Webstrategiestopromoteinternetshopping_withmarginnotes

Sia et al./Web Strategies to Promote Internet Shopping

RESEARCH ARTICLE

WEB STRATEGIES TO PROMOTE INTERNET SHOPPING:

IS CULTURAL-CUSTOMIZATION NEEDED?1

By: Choon Ling Sia

City University of Hong Kong

83 Tat Chee Avenue

Kowloon, Hong Kong

CHINA

iscl@cityu.edu.hk

Kai H. Lim

City University of Hong Kong

83 Tat Chee Avenue

Kowloon, Hong Kong

CHINA

iskl@cityu.edu.hk

Kwok Leung

City University of Hong Kong

83 Tat Chee Avenue

Kowloon, Hong Kong

CHINA

mgkleung@cityu.edu.hk

Matthew K. O. Lee

City University of Hong Kong

83 Tat Chee Avenue

Kowloon, Hong Kong

CHINA

fbmatlee@cityu.edu.hk

Wayne Wei Huang

College of Business

Ohio University

Athens, OH 45701

U.S.A.

huangw@ohio.edu

1Carol Saunders was the accepting senior editor for this paper. France

Belanger served as the associate editor.

Izak Benbasat

Sauder School of BUsiness

University of British Columbia

Vancouver, BC V6T 1Z2

CANADA

izak.benbasat@sauder.ubc.ca

Abstract

Building consumer trust is important for new or unknown

Internet businesses seeking to extend their customer reach

globally. This study explores the question: Should website

designers take into account the cultural characteristics of

prospective customers to increase trust, given that different

trust-building web strategies have different cost implications?

In this study, we focused on two theoretically grounded

practical web strategies of customer endorsement, which

evokes unit grouping, and portal affiliation, which evokes

reputation categorization, and compared them across two

research sites: Australia (individualistic culture) and Hong

Kong (collectivistic culture). The results of the laboratory

experiment we conducted, on the website of an online book-

store, revealed that the impact of peer customer endorsements

on trust perceptions was stronger for subjects in Hong Kong

than Australia and that portal (Yahoo) affiliation was effective

only in the Australian site. A follow-up study was conducted

as a conceptual replication, and provided additional insights

on the effects of customer endorsement versus firm affiliation

on trust-building. Together, these findings highlight the need

to consider cultural differences when identifying the mix of

web strategies to employ in Internet store websites.

Keywords: Cross-cultural study, Internet shopping, trust,

web strategies

MIS Quarterly Vol. 33 No. 3, pp. 491-512/September 2009 491

Sia et al./Web Strategies to Promote Internet Shopping

Introduction

With the globalization of businesses and the advent of the

Internet, many firms have set up websites for each country/

city in which they have a local presence, featuring the respec-

tive local language and contents, but typically with uniform

website designs and features. To build trust in Internet stores,

particularly if they are new or relatively unknown, numerous

web strategies/design features have been proposed in the

practitioner literature (e.g., Cheskin Research and Studio

Archetype/Sapient 1999). Trust is a critical concern of Inter-

net consumption behaviors (Brynjolfsson and Smith 2000;

Keen 1997; Sin 2003; Van Slyke et al. 2004), and has

repeatedly been shown to enhance Internet shopping inten-

tions and behaviors (e.g., Jarvenpaa et al. 2000; Lim et al.

2006).

Much of the extant literature on Internet trust has adopted a

universalist approach by assuming that trust (particularly its

antecedents) is built uniformly across different national cul-

tures (e.g., Jarvenpaa et al. 2000; Lim et al. 2006). Although

traditional marketing scholars (e.g., Aaker and Maheswaran

1997; de Mooij and Hofstede 2002; Straughan and Albers-

Miller 2001) and Internet practitioners (e.g., A. C. Nielsen

2005) have recognized the importance of culture in

influencing consumer behaviors, still, typically similar

constructs and relationships are investigated in Internet

shopping research regardless of culture. However, scholars

have begun to question whether this is appropriate; for

example, Gefen and Heart (2006) argued that trust could be

formed differently in different cultures. Research also

suggests that inherent differences in how people of different

cultures build trust could influence the relative effectiveness

of trust-building strategies in Internet shopping (Lim et al.

2004). Hence, there exists a gap in the literature: although

online trust has been extensively researched, building online

trust across cultures has not. Thus, our pioneering study

investigates the important yet still unanswered question:

Should similar trust building strategies be adopted in studying

the impact of trust building on Internet shopping regardless of

national cultures, or is a contingent approach customized to

specific cultures more effective?

Two web strategies, whose effectiveness in building initial

trust could be heavily influenced by culture, are reputable

website affiliation and customer endorsement. Reputable

website affiliation refers to a website without an established

reputation associating itself with another reputable third party

website, such as Yahoo, and is one of the most commonly

used web strategies to build online trust (Cheskin Research

and Studio Archetype/Sapient 1999; Stewart 2003). Repu-

table website affiliation is used in many new U.S.-based

Internet stores such as Buydig.com (Yahoo affiliation) and

millionbuy.com (Yahoo affiliation) and some Asian-based

web-stores such as Yesasia.com (Yahoo affiliation). Yet, in

a ranking of peers versus organizations in Hong Kong, local

peers (both recognizable and non-recognizable) are trusted

significantly more than reputable organizations (Yahoo.com

and Amazon.com) (Lim et al. 2006). Thus, it is not evident

whether a trust-building web strategy using a third-party web-

site would be effective in collectivistic cultures (e.g., in Asia),

which tend to have more favorable attitudes toward in-group

(e.g., local peers, family members, friends, etc), than out-

groups entities (e.g., foreigners, foreign organizations, etc.).

Another web strategy that is less commonly used by Internet

stores is customer endorsements (Cheskin Research and

Studio Archetype/Sapient 1999), despite being ubiquitously

used in brick-and-mortar firms (Bearden and Etzel 1982;

Blythe 1997). It involves using existing customers to provide

positive recommendations to potential ones (e.g.,

Chinaetravel.com, COL.com.hk). Still, it is not clear if a web

strategy based solely on positive feedback or comments

would work in an online environment, due to the tendency

among Internet users to be suspicious of such comments.

Further, in a collectivistic culture, local peers are trusted

significantly more than foreigners of the same age-group (Lim

et al. 2006). Thus, cultural differences could affect the way

people view such positive feedback when provided by

customers perceived to be socially close peers (in-group

members). These two web strategies are selected because

they are good exemplars of two of the three initial trust

building mechanisms (reputation categorization and unit

grouping) noted in McKnight et al. (1998). The third mecha-

nism of institutional assurance is controlled for in the

research. The research question investigated in this study is:

What is the impact of national culture on the effectiveness of

the two trust building web strategies of portal affiliation and

peer customer endorsements in building trust and enhancing

purchasing intentions and behaviors?

In this early attempt to study the impact of trust-building web

strategies across cultures, we focused on individualism–

collectivism since it is the most significant cultural dimension

(Triandis 2001). How people respond differently to in-groups

versus out-groups is a key factor differentiating individualists

from collectivists. The two strategies, portal (Yahoo) affilia-

tion and peer customer endorsement, could invoke an in-

group/out-group categorization among subjects based on

individualism–collectivism trait differences of people in two

different cultures (Australia and Hong Kong). Hence, this

permits us to investigate how in-group versus out-group

categorizations underlying different web strategies could

affect trust building in different cultures. Portal affiliation

492 MIS Quarterly Vol. 33 No. 3/September 2009

Sia et al./Web Strategies to Promote Internet Shopping

involves a U.S.-based portal (an entity external to the research

sites of Australia and Hong Kong), versus local peer customer

endorsements. This is one of the first attempts to investigate,

through a laboratory study, whether trust is built similarly in

different cultures through a uniform set of web strategies, or

built differently through specific web strategies, with cogni-

tive and behavioral consequences. It extends beyond studying

the macro-level relationship between culture and Internet

shopping (e.g., Choi and Geistfeld 2004; Park and Jun 2003)

by probing more deeply into the impact of culture on specific

practical trust-building web strategies that could evoke

different culturally based responses. Our results enable web-

site designers to make informed decisions about the cultural

impact of each specific trust building web strategy to achieve

global presence, an important Information System culture

research issue (Leidner and Kayworth 2006).

Theoretical Background

and Research Model

Trusting Beliefs and Intentions

Trust is defined as a consumer’s perception that an Internet

store can be relied upon to engage in generally acceptable

business practices and will deliver the promised products/

services, despite the possibility of exposure to loss during an

Internet shopping transaction (Mayer et al. 1995; McKnight

et al. 2002). This is a widely held view of trust in established

works on Internet shopping (e.g., Gefen et al. 2003; Jarvenpaa

et al. 2000). Trust can be delineated into trusting beliefs and

trusting intentions (Gefen et al. 2008; McKnight et al. 1998).

With trusting beliefs, a potential consumer perceives that the

Internet store can be counted on to adhere to good business

practices in the shopping transaction to deliver the promised

goods/services, even though doing so could involve risks

(Mayer et al. 1995; McKnight et al. 1998). With trusting

intentions, the potential consumer, having formed trusting

beliefs, is willing to be exposed to the risk of loss by

intending to make purchases at the Internet store (McKnight

et al. 1998). Trusting beliefs is the key dependent variable of

interest in this study. The next section presents the research

model, which describes the antecedents of interest and the

nomological net of trusting beliefs.

Research Model

The research model is shown in Figure 1. When interacting

with new or unknown Internet stores, people could engage in

several cognitive processes to build initial trust (Bigley and

Pearce 1998; McKnight et al. 2002; McKnight et al. 1998;

Stewart 2003; Zucker 1986). They include reputation cate-

gorization, unit grouping, and institutional assurances.

Positive reputation categorization occurs when the trustor

associates the trustee (e.g., an Internet store) with a brand of

good reputation (Stewart 2003). Unit grouping is the extent

that the trustee is perceived to share common characteristics

and identity with the trustor. Members of one’s unit group are

viewed more positively, and are considered more trustworthy

than nonmembers (Kramer et al. 1996; Zucker et al. 1996).

When the trustee is an inanimate object (e.g., a firm), trust can

still be formed using unit-grouping through a trust trans-

ference process involving an animate entity (Stewart 2003;

Swan and Nolan 1985) that shares commonalities with the

trustor. Institutional assurances refer to the regulations, guar-

antees by accreditation agencies (e.g., TRUSTe, VeriSign),

and legal recourses available. Thus, web strategies that

employ reputation categorization, unit grouping, or institu-

tional assurances should lead to increase in initial trust

(McKnight et al. 1998). We controlled for institutional

assurances in this study, because one needs an adequate level

of such assurances (e.g., privacy protection, TRUSTe) to be

willing to make online purchases (Cheskin Research and

Studio Archetype/Sapient 1999).

Reputable Website Affiliation Strategy

and Impact of Culture

Reputation-building mechanisms derived from reputation-

categorization cognitive processes could build trusting beliefs

(McKnight et al. 1998). In the absence of direct information

on a trustee’s reputation (e.g., a new online store), trust can be

built through reputable third parties such as Yahoo (Pavlou

and Gefen 2004; Stewart 2003). Specifically, according to

attribution theory (Kelley 1973), a trustor (e.g., potential

shopper) could transfer his/her trust in an associated reputable

entity (e.g., reputed physical store or Internet firm) to the

trustee (e.g., unknown Internet store) (Stewart 2003; Swan

and Nolan 1985).

Lim et al. (2004), conducting a multiple-country analysis,

found that Hofstede’s (2001) cultural dimension of indi-

vidualism–collectivism is a major predictor of Internet

shopping rates. Indeed, individualism–collectivism has been

considered to be the most significant dimension explaining

differences across cultures (Triandis 2001), and referred to as

the “deep structure” underlying differences across cultures

(Greenfield 2000). It is also the central theme in almost 100

studies published annually on cultural differences (Triandis

and Suh 2002).

MIS Quarterly Vol. 33 No. 3/September 2009 493

Sia et al./Web Strategies to Promote Internet Shopping

Figure 1. Research Model

Individualism–collectivism is a country-level measure of the

extent to which the self-concept of people in a country

revolves around that of the individual, or of a member of a

group (Clark 1990; Hofstede 1991). Table 1 summarizes the

characteristics and behaviors of individualists and collectivists

identified in past cultural research.

Portal affiliation involves the display of a reputable third-

party portal logo on the website of an Internet store as a

hyperlink to permit verification of the relationship. Being a

widely recognized and reputable portal carries a certain

amount of assurance about its quality. A reputable portal that

has been popularly used for portal affiliation to build trust is

Yahoo (Cheskin Research and Studio Archetype/Sapient

1999) in websites such as Buydig.com, Millionbuy.com,

Diamondsafe.com, Kidsmartliving.com, Yesasia.com, etc.

Despite being reputable, portals like Yahoo and MSN are

organizations that are likely to be viewed by people outside

the United States as U.S.-based portals, and thus as external,

out-group entities. Collectivists tend to have lesser trust and

positive opinions of out-group members. They also tend to

behave less favorably towards out-group entities, compared

to individualists. Thus, the overall impact of portal (e.g.,

Yahoo) affiliation on trust building is likely to be dampened

somewhat among collectivists. In contrast, individualists tend

to view both in-group and out-group members more equitably,

and are unlikely to view and treat out-group members less

favorably (Doney et al. 1998; Hui et al. 1991; Iyengar et al.

1999; Lee and Ward 1998; Triandis 1972; Triandis and Suh

2002). Thus, although portal (e.g., Yahoo) affiliation is

expected to be effective for building trusting beliefs in both

individualists and collectivists, given the differential treatment

that individualists and collectivists are likely to accord to out-

group entities, it is expected that

H1: The positive impact of portal (Yahoo) affiliation

on trusting beliefs will be stronger in individualistic

cultures than in collectivistic cultures.

Peer Customer Endorsement Strategy and

Impact of Culture

Peer customer endorsement to deliver endorsement messages

is a web strategy that invokes unit grouping, which appeals to

social similarity to build initial trust (McKnight et al. 1998;

Zucker 1986). People tend to exhibit more positive attitudes

and trust toward those who share the same characteristics as

they do (Doney et al. 1998; Kramer et al. 1996; Zucker et al.

1996). When members of peer groups, whom people deem to

be trustworthy, trust a certain entity, other members are likely

to transfer their trust to that entity (Stewart 2003; Swan and

Nolan 1985), particularly if these peers had interacted with

that entity in the past (e.g., as customers). This trust-building

process, through positive buying experiences, could also

apply to the Internet (Ba and Pavlou 2002).

Social psychologists (Prentice et al. 1994) have noted the

existence of specific (common-bond) in-groups and general-

ized (common-identity) in–groups. Specific in–groups are

those with personal ties to and relationships with the indi-

vidual, such as family members, relatives, and friends.

Generalized in-groups, are based on large group memberships

and social categories to which an individual has a symbolic

attachment, such as fellow university students or other local

Portal

Affiliation

Peer Customer

Endorsement

Trusting

Beliefs

Attitude

Intention to

Buy

Actual

Buying

Culture

(Individualist versus Collectivist)

H1

H2

Portal

Affiliation

Peer Customer

Endorsement

Trusting

Beliefs

Attitude

Intention to

Buy

Actual

Buying

Culture

(Individualist versus Collectivist)

H1

H2

494 MIS Quarterly Vol. 33 No. 3/September 2009

Sia et al./Web Strategies to Promote Internet Shopping

Table 1. Characteristics of Individualism and Collectivism

Individualism Collectivism Reference

The needs, values, and goals

of the individual take

precedence over the groups’

(individual focus).

The needs, values, and goals of the group

take precedence over the individual (group

focus).

Gudykunst 1997; Gudykunst et

al. 1992

Collectivists tend to emphasize inter-

dependence and sociability. They tend to

have more positive attitudes towards and

trust people from their in-group (family

members or peers) than those from their

out-group.

Doney et al. 1998; Lee and

Ward 1998

Collectivists are more strongly influenced by

the opinions of referent others, and respond

more positively to social proof arguments

that their peers have also acted similarly

(e.g., “your peers have complied with this

request”) than people from individualistic

cultures.

Cialdini et al. 1999

Individualists tend to view both

in-group and out-group

members more equitably, and

are unlikely to view and treat

out-group members less

favorably than collectivists.

Collectivists behave more favorably toward

in-group members.

Doney et al. 1998; Hui et al.

1991; Iyengar et al. 1999; Lee

and Ward 1998; Triandis 1972;

Triandis and Suh 2002

Collectivists have greater preference for

interacting with in-group members.

Gudykunst et al. 1992;

Oyserman et al. 2002; Wheeler

et al. 1989

Individualists are characterized

by autonomy, self-reliance, and

emotional distance from in-

groups. They emphasize

efficiency and directness.

Grimm et al. 1999; Triandis

2001

Individualists value consistency

and stability of attitudes more

than collectivists.

Iyengar et al. 1999; Triandis

1972; Triandis and Suh 2002

Individualists display greater

willingness to trust out-group

members than collectivists.

Yamagishi 1988; Yamagishi et

al. 1998

Collectivists tend to make group-based

decisions rather than individual decisions.

They tend to place greater importance on,

and react in accordance to, the opinions of

in-group members, such as their peers.

Triandis et al. 1985

At the individual level, people who exhibit predominantly individualistic characteristics are termed idiocentrics, whereas those with collectivistic

characteristics are called allocentrics (Triandis et al. 1985).

MIS Quarterly Vol. 33 No. 3/September 2009 495

Sia et al./Web Strategies to Promote Internet Shopping

youths. The advertising industry has relied heavily on gener-

alized in-group endorsements (Bearden and Etzel 1982;

Blythe 1997).

Peer customer endorsements involving generalized in-group

endorsements could help people, regardless of whether they

are individualists or collectivists, to build trust in Internet

stores, because they could provide valuable advice in the

absence of first-hand experiences. Nevertheless, people in

collectivistic cultures tend to respond even more positively to

endorsements by both specific and generalized in-group peers,

than those in individualistic cultures. This is because they

respond to social proof arguments (that their peers have also

acted similarly) to a greater extent than individualists

(Cialdini et al. 1999). In contrast, people in individualistic

cultures are relatively more autonomous in decision-making,

and are likely to be comparatively less responsive to peer

endorsements than collectivists. Thus,

H2: The positive impact of peer customer endorse-

ment on trusting beliefs will be stronger in collecti-

vistic cultures than in individualistic cultures.

Extending the Nomological Network: Trusting

Beliefs, Attitudes, and Behavior

Trusting beliefs in an Internet store could significantly affect

consumers’ attitudes toward and willingness to engage in

Internet shopping (Hoffman et al. 1999; Keen 1997; Ratna-

singham 1998; Taylor Nelson Sofres 2001). Research has

shown that for many Internet stores (e.g., bookstores, travel

agents, etc.), increased trust leads to a more positive attitude,

positive attitude leads to greater intention to purchase from

the Internet stores (Jarvenpaa et al. 2000), and greater inten-

tion leads to more actual buying behavior (Lim et al. 2006).

Furthermore, a direct causal link between behavioral beliefs

and behavioral intentions was found in several studies on

Internet shopping (Gefen et al. 2003; Koufaris 2002;

McKnight et al. 2002). Thus, while the main focus of this

research is on the impact of culture on trust-building, for

nomological completeness, we extended the research model

to also include the impacts of trusting beliefs on attitudes,

attitudes on buying intentions, trusting beliefs on buying

intentions, and intentions on actual buying behavior in both

cultures (see Figure 1).

Research Methodology

Identical experimental procedures, task, and web technologies

were used in each research site (large universities in Australia

and Hong Kong) to ensure comparability of results.

Experimental Task

The task asked subjects (students) to decide on whether or not

to purchase required course textbooks from an online book-

store, iBook, a realistic activity for them. Subjects were

offered incentives to purchase a textbook, yet faced a certain

amount of risk and uncertainty, to create a realistic online

shopping experience. The benefits included a discount of 5

percent off the university bookstore’s purchase price, the

convenience of classroom delivery, and a lucky draw with a

3 percent chance of winning. The risks included the unknown

reputation of iBook, the need to reveal some personal infor-

mation including, most importantly, one’s bank account

number, and the possibility that iBook may renege on its

promise to deliver the textbooks. The experiments were con-

ducted during the first week of the new semesters. The

research protocol required the administrator (a research asso-

ciate) to introduce iBook as a relatively new online bookstore.

Subjects were explicitly told that the university was not

affiliated with iBook and was not liable for its actions. After

browsing through the website, subjects had to individually

decide whether or not they would purchase the textbooks

(alternative ways to obtain the textbooks included buying

from the university or other bookstores, or buying used copies

from senior students).

Subjects

Subjects were second- and third-year undergraduate IS-major

business school students recruited from each research site.

The research sites comprised a large university in Australia

and another in Hong Kong. Subjects belonged to the age

group that formed a substantial portion of Internet shoppers

(i.e., 18 to 34 years old) (Scarborough Research 2008), and

were likely to exhibit the behaviors of actual Internet

shoppers. To encourage participation, all subjects were paid

the equivalent of one hour of a research student’s pay (about

U.S.$7) for every hour of their time. In all, 166 and 128

subjects were recruited from the Australian and Hong Kong

sites respectively.

Research Design

Hypothesis H1 is tested using Yahoo affiliation, and

Hypothesis H2 is tested using peer endorsement (generalized

in-group peers), both in the context of book buying at iBook

across Australia (individualistic culture) and Hong Kong

(collectivistic culture).

496 MIS Quarterly Vol. 33 No. 3/September 2009

Sia et al./Web Strategies to Promote Internet Shopping

The experiment was a 2 × 2 × 2 factorial design: culture

(individualism: Australia, versus collectivism: Hong Kong),

portal affiliation (Yahoo affiliation versus no Yahoo affilia-

tion) and peer customer endorsement (endorsement versus no

endorsement). Trust building across cultures was investigated

through the theoretical lens of the individualism–collectivism

dimension, which is among Hofstede’s (1991) five dimen-

sions of culture. This is because the practical web strategies

of Yahoo portal affiliation and peer customer endorsement are

associated with out-group versus in-group recommendations,

which could invoke people’s individualistic/collectivistic

tendencies. Australia and Hong Kong were chosen as they

differ widely on the individualism index (Australia: 90; Hong

Kong: 25; difference of 65) (Hoftsede 2001). Each peer

endorsement and portal affiliation treatment in Australia had

between 41 and 42 subjects, while each treatment in Hong

Kong had 32 subjects. Appendices A and B show screen

shots of the condition with both peer customer endorsement

and portal affiliation used in the Australian and Hong Kong

research sites, respectively. To provide a certain level of

institutional assurances (Cheskin Research and Studio Arche-

type/Sapient 1999), security seals of approval (TRUSTe) and

privacy policy statements were controlled by having all

website versions (at both sites) display the same TRUSTe

logo and privacy statement.

There were three independent variables. Individualism–

Collectivism is varied using Australia and Hong Kong sub-

jects. Peer Customer Endorsement is varied by having web-

for an actual purchase made if a subject completed the direct

debit authorization form, “0” otherwise). Items measuring

trusting beliefs, attitude, and buying intention were adapted

to the context of iBook from Jarvenpaa et al. (2000) and Lim

et al. (2006) to ensure content validity. Actual buying

behavior is an objective measure. All items were reflective

indicators of the respective underlying constructs, with the

exception of the trusting beliefs construct that was formative

in nature (i.e., the items form a latent construct and could vary

independently from one another within the construct) (Chin

and Gopal 1995; Cohen et al. 1990; Petter et al. 2007). This

is because the latter included measures that probed the

subjects’ beliefs about iBook concerning benevolence,

integrity, and ability, the three subdimensions of trusting

beliefs (Mayer et al. 1995). All items were measured using a

seven-point Likert scale.

Variables such as age and gender of subjects, Internet

experience, and Internet usage, which could potentially affect

Internet behaviors, were controlled through randomization;

and experience with Internet shopping, which could affect

trusting beliefs (Gefen 2000), was included in the research

model as a control variable.

Experimental Procedures

Each session, held in a computer laboratory, had between 20

and 30 subjects. The experimental procedure consisted of a

preexperimental briefing, the experiment proper, and a