor is high,
consumers can place orders through telephone, fax, Internet
Website, or buy directly from physical stores where they can
touch and feel the products and take them home immediately. They can call to check the status of the products they
have ordered online. They can also return the products they
have bought online to any of the Websites corresponding
physical stores and get the refund. Meanwhile, consumers
can request after-sales services online for any product they
have bought from the vendors physical stores or join online
communitys discussion about the e-commerce vendor and
its products. This indicates that the vendor must have strong
self-condence in its products and the quality of its products
should be good enough to cater to picky customers needs.
Therefore, the fully integrated contact channels of an ecommerce vendor will increase consumers condence in
the vendor, which in turn would raise the level of consumers
trust in the vendor. Conversely, less integration of interaction
channels tend to decrease consumers trust in the vendor

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

because of less condence and communication [38]. In a

similar vein, Bendoly et al. [6] provided evidence regarding
the role of perceived multi-channel integration on loyalty
to a multi-channel rm. Hence, the following hypothesis is
proposed:
H3. The level of multi-channel integration of an ecommerce vendor is positively related to the level of
consumers trust in the vendor.
2.1.3. System assurance
We dene system assurance as the dependability and
security of a vendors online transaction system, which
enables transactions through the Internet be secure and
successful. Previous research has demonstrated that a substantial risk perceived by consumers is related to sharing
credit or debit card information over the Internet. Ambrose
and Johnson [39] found that insufcient trust in the security
and reliability of the transactions over the Internet is a commonly expressed concern of consumers. In addition, Kini
and Choobineh [40] argued that the assurance properties of
the system that consumers interact with are critical in developing and maintaining consumers trust. It follows that:
H4. The system assurance of an e-commerce vendor is
positively related to the level of consumers trust in the
vendor.

2.2. Characteristic of trustors (consumers)

2.2.1. Propensity to trust
In terms of trustor attributes, propensity to trust is the
general willingness to trust other people and a measure of
an individuals tendency to trust or distrust. It is inuenced
by cultural background, personality type and previous experiences [2]. Prior research has provided sufcient evidence
that individuals differ considerably in their general propensity to trust other people. A person may have dispositional
trust because they either believe in the general good nature
of people, or they believe that they will achieve better outcomes by tending to trust people [10]. Existing research has
revealed that an individuals propensity to trust has a major
inuence on his/her trust (e.g., [1]). Hence, we hypothesize
that:
H5. Propensity to trust is positively related to the level
of consumers trust in the e-commerce vendor.

2.3. Consequences of consumer trust

2.3.1. Consumer trust, attitude, and willingness to buy
Ajzen and Fishbein [41] proposed the theory of reasoned
action (TRA) to analyze the psychological processes that
reect observed relationships among beliefs, attitudes, intentions, and behaviors. The theory asserts that intention to

25

perform behavior is determined by the individuals attitude

toward the behavior, and a persons attitude is affected by
his/her beliefs. TRA has been widely accepted and applied
to a broad range of disciplines and contexts. Existing empirical research has revealed that trust is signicantly related to attitude, and attitude positively signies peoples
purchase intention [32,42]. TRA is also applied as the theoretical base in recent studies on trust formation [10,43],
especially in the context of e-commerce [3,5]. Since trust
can be seen as a belief, condence, sentiment, or expectation about an exchange partners intention and/or likely
behavior, it is posited to be directly related to the attitudes
toward purchasing from a vendor and indirectly related to
consumers willingness to buy through purchasing attitudes.
Trust may lead consumers to focus more on the positive motivation because of a sense of afliation and identication
with the vendor. It follows that:
H6. Consumer trust toward an e-commerce vendor is
positively related to favorable attitudes toward purchasing from the vendor.
H7. Favorable attitudes toward an e-commerce vendor
is positively related to consumers willingness to purchase from the vendor.

2.3.2. Consumer trust and risk perception

In an online market, issues such as security, privacy and
risk perceptions are important factors affecting consumers
purchasing decision. Unlike the physical market, consumers
may be dealing with remote vendors they have never met and
products that cannot be touched and felt. Hence, consumers
tend to be reluctant to conduct businesses based only on the
information provided by e-commerce vendors because such
information may not be reliable. Further, in the traditional
marketing literature, researchers maintained that the theory
underlying the relationship marketing concept suggested that
trust is negatively related to perceived risk [44]. Ganesan
[30] argued that trust can reduce the consumers perception
of risk associated with opportunistic behavior by the seller.
It follows that:
H8. Consumer trust toward an e-commerce vendor is
negatively related to the perceived risks involved in
purchasing from the vendor.

2.3.3. Risk perception, attitude, and willingness to buy

Bauer [45] argues that once a risk has been perceived
in a purchase situation, there seems to be some reasonable
evidence that subsequent consumer behavior is shaped by
this risk perception. Similar to trust, perceived risk could
also be regarded as a belief about situations. For example, Mayer et al. [2, p. 726] dened risk perception as the
trustors belief about likelihoods of gains and losses outside of considerations that involve the relationships with the
particular trustee. Therefore, in accordance with TRA [41],

26

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

Table 1
Measures of model variables
Description

Source

Perceived reputation (PR)

This e-commerce vendor:
Is well known
Has a good reputation in the market
Has a bad reputation (reverse coded)
Has a reputation for being honest
Has a reputation for being fair
Has a reputation for being consumer-oriented

Doney and Cannon [26];

Ganesan [30];
Jarvenpaa and Tractinsky [4];
Jarvenpaa et al. [3]

Perceived size (PS)

This e-commerce vendor:
Is a very large company
Is one of the industrys biggest suppliers
on the Web
Is a small player in the market (reverse)
Has regional presence
Has global presence

Doney and Cannon [26];

Jarvenpaa and Tractinsky [4];
Jarvenpaa et al. [3]

Multi-channel integration (MI)

I can order online and collect products in the
physical stores of this e-commerce vendor.
I can return the products I have bought online
to any of this vendors physical stores and get the refund.
I can request after-sales services ofine for any
product I have bought online from this vendor.
I can request after-sales services online for any
product I have bought ofine from this vendor.
The pricing of this e-commerce vendors online
and physical stores is usually the same.
I can choose the most convenient way of
interacting with this vendor (e.g., searching
and purchasing products) ofine or online.

Gulati and Garino [34];

ICSC [64]

System assurance (SA)

The online transaction system of this
e-commerce vendor is:
Stable
Reliable
Dependable
Secure

Kini and Choobineh [40]

Propensity to trust (PTT)

I feel that people are generally reliable.
I feel that people are generally dependable.
I feel that people are generally trustworthy.
I generally trust other people unless they
give me reason not to.

Gefen [1]

Consumer trust (CT)

This e-commerce vendor is trustworthy.
This e-commerce vendor provides reliable information.
This e-commerce vendor keeps promises and commitments.
This e-commerce vendor keeps my best interests in mind.
This e-commerce vendors behavior meets my expectations.

Doney and Cannon [26]; Gefen [1];

Jarvenpaa et al. [3];
Jarvenpaa and Tractinsky [4];
Jarvenpaa et al. [3];
Macintosh and Lockshin [42]

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

27

Table 1 (continued)
Description

Source

Attitude (ATTD)
Using the Internet to shop from this vendor is pleasant.
Using the Internet to shop from this vendor is a good idea.
The idea of using the Internet to shop from this vendor
is appealing.
I like the idea of using the Internet to shop from this vendor.

Jarvenpaa and Tractinsky [4];

Jarvenpaa et al. [3];
Taylor and Todd [65]

Willingness to buy (WTB)

The likelihood that I would return to this vendors Website is:
The likelihood that I would purchase online from this
e-commerce vendor is:
The likelihood that I would purchase online from this e-commerce
vendor in the next 3 months is:
The likelihood that I would purchase online from this e-commerce
vendor in the next year is:
My willingness to buy product(s)/service(s) online from this vendor is:
The probability that I would consider buying online from this vendor is:

Jarvenpaa and Tractinsky [4];

Jarvenpaa et al. [3];
Macintosh and Lockshin [42]

Perceived risk (PRISK)

I believe that the risk of purchasing online from this e-commerce
vendor is very high.
There is a high probability of losing a great deal by purchasing online
from this e-commerce vendor.
There is great uncertainty associated with purchasing online
from this e-commerce vendor.
Overall, I would label the option of purchasing
online from this e-commerce vendor as something negative.

Houghton et al. [66];

Simon et al. [67]

consumers perceived risk might have a negative relationship

with their attitudes toward the purchasing behavior. Recent
research has conrmed that risk perception and attitudes are
closely related. For example, Ruyter et al. [46] empirically
veried that perceived risk has an impact on consumers
attitudes toward e-service. McKnight et al. [43] stated that
trusting intention is likely to be fragile if the perceived risk
is high. Consequently, purchase likelihood tends to be low.
Therefore, we hypothesize that:
H9. Consumers perceived risk associated with purchasing from an e-commerce vendor is negatively related to favorable attitudes toward buying from the
vendor.
H10. Perceived risk associated with purchasing from an
e-commerce vendor is negatively related to consumers
willingness to buy from the vendor.

3. Method
An online survey, targeted at general Internet users, was
utilized to collect data. The survey Websites were hosted
at three homepages designed for respondents in the US,

Singapore, and China respectively. In order to attract more

respondents to participate in this survey, a lucky draw with
prizes up to US$120 was offered. In the survey, items of various constructs were developed by adapting existing measures to the research context. Table 1 summarizes the list
of items and the source. All items were scored on a sevenpoint Likert-type scale ranging from (1) Strongly Disagree
to (7) Strongly Agree.
The survey questionnaire consisted of three sections. In
the rst section, respondents were asked to answer questions on their online purchasing experience (i.e., whether
they had bought products/services online), the possible
types of products/services that they had bought or would
buy online, and the average price of the products/services
that they had bought or would buy online. In the second
section, respondents were asked to choose one of the three
categoriesbooks, music, or travel as the frame of reference when they answered questions in this part. The category chosen was preferably the one that they had recently
purchased. If they had never bought any product/service
online, they were asked to choose a commercial Web vendor that they were familiar with. Respondents were asked
to indicate the extent to which they agree or disagree
with the statements about this specic e-commerce vendor.

28

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

The third section contained basic demographic characteristics including age, gender, nationality, education level,
occupation, monthly personal income, Internet experience,
etc. Respondents were required to provide a valid email
address for the lucky draw. They were also encouraged
to ll in any suggestions and comments at the end of the
questionnaire.
As the survey was conducted in three countries (US
and Singapore being predominantly English-speaking, and
China being predominantly Chinese-speaking), two versions
of the questionnaire were administered. The questionnaire,
originally written in English, was translated into Chinese
by a bilingual person whose native language is Chinese.
The Chinese questionnaire was then translated back in English by another bilingual person whose native language is
English. These two English versions were then compared
and no item was found to pertain to a specic cultural
context in terms of language. This process was conducted
not only because it can prevent any distortions in meaning
across cultures where necessary, but also because it can
enhance the translation equivalence [47]. The questionnaire
was pre-tested with 20 Internet users and there were no major problems with understanding, wording, etc. The same
data-collection procedures were used in the three countries and by the same researchers to improve measurement
equivalence across cultures. The survey was announced
through email, newsgroups, online forums, search engines,
and online banner.
4. Results
4.1. Demographic prole of respondents
The prole of the respondents is shown in Table 2. The
sample consisted of 544 (US), 1381 (Singapore), and 988
(China) valid responses. The gender of the US and Singapore respondents was evenly mixed (US: 50% males; Singapore: 52% males), while majority of the China respondents (73.1%) were males. The gender percentage of the
US respondents was comparable to the results of Spooner
[48], which revealed that 52% of the Internet users in the
US were males. The gender distribution of the Singapore respondents was similar to the ndings of Nielsen/NetRatings
[49], which showed that 58% of the Singapore Internet users
were males. The gender percentage of the China respondents was comparable to the result of China Internet Network Information Center [50], which reported that 74.7%
of the respondents were males.
In addition, most of the respondents of the three groups
were young (say, < 40 years old). About 60% of the US respondents and majority of the Singapore (93.9%) and China
(97.2%) respondents were in the age of 1640 years old.
The US had a much higher percentage of respondents in the
age range of over 40 years old than Singapore and China
(US: 40.4%, Singapore: 4.9%, China: 2.6%). The gender
and age distributions of these three countries conrm the

ndings of Graphic, Visualization, and Usability Centers

(GVU) WWW User Survey [51] that there exists a trend
for the Web towards more balanced gender ratios and an increasing percentage of older users (say, > 40 years old), revealing that the development of the Internet is more mature
in the US than in Singapore and China.
Moreover, a majority of the respondents was well educated. The education level of 95.6% of the US respondents,
92.7% of the Singapore respondents, and 99.4% of the China
respondents was higher than junior college (including junior
college). Results also showed that the US and Singapore
respondents had more Internet experience than their China
counterpart. Only 7.9% and 19.6% of the US and Singapore respondents, respectively reported that they had used
the Internet for less than 4 years, compared to 49.3% of the
China respondents. Conversely, a higher percentage of the
US (42.8%) and Singapore (18.2%) respondents reported
that they had over 7 years of Internet experience compared
to China (3.5%). Moreover, about two-thirds of the respondents in the three countries said their daily usage of the
Internet was about 15 h per day.

4.2. Test for common method variance

As both dependent and independent variable data were
collected from a single informant, common method variance
could be a potential problem. Following Podsakoff and Organ [52], we used the Harmans one-factor test to examine
the extent of the bias. The results of principal component
analyses indicate common method variance is not a problem because several factors with eigenvalue greater than 1
were identied and because no factor account for almost all
the variance [52].

4.3. Means, standard deviations, and correlation matrices

Tables 3ac show the means, standard deviations, and
bivariate correlation coefcients of the constructs. According to Judge et al. [53, p. 620], multi-collinearity typically
is considered to be a serious problem only if the correlation coefcient between the values of two regressors
is greater than 0.8 or 0.9. All correlation coefcients in
this study were less than 0.8. Another indicator of multicollinearityVariance Ination Factors (VIF) was used in
this study. VIF is the reciprocal of the tolerance value. As
VIF increases, so does the variance of the regression coefcient, making it an unstable estimate. Large VIF values
indicate a high degree of collinearity or multi-collinearity
among the independent variables. Results in Table 4 show
that all independent variables have VIFs less than 2.0, which
is well below the usual cutoff level of 10 [54]. This suggests that no multi-collinearity problem among independent
variables exists in this study.

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

29

Table 2
Demographic prole of respondents
Number (percent)
US

Singapore

China

Total

Gender

Male
Female

272
272

(50.0)
(50.0)

718
663

(52.0)
(48.0)

722
266

(73.1)
(26.9)

1712
1201

(58.8)
(41.2)

Age

Under 15
1620
2125
2630
3135
3640
4145
4650
Above 50

0
41
108
69
53
53
56
61
103

(0)
(7.5)
(19.9)
(12.7)
(9.7)
(9.7)
(10.3)
(11.2)
(18.9)

15
333
645
167
90
62
25
24
20

(1.1)
(24.1)
(46.7)
(12.1)
(6.5)
(4.5)
(1.8)
(1.7)
(1.4)

2
67
486
274
107
26
12
6
8

(0.2)
(6.8)
(49.2)
(27.7)
(10.8)
(2.6)
(1.2)
(0.6)
(0.8)

17
441
1239
510
250
141
93
91
131

(0.6)
(15.1)
(42.5)
(17.5)
(8.6)
(4.8)
(3.2)
(3.1)
(4.5)

Education level

Primary
Secondary
Junior college
Poly/Diploma
Bachelor
Master
PhD
Others

0
21
16
123
140
152
90
2

(0)
(3.9)
(2.9)
(22.6)
(25.7)
(27.9)
(16.5)
(0.4)

6
81
663
148
364
91
14
14

(0.4)
(5.9)
(48.0)
(10.7)
(26.4)
(6.6)
(1.0)
(1.0)

1
4
31
117
496
279
60
0

(0.1)
(0.4)
(3.1)
(11.8)
(50.2)
(28.2)
(6.1)
(0)

7
106
710
388
1000
522
164
16

(0.2)
(3.6)
(24.4)
(13.3)
(34.3)
(17.9)
(5.6)
(0.5)

Internet experience (Years)

<4
4 to < 7
7 to < 10
> = 10

43
268
130
103

(7.9)
(49.3)
(23.9)
(18.9)

271
859
206
45

(19.6)
(62.2)
(14.9)
(3.3)

487
466
33
2

(49.3)
(47.2)
(3.3)
(0.2)

801
1593
369
150

(27.5)
(54.7)
(12.7)
(5.1)

Daily usage of Internet (Hours)

<1
1 to < 3
3 to < 5
5 to < 7
7 to< 9
>=9

112
240
121
35
17
19

(20.6)
(44.1)
(22.2)
(6.4)
(3.1)
(3.5)

249
626
280
98
43
85

(18.0)
(45.3)
(20.3)
(7.1)
(3.1)
(6.2)

87
443
201
113
78
66

(8.8)
(44.8)
(20.3)
(11.4)
(7.9)
(6.7)

448
1309
602
246
138
170

(15.4)
(44.9)
(20.7)
(8.4)
(4.7)
(5.8)

4.4. Structural equation modeling (SEM)

SEM was conducted using AMOS 4. A multiple-group
conrmatory factor analysis (CFA) [55] was conducted to
test the cross-cultural measurement invariance and the hypothesis that the same model form and parameter structure
applies to the United States, Singapore, and China samples.
First, the three datasets were tested separately to ensure that
the model t the data for each group. Then the multiplegroup models without constraints on the parameters were estimated simultaneously to test if the same model form holds
across groups. This unconstrained model was then compared
with a constrained one, in which equality constraints on factor structure were placed across the three samples. If the
t of the unconstrained model is not signicantly different
from the t of the constrained one, then the factor structure

does not differ across the samples [55,56]. The results indicate that the t of the unconstrained and constrained were
not signicantly different, thereby implying similar factor
structure across samples.
4.4.1. Measurement equivalence
Establishing data equivalence prior to the interpretation
of results is important to cross-cultural comparison. Measurement equivalence examines whether the same model
hold across different cultures [57]. Translation equivalence,
sample equivalence, and metric equivalence are three main
types of measurement equivalence [24]. Translation equivalence seeks to assure that the translated version of the questionnaire has the same meaning with the version in original
language [57]. This is usually satised by back-translation
[24]. Other methods of testing translation equivalence

0.990
1.410
1.250
1.174
1.055
1.080
1.215
1.370
1.297

1.056
1.350
1.403
1.239
1.261
1.148
1.254
1.267
1.353

(b) Singapore sample

1. PR
5.302
2. PS
5.050
3. IIC
3.858
4. SA
5.327
5. PTT
4.431
6. CT
5.167
7. ATTD
5.146
8. WTB
4.897
9. PRISK
2.988

(a) China sample

1. PR
5.396
2. PS
4.589
3. IIC
3.812
4. SA
5.428
5. PTT
4.693
6. CT
5.338
7. ATTD
5.294
8. WTB
5.420
9. PRISK
2.764

0.372**
0.198**
0.643**
0.267**
0.691**
0.566**
0.499**
0.365**

0.488**
0.130**
0.629**
0.325**
0.669**
0.521**
0.437**
0.288**

0.4 0**
0.056
0.653**
0.323**
0.676**
0.537**
0.456**
0.411**

Correlation is signicant at the 0.01 level (2-tailed).

Correlation is signicant at the 0.05 level (2-tailed).

1.085
1.405
1.697
1.077
1.193
1.043
1.088
1.085
1.199

5.988
5.423
3.438
6.184
4.986
5.899
6.105
6.146
1.906

(a) US sample
1. PR
2. PS
3. IIC
4. SA
5. PTT
6. CT
7. ATTD
8. WTB
9. PRISK

SD

Mean

Measure

Table 3
Means, standard deviations and correlations

0.193**
0.324**
0.143**
0.290**
0.230**
0.178**
0.061

0.010
0.341**
0.208**
0.351**
0.252**
0.093**
0.052

0.100*
0.420**
0.164**
0.286**
0.321**
0.340**
0.201**

0.212**
0.139**
0.228**
0.162**
0.100**
0.127**

0.167**
0.152**
0.178**
0.137**
0.116**
0.046

0.062
0.002
0.005
0.086*
0.130**
0.093*

0.379**
0.777**
0.675**
0.556**
0.368**

0.342**
0.769**
0.653**
0.565**
0.383**

0.401**
0.772**
0.718**
0.640**
0.564**

0.406**
0.393**
0.289**
0.117**

0.425**
0.329**
0.249**
0.105**

0.461**
0.395**
0.247**
0.279**

0.736**
0.607**
0.407**

0.720**
0.591**
0.372**

0.724**
0.578**
0.572**

0.652**
0.371**

0.723**
0.396**

0.707**
0.592**

0.375**

0.403**

0.596**

30
T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

31

Table 4
Collinearity statistics
Independent
variable

US

PR
PS
IIC
SA
PTT

Singapore

China

Tolerance

VIF

Tolerance

VIF

Tolerance

VIF

0.527
0.758
0.989
0.515
0.831

1.899
1.319
1.011
1.941
1.203

0.511
0.754
0.957
0.577
0.852

1.958
1.327
1.045
1.733
1.174

0.555
0.837
0.932
0.531
0.852

1.803
1.194
1.073
1.884
1.174

Table 5
Construct reliability analysis
Construct

PR
PS
MI
SA
PTT
CT
ATTD
WTB
PRISK

No. of items

4
3
2
4
3
4
4
5
4

Composite reliability

Variance extracted

US

Singapore

China

US

Singapore

China

0.91
0.88
0.76
0.96
0.96
0.95
0.96
0.94
0.92

0.89
0.84
0.77
0.96
0.94
0.93
0.94
0.94
0.91

0.90
0.81
0.67
0.94
0.96
0.92
0.91
0.92
0.91

0.72
0.71
0.61
0.84
0.90
0.81
0.85
0.75
0.74

0.67
0.65
0.63
0.85
0.84
0.78
0.81
0.77
0.72

0.70
0.59
0.50
0.79
0.88
0.75
0.72
0.71
0.72

include visual examination of factor patterns for similarity

[47] and common form test of multiple group measurement
models [55]. Sample equivalence addresses the extent to
which the samples from different cultures are comparable, such that cross-cultural difference cannot be attributed
to heterogeneous sample characteristics. Sample equivalence can be satised by using similar sampling frames
and sampling methods in all cultural groups [24]. Metric
equivalence examines whether the psychometric properties
of data from various groups exhibit the same coherence
and structure. If measures satisfy the requirement of metric
equivalence, conclusions made would not be inuenced by
measure unreliability or differing dimensionality. Metric
equivalence can be tested by constraining the loadings to
be the same across countries [58]. In this study, translation
equivalence and sample equivalence were enhanced by the
deliberate design of the translation and back-translation
procedure and the same sampling methods across the three
countries. Metric equivalence was tested via multiple-group
CFA.
4.4.2. Measurement model
In order to assess the validity of the measures, Bollen [55]
suggested examining the factor loadings and the squared
multiple correlations (SMC, which is equivalent of R 2 in linear regression) between the items and the constructs. All fac-

tor loadings were highly signicant (p < 0.01 for all cases).
Items that had SMC lower than 0.4 were dropped. The same
items were eliminated from the three samples to ensure that
the form structure was equivalent across countries. Composite reliability and variance extracted measures were used to
test the reliability of model constructs. All composite reliabilities except multi-channel integration for China sample
are above 0.7 while variance extracted measures were above
0.5 suggested by Hair et al. [54] (see Table 5). Hence, the
model constructs are deemed reliable.
We estimated two sets of measurement models for exogenous and endogenous variables respectively and the following model t indices were employed: Goodness-of-Fit Index
(GFI), Adjusted Goodness-of-Fit Index (AGFI), Normed Fit
Index (NFI), Tucker-Lewis Index (TLI), Comparative Fit
Index (CFI), and Root Mean Square Error of Approximation (RMSEA). The obtained GFIs were higher than the 0.9
benchmark recommended by Bagozzi and Yi [59] for the
three samples. AGFIs were also higher than the threshold
of 0.8. In addition, the obtained NFIs, TLIs, CFIs were all
above the benchmark of 0.9. An RMSEA of 0.05 or less
had been suggested as an indicator of close t, whereas
values between 0.05 and 0.08 suggested a reasonable t
of the model to the data [60]. All obtained RMSEAs fall
within the acceptable standards. Therefore, all the model
t indices are indicative of good t for the exogenous and

32

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

Table 6
Fit indices of multiple-group CFA analysis for measurement models
Fit indices

2
df
p
GFI
AGFI
NFI
TLI
CFI
RMSEA

Exogenous

Endogenous

Model A

Model B

Model C

Model A

Model B

Model C

1658.36
282
< 0.05
0.93
0.90
0.96
0.95
0.96
0.04

1744.30
304
< 0.05
0.93
0.91
0.95
0.96
0.96
0.04

1882.72
334
< 0.05
0.92
0.91
0.95
0.96
0.96
0.04

1998.90
336
< 0.05
0.92
0.90
0.96
0.96
0.97
0.04

2081.63
362
< 0.05
0.92
0.90
0.96
0.96
0.96
0.04

2285.69
384
< 0.05
0.92
0.90
0.95
0.96
0.96
0.04

Model A: unconstrained model.

Model B: factor loadings are constrained to be equal.
Model C: factor loadings, factor variances, and factor covariances are constrained to be equal.

endogenous parts of the measurement models for the three

countries.
Next, the multiple-group CFA analytic procedure was employed to test the cross-national invariance of the instrument.
We chose the nested testing hierarchy suggested by Bollen
[55]. An unconstrained model (Model A), in which no constraints were placed across groups, was rst estimated. Table
6 shows that although
2 was signicant for both the exogenous and endogenous parts, all other indices of the model
t fall within the recommended ranges, indicating that the
form of measurement models was appropriate for all the
three countries.
The unconstrained model (Model A) was then compared
with a constrained one (Model B), in which the factor loadings were constrained to be invariant across samples [55].
The differences of
2 between Model B and Model A for
exogenous and endogenous parts were signicant (

2 =
85.943,
df = 22, p < 0.05;

2

Endo

Exo

= 82.737,
df = 26,

p < 0.05). However, because
2 difference test is sensitive

to the sample size, we compared the differences in CFI.
The differences of CFI between Model B and Model A for
the exogenous and endogenous parts were 0.002 and 0.001,
respectively, which fall within the 0.01 benchmark recommended by Cheung and Rensvold [61], indicating that the
factor loadings were the same for the three groups.
Then we estimated Model C, in which all factor loadings, factor variances, and factor covariances were declared invariant across groups. Since Model C was nested
within Model B, Model C was subsequently compared
with Model B using the test of differences in CFI values.
The differences in CFI between Model C and Model B
for the exogenous and endogenous parts were 0.003 and
0.004, respectively. Therefore, the model structure was
deemed the same across the US, Singapore, and China
groups, indicating that the measurement equivalence was
satised.

Table 7
Fit indices of multiple-group SEM analysis for structural model
Fit indices

Model A

Model B

2

4208.28
1392
< 0.05
0.92
0.90
0.96
0.97
0.97
0.03

4294.27
1412
< 0.05
0.92
0.90
0.95
0.96
0.97
0.03

df
p
GFI
AGFI
NFI
TLI
CFI
RMSEA

Model A: unconstrained model.

Model B: structural paths are constrained to be equal.

4.4.3. Structural model

After establishing measurement invariance across samples, the structural model was then tested. First, the US, Singapore, and China samples were estimated separately. Overall, the t statistics indicate that the consumer trust model
provides a good t to the data for the three countries.
Similar to the procedure conducted for measurement models, multiple-group SEM was then used to determine whether
the strengths of relationships among constructs hold across
countries [56]. We rst estimated an unconstrained model
(Model A), in which data of the three groups were analyzed
simultaneously and no constraints were imposed. Again all
t statistics fall within the accepted ranges except
2 , indicating that the same structural model form applies to both
countries (see Table 7). Next, a constrained model (Model
B) was estimated in which all the structural paths were imposed to be equal across the three samples [55]. The difference of CFI between Model A and Model B is only 0.001,
supporting the hypothesis that the structural paths are the
same in the three countries.

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

33

Characteristics of Trustees
(E-Commerce Vendors)

Perceived Reputation
0.394**
0.568**
0.392**
Perceived Size

Multichannel Integration

System Assurance

0.067
0.012
0.009
0.014
0.022
0.006
0.673**
0.550**
0.567**

0.095**
0.115**
Characteristic of Trustors (Consumers)
0.097**
Propensity to Trust

0.80
Consumer Trust 0.77
0.76
0.699**
0.502**
0.510**

0.504**
0.656**
0.560**

0.66
Attitude 0.64
0.68

0.616**
0.807**
0.865**

0.229**
0.103**
0.049*

0.68
Willingness to Buy 0.62
0.55

0.44
Perceived Risk 0.19
0.20

US Sample
0.900
GFI
0.879
AGFI
NFI
0.949
CFI
0.972
PNFI
0.834
RMSEA 0.047
n
544

Singapore Sample
0.931
0.916
0.961
0.972
0.845
0.044
1381

0.250**
0.150**
0.132**
China Sample
0.915
0.897
0.946
0.962
0.832
0.048
988

Fig. 2. Structural model. Unstandardized path coefcients appear on arrows and squared multiple correlations appear inside each latent
variables ellipse. The estimates of the US sample are shown above those of the Singapore and China samples. p < 0.01, p < 0.05.

The SMC results indicate that the ve antecedents of consumer trust explain 80%, 77%, and 76% of the variance of
consumer trust for the US, Singapore, and China samples respectively. Sixty-six percent (US sample), 64% (Singapore
sample), and 68% (China sample) of the variance of attitude
is accounted for by the variance in consumer trust and perceived risk. Sixty-eight percent, 62%, and 55% of the variance in willingness to buy can be indicated by attitude and
perceived risk for the US, Singapore, and China groups, respectively. In addition, 44% (US sample), 19% (Singapore
sample), and 20% (China sample) of the variance in perceived risk is explained by consumer trust. Hence, the consumer trust model proposed in this study can be deemed to
explain a large proportion of the variance in consumer trust,
attitude, and willingness to buy.
4.5. Hypotheses testing
The unstandardized weights were used to discuss the results because although the standardized coefcients are useful for determining relative importance, they are sample
specic and not comparable across samples. Conversely,
the unstandardized coefcients are comparable across samples and retain their scale effect [54]. The results in Fig. 2
show that three of the ve hypotheses regarding antecedents

of consumer trust are supported by the data of the three

samples.
As predicted, perceived reputation (H1 : US = 0.394,
Singapore = 0.568, China = 0.392, p < 0.01), system assurance (H4 : US = 0.673, Singapore = 0.550, China = 0.567,
p < 0.01), and propensity to trust (H5 : US = 0.095,
Singapore = 0.115, China = 0.097, p < 0.01 ) have signicantly positive relationships with consumer trust in
e-commerce vendors. The high s of H4 suggest that the
system assurance of an e-commerce vendor have a major
inuence on consumers trust in the vendor. In contrast,
consumers perceived size (H2 : US = 0.067, Singapore =
0.012, China = 0.009, p > 0.05) does not show positive
relationship with consumer trust for the three samples.
In addition, multi-channel integration (H3 : US = 0.014,
Singapore = 0.022, China = 0.006, p > 0.05) is unrelated
to consumer trust for the three samples.
Consumers trust has a signicant positive relationships
with their attitude toward the vendor (H6 : US = 0.616,
Singapore =0.807, China =0.865, p < 0.01). Subsequently,
consumers attitude has a strong positive relationship with
their willingness to buy from the vendor (H7 : US = 0.504,
Singapore = 0.656, China = 0.560, p < 0.01). As expected,
the negative effect of consumer trust on their perceived
risk associated with purchasing from the vendor is also

34

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

highly signicant (H8 : US = 0.699, Singapore = 0.502,

China = 0.510, p < 0.01). The ndings for H9 (US =
0.229, Singapore = 0.103, p < 0.01; China = 0.049,
p < 0.05) suggest that the negative relationship between
consumers risk perception and their attitude exists for the
three samples. Perceived risk is also revealed to have a strong
negative relationship with willingness to buy (H10 : US =
0.250, Singapore = 0.150, China = 0.132, p < 0.01)
for the three samples.

5. Discussion
Overall, the results provide support for the proposed
model of consumer trust in e-commerce vendors for the
US, Singapore, and China groups. As expected, characteristics of both e-commerce vendors (perceived reputation and
system assurance) and consumers (propensity of trust) are
found to be determinants of consumer trust in Web vendors
and have similar loadings across countries.
In addition, results of the three countries are consistent
with the ndings of Jarvenpaa et al. [3], which revealed that
perceived reputation is positively related to consumer trust
but perceived size is not related to trust. One possible explanation for the lack of support for the positive relationship
between perceived size and consumer trust could be that the
perceived size of an e-commerce vendor does not inuence
consumers trust as heavily as the perceived size of a traditional business does. Size of an online vendor is less easily
perceived on the Web than size of a physical store. In the
physical world, it is easy for consumers to assess a companys size themselves through its physical presence. That
is, they can personally judge the size of a vendor. But in
the online world, this becomes difcult as size of a vendor
cannot be easily and correctly judged through its website.
Hence, consumers may not care much about the size of ecommerce vendors.
Results on consequences of consumer trust are also consistent with the ndings of Jarvenpaa et al. [3]. Positive relationships exist between consumer trust and their attitude
toward a vendor and between consumers attitude and their
willingness to buy from the vendor. Consumer trust mitigates the perceived risk associated with online purchase
while perceived risk is negatively related with consumers
attitude and their willingness to buy. TRA is conrmed by
the results of this study, which reveal that consumers trust
toward an e-commerce vendor is signicantly positively related to their attitude toward the vendor, and favorable attitude is positively related with customers willingness to buy.
This nding is consistent with the results reported in previous empirical studies (e.g., [3,32]). As predicted, results
show that there is a strong negative relationship between
consumer trust and their risk perception across countries. In
other words, high level of consumers trust in an e-commerce
vendor decreases consumers perception of risk associated
with purchasing online from the vendor. This result is con-

sistent with that of Jarvenpaa et al. [3] and Shemwell et al.

[44]. Moreover, perceived risk is found to be negatively associated with consumers attitude toward the vendors and
their willingness to buy from the vendor.
Another interesting nding is that risk perception has the
least negative relationships with attitude and with willingness to buy for the China sample, followed by the Singapore and the US samples. The risk attitude of China consumers has long been a controversial topic. Yamagishi and
Yamagishi [22] suggested that collectivists (e.g., Chinese)
are less trusting and more risk-averse. Conversely, Chinese
respondents were found to be signicantly less risk-averse
and more risk-seeking than Americans and other Western
respondents in their choices between risky options and sure
outcomes [62]. The individualism index of Singapore is
higher than that of China, but lower than that of the US.
Hence, the negative effect of risk perception on consumers
attitude of the Singapore sample is at the middle of the three
samples.
In addition, the different development stages of ecommerce in the three countries may lead to the fact that
risk perception has different effect on attitude and purchase
intention. The online transaction system is the most mature
in the US among the three countries, followed by Singapore
and China. Consequently, US and Singapore consumers
generally pay online if they purchase some products from
Web vendors. Conversely, because credit card penetration
in China is still very low, the main way for China consumers to utilize the advantages of the Internet is to search
and order online and then pay ofine when they receive the
products. Most of the China consumers regard this procedure (order online and pay ofine) as purchase online.
Because of this special method of conducting e-commerce
transactions, it is possible that even if China consumers
observe risks, they still have a comparably positive attitude
toward e-commerce vendors and are more willing to buy
from them. Moreover, the continuous effort of the China
Government to develop e-commerce in China and the mass
positive reports published in the media could be another
possible reason to explain the less negative relationships
between risk perception and attitude and between risk
perception and willingness to buy in the China group.
Results indicate that system assurance has the strongest
inuence on consumer trust among all the antecedents of
trust for the US and China samples. The structural loadings
of system assurance are > 0.55 for the three countries, followed by those of perceived reputation (> 0.39) and consumers propensity to trust (> 0.09). This reveals that the
reliability and security of an e-commerce vendors online
transaction system is a vital factor in developing and maintaining consumers trust in Internet vendors. For consumers
who conduct transactions online, privacy and security are
prime concerns. Although many researchers have emphasized the important role that the properties of online vendors transaction system plays in e-commerce, little empirical research has been done to test the relationship between

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

system assurance and consumer trust. Our study lls in the

gap and facilitates future research in trust, especially in the
context of Internet shopping.
Results indicate that consumers propensity to trust has
a signicantly positive relationship with their trust in Web
vendors for the three countries. This is consistent with previous research e.g., Gefen [1] empirically revealed that individuals general trusting propensity, which is the product
of a lifelong socialization process, is positively related to
individuals trust. This is because individuals general experiences and perceptions of whether others can be relied on
to deliver promised outcomes will play an important role in
inuencing their trust in others. Hence, individuals who are
trusting will be more likely to be open with others and trust
others.
Surprisingly, multi-channel integration is not related to
consumer trust. It is possible that the low price of the products/services that most consumers tend to buy online may decrease consumers tendency to assort communication from
vendors across different interaction channels. In addition,
because company size and multi-channel integration are related, consumers may not care much about the integration
of interaction channels because they do not care about the
size of the vendors. Another possible reason may be that
although companies increasingly recognize the importance
of integrating interaction channels with consumers, the efforts made on increasing the integration level is still much
less than those emphasized on other aspects (e.g., product
information, site design). Consequently, most consumers do
not have a clear concept on this special service, even if the
vendor has fully integrated their communication channels.
All these lead to the ndings that consumers focus more on
the reputation or branding and the system assurance of ecommerce vendors instead of the perceived size and multichannel integration.

6. Implications and conclusions

As some researchers have argued, theory-guided empirical studies of consumer trust in Internet shopping are relatively rare, which critically impedes an understanding of
consumer trust in Internet shopping [13]. Our research lls
in the gap by proposing a theoretical model regarding the
formation of consumer trust and the consequences of it in
e-commerce, empirically testing if it holds across countries,
and comparing consumer-related attributes among the US,
Singapore, and China. It mainly contributes to trust literature in several aspects.
First, this study extends prior research in trust by including characteristics of both trustors (e-commerce vendors)
and trustees (consumers) in the same empirical study and
exploring the formation of trust in a new context. Our results
show that the characteristics of the trustees (i.e., perceived
reputation and system assurance of an e-commerce vendor)
and the characteristic of trustors (i.e., consumers propensity

35

to trust) are important determinants of trust across countries.

These ndings conrm the conceptual model proposed by
Mayer et al. [2], which described the effect of dyadic attributes of trustees and trustors on trust. This study is one
of the rst to empirically examine attributes of both vendors and consumers in e-commerce application. In addition,
although some researchers have emphasized the important
role that the properties of online transaction system plays
on the formation of trust [40] in e-commerce, previous research has not involved investigating the effect of system
assurance. Our study includes system assurance as a unique
determinant of consumer trust in the model and shows that
system assurance of a Web vendor has the strongest inuence on consumer trust among all the antecedents of trust,
which reveals a distinct difference from the antecedents of
trust in the traditional marketing literature. We also propose
a new constructmulti-channel integration as a characteristic of e-commerce vendors in the research model. Though
the integration level of multiple communication channels
of vendors has drawn researchers and practitioners attention, previous research has not involved examining its effect
on consumers trust. Although the results of this study did
not show positive relationship between multi-channel integration and consumer trust, our model could be a reasonable foundation for future research to explore the effect of
multi-channel integration on trust under certain situations
(e.g., purchase of expensive products/services) or in more
advanced development stage of e-commerce.
Second, this study conrms that consumers attitude and
intention to participate in e-commerce behaviors can be studied effectively using the traditional psychological theory of
TRA. Specically, the results show that consumers trust in
e-commerce vendors and their risk perception can also be
regarded as behavioral beliefs that affect consumers behavioral attitude. Our empirical results for the three countries
are consistent with TRA, indicating its generalizability in
e-commerce application.
Third, researchers have stated that in the context of Internet shopping, whether and how trust affects perceived risk
is still very much an open issue [13]. However, similar to
our study, a recent study by Pavlou and Gefen [63] found
that trust reduces perceived risk. Our study is one of the
few studies to examine both trust and risk in one theoretical
model in the context of e-commerce. Moreover, the structural weights of the three samples are all over 0.5, revealing
that the effect of consumer trust on mitigating risk perception
is strong. Hence, this study could also be a reasonable starting point for exploring the relationship between consumer
trust and risk perception in the context of e-commerce.
Fourth, for Internet vendors who want to leverage the
benets of online transactions, the study provides a number
of strategies that they might deploy. Since system assurance
is a unique factor in online business and is found to have
a strong inuence on consumers trust across countries, online retailers may need to stress the security and reliability
of their online transaction systems. In addition, they may

36

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

increase their perceived reputation through advertising and

publicity. The design of the Websites should also focus on
the information that conveys good reputation of the company and measures taken to protect consumers privacy and
provide security associated with conducting transactions online.
Fifth, in order to encourage consumers to conduct transactions online, vendors also need to use various strategies to
build trust. For example, vendors should provide prompt delivery, better after-sales service, and more accurate product
information. Vendors should also make an effort to decrease
consumers risk perception to promote online purchase.
Although our empirical evidence does not support the
relationship of consumers perceived size and multi-channel
integration of a vendor with consumer trust, further research
may need to explore the in-depth reasons. For example,
the insignicant relationship between the perceived size and
consumer trust of this research is inconsistent with the results
of traditional marketing research. Does perceived size only
inuence consumer trust in traditional business? Is there
any other factor that restricts the effect of perceived size on
trust in e-commerce vendor (e.g., product catalog, product
value, etc.)? Although multi-channel integration does not
show the proposed relationship with consumer trust in this
study, it is not clear if this construct has a signicant effect
on trust in other countries. Hence, further examination of
the consumer trust model in other countries is necessary to
check its generalizability. In addition, further research using
a longitudinal design may give us a better understanding
of the dynamics of the development of trust in e-commerce
applications.
An interesting nding of this study is that similar results
were obtained across three cultures. This demonstrates the
generalizability of the model across cultures. Future research
can examine the proposition that culture does not make a
difference on trust among educated people from various cultures. This proposition, though contrary to Doney et al. [8]
arguments on the importance of national culture on trust,
warrants further investigation. In fact, in a study on consumer trust, Jarvenpaa and Tractinsky [4] found that trust
seems to work the same in the US and Israel. Similarly, in
a study of cultural diversity and trust in IT adoption, Gefen
et al. [23] found that trust seems to work almost the same
in both South Africa and the United States.

References
[1] Gefen D. E-commerce: the role of familiarity and trust. Omega
2000;28(6):72537.
[2] Mayer RC, Davis JH, Schoorman FD. An integrative model
of organizational trust. Academy of Management Review
1995;20(3):70934.
[3] Jarvenpaa SL, Tractinsky N, Vitale M. Consumer trust in
an internet store. Information Technology and Management
2000;1(12):4571.

[4] Jarvenpaa SL, Tractinsky N. Consumer trust in an internet

store: a cross-cultural validation. Journal of ComputerMediated Communication 1999;5(2):135 Available from:
http://www.ascusc.org/jcmc/vol5/issue2/jarvenpaa.html.
[5] McKnight DH, Choudhury V, Kacmar C. Trust in e-commerce
vendors: a two-stage model. In: Proceedings of international
conference on information systems (ICIS2000), Australia,
December, 2000.
[6] Bendoly E, Blocher D, Bretthauer K, Krishnan S,
Venkataramanan MA. Impacts of availability and perceptions
of integration in multi-channel operations. Journal of Service
Research, 2005, forthcoming. Available from: http://www.fc.
bus.emory.edu/elliot_bendoly/JSR_Channels.pdf.
[7] Hofstede G. Cultures consequences: international differences
in work-related values. Beverly Hills, CA: Sage; 1980.
[8] Doney PM, Cannon JP, Mullen M. Understanding the
inuence of national culture on the development of trust.
Academy of Management Review 1998;23(3):60120.
[9] Chai L, Pavlou PA. Customer relationship management.com: a
cross-cultural empirical investigation of electronic commerce.
In: Proceedings of the eighth Americas conference on
information systems, Dallas, 2002. p. 48391.
[10] Mcknight DH, Chervany NL. What trust means in
e-commerce customer relationships: an interdisciplinary
conceptual typology. International Journal of Electronic
Commerce 2001;6(2):3559.
[11] Bhattacherjee A. Individual trust in online rms: scale
development and initial test. Journal of Management
Information Systems 2002;19(1):21141.
[12] Pennington R, Wilcox HD, Grover V. The role of system trust
in business-to-consumer transactions. Journal of Management
Information Systems 2003;20(3):197226.
[13] Lee MKO, Turban E. A trust model for consumer internet
shopping. International Journal of Electronic Commerce
2001;6(1):7591.
[14] Tan YH, Thoen W. Toward a generic model of trust for
electronic commerce. International Journal of Electronic
Commerce 2000;5(2):6174.
[15] Tan YH, Thoen W. Electronic contract drafting based on
risk and trust assessment. International Journal of Electronic
Commerce 2003;7(4):5571.
[16] Gefen D, Karahanna E, Straub DW. Trust and TAM in online
shopping: an integrated model. MIS Quarterly 2003;27(1):
5190.
[17] Pavlou PA. Consumer acceptance of electronic commerce:
integrating trust and risk with the technology acceptance
model. International Journal of Electronic Commerce
2003;7(3):10134.
[18] Koufaris M, Hampton-Sosa W. The development of initial
trust in an online company by new customers. Information
and Management 2004;41(3):37797.
[19] Walczuch R, Lundgren H. Psychological antecedents of
institution-based consumer trust in e-retailing. Information &
Management 2004;42(1):15977.
[20] Gefen D, Straub DW. Consumer trust in B2C e-Commerce and
the importance of social presence: experiments in e-Products
and e-Services. Omega 2004;32(6):40724.
[21] Kennedy MS, Ferrell LK, LeClair DT. Consumers trust of
salesperson and manufacturer: an empirical study. Journal of
Business Research 2001;51(1):7386.

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

[22] Yamagishi T, Yamagishi M. Trust and commitment in
the United States and Japan. Motivation and Emotion
1994;18(2):12966.
[23] Gefen D, Rose GM, Warkentin M, Pavlou P. Cultural diversity
and trust in IT adoption: a comparison of USA and South
African e-voters. Journal of Global Information Management
2005;13(1):5478.
[24] Sin LYM, Cheung GWH, Lee R. Methodology in crosscultural consumer research: a review and critical assessment.
Journal of International Consumer Marketing 1999;11(4):
7596.
[25] Sekaran U. Methodological and theoretical issues and
advancements in cross-cultural research. Journal of
International Business Studies 1983;14(2):6173.
[26] Doney PM, Cannon JP. An examination of the nature of
trust in buyerseller relationships. Journal of Marketing
1997;61(2):3551.
[27] Yaniv I, Kleinberger E. Advice taking in decision
making: egocentric discounting and reputation formation.
Organizational Behavior and Human Decision Processes
2000;83(2):26081.
[28] Kartalia J. Reputation at risk?. Risk Management 2000;4(7):
518.
[29] Chiles TH, McMackin JF. Integrating variable risk
preferences, trust, and transaction cost economies. Academy
of Management Review 1996;21(1):7399.
[30] Ganesan S. Determinants of long-term orientation in
buyerseller relationships. Journal of Marketing 1994;58(2):
119.
[31] Grifn A, Hauser JR. The voice of the customer. Marketing
Science 1993;12(1):127.
[32] Chow S, Holden R. Toward an understanding of loyalty:
the moderating role of trust. Journal of Managerial Issues
1997;9(3):27598.
[33] Teo TSH, Pian Y. A model for web adoption. Information and
Management 2003;41(4):45768.
[34] Gulati R, Garino J. Get the right mix of bricks and clicks.
Harvard Business Review 2000; MayJune:10714.
[35] Daniel EM, Wilson HN. The role of dynamic capabilities in
e-business transformation. European Journal of Information
Systems 2003;12(4):28296.
[36] Gurau C, Ranchhod A, Hackney R. Internet transactions
and physical logistics: conict or complementarity?. Logistics
Information Management 2001;14(2):3343.
[37] Shankar V, Urban GL, Sultan F. Online trust: a stakeholder
perspective, concepts, implications, and future directions.
Journal of Strategic Information Systems 2002;11(34):
32544.
[38] Teo TSH, Liu J. Consumer trust in e-commerce: a crosscultural study. In: Proceedings of academy of management
2002 conference, Denver, August, 2002.
[39] Ambrose PJ, Johnson GJ. A trust based model of buying
behavior in electronic retailing. In: Proceedings of the fourth
Americas conference on information systems, Baltimore, MA,
August, 1998.
[40] Kini A, Choobineh J. Trust in electronic commerce: denition
and theoretical considerations. In: Proceedings of the 31st
Annual Hawaii International Conference on System Sciences
4. Los Alamitos, CA: IEEE Computer Society Press; 1998.
p. 5161.
[41] Ajzen I, Fishbein M. Understanding attitudes and predicting
social behavior. Englewood Cliffs, NJ: Prentice-Hall; 1980.

37

[42] Macintosh G, Lockshin LS. Retail relationships and store

loyalty: a multi-level perspective. International Journal of
Research in Marketing 1997;14(5):48797.
[43] McKnight DH, Cummings L, Chervany N. Initial trust
formation in new organizational relationship. Academy of
Management Review 1998;23(3):47390.
[44] Shemwell DJ, Cronin JJ, Bullard WR. Relational exchange
in services: an empirical investigation of ongoing customer
service-provider relationships. International Journal of Service
Industry Management 1994;5(3):5768.
[45] Bauer RA. Consumer behavior as risk taking. In: Cox DF,
editor. Risk taking and information handling in consumer
behavior. Boston, MA: Harvard University Press; 1967.
p. 2333.
[46] Ruyter KD, Wetzels M, Kleijnen M. Customer adoption of eservice: an experimental study. International Journal of Service
Industry Management 2001;12(2):184207.
[47] Douglas S, Craig CS. International Marketing Research.
Englewood Cliffs, NJ: Prentice-Hall; 1983.
[48] Spooner T. Asian Americans and the internet: the
young and the connected. Pew Internet and American
Life Project, 2001. Available from: http://www.pewinternet.
org/reports/pdfs/PIP_Asians_Report.pdf.
[49] Nielsen/NetRatings. Women a formidable force on the
Web 2001. Available from: http://www.nielsen-netratings.
com/pr/pr_010628_au.pdf.
[50] China Internet Network Information Center. Semiannual
Survey Report on the Development of Chinas Internet,
July, 2000. Available from: http://www.cnnic.net.cn/develst/
e-cnnic200007.shtml.
[51] Graphic, Visualization, and Usability Centers (GVU)
WWW User Survey, 1998. Available from: http://www.gvu.
gatech.edu/user_surveys/.
[52] Podsakoff PM, Organ DW. Self-reports in organizational
research: problems and prospects. Journal of Management
1986;12(4):53144.
[53] Judge GG, Hill RC, Grifths W, Lutkepohl H, Lee TC.
Introduction to the theory and practice of econometrics. New
York: Wiley; 1982.
[54] Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate
data analysis with readings. 5th ed., Englewood Cliffs, NJ:
Prentice-Hall; 1998.
[55] Bollen KA. Structural equations with latent variables. New
York: Wiley; 1989.
[56] Durvasula S, Andrews JC, Lysonski S, Netemeyer RG.
Assessing the cross-national applicability of consumer
behavior models: a model of attitude toward advertising in
general. Journal of Consumer Research 1993;19(4):62636.
[57] Mullen MR. Diagnosing measurement equivalence in crossnational research. Journal of International Business Studies
1995;26(3):57396.
[58] Steenkamp JEM, Baumgartner H. Assessing measurement
invariance in cross-national consumer research. Journal of
Consumer Research 1998;25(1):7890.
[59] Bagozzi RP, Yi Y. On the evaluation of structural equation
models. Journal of the Academy of Marketing Research
Marketing Science 1988;16(1):7494.
[60] Browne MW, Cudeck R. Alternative ways of assessing model
t. In: Bollen KA, Long JS, editors. Testing structural equation
models. Newbury Park, CA: Sage; 1993. p. 13662.

38

T.S.H. Teo, J. Liu / Omega 35 (2007) 22 38

[61] Cheung GW, Rensvold RB. What constitutes signicant

differences in evaluating measurement invariance? 1999
Conference of the Academy of Management, Chicago, 1999.
[62] Weber EU, Hsee CK. Cross-cultural differences in risk
perception but cross-cultural similarities in attitudes towards
risk. Management Science 1998;44(9):120517.
[63] Pavlou PA, Gefen D. Building effective online marketplaces
with institution-based trust. Information Systems Research
2004;15(1):3759.
[64] ICSC, Multi-channel retailing. ICSC Research Quarterly
1999;6(3): 15 Available from: http://www.icsc.org/
ecommerce/d.pdf.

[65] Taylor S, Todd PA. Understanding information technology

usage: a test of competing models. Information Systems
Research 1995;6(2):14476.
[66] Houghton SM, Simon M, Aquino K, Goldberg C. No safety
in numbers: the effects of cognitive biases on risk perception
at the team level. Group and Organization Management
2000;25(4):32553.
[67] Simon M, Houghton SM, Aquino K. Cognitive biases,
risk perception, and venture formation: how individuals
decide to start companies?. Journal of Business Venturing
2000;15(2):11334.