1484
An Empirical Investigation of the Role of Trust and Power in Shaping the Use of Electronic Markets
Ratnasingam, P. (2005) Trust in inter-organi-
zational exchanges: A case study in business to
business electronic commerce. Decision Support
Systems, 39(3), 525-544.
Ratnasingam, P., Gefen, D., & Pavlou, P. A.
(2005). The role of facilitatinh conditions and
institutional trust in electronic marketplaces.
Journal of Electronic Commerce in Organisa-
tions, 3(3), 69-82.
Sako, M. (1992). Prices, quality and trust: Inter-
¿UPUHODWLRQVLQ%ULWDLQDQG-DSDQ. Cambridge:
Cambridge University Press.
Salancik, G. R., & Pfeffer, J. (1974). Organisational
decision making as a political process: The case
of a university budget. Administrative Science
Quarterly, 19(2), 135-151.
Smith Ring, P., & Van de Ven, A. H. (1992).
Structuring cooperative relationships between
organisations. Strategic Management Journal,
13(7), 483-498.
Soliman, K. S., & Janz, B. D. (2004). An explor-
atory study to identify the critical factors affecting
the decision to establish Internet-based interor-
ganizational information systems. Information
& Management, 41(6), 696-706.
S t a ke , R . E . (19 9 5 ) . The art of case study research.
Thousand Oaks, CA: Sage Publications.
Turban, E., King, D., Viehland, D., & Lee, J.
(2006). Electronic commerce. A managerial

perspective. NJ: Prentice Hall International.
Vlosky, R. P., Fontenot, R., & Blalock, L. (2000).
Extranets: Impacts on business practices and
relationships. Journal of Business & Industrial
Marketing, 15(6), 438-457.
Webster, J. (1995). Networks of collaboration or
F R Q À LF W"(O H F W UR Q LF G D W DL QW H U F K D Q J H D Q G S RZH U L Q 
the supply chain. Journal of Strategic Information
Systems, 4(1), 31-42.
Yin, R. (1994). Case study research: Design and
methods. Second edition. Beverly Hills, CA: Sage
Publishing.
Zaheer, A., McEvily, B., & Perrone, V. (1998). Does
trust matter? Exploring the effects of interorgani-
zational and interpersonal trust on performance.
Organization Science, 9(2), 141-159.
ENDNOTES
1
See Gerst and Bunduchi (2005) and Ratnas-
ingam (2000) for some exceptions.
2
Transactional relationship trust and power
DUH QRW FRQVLGHUHG VLJQL¿FDQW /DPEH
Wittmann, & Spekman, 2001; Morgan &
Hunt, 1994; Sako, 1992).
3
That is the Suppliers Directive 93/36/EC
and the Utilities Directive 93/38/EC.
4
It is interesting to note that although such

regulations exist within the national legis-
lation as well, the interviewees focus only
on the EU legislation. Such an emphasis on
EU, rather than national legislation, can be
explained by the fact that where EU law is
applicable, it prevails over national law. EU
legislation is enforced more severely, and the
sanctions are harsher than in the case of the
national legislation. The threat of substan-
WLDO¿QHVIRUDEUHDFKRIWKH(8UHJXODWLRQV
increases Utilia’s incentives to assure com-
pliance with EU legislation. Additionally,
the emphasis on the EU also could be the
result of the extension of Utilia’s strategic
interest, which does not focus exclusively on
the domestic market but looks to compete
in an extended EU energy market.
5
Just the price of the item, or the total lifecy-
cle cost of the product, that includes several
cost variables.
1485
An Empirical Investigation of the Role of Trust and Power in Shaping the Use of Electronic Markets
6
For an exception, see the study by Allen et
al. (2000) of an IOS between the UK’s motor
vehicle leasing and contract hire companies
and their repair agents.
This work was previously published in Social Implications and Challenges of E-Business, edited by F. Li, pp. 159-172, copyright
2007 by Information Science Reference (an imprint of IGI Global).

1486
Copyright © 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
Chapter 5.7
How Well Do E-Commerce Web
Sites Support Compensatory
and Non-Compensatory
Decision Strategies?
An Exploratory Study
Naveen Gudigantala
Texas Tech University, USA
Jaeki Song
Texas Tech University, USA
Donald R. Jones
Texas Tech University, USA
ABSTRACT
The burgeoning growth of online retailing is
forcing businesses to provide better support for
consumer decision making on e-commerce Web
sites. Consequently, researchers in information
systems and marketing have been focusing on
investigating the effectiveness of Web-based
decision support systems (WebDSS) in providing
accurate and satisfying choices for customers. We
consider WebDSS implementation based on com-
pensatory, non-compensatory decision strategies
and synthesize the existing literature. The results
of synthesis show that compensatory WebDSS
perform better than non-compensatory WebDSS
in terms of decision quality, satisfaction, effort,
DQGFRQ¿GHQFH:HWKHQLQYHVWLJDWHWKHOHYHORI

Web site support provided for consumers’ execu-
tion of compensatory and non-compensatory
strategies. We examined 375 U.S based company
Web sites and found that though moderate levels
of support exists for consumers to implement
1487
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
non-compensatory choice strategies, virtually no
support exists for executing multi-attribute-based
compensatory choice strategies.
INTRODUCTION
The advent of the World Wide Web and search
engines caused a revolution in the way people
search for information. The Internet is now used
by 73% of all American adults as of 2006 (Pew
Internet and American Life, 2006). In addition,
the rapid growth of e-commerce has resulted in
digital marketplaces offering a wide variety of
product alternatives, elaborate product related
information, and great convenience for visitors.
Consequently, ever-greater numbers of individu-
als are interacting with online environments to
search for product related information and to buy
products and services (Xiao & Benbasat, 2007).
In fact, searching for product or service related
information was the second most popular activ-
ity on the Internet in 2003 after e-mail or instant
messaging (U.S. Department of Commerce, 2004).
Concurrently, online sales are expected to reach
$331 billion by 2010, according to a report by

Forrester research.
1
The Web retailers with their
retailer innovations and Web site improvements
are expected to account for 13% of total retail sales
in 2010, up from 7% in 2004. These statistics sug-
gest that e-commerce is growing at a rapid pace,
and individuals are increasingly using digital
market places at every phase of their decision
making process from search to choice.
Nonetheless, access to abundant information
on the Web can be a blessing and a curse at the
same time (Hauble & Murray, 2003). It is a bless-
ing in that consumers now have access to a huge
amount of information from several sources, and
a curse because human beings are limited in their
information processing capabilities (Simon, 1955).
Therefore, many Web retailers are incorporating
WebDSS to assist consumers with their decision-
making process (Grenci & Todd, 2002). WebDSS
capture individual user preferences for products
either explicitly or implicitly and provide recom-
mendations based on such preferences (Xiao &
Benbasat, 2007). WebDSS have the potential to
ease consumers’ information overload and to re-
duce search complexity in addition to improving
their decision quality (Haubl & Trifts, 2000).
The research in the area of consumer decision
support on e-commerce Web sites is rapidly be-
coming interdisciplinary. Researchers from com-

puter science, library sciences, social psychology,
marketing, management, and information systems
are beginning to make important contributions to
this area of research. Consequently, the decision
technology implemented on e-commerce Web
sites is known with different names although they
all refer to the same tool to be used by the consum-
ers. Examples include intelligent agents, electronic
product recommendation agents, recommendation
systems, and WebDSS. In their extensive review
of electronic recommendation agents, Xiao and
Benbasat (2007) categorized recommendation
D JHQWV 5 $VL Q W RW K UH H W \ S H V  7 K H ¿ U VW W \ S H RI5 $ V 
LQFOXGHVFRQWHQW¿OWHULQJFROODERUDWLYH¿OWHULQJ
and hybrid agents. The second type includes
feature-based and need-based RAs. Finally, the
third type of RAs includes compensatory and
non-compensatory-based systems.
The present article considers only compensa-
tory and non-compensatory WebDSS and investi-
gates the level of consumer support provided on
commercial Web sites to execute compensatory
or non-compensatory strategies. We present a
synthesis of literature concerning the effective-
ness of implementing compensatory versus non-
compensatory DSS, and then examine whether
RUQRWVXFK¿QGLQJVKDYHPDGHWKHLUZD\LQWRWKH
design of commercial Web sites. We believe that
understanding the reality of the extent to which
e-commerce Web sites support compensatory

and non-compensatory strategies is important
for several reasons. From a practical standpoint,
LI ZH ¿QG WKDW UHODWLYHO\ D VPDOOHU IUDFWLRQ RI
Web sites provide compensatory-based support
1488
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
GHVSLWHDZHOOVXSSRUWHG¿QGLQJWKDWVXFKVXSSRUW
is normatively better, then that would highlight
an opportunity for the Web retailers to increase
the support levels to their customers. From a
WKHRUHWLFDOVWDQGSRLQWVXFK¿QGLQJZRXOGUDLVH
further questions concerning the factors affect-
ing the implementation of non-compensatory and
compensatory WebDSS.
The rest of the article is organized as follows.
:H¿UVWSUHVHQWEDFNJURXQGVFRQFHUQLQJ:HE'66
types and research concerning the effectiveness
of compensatory versus non-compensatory DSS.
We then describe the methodology used for the
study and present the results. We conclude with a
discussion on managerial implications, directions
for future research, and conclusion.
BACKGROUND: TYPES OF
WEBDSS
One of the common implementations of WebDSS
XVH ¿OWHULQJEDVHG PHWKRGV &RQWHQW¿OWHULQJ
WebDSS consider users’ most desired attributes
and provides recommendations accordingly. Some
of the commercial implementations of content
¿OWHULQJ:HE'66LQFOXGH$FWLYH%X\HUV*XLGH

and MySimon (Xiao & Benbasat, 2007). Col-
ODERUDWLYH¿OWHULQJ:HE'66XVHWKHVXJJHVWLRQV
provided by like-minded consumers to provide
recommendations (Ansari, Essegaier, & Kohli,
2000). Amazon, CD Now provides collaborative
¿OWHULQJ:HE'66RQWKHLU:HEVLWHV
On the other hand, WebDSS can also be imple-
mented using decision strategies. Research inves-
tigating the decision strategies used by individual
decision makers has a long history. Much of the
knowledge acquired from the research domain
of traditional DSSs is now guiding the research
that examines the effectiveness of WebDSS
implementing using different decision strategies.
The scope of our study is limited to studying the
commercial implementation of compensatory and
non-compensatory WebDSS, and the following
section provides an in-depth treatment of the
related concepts.
Decision strategies refer to the rules employed
by individuals to arrive at decisions (Hogarth
1987, p. 72). The decision strategies can be clas-
VL¿HGLQWRFRPSHQVDWRU\DQGQRQFRPSHQVDWRU\
decision strategies. These strategies are discussed
using an example of renting an apartment (see
Table 1).
Non-Compensatory WebDSS
A non-compensatory WebDSS implements one of
the many non-compensatory decision strategies.
Alterna-

tives
Attributes
Rent ($) A/C
Covered
Parking
Washer and
Dryer
Dish
Washer
A 350 Yes No No No
B 400 Yes No Yes No
C 450 Yes No Yes Yes
D 500 Yes Yes Yes Yes
Table 1. Four apartment alternatives
1489
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
The use of a non-compensatory strategy avoids
FRQIURQWLQJWKHFRQÀLFWVLQKHUHQWLQWKHFKRLFH
situation and does not allow the decision maker
to trade off a low value on one attribute against a
high value on another attribute (Hogarth, 1987).
In the apartment rental example, if a university
student decides that covered parking represents
the most important attribute, alternatives A, B,
and C are immediately eliminated. Irrespective of
the attractiveness of alternatives A, B, and C, the
use of one attribute to make the decision results
in their elimination.
There are many examples of non-compensa-
tory strategies. The conjunctive strategy sets cut

off points for certain attributes, and the alterna-
tives that do not meet all of these thresholds are
eliminated. The use of the disjunctive strategy al-
lows an alternative to remain under consideration
so long as either of two attributes has a value that
PHHWVLWVVSHFL¿HGFXWRII([HFXWLQJDQelimina-
tion-by-aspects strategy (EBA) requires select-
ing an attribute at every stage and eliminating
the alternatives that do not include such aspect.
The process continues until a winner is selected.
The VDWLV¿FLQJ strategy compares each attribute
value with a predetermined cut-off level and the
alternative that fails to meet the cut-off level is
rejected (Hogarth, 1987).
Computer support for non-compensatory
strategies focuses on allowing the consumer to
have some control over viewing and manipulating
features of interest. For example, a Web site could
allow the consumer to specify the value of an
attribute (e.g., enter text for searching a Web site
or choosing from a list of criteria) to help select
a set of products for further investigation. Even
more support is given when the consumer can sort
products based on the value of an attribute (as in
sorting by price, weight, or category). Nonethe-
less, many Web sites do not give even this level of
VXSSRUWIRUFLQJWKHXVHUWR³GULOOGRZQ´LQHDFK
SURGXFWWR¿QGLWVIHDWXUHVDQGWKHLUYDOXHV
In general, the use of non-compensatory
WebDSS results in presentation of options that

meet the predetermined thresholds set by the
decision maker. However, the use of non-com-
pensatory WebDSS by decision makers may not
always yield the recommendations suggested
by normative decision-making models. First,
non-compensatory WebDSS do not consider
a consumer’s preference function for multiple
attributes. Second, at times, the use of non-com-
pensatory WebDSS may result in the elimination
of some alternatives based on criterion set on
one attribute, though such options may be very
attractive on the other attributes.
Compensatory WebDSS
A compensatory WebDSS implements one of
the many compensatory decision strategies on
a Web site. The use of a compensatory strategy
PDQGDWHVFRQIURQWLQJWKHFRQÀLFWVLQKHUHQWLQWKH
choice situation and allows the decision maker to
trade off a low value on one attribute against a
high value on another attribute (Hogarth, 1987).
Using the example presented in Table 1, if a uni-
versity student decides to rent an apartment, the
use of a compensatory strategy would facilitate
balancing the lack of some apartment features
of alternative A against high value (low rent) of
another apartment feature.
Compensatory strategies enable desirable
values of one attribute of a product or service to
compensate for undesirable values of another.
Compensatory strategies require computations

or judgmental assessments that combine multiple
variables for each product being considered.
An example of a compensatory strategy is a
weighted-additive strategy in which a weight
(relative importance) is assigned to each attribute
and multiplied by its value. These products are
summed to provide a score for each product. Then
the product with the highest score is selected.
Variations of the weighted-additive strategy use
judgmental assessment of these trade-offs rather
than an actual calculation. Compensatory strate-
gies are more accurate than non-compensatory
1490
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
strategies when potential trade-offs exist between
variables, such as when the consumer would say,
³ , ZRX OG U D W K H U K DYH D F R QYH U W L EOH  E XWQRW L I , K D YH
to pay too much extra to get it.” Unfortunately,
FRPSHQVDWRU\VWUDWHJLHVDUHVRGLI¿FXOWWKDWWKH\
are rarely used except for small sets of products
or if some sort of computer-based assistance is
provided (Johnson & Payne, 1985; Todd & Ben-
basat, 1994).
Computer support for compensatory strategies
would also make it easier for consumers to see
several attributes for several products at a time,
such as when a Web site displays an array of rows
of products with columns for various attributes.
Such an array makes it easier for the consumer
to identify and assess the trade-offs inherent in

compensatory strategies. Some Web sites provide
the display of such an array. Support for com-
pensatory strategies could also enable the selec-
tion of products based on multiple criteria. Still
further support would be to assist the consumer
in evaluating trade-offs by providing some sort
of scoring computation of the values of various
attributes. Such a model would best serve the
consumer if the consumer were allowed to say
which attributes were important and how much
w e i g h t t o g iv e e a c h a t t r ib u t e . F i n al l y t he c o n su me r
ZRX OG Q H H G D ZD\WRH D V L O\¿ QGRXW Z K L F KSURGXFWV
or services obtained the highest scores, such as
having the system sort or select products based
on scores provided by the model.
Partial support for compensatory strategies can
come from allowing the consumer to see external
ratings of products (such as Consumer Reports).
External ratings have the advantages of being
an overall assessment of multiple criteria (i.e.,
they are compensatory) and of being objective.
They have the disadvantage that they are based
on someone else’s assumption about the relative
importance of various criteria. They do not capture
the consumer’s own preferences.
(GZDUGVDQG)DVRORQRWHWKDW³WKHLGHD
of a procedure for making important decisions
that does not depend heavily on human inputs
seems unlikely as well as unattractive. Selection,
training, and elicitation of responses from the

person…become crucial” (p. 588). Compensatory
:HE'66DUHVSHFL¿FDOO\GHVLJQHGWRLPSOHPHQW
such an idea. As opposed to a non-compensatory
WebDSS, compensatory WebDSS need to draw
on the processing capacity and storage abilities
of the computers to implement the normative
algorithms such as multi-attribute utility theory,
Bayesian nets, and subjective expected utility
WKHRU\ WKDWDUH RWKHUZLVH YHU\GLI¿FXOWIRUWKH
unaided decision maker to implement (Lar-
rick, 2004). Compensatory WebDSS execute
algorithms in the background and also perform
consistency checks on user provided weights
making the decision tools more appealing to end
users (Larrick, 2004).
Research Comparing the
Effectiveness of Compensatory
Versus Non-Compensatory WebDSS
How effective are compensatory WebDSS com-
pared to non-compensatory WebDSS? We address
this question in this section based on empirical
U H V X O W V I UR P ¿ YHV W XG LHVWKDWF R P SD U H G F R PSH Q V D -
tory and non-compensatory WebDSS. The results
are summarized in Table 3. The compensatory
WebDSS used in these studies elicit weights from
users on attributes and present alternatives with
¿QDOVFRUHVEDVHGRQH[SHFWHGYDOXHV7KHQRQ
compensatory WebDSS used were based on any
RIWKHFRQMXQFWLYHGLVMXQFWLYH(%$RUVDWLV¿F-
ing strategies. The explanations of the variables

used to compare the two types of WebDSS are
provided in Table 2.
Satisfaction
Previous research suggests that users experience
more satisfaction with the use of compensatory
WebDSS compared to that of non-compensatory
WebDSS (Fasolo, McClelland, & Lange, 2005;
Olson & Widing, 2002; Pereira 2001; Song,
1491
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
Jones, & Gudigantala, 2007). Widing and Talar-
zyk (1993) found that no such differences exist
between the two formats. However they used only
one item to measure satisfaction. The result by
Fasolo et al. (2005) concerning the effectiveness
of compe n sa tor y WebDSS i s note wor thy bec au se
their research study employed an alternative set
with negative inter-attribute correlations. Inter-
attribute correlation is obtained by calculating
the average correlation among the all the pair of
Variable Explanation
Satisfaction The user’s satisfaction with the WebDSS in supporting the decision making process
Decision
Quality/ Ac-
curacy
Preference matching: The extent to which the choice selected by the user matches
her stated preferences
Product switching: Once a purchase decision is made, will the user change his mind
and switch to another choice given a chance?
Effort

The amount of cognitive resources exerted by the user in processing the information
to arrive at the choice
&RQ¿GHQFH 7KHGHJUHHWRZKLFKDXVHUKDVFRQ¿GHQFHLQ:HE'66¶VUHFRPPHQGDWLRQV
Decision Time 7KHWLPHWDNHQWRDUULYHDWWKH¿QDOFKRLFH
Table 2.Variables used for comparing the effectiveness of compensatory and non-compensatory
WebDSS
Fasolo et al.
(2005)
Song et al. (2007)
Olson and wi-
ding (2002)
Periera (2001)
Widing and Ta-
larzyk (1993)
Satisfaction
Compensatory
WebDSS are
better
Compensatory
WebDSS are bet-
ter
Compensatory
WebDSS are
better
Compensatory
WebDSS are
better
No Difference
Decision Qua-
lity/ Accuracy

Compensatory
WebDSS are
better
Compensatory
WebDSS are bet-
ter
Compensatory
WebDSS are
better
Compensatory
WebDSS are
better
Compensatory
WebDSS are
better
Effort
Compensatory
WebDSS are
better
Compensatory
WebDSS are bet-
ter
Compensatory
WebDSS are
better
&RQ¿GHQFH
Compensatory
WebDSS are
better
Decision Time No Difference

Non-Compen-
satory WebDSS
took more time
Table 3. Studies that compared the effectiveness of compensatory and non-compensatory WebDSS
1492
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
attributes. With positive inter-attribute correla-
tions (friendly environment), the alternative that is
favorable on one attribute tends to be favorable on
others. With negative inter-attribute correlations
(unfriendly environment), the more attractive level
of one attribute is associated with less attractive
level on the other. Alternatives characterized by
negative inter-attribute correlations such as those
between cost and quality, are common and lower
FRQVXPHUFRQ¿GHQFHDQGVDWLVIDFWLRQ)DVRORHW
al., 2005). Therefore, consumers require more
support when dealing with alternatives with nega-
tive inter-attribute correlations. Hence, existing
evidence overwhelmingly supports the notion
that compensatory WebDSS contribute to better
satisfaction ratings compared to those of non-
compensatory WebDSS.
Decision Quality
Decision quality has been measured in various
ways in literature. Preference matching measures
the extent to which the choice selected by the user
matches his/her stated preferences (Pereira, 2001)
whereas product switching measures the extent
to which the user is likely to change his/her mind

and switch to alternative choice given that he/she
already made a purchase decision (Widing &
Talarzyk, 1993). The existing evidence strongly
suggests that compensatory WebDSS provide
better decision quality to users compared to non-
compensatory WebDSS (Fasolo et al., 2005; Olson
& Widing, 2002; Pereira 2001; Song et al., 2007;
Widing & Talarzyk, 1993).
Effort
The existing evidence supports the idea that non-
compensatory WebDSS requires more effort from
users than that of compensatory WebDSS (Fasolo
et al., 2005; Pereira, 2001; Song et al., 2007).
&RQ¿GHQFH
Research by Pereira (2001) found out that users
IHOWPRUHFRQ¿GHQWZKHQPDNLQJFKRLFHVZLWK
compensatory WebDSS as opposed to a non-
compensatory WebDSS.
Decision Time
While research by Widing and Talarzyk (1993)
suggests that non-compensatory DSS took more
time to arrive at a choice, Olson and Widing (2002)
found that no such differences exist. Therefore,
the existing evidence is inconclusive whether the
use of compensatory WebDSS saves time.
Therefore, overwhelming evidence supports
the notion that compensatory WebDSS are better
than non-compensatory WebDSS on important
variables such as satisfaction, decision quality,
HIIRUWDQGFRQ¿GHQFH,QDGGLWLRQHYHQLQWKH

absence of decision tools implementing compensa-
tory decision strategies, Web sites that facilitate
the comparison of alternatives contribute to an
increased use of compensatory strategies by us-
ers (Jedetski, Adelman, & Yeo, 2002). However,
research also suggests that non-compensatory
WebDSS support consumer decision making
better than Web sites that just provide products
by alphabetical order (Widing & Talarzyk, 1993).
When technology makes compensatory strategies
easier, consumers are more likely to use them
(Todd & Benbasat, 1994). Moreover, consumers
are more likely to use ratings based on their own
weightings of the features, rather than someone
else’s preferences (Jones & Brown, 2003). Hence,
based on the empirical results that non-compensa-
tory WebDSS are better than Web sites providing
alphabetical order of products, and compensa-
tory WebDSS are better than non-compensatory
WebDSS, we set out to investigate how well the
commercial Web sites provide support to com-
pensatory and non-compensatory strategies.
1493
How Well E-Commerce Web Sites Support Compensatory and Non-Compensatory Decision Strategies
METHODOLOGY
Research on evaluation of Web-based information
systems in the context of e-commerce has been
ongoing. Previous work has examined e-com-
merce Web sites based on concordance analysis
(Jinling & Guoping, 2005) and the factors that

LQÀXHQFH:HEEDVHGLQIRUPDWLRQV\VWHPVVXF-
cess (Garrity, Glassberg, Kim, Sanders, & Shin,
2005). Garrity et al. (2005) found that decision
support satisfaction plays an important role in
Web-based information system success. Given
WKLV¿QGLQJDQGWKHHYLGHQFHWKDWFRPSHQVDWRU\
and non-compensatory WebDSS provide varying
levels of decision support satisfaction (with com-
pensatory WebDSS providing more satisfaction),
we set out to evaluate retail Web sites based on
three criteria:
1. D oes t h e We b si te h ave use f u l fe at u res c o m
-
monly found on competitors’ Web sites?
2. How much support does the Web site give
to consumers’ non-compensatory decision
strategies?
3. How much support does the Web site give
to compensatory strategies in a way that
captures consumers’ own preferences and
weightings of product features?
Selection of Data Source
,QLGHQWLI\LQJEXVLQHVV¿UPVXVHGLQWKLVVWXG\
(as opposed to personal or very small operations),
we used the Business and Company Center da-
tabase.
2
This is one of the most comprehensive
Web-based business databases available today
that offers extensive information on hundreds of

thousands of companies worldwide. We focused
on retail and service industry based on four-digit
industry codes. The scope of our analysis is re-
stricted to the U.S based companies. The database
contains approximately 2,600 U.S. companies out
of approximately 7,600 worldwide companies.
We selected about 25% of the U.S. companies by
selecting every fourth company from the database
consisting of 2,600 companies.
Questionnaire Preparation and Pilot
Testing
7KH ¿UVW SDUW RI WKH GDWD FROOHFWLRQ SURFHGXUH
involved preparation of the questionnaire to be
used by the researcher evaluating a Web site. The
questionnaire elicited information concerning
different kinds of decision support provided by
Web retailers. The questionnaire was intended to
capture information concerning the support pro-
vided to help users locate, evaluate, and compare
products. In addition, information concerning the
provision of a multi-attribute model that would
elicit user preferences as well as the provision of
others ratings’ about products was captured. The
information concerning the communication of
privacy policy was also captured. The question-
naire is included in Appendix A. In a nutshell, the
questionnaire was intended to capture Web site
support for executing non-compensatory strate-
gies, compensatory strategies, product related
information, and security and privacy based

information.
$ SLORW VWXG\ ZDV FRQGXFWHG WR UH¿QH WKH
questionnaire and to examine the agreement about
the information collected. The three authors indi-
vidually visited 30 Web sites and gathered data.
The authors found 90% agreement on the data
collected. Based on discussions about the sources
of the few disagreements, further revisions were
made to the questionnaire to remove ambiguities
in the questions.
Data Collection
We selected about 25% of the 2,600 U.S. compa-
nies, and one of the authors visited 610 Web sites
to collect data.
3
In order to reduce the possible bias
in the sample, we avoided any duplication, such
as companies listed in multiple SIC industries. In
addition, the URLs of many companies are out