This article was downloaded by: [128.59.226.

54] On: 09 December 2014, At: 16:18

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA

Organization Science
Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org

Trust Transfer on the World Wide Web

Katherine J. Stewart,

To cite this article:

Katherine J. Stewart, (2003) Trust Transfer on the World Wide Web. Organization Science 14(1):5-17. http://
dx.doi.org/10.1287/orsc.14.1.5.12810

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact permissions@informs.org.

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.

© 2003 INFORMS

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
 Trust Transfer on the World Wide Web
Katherine J. Stewart
University of Maryland, Robert H. Smith School of Business, College Park, Maryland 20742
kstewart@rhsmith.umd.edu
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

Abstract extent to which a target is likely to behave in a way that

The World Wide Web (WWW) has been touted as providing is “benevolent, competent, honest, [and] predictable in a
great opportunities for small businesses to compete and thrive. situation” (McKnight et al. 1998, p. 474). Based on the
Concerns about trust have been identified as a barrier to such theory of reasoned action (Fishbein and Ajzen 1975),
businesses’ success. This research explores how consumers’ McKnight et al. (1998) distinguish trusting beliefs from
initial trust judgments about organizations they encounter on trusting intention, defined as the extent to which an in-
the Web may be influenced by hypertext links from trusted dividual is willing to make himself vulnerable to a tar-
websites and associations with the more trust-inducing tradi- get’s actions. Research supports the proposition that posi-
tional retail channel. This paper develops and tests a cognitive tive beliefs regarding an action have a positive effect on
model of the trust transfer process, arguing that trust is trans-
intentions to perform that action (Ajzen 1988), and
ferred across hypertext links based on the perceived interaction
and similarity of the linked organizations, and that institution-
McKnight et al. (1998) propose a positive relationship
based trust is transferred from the traditional shopping channel between trusting beliefs and trusting intentions. Such a
to a Web-based organization based on evidence that the Web- relationship is logical in that the more one believes an-
based organization has a physical store. An experimental study other to be trustworthy, the more one is likely to be will-
shows that a hypertext link from one website to another in- ing to depend on that other, even when negative conse-
creased the extent to which the linked organizations were per- quences are possible. Nonetheless, it is important to
ceived to have a business relationship and be similar, and these distinguish between trusting beliefs and trusting inten-
perceptions had a positive influence on trusting beliefs regard- tions, because it is possible that one might hold trusting
ing the linked site. Associating with the physical shopping beliefs about another and still be unwilling to make one-
channel by showing a picture of a building on a website in- self vulnerable to the other’s actions. For example, there
creased the extent to which subjects reported intention to buy could be risks inherent in a situation that are outside the
from the site. The study provided empirical evidence that trust-
control of both the trustor and the target (Mayer et al.
ing beliefs regarding the website had a significant positive effect
on intention to buy from it. This paper discusses further devel-
1995).
opment of the trust transfer model based on the social percep- There may be many intended actions that represent
tion literature and explores implications for future research. trusting intentions. Previous work has found effects of
(Trust Transfer; Consumer Trust; Trust in Websites; Entitativity; Hypertext trusting beliefs on the likelihood of continuing a relation-
Links) ship with a target (Crosby et al. 1990), a trustor’s long-
term orientation toward future goals involving the target
(Ganesan 1994), and intent to make a purchase from the
target (Doney and Cannon 1997, Jarvenpaa et al. 2000),
all of which involve a trustor making himself vulnerable
Trust is an important factor in the decision when people to a target’s actions. The trusting intention of interest in
choose others with whom to interact (McKnight et al. this study is intention to buy from the target. Intending
1998, Zaheer et al. 1998); an organization must be trusted to buy from a target may represent a willingness to be
for it to enter a consideration set of potential exchange vulnerable in many ways: The product purchased could
partners (Doney and Cannon 1997). But how does an in- be faulty, and the target could refuse to replace or ex-
dividual, when first encountering an unknown other, de- change it; the product may require service in the future,
cide how much to trust? This paper examines trust trans- and the target could fail to provide such service in an
fer as a means of establishing initial trust in organizations adequate manner; or the target could collect sensitive per-
doing business on the WWW. The main goal of this study sonal information about the trustor during the transaction
is to develop and test a cognitive model of trust transfer. and then release that information to others (Hoffman et
Trusting beliefs are an individual’s beliefs about the al. 1999).

1047-7039/03/1401/0005/$05.00 ORGANIZATION SCIENCE, 䉷 2003 INFORMS

1526-5455 electronic ISSN Vol. 14, No. 1, January–February 2003, pp. 5–17
 KATHERINE J. STEWART Trust Transfer on the World Wide Web

HYPOTHESIS 1. Trusting beliefs about a target will have that “[a] third party offers to each person a definition of
a positive influence on intention to buy from that target. the other as trustworthy. Each accepts or rejects this def-
inition . . . largely on the basis of his trust for the third
Several factors may explain why high initial levels of
party’s judgment.”
trusting beliefs and intentions might be observed. These
Other studies argue that transfer may be made from a
include individuals’ disposition to trust (Rotter 1967), the
place (Henslin 1968) or an industry association (Milliman
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

existence of assurance mechanisms (Zucker 1986), se-

and Fugate 1988) to an individual. For example, Milliman
curity felt in the situation in which the trust target is en-
and Fugate (1988) found that a salesman could transfer
countered (Lewis and Weigart 1985), calculations re-
the burden of establishing trust from himself to a “proof
garding the incentives and penalties to the target of acting
source”—specifically that the salesman could co-opt a
in a trustworthy manner (Lewicki and Bunker 1996), and
prospect’s trust in an industry association. They explained
cognitive processes such as stereotyping and categoriza-
that the proof source offered verifiable evidence of the
tion (McKnight et al. 1998). Considering trust transfer as
salesperson’s claims and therefore led to a greater inten-
a cognitive process, transfer occurs when a person (the
tion to buy on the part of the client.
trustor) bases initial trust in an entity (a person, group, or
In addition to showing that transfer may occur from
organization referred to as the target) on trust in some
different kinds of sources, previous work suggests that
other related entity, or on a context other than the one in
transfer may work through different processes. It may oc-
which the target is encountered (e.g., a different place).
cur based on a communication process in which either
While the transfer of trust has been observed in prior
the target or a trusted third party exerts direct influence
studies, the mechanisms underlying transfer have not re-
on the trustor, or be based on a cognitive process in which
ceived a great deal of attention. The next section sum-
the mere knowledge of the relationship between the target
marizes previous studies of trust transfer, develops hy-
and another source of trust induces transfer. The present
potheses regarding the cognitive process underlying trust
research focuses on the cognitive process of trust transfer
transfer, and presents the research model in Figure 1. Sec-
and considers both the possibility of transfer from one
tion 3 discusses the context of the study. Section 4 de-
individual entity to another, and from a context to an in-
scribes the experiment used to test the model. Analyses
dividual entity. Here, context refers to the situation in
and results are discussed in §5. The paper concludes with
which a target is encountered, specifically the institutional
a discussion of contributions, limitations, and implica-
structures in the situation. Context may be represented by
tions for research.
a place such as a neighborhood or, as in this case, by a
channel such as the WWW.
Trust Transfer
Trust may be transferred from different kinds of sources. Trust Transfer Between Targets
Some studies focus on individuals (known targets) who Trust transfer from one entity to another relies on the
serve as the source of trust transfer to unknown targets unknown target being perceived as related to the source
(e.g., Uzzi 1996). Strub and Priest (1976, p. 408) found of the transferred trust. Campbell (1958) suggested that
that a marijuana user would decide to trust an unknown such perceptions are based on the similarity, proximity,
person if a third person, trusted by the marijuana user, and common fate of the entities. He introduced the term
vouched for the unknown party. Their explanation was “entitativity” to describe the degree to which a collection

Figure 1 Research Model

6 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

of individuals is perceived as forming a group. The con- (Wilder and Simon 1998). In this case, a perceiver may
cept of entitativity allows for the study of collections of construct a group with or without the participation of
individuals who vary along a continuum in the extent to members. Alternatively, in forming a group based on in-
which they are perceived as forming a cohesive unit, teraction among the members, perceptions are more
rather than forcing such collections to be categorized in closely bound to observation of members’ behavior
a dichotomous fashion as forming a group or not. Re- (Wilder and Simon 1998). Different types of interaction
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

search on entitativity helps to illuminate how perceptions can cause perceptions of entitativity, including standing
of relatedness may cause transfer to occur by examining in a line at a bank, working with others on a project, and
the manner in which perceived entitativity influences in- playing on a sports team (Lickel et al. 2000). Previous
formation processing. In processing information about in- transfer studies have included a number of different kinds
dividuals, perceivers tend to expect consistency (Hamilton of interactions ranging from sharing illegal drugs (Strub
and Sherman 1996). Perceivers form an impression and Priest 1976) to forming business partnerships (Uzzi
quickly (Hastie and Park 1986) and as more information 1996).
is gathered they seek to confirm that impression, causing HYPOTHESIS 2A. The greater the perceived interaction
new information to be interpreted in a way consistent with between an unknown target and a trusted target, the more
the already formed impression (Hamilton et al. 1989). In positive the initial trusting beliefs about the unknown tar-
contrast, findings on information processing relating to get will be.
groups have tended to indicate that perceivers expect less HYPOTHESIS 2B. The greater the perceived similarity
consistency (Hamilton and Sherman 1996). People do not between an unknown target and a trusted target, the more
form an impression quickly as information is received; positive the initial trusting beliefs about the unknown tar-
rather, when called upon to make a judgment about the get will be.
group, they form the impression based on memory
Supposing these hypotheses hold, how might targets
(Hamilton et al. 1985).
influence perceptions of interaction and similarity? Evi-
Hamilton and Sherman (1996) suggested differences in
dence of a tie between parties may be expected to increase
information processing strategies for individuals versus
the extent to which an observer perceives them to be in-
groups are due to differences in the expected entitativity
teracting with each other, and therefore having some de-
of individuals versus groups. They extended the idea of
gree of entitativity.1 Hamilton and Sherman (1996) spec-
entitativity from groups to individuals, suggesting that
ulate that having such a cue to entitativity may increase
entitativity implies “expectations of unity and consis- perceptions of other aspects of entitativity, including sim-
tency” (Hamilton and Sherman 1996, p. 346), which can ilarity. For example, once the existence of a group is per-
vary for both individuals and groups. By experimentally ceived based on interaction cues, individuals within it
varying subjects’ expected entitativity of individuals and may tend to be perceived as being more similar to one
groups, McConnell et al. (1997) found information pro- another. Lickel et al. (2000) found a positive correlation
cessing was performed following the impression forma- between perceived interaction and perceived similarity.
tion strategy for both individuals and groups high in en- Lickel et al. (2000) point out that any collection of
titativity. Impressions were formed based on memory for individuals, ranging from those with seemingly little re-
both individuals and groups low in entitativity. The im- lation to one another (such as people standing in a line at
pression formation information-processing strategy for a bank) to those with seemingly strong relations to one
groups implies that once an initial impression of one another (such as members of a family), may be perceived
group member is formed, other group members are per- as forming a group. Groups vary in the extent to which
ceived in terms of that impression, and information about they are perceived as high in entitativity. The nature of
them is processed in such a manner as to try to confirm the tie displayed, as an indicator of how much interaction
the impression (McConnell et al. 1997). Thus, I propose the group members have (e.g., little for people standing
that if one group member is believed to be trustworthy, in a line, a lot for members of a family), may be one
when a new group member is encountered, that group influence on how the group is perceived. Holding con-
member may be assumed to be trustworthy to the extent stant the nature of the tie, there is some evidence to in-
that the group is perceived to be high in entitativity. dicate that the number of others to whom a tie is displayed
Most definitions of a group suppose two or more en- may also influence the extent to which the tied parties are
tities interacting in some way (DeLamater 1974, Homans perceived as forming a group. In studying perceived en-
1950) and/or sharing some similarities (Wilder and titativity, Lickel et al. (2000) observed negative correla-
Simon 1998). When similarity is used as the criterion for tions between group size and all other indicators of per-
group formation, a group is akin to a social category ceived entitativity. If displaying a tie to one other

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 7

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

increases perceptions of interaction and similarity, then for high initial trusting intentions toward a target en-
what of displaying ties to multiple others? countered in that situation (McKnight et al. 1998). Fur-
Displaying ties to multiple others increases the size of ther, McKnight et al. (1998) propose that institutional
the implied group. Mullen (1991) and his colleagues have trust will effect trusting beliefs about a target because
analyzed evidence for the effects of group size on per- during initial encounters, beliefs about a target (i.e.,
ceptions of a group and its members in meta-analyses in trustworthiness) and beliefs about the situation in which
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

which they studied the emergence of group leaders, in- the target is encountered (i.e., the institutional factors in
group bias, social projection effects, and the heteroge- force) are yet to be differentiated, hence the tendency
neity effect (Mullen 1991). Based on results of these toward cognitive consistency may cause a positive re-
meta-analyses, Mullen (1991) argued that there is strong lationship between the two.
evidence that as the size of a group decreases, its salience The context of an encounter implies the set of institu-
to both members and observers increases. Increased sa- tional structures in place during the encounter. Previous
lience exaggerates evaluative perceptions in whatever di- work has demonstrated that different types of contextual
rection they initially tend (Taylor et al. 1978). If being factors may influence trust. For example, effects on gen-
perceived as a group causes a tendency for members to eral measures of trust in a society have been found to be
be perceived as more similar, and decreases in group size based on its religious makeup (La Porta et al. 1997) and
heighten that tendency, then members of smaller groups the communications infrastructure (Fisman and Khanna
should be perceived to be more similar to one another 1999) of the society. Zaheer et al. (1998) found a strong
than members of larger groups. correlation between trust in an organization and trust in
It has been argued that a target may be perceptually an individual within that organization. Henslin’s (1968)
grouped with others to which it displays a tie, increasing study of cab drivers’ trust in passengers demonstrated that
perceptions of interaction and similarity. The greater the the location involved in an encounter influenced trust in
number of others to whom the target displays a tie, the the target.
larger the implied group. Similarity is an evaluative per- To suggest that trust gets transferred from one context
ception of group members. Thus, the greater the number to another means that a trustor bases trust not only on the
of ties displayed, the less similar any one tied target may institutional structures in place in the context of the en-
be considered to be to the target displaying the tie. counter, but also on the structures in place in some other
context. To illustrate, suppose Context A has weak insti-
HYPOTHESIS 3A. An observer will perceive greater in- tutional structures and therefore generates low institution-
teraction between two targets when one of the targets based trust. Context B has strong structures, therefore
displays a tie to the other than when no tie is displayed. institution-based trust in Context B is high. A rational
assessment of potential outcomes will lead a trustor to be
HYPOTHESIS 3B. An observer will perceive greater sim- more willing to trust a given target in Context B than in
ilarity between two targets when one of the targets dis- Context A. Transfer occurs if the target is encountered in
plays a tie to the other than when no tie is displayed. A but manages to associate itself with B, such that trust
in the target in A is increased.
HYPOTHESIS 3C. The greater the number of others to Bacharach’s and Gambetta’s (2001) discussion of trust
whom a target displays a tie, the lower will be the per- signals provides a basis for understanding when and why
ceived similarity between the target and any one of those this might occur. They posit that a manifesta is an ob-
others. servable signal that a target possesses some krypta, which
is a trust-inducing property. For example, a target’s pres-
Trust Transfer from a Context ence in a certain location (a contextual factor of an en-
McKnight et al. (1998, p. 475) define institution-based counter) may be a manifesta that the target abides by the
trust by saying it “reflects the security one feels about a norms in force in that location, which is a property that
situation because of guarantees, safety nets, or other may increase trust. To induce the transfer of trust from
structures.” Such trust stems from the belief that imper- one context to another, a target may display a manifesta
sonal structures are in place to facilitate and encourage to indicate he abides by the norms in force in the trusted
trustworthy behavior in a given situation (Zucker 1986). context, B, although he is being encountered in A. Such
For example, enforcement mechanisms in place in a sit- a manifesta could signal that the target is subject to the
uation may lead a trustor to assess the likelihood of a institutional structures in Context B even though he is
target upholding commitments as higher than he would being encountered in Context A, where those structures
in the absence of those mechanisms, thereby allowing are not in place. This signal may trigger the trustor to

8 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

calculate the costs and benefits of interacting with the related to the context and thereby hinder trust develop-
target using the facts known about B in addition to the ment in this context (Liang and Huang 1998). I assume
facts known about A, thereby increasing trust. that consumers tend to have less trust in organizations
encountered on the WWW than in those encountered in
HYPOTHESIS 4A. If a target signals association with a the physical world, and Hypotheses 4a and 4b are tested
trusted context, although the target is not being encoun- by creating an association between a WWW vendor and
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

tered within that context, trusting beliefs regarding the the physical shopping context. By associating its website
target will be greater than if the target does not signal with a physical business, an organization may signal that
such an association. it is enmeshed in the more traditional economic and social
HYPOTHESIS 4B. If a target signals association with a systems with which consumers are familiar, although
trusted context, although the target is not being encoun- those institutions do not necessarily hold sway over in-
tered within that context, intention to buy from the target teraction on the WWW. If a WWW site is associated with
will be greater than if the target does not signal such an a physical business, the trust engendered by the tradi-
association. tional physical retail presence may be transferred to the
website.

The World Wide Web Context Method

Some (e.g., Keen 1997) have argued that a lack of con-
Hypotheses were tested in an experimental setting. Sub-
sumer trust in companies doing business on the Internet
jects were asked to imagine that they would be purchasing
acts as a significant barrier to realizing the opportunities
a laptop computer in the near future. The website of a
presented by the WWW. There are at least two potential
computing magazine was selected to act as a source of
reasons for such a trust barrier. First, consumers may fear
trust to be transferred. Based on pretests in which subjects
that companies on the Web do not respect the privacy of
indicated that they were familiar with the magazine and
personal information needed to complete transactions,
trusted it, this source site was expected to elicit a high
and that they will use that information for direct market-
level of knowledge-based trust.2 Pages from the maga-
ing or sell it to third parties (Hoffman et al. 1999, Tweney
zine’s website were copied and modified so that they con-
1988). Therefore, trusting beliefs regarding Web-based
tained general information about laptop computers and
companies may be low. As implied by the hypotheses set
zero links, one link, or nine links to fictitious vendor sites
forth above, organizations might attempt to overcome
created for the experiment (this manipulation is referred
barriers erected by a lack of trusting beliefs by associating
to below as links).
their websites with other trusted targets, encouraging con-
A fictitious vendor website was created to serve as the
sumers to transfer trust to them. The WWW allows or-
unknown target site. After browsing the material on the
ganizations to easily associate themselves with one an-
magazine site, subjects viewed one of two versions of this
other by sending hypertext links between their sites.
target site. The versions differed only in whether they
Hypertext linkages between websites A and B may cause
indicated that the company had a physical store by show-
the consumer to perceive a positive A-B relation because
ing a picture of a building and providing a street address
a link may provide a salient cue that can be used to infer
(this manipulation is referred to below as store).
that the two organizations interact. Links may represent
Data were collected using online surveys. Seven-point
business relationships among the linked organizations in
Likert scales were used to measure trusting beliefs re-
the form of advertising agreements or endorsements. In
garding the vendor website, perceived similarity of the
this study, hypertext links are the manner in which ties
websites, perceived business tie between the sites, and
are displayed and perceived interaction is operationalized
intention to buy from the target website. Where possible,
as the perception that two organizations are engaged in a
scales were adapted from previous research. Otherwise,
business relationship.
they were developed during pretesting. Scales are shown
A second factor underlying the trust barrier may be the
in the Appendix.
novelty of the channel, which may imply weaker struc-
tural assurances, rendering companies doing business Control Variables
through it less able to generate institution-based trust than Many factors may influence a consumer’s perception of
companies doing business using more traditional chan- the similarity or business relationship between two web-
nels. Information asymmetry between consumers and sites, trusting beliefs, or intention to buy from a website.
sellers (e.g., due to the inability of consumers to physi- These include factors related to the companies’ Web
cally inspect goods online) may also increase uncertainty pages as well as individual differences among consumers.

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 9

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

The first set of factors was experimentally controlled in taken to complete the study. During pretesting, the min-
this research. However, due to repeated mention during imum time to complete the study was 13.6 minutes (mean
pretesting, subject perceptions of some of these factors ⳱ 34.4, maximum ⳱ 69.7). However, the minimum
were measured and statistically controlled. Items were in- among the full sample of 221 was 7.4 minutes, indicating
cluded to measure perceived ease of use of the website, that some subjects may not have taken the study seri-
the extent to which the subject found the site to be graph- ously, and may not have read the items before answering.
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

ically pleasing (graphics), and the perceived technologi- To avoid the increased error variance introduced by such
cal sophistication of the target site. subjects, 13.6 minutes was used as a cutoff for inclusion
Age, income, and gender (Dholakia et al. 1995) may in the analyses.3
be related to familiarity with the channel, attitudes toward
shopping, and intention to make purchases. Experience Procedure
with computers and the Web may be important, as ex- Subjects accessed study materials through a Web page
perienced users may be more discerning in their evalua- that provided an explanation of the procedures. Instruc-
tions of websites, more comfortable using the Web to tions explained that subjects would be asked to engage in
make a purchase, or have stronger preexisting notions a shopping exercise related to laptop computers. This task
about the product category used in the study. Data on how involved first viewing general information about laptops
long the subject had been using the Web (months using to help participants determine their shopping criteria and
Web), frequency with which the subject used the Web then browsing a vendor site to select a product. Although
(Web use), frequency of computer use (computer use), the procedure appeared the same to all subjects, the pages
and months of full-time information technology (IT) work viewed varied depending on the experimental condition
experience were collected to control for effects related to to which they had been randomly assigned.
experience with the channel and product. After viewing instructions, subjects completed a survey
Measurements of subject propensity to trust companies measuring computer use, months using the Web, Web
encountered on the Web (propensity to trust websites) use, propensity to trust websites, and propensity to trust
and companies encountered in a physical retail-shopping stores. Subjects were then directed to one of the three
context (propensity to trust stores) were included to test versions of the source and asked to use information on
the assumption that the latter would, on average, be the site to help them determine the most important factors
higher than the former, and because the trust literature in purchasing a laptop. After inputting those factors, sub-
indicates that propensity may be an important factor de- jects completed a survey measuring trusting beliefs re-
termining trust in a target (McKnight et al. 1998). The garding the source website and control variables related
extent to which the subject enjoyed using the Web (Web to the source site.
affect) was measured because pretests indicated that it Subjects were then directed to follow a link to the un-
may influence intention to use the Web to make a pur- known target website. This was the first link on the source
chase. Finally, data on the subject’s previous knowledge site or, for those who viewed the source site without any
of and experience with the source website were collected links, it was a link in an instruction bar maintained (in all
(heard of source, visited source), both to confirm that the conditions) in a frame at the top of the browser. When
source was well known, and because such knowledge they followed this link, subjects viewed one of the two
may influence the interpretation of links placed on the versions of the target site. They were asked to browse the
website. site, pick a product that most closely matched the criteria
that they had established in the first part of the exercise,
Subjects and input the model and price of that product. After fin-
Two hundred twenty-one subjects were randomly as- ishing with the target site, subjects completed instruments
signed to six experimental groups created by crossing the measuring control variables related to the target, trusting
links and store factors. Subject demographics are shown beliefs, intention to buy, perceived similarity, and per-
in Table 1. Subjects were consumers offered a chance to ceived business tie.
win a $300 lottery as an incentive to participate. They A separate survey asked manipulation check questions.
were recruited from announcements made in courses at Subjects were asked to indicate whether or not “[target]
five U.S. universities, e-mail solicitations, and fliers sells their notebooks in their own retail store,” and
posted in the vicinity of one university. Approximately whether or not “there was a link between the [source]
70% reported “student” as their primary occupation. website and the [target] website.” Finally, subjects com-
Of the subjects, 187 were included in the analyses. Oth- pleted a survey to collect data on demographic control
ers were eliminated from the sample based on the time variables.

10 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

Table 1 Descriptive Statistics and Correlation Matrix1

Mini- Maxi-
Variable Mean S.D. mum mum 1 2 3 4 5 6 7 8 9 10 11

1. Age 25.818 8.385 18 79

2. Gender2 0.417 0.494 0 1 0.069
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

3. Income3 2.489 1.753 1 7 0.172* ⳮ0.098

4. IT Work Experience 7.032 22.731 0 216 0.208** ⳮ0.046 0.246**
5. Months Using Web 47.615 20.401 3 120 ⳮ0.106 ⳮ0.138 0.078 0.127
6. Link 1 0.310 0.464 0 1 0.118 0.019 0.097 ⳮ0.128 ⳮ0.079
7. Link 9 0.310 0.464 0 1 ⳮ0.089 ⳮ0.005 0.007 0.030 0.033 ⳮ0.450**
8. Perceived Similarity 12.086 3.620 4 21 ⳮ0.096 ⳮ0.002 ⳮ0.095 ⳮ0.147* ⳮ0.122 0.180* 0.035
9. Perceived Business Tie 11.925 3.545 3 21 ⳮ0.153* 0.058 ⳮ0.062 ⳮ0.143 ⳮ0.113 0.103 0.195** 0.456**
10. Store 0.476 0.501 0 1 ⳮ0.055 0.062 ⳮ0.022 ⳮ0.011 ⳮ0.021 ⳮ0.060 ⳮ0.083 ⳮ0.136 ⳮ0.026
11. Trust in Target 13.342 2.887 3 21 ⳮ0.144* ⳮ0.082 ⳮ0.161* 0.008 0.212** ⳮ0.092 0.093 0.369** 0.303** 0.050
12. Intention to Buy 14.722 5.688 4 27 0.017 0.093 ⳮ0.128 ⳮ0.170* 0.212** ⳮ0.020 0.114 0.379** 0.239** 0.156* 0.405**

1
Some control variables are not included in the table due to space constraints. The complete table is available from the author upon request.
2
Male ⳱ 0; female ⳱ 1.
3
Subjects indicated income categories, 1 ⬍$20,000, 2 is $20,001 to 40,000, . . . 7 ⬎$120,000.
*p ⬍ 0.05; **p ⬍ 0.01.

Analyses and Results perceived similarity and perceived business tie were en-
Analyses indicate that answers to manipulation check tered as the dependent variables, the manipulations (links
questions were influenced as expected by the condition and store) were entered as factors, and control variables
the subject was assigned to (X 21 ⳱ 37.20, p ⬍ 0.001 for were entered as covariates. One-way orthogonal planned
links; X21 ⳱ 6.90, p ⬍ 0.01 for store). An assumption was contrasts were conducted in which the estimated mean
that, on average, subject trust in the source website would values of perceived similarity and perceived business tie
be higher than trust in the target website. A one-tailed were compared between the zero link and nonzero link
paired samples t-test comparing the overall means con- conditions (contrast coefficients: 2,–1,–1) and between
firmed the validity of this assumption (t186 ⳱ 12.10, p ⬍ the one link and nine link conditions (contrast coeffi-
0.001). A second assumption was that subject propensity cients: 0, 1, –1). The MANCOVA yielded a significant
to trust websites would, on average, be lower than pro- multivariate effect for links (F4,334 ⳱ 6.112, p ⬍ 0.001).
pensity to trust stores. This assumption was tested and Univariate effects on perceived similarity (F2,167 ⳱
supported by a one-tailed paired samples t-test (t186 ⳱ 5.818, p ⬍ 0.01) and perceived business tie (F2,167 ⳱
2.16, p ⬍ 0.05). 8.509, p ⬍ 0.001) were also significant. Results of mul-
Factor analyses were performed and Cronbach’s alpha tivariate and, where the multivariate effect is significant,
calculated to assess the validity and reliability of the Lik- univariate tests for all variables in the model are displayed
ert scale measures (see the Appendix). To maintain a suf- in Table 2.
ficient observation/item ratio, the items were split into The first planned contrast provided support for Hy-
two groups subject to the criteria of keeping all items potheses 3a and 3b by showing that perceived similarity
from a single scale in the same analysis and keeping (difference ⳱ ⳮ2.752, p ⬍ 0.01) and perceived business
scales representing similar constructs in the same analy- tie (difference ⳱ ⳮ4.185, p ⬍ 0.001) were both higher
sis.4 Factors with an eigenvalue greater than one were in the link conditions than in the zero link condition. The
extracted and subjected to an orthogonal (varimax) rota- second planned contrast provided support for Hypothesis
tion. Items generally loaded highly on their factors with 3c by showing that perceived similarity was significantly
low crossloadings. For scales used in hypothesis testing, lower in the nine-link condition than in the one-link con-
the minimum reliability was 0.70. Descriptive statistics dition (difference ⳱ 1.271, p ⬍ 0.05). Results are dis-
and correlations are displayed in Table 1. played in Table 3. There was no significant difference
between the means of perceived business tie in the one-
Hypothesis Tests link versus nine-link conditions.
Number of Links, Perceived Similarity, and Perceived Perceptions, Trusting Beliefs, Store, and Intention to
Business Tie. Hypotheses 3a, 3b, and 3c were tested using Buy. Hypotheses 1, 2a, 2b, 4a, and 4b were tested using
a multiple analysis of covariance (MANCOVA) in which path analysis (Asher 1976). Though hypotheses regarding

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 11

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

Table 2 MANCOVA Predicting Perceived Similarity (P. Sim) Table 3 Estimated Marginal Means of Perceived Similarity and
and Perceived Business Tie (P. Tie) (N ⳱ 185) Perceived Business Tie

Univariate F Estimated Lower Upper

Variable Links Mean Bound* Bound
Source Multivariate F df P. Sim/P. Tie df
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

Perceived Similarity 0 11.1 10.3 11.9

Links 6.112*** 4 5.818**/8.509*** 2/2 1 13.1 12.3 14.0
Store 2.027 2 9 11.9 11.0 12.7
Links*Store 0.871 4 Perceived Business Tie 0 10.6 9.8 11.4
Age 3.052* 2 4.155*/4.230* 1/1 1 12.5 11.6 13.4
Gender 1.285 2 9 12.9 12.0 13.8
Income 1.023 2
IT Work Experience 0.385 2 *95% Confidence Interval.
Months Using Web 1.140 2
Web Use 2.362 2
Computer Use 2.262 2 perceived business tie to trusting beliefs. Hypothesis 4a
Heard of Source 1.858 2 was not supported. Hypothesis 4b was supported, as in-
Visited Source 0.744 2 dicated by the significant positive coefficient on the path
Difference in Ease 5.168** 2 9.385**/0.034 1/1 from store to intention to buy. In support of Hypothesis
Difference in Graphics 4.644* 2 5.489**/7.206** 1/1 1, the model shows that trusting beliefs had a significant
Web Affect 0.739 2 positive influence on intention to buy.
R 2 (adjusted R 2) P. Sim 0.288*** (.215)
The fit of the path analysis model was tested by esti-
P. Tie 0.222*** (.142)
mating omitted paths. The model was re-estimated, in-
*p ⬍ 0.05; **p ⬍ 0.01; ***p ⬍ 0.001. cluding a logical set of omitted paths based on the theory
and subject to the constraint of maintaining recursiveness.
This re-estimation, displayed in Model 2 of Tables 4 and
5, did not result in the inclusion of any new paths.
the influence of links have been tested above, to fully
represent the research model two variables were included
to model the effect of the links manipulation, Link1 (set Discussion
to 1 if the subject was in the one-link condition and 0 The main goal of this paper was to develop and test a
otherwise) and Link9 (set to 1 if the subject was in the cognitive explanation of trust transfer. Results indicated
nine-links condition and 0 otherwise). The path diagram that observing a link between two organizations’ web-
is shown in Figure 2. To present an uncluttered picture sites led subjects to perceive interaction and similarity
of the model, neither control variables nor error terms are between them, and that those perceptions had a positive
included in this figure. Effects of control variables are impact on the perceived trustworthiness of the unknown
shown in Tables 4 and 5, which (in Model 1 of each table) site. Thus, the notion that trust transfer from one target
display the results of the analyses underlying the path to another occurs based on perceived entitativity of the
diagram. targets was supported. I suggested that increasing the
Hypotheses 2a and 2b were supported by the significant number of ties displayed by a target would have a neg-
positive path coefficients from perceived similarity and ative influence on perceived similarity because of the

Figure 2 Path Diagram of Research Model

12 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

Table 4 OLS Regressions Predicting Intention to Buy Table 5 OLS Regressions Predicting Trust in Target
(N ⳱ 185) (N ⳱ 185)

Variable 1 2 Variable 1 2

Age 0.132* 0.152* Age ⳮ0.016 0.004

Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

Gender 0.042 0.051 Gender ⳮ0.057 ⳮ0.049

Income ⳮ0.016 ⳮ0.022 Income ⳮ0.115 ⳮ0.099
IT Work Experience ⳮ0.162* ⳮ0.153* IT Work Experience 0.065 0.042
Months Using Web ⳮ0.055 ⳮ0.025 Months Using Web 0.236*** 0.235***
Web Use 0.254** 0.251** Web Use 0.190* 0.187*
Computer Use ⳮ0.294*** ⳮ0.290** Computer Use ⳮ0.158 ⳮ0.144
Ease of Use—Target 0.221** 0.196** Ease of Use—Target ⳮ0.054 ⳮ0.076
Graphics—Target 0.004 ⳮ0.002 Graphics—Target 0.070 0.085
Technological Sophistication—Target 0.241** 0.184 Technological Sophistication—Target 0.226* 0.203*
Web Affect 0.110 0.110 Web Affect 0.039 0.018
Propensity to Trust Websites ⳮ0.074 ⳮ0.049 Propensity to Trust Websites 0.226*** 0.251***

Trust in Target 0.271*** 0.218** Perceived Similarity 0.228** 0.267***

Store 0.120* 0.148* Perceived Tie 0.177** 0.168*
Store 0.053 0.055
Perceived Similarity 0.111
Perceived Business Tie 0.032 Link 1 ⳮ0.112
Link 1 0.016 Link 9 0.041
Link 9 0.065
R2 0.412 0.430
R2 0.425 0.439 Adjusted R 2 0.363 0.372
Adjusted R 2 0.378 0.378 F 8.502*** 7.412***
F 8.986*** 7.212***
*p ⬍ 0.05; **p ⬍ 0.01; ***p ⬍ 0.001.
*p ⬍ 0.05; **p ⬍ 0.01; ***p ⬍ 0.001.

the positive influence of trusting beliefs on trusting inten-

tendency for smaller group sizes to increase salience,
tions.
which in turn exaggerates evaluative perceptions. The
finding that the multiple-links condition was associated Theoretical Implications and Future Research
with lower perceived similarity than the one-link con- Directions
dition supported this proposition.
Developing a Theory of Trust Transfer. This study fo-
Regarding trust transfer from a context, results showed
cused on perceived interaction and similarity as compo-
that the store manipulation had a significant positive ef-
nents of entitativity and contributes to theory by dem-
fect on intention to buy from the target, but no effect on
onstrating that these perceptions play a causal role in the
trusting beliefs, indicating that, at least in this instance,
trust transfer process. Lickel et al. (2000) propose other
institutional factors were more important to trusting in- factors that may be relevant to entitativity perceptions,
tentions than to trusting beliefs. This pattern of results and therefore to a theoretical model of trust transfer.
seems to indicate that associating with a trusted context These include contextual factors such as the amount of
enhanced trust in a target by making salient the assur- competition between groups, perceiver differences such
ances and safeguards that help to ensure trustworthy be- as the need for closure, and differences in properties of
havior, and thereby increasing willingness to be vulner- the group such as permeability and duration. Incorporat-
able to the target, rather than acting through cognitive ing these factors into a theory of trust transfer may help
consistency to enhance beliefs about the target’s benev- to further explain how initial trust is developed. For ex-
olence, honesty, competence, and integrity. Finally, this ample, qualitatively different kinds of ties may imply dif-
research focused on the intention to buy as a kind of trust- ferent group properties. Here, a relatively neutral tie ma-
ing intention that may be of interest in many different nipulation was used. Less neutral ties such as advertising
business settings, and provided one of the first tests sup- and cobranding links are ubiquitous. Given that such links
porting McKnight et al.’s (1998) proposition regarding may represent a financial relationship between the linker

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 13

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

and the linkee, the extent to which they are perceived as 2001). For example, instead of merely indicating that a
implying a business relationship might be higher than Web vendor is associated with a physical location, the
other kinds of links (e.g., “cool sites”). vendor might indicate that the online store follows the
The direction of the tie between source and target may same policies as the physical store, or that goods pur-
be another important factor in how their relationship is chased online may be returned to the physical store. It is
perceived. In this study, subjects viewed a trusted website relatively inexpensive to put a picture of a building on a
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

and then an unknown website, with the trusted site pro- website—this signal may easily be displayed whether
viding the connection. While the theory would suggest there is any relationship between the site and the building
that ties displayed by the unknown site should have simi- or not. A more expensive signal, such as a video tour of
lar effects on perceived similarity and interaction as those the store or a live Webcam broadcast from the store could
displayed by the known site, this suggestion is untested be even more effective. A less expensive or less notice-
here. Previous work on entitativity has generally focused able signal, such as a statement that there is a store un-
on groups in which targets may be observed together, but accompanied by a photo, could fail to produce any effect
the choices behind their associations are not displayed. In at all.
this study, the trusted target clearly chose to associate The nature of trust may also be important to under-
with the unknown target. Had the choice been in the op- standing trust transfer. An issue not considered in this
posite direction (i.e., the unknown target chose to asso- study is that trust in the source may have different bases;
ciate with the known target), effects may have differed. e.g., Lewicki and Bunker (1996) suggest trust may rest
Assuming that the target providing the link is encountered on calculus, knowledge, or identification. The extent to
before the linked target, the information-processing ex- which trust can be transferred from one target to another
planation implies that evaluations of low trustworthiness may depend on how that trust has developed. For exam-
could be propagated in the same fashion as evaluations ple, trust based on calculus may transfer from one target
of high trustworthiness. Thus, judgments about the un- to another less readily than other types of trust because it
known target may be transferred to the known target. Al- depends on a rational assessment of the benefits and costs
ternatively, previous literature on trust transfer would ar- to the target and those benefits and costs may not nec-
gue that transfer occurs from better known to less essarily be seen as applicable to associates.
well-known targets, implying that the direction of the tie
may not matter, and perhaps an unknown target could Social Networks and Alliances. Research on the influ-
increase perceived trustworthiness by sending a link to a ences of ties in the social network literature has not often
trusted target. focused on cognitive effects. Exceptions are Krackhardt
These possibilities pose interesting questions for future (1990), who showed that people’s perceptions of others’
research: Can trust transfer be reversed so that the per- ties vary, and Kilduff and Krackhardt (1994), who pro-
ception that a known and an unknown target are related vided evidence that perceptions of ties are more important
serves to decrease trust in the known target? Results of a than actual ties to evaluations of others. Results of this
study examining this question could help define the study add to the literature by supporting the importance
boundaries of the model proposed herein by shedding of perceptions, and also by providing some logic for how
light on the dynamics of the transfer of distrust, as well the number of ties observed may influence those percep-
as trust. Alternatively, if the direction of the tie does not tions, and how perceptions may affect evaluations of the
matter, can unknown organizations increase perceived tied parties through their influence on information pro-
trustworthiness by sending ties to more trusted organi- cessing.
zations? The technology of the WWW allows links to be Gulati (1998) points out that the alliance literature, gen-
sent with or without the consent of the linkee, and leg- erally concerned with the formation, governance, and per-
islation regulating such practices is problematic (Dawson formance implications of firms’ alliances, has often fo-
2000, Sableman 2000). cused on dyadic exchanges, and he argues that a network
As the nature of a tie may influence the transfer process perspective may expand researchers’ understanding of al-
from one target to another, so may the nature of a con- liances. In keeping with much of the network literature,
textual association influence the extent to which trust can Gulati (1995) used a communication argument to explain
be transferred from a context to a target in a different the influence of network ties on alliance formation pat-
context. I suggested that a signal associating a target with terns. Results of this study suggest that there may also be
a more trusted context may influence trust by indicating a cognitive explanation for network effects. In addition
that the target is subject to the structural assurances in the to the influence that ties will have on what information is
trusted context. The strength and cost of such a signal communicated to and between potential partners, this
may influence its effectiveness (Bacharach and Gambetta study implies that observation of other firms’ patterns of

14 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

ties may influence choices in forming alliances. Such in- sis. While the data seem to support the entitativity expla-
fluences could be particularly important when new firms nation without providing evidence for an economic ef-
are involved, given that such firms may not have many fect, nonetheless, it is not possible to rule out the
preestablished communication channels to others, but existence of such an effect.
they may be able to observe others’ ties. The transfer
Product. The tasks that subjects performed during the
model may be useful in addressing the question of how
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

experiment focused on buying a laptop computer. The

new firms choose alliance partners.
choice of this product imposes some limitations on the
Practical Implications generalizability of the results. There are several dimen-
While there has been much verbiage in the media about sions of the product (e.g., price, importance of branding,
the importance of trust to Internet retailers, there has not complexity) that might influence trust. Using two targets
been much empirical work. This study contributes to that were both naturally associated with the product cate-
practice by providing evidence for the relationship be- gory may have served to increase their perceived entita-
tween trusting beliefs and intention to buy, and by point- tivity. Industry association could represent what Lickel et
ing out some of the potential implications of the network al. (2000) called a contextual factor, which has the po-
of links created on the Web. Results of this research could tential to moderate the influence of links on perceptions.
help website owners to look beyond metrics like click- Without this contextual factor implying a base level of
through rates to understand how links on their pages may entitativity, effects of links could be weaker. For exam-
influence consumers’ perceptions about their relation- ple, if the source had been a gardening magazine Web
ships with others, trusting beliefs regarding the organi- site rather than a personal computing magazine site, a link
zation, and intention to buy from it. to a laptop vendor may not have been perceived as rep-
There has been controversy over the legal implications resenting as high a level of interaction and similarity.
of linking on the WWW. As Dawson (2000) puts it,
“Long-time Web initiates take as bedrock the right to link
to any URL they please. Lawyers assume no such Conclusion
right. . . . No U.S. court has ruled decisively on the many This research has provided a starting point for studying
issues. . . .” This controversy is concerned with issues the cognitive process underlying trust transfer by explain-
such as trademark infringement. These issues are directly ing how perceptions underlie that process and by provid-
related to how consumers perceive links. Sableman ing an empirical test of the process. The model proposed
(2000) explains that there may be a case for infringement herein expands upon McKnight et al.’s (1998) work by
when “confused consumers think there is a connection adding trust transfer to the set of cognitive processes that
with the trademark owner . . . or a negative connotation they suggest to be influential in initial trust development.
is added to the mark.” This study may help inform the The results of this study help to support their more gen-
controversy over links by providing empirical evidence eral theoretical model by providing empirical evidence of
of at least one way in which links do influence consum- the importance of cognitive processes as well as the effect
ers’ perceptions of websites. of trusting beliefs on trusting intentions.

Limitations Acknowledgments
There is a potential economics-based explanation for the This paper is based on the author’s dissertation. The author is indebted to
effect of displaying a tie on trust in the target. If a trustor Sirkka Jarvenpaa; Hank Lucas; Jeffrey Polzer; Prabhudev Konana; Marc-
believes it is costly to the source to display a link to an David Seidel; James Westphal; Organization Science guest editors Bill
McEvily, Vincenzo Perrone, and Akbar Zaheer; participants in the June
untrustworthy target, then trust in that target may be in-
2001 Organization Science conference on Trust in an Organizational Con-
creased by the display of a link regardless of the percep-
text; and anonymous Organization Science reviewers for their helpful com-
tion of entitativity among the targets. If this were the case ments and advice. This study was funded by the University of Texas Center
here, one might expect the links manipulation to have had for Customer Insight and the Dora and Eugene Bonham Fund.
an effect on trust that was not explained by the entitativity
measures. As can be seen in Model 2 of Tables 4 and 5, Appendix. Items, Factor Loadings, and Scale
no such effect was present in the data. Further, when sub- Reliabilities
jects were asked to write any comments they had about Factor
the source, several volunteered that their trust in the Items ␣ Loading
source was based on their familiarity with it (e.g., “[the Perceived Similarity 0.79
magazine] has always been a reliable source for me”), [source] would probably give a high rating to 0.747
while none indicated that their trust had an economic ba- the [target] website.

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 15

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

The [target] website has a lot in common with 0.863 Endnotes

the [source] website. 1
The term “displaying a tie” will be used to refer to an indication by a
The [source] and [target] websites are similar. 0.826 target that it has a positive relationship with another target. This study
Perceived Business Tie 0.70 focuses on hypertext links on Web pages as a means of displaying ties.
[source] is not connected to [target]. (r) 0.802 In all cases, the use of the plural, ties, is meant to imply ties to multiple
[source] is not likely to recommend [target] to 0.559 others rather than multiple ties to a single other.
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

2
an individual who wants to buy a notebook Lewicki and Bunker (1996) suggest that knowledge-based trust rests on
computer. (r) predictability, which requires information about the target and may come
[source] and [target] have a business 0.866 from repeated interactions with the target. Data indicated that subjects
relationship with one another. held this kind of trust in the source in that 90.4% reported having at least
heard of the company before, and several subjects made statements such
Propensity to Trust Stores1 0.69
as “[The magazine] has always been a reliable source for me” and “[The
Most retail stores are run competently. 0.649
magazine] is a well-known computer information source.”
In most retail stores, you will get honest replies 0.768 3
This cutoff was changed to 12 and 15 minutes without significant
to your questions and concerns.
differences in the pattern of results discussed below.
In most retail stores, you can get an honest 0.768 4
Several different analyses were run regrouping the scales, and the
description of the store’s business and
pattern of loadings was the same.
motives.
Propensity to Trust Websites1 0.67 References
Most websites are run competently. 0.793 Ajzen, I. 1988. Attitudes, Personality, and Behavior. Dorsey Press,
On most websites, you will get honest replies to 0.699 Chicago, IL.
your questions and concerns. Asher, H. B. 1976. Causal Modeling. Sage Publications, Beverly Hills,
On most websites, you can get an honest 0.613 CA.
description of the site’s business and motives. Bacharach, M., D. Gambetta. 2001. Trust in signs. K. Cook, ed. Trust
in Society. Sage Publications, New York, 148–184.
Web Affect 0.72
Campbell, D. T. 1958. Common fate, similarity, and other indices of
I like using the World Wide Web. 0.830
My experiences using the World Wide Web 0.634 the status of aggregates of persons as social entities. Behavioral
have generally been positive. Sci. 3 14–25.
I do not enjoy using the World Wide Web. (r) 0.792 Crosby, L. A., K. R. Evans, D. Cowles. 1990. Relationship quality in
services selling: An interpersonal influence perspective. J. Mar-
2
Intention to Buy 0.84 keting 54 68–81.
I probably would not buy from the [target] 0.851 Dawson, K. 2000. An inalienable right to link? 具www.digitalmass.com/
website. (r) columns/internet/0412.html典.
It is likely I would consider purchasing from the 0.894 DeLamater, J. 1974. A definition of “group.” Small Group Behavior 5
[target] website. 30–44.
It is unlikely I would return to the [target] 0.727 Dholakia, R. R., B. Pederson, N. Hikmet. 1995. Married males and
website before making a purchase decision. shopping: Are they sleeping partners? Internat. J. Retail and Dis-
(r) tribution Management 23 27–33.
It is likely I would use the [target] website to 0.776
Doney, P. M., P. Cannon. 1997. An examination of the nature of trust
make a purchase.
in buyer-seller relationships. J. Marketing 61 35–51.
Trust in Target3 0.77 Fishbein, M., I. Ajzen. 1975. Belief, Attitude, Intention, and Behavior:
The [target] website can be relied upon to be 0.781 An Introduction to Theory and Research. Addison-Wesley, Read-
sincere. ing, MA.
Promises and commitments made on the [target] 0.715 Fisman, R., T. Khanna. 1999. Is trust a historical residue? Information
website are not likely to be kept. (r) flows and trust levels. J. Econom. Behavior and Organ. 38 79–
The [target] website is trustworthy. 0.865 92.
Trust in Source 3
0.66 Ganesan, S. 1994. Determinants of long-term orientation in buyer-
Promises and commitments made on the 0.428 seller relationships. J. Marketing 58 1–19.
[source] website are not likely to be kept. (r) Gulati, R. 1995. Social structure and alliance formation patterns. Ad-
The [source] website can be relied upon to be 0.904 min. Sci. Quart. 40 619–652.
sincere. ——. 1998. Alliances and networks. Strategic Management J. 19 293–
The [source] website is trustworthy. 0.896 317.
Hamilton, D., S. Sherman. 1996. Perceiving persons and groups. Psych.
(r) ⳱ reverse coded. Rev. 103 336–355.
1 ——, D. M. Driscoll, L. T. Worth. 1989. Cognitive organization of
Based on Mayer and Davis (1999)
2
Jarvenpaa et al. (2000) impressions: Effects of incongruency in complex representations.
3
Doney and Cannon (1997) J. Personality and Soc. Psych. 57 925–939.

16 ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003

 KATHERINE J. STEWART Trust Transfer on the World Wide Web

——, P. M. Dugan, T. K. Troiler. 1985. The formation of stereotypic ——, ——, F. D. Schoorman. 1995. An integrative model of organi-
beliefs: Further evidence for distinctiveness-based illusory cor- zational trust. Acad. Management Rev. 20 709–734.
relations. J. Personality and Soc. Psych. 48 5–17. McConnell, A. R., S. J. Sherman, D. L. Hamilton. 1997. Target enti-
Hastie, R., B. Park. 1986. The relationship between memory and judg- tativity: Implications for information processing about individual
ment depends on whether the judgment is memory-based or on- and group targets. J. Personality and Soc. Psych. 72 750–762.
line. Psych. Rev. 93 258–268. McKnight, D. H., L. L. Cummings, N. L. Chervany. 1998. Initial trust
Downloaded from informs.org by [128.59.226.54] on 09 December 2014, at 16:18 . For personal use only, all rights reserved.

Henslin, J. M. 1968. Trust and the cab driver. M. Truzzi, ed. Sociology formation in new organizational relationships. Acad. Management
and Everyday Life. Prentice Hall, Englewood Cliffs, NJ, 138–158. Rev. 23 473–490.
Hoffman, D. L., T. P. Novak, M. Peralta. 1999. Building consumer Milliman, R. E., D. L. Fugate. 1988. Using trust-transference as a per-
trust online. Comm. ACM 42 80–85. suasion technique: An empirical field investigation. J. Personal
Homans, G. C. 1950. The Human Group. Harcourt, Brace, New York. Selling and Sales Management 8 1–7.
Jarvenpaa, S. L., N. Tractinsky, M. Vitale. 2000. Consumer trust in an Mullen, B. 1991. Group composition, salience, and cognitive represen-
Internet store. Inform. Tech. and Management 1 45–71. tations: The phenomenology of being in a group. J. Experiment.
Keen, P. 1997. Are you ready for the trust economy? Computerworld Soc. Psych. 27 297–323.
80 (April 21). Rotter, J.1967. A new scale for the measurement of interpersonal trust.
Kilduff, M., D. Krackhardt. 1994. Bringing the individual back in: A J. Personality 651–665.
structural analysis of the internal market for reputation in orga- Sableman, M. 2000. Link law: The emerging law of Internet hyper-
links. 具www.ldrc.com/cyber2.html典.
nizations. Acad. Management J. 37 87–108.
Strub, P. J., T. B. Priest. 1976. Two patterns of establishing trust: The
Krackhardt, D. 1990. Assessing the political landscape: Structure, cog-
marijuana user. Sociological Focus 399–411.
nition, and power in organizations. Admin. Sci. Quart. 35 342–
Taylor, S. E., S. T. Fiske, N. L. Etcoff, A. J. Ruderman. 1978. Cate-
369.
gorical and contextual bases of person memory and stereotyping.
La Porta, R., F. Lopez-de-Salinas, A. Shleifer, W. Vishny. 1997. Trust
J. Personality and Soc. Psych. 36 778–793.
in large organizations. Amer. Econom. Rev. Papers and Proc. 87
Tweney, D. 1998. Lack of trust hurts consumer commerce for online
333–338.
retailers. Infoworld 20 77.
Lewicki, R. J., B. B. Bunker. 1996. Developing and maintaining trust
Uzzi, B. 1996. The sources and consequences of embeddedness for the
in work relationships. R. M. Kramer and T. Tyler, eds. Trust in economic performance of organizations: The network effect.
Organizations: Frontiers of Theory and Research. Sage Publi- Amer. Sociological Rev. 61 674–698.
cations, Thousand Oaks, CA, 114–139. Wilder, D., A. F. Simon. 1998. Categorical and dynamic groups: Im-
Lewis, D. J., A. Weigart. 1985. Trust as social reality. Soc. Forces 63 plications for social perception and intergroup behavior. C. Se-
967–985. dikides, J. Schopler, and C. A. Insko, eds. Intergroup Cognition
Liang, T., J. Huang. 1998. An empirical study on consumer acceptance and Intergroup Behavior. Lawrence Erlbaum Associates, Mah-
of products in electronic markets: A transaction cost model. De- wah, NJ, 27–44.
cision Support Systems 24 29–43. Zaheer, A., B. McEvily, V. Perrone. 1998. Does trust matter? Exploring
Lickel, B., D. L. Hamilton, G. Wieczorkowska, A. Lewis, S. J. Sher- the effects of interorganizational and interpersonal trust on per-
man, A. N. Uhles. 2000. Varieties of groups and the perception formance. Organ. Sci. 9 141–159.
of group entitativity. J. Personality and Soc. Psych. 78 223–246. Zucker, L. G. 1986. Production of trust: Institutional sources of eco-
Mayer, R. C., J. H. Davis. 1999. The effect of the performance ap- nomic structure, 1840–1920. B. M. Staw and L. L. Cummings,
praisal system on trust for management: A field quasi-experiment. eds. Research in Organizational Behavior. Sage Publications,
J. Appl. Psych. 84 123–136. Thousand Oaks, CA, 53–111.

ORGANIZATION SCIENCE /Vol. 14, No. 1, January–February 2003 17