-
An experimental investigation of W
Abstract
electronic marketplace is having a profound impact on
the world economy and the way business is con-
ducted. In 1996, online sales were approximately
[email protected] (B. Glassberg)8 [email protected](Y.J. Kim)8 [email protected] (G.L. Sanders)8
Decision Support Systems 39 ([email protected] (S.K. Shin).1 Tel.: +1 716 888 2267; fax: +1 716 888 2525.In this paper, we examined Web-based information systems (WIS) success and focused on User Satisfaction in the context of
a consumer purchasing decision. The results indicate strong support for the research model consisting of three fundamental User
Satisfaction components: Task Support Satisfaction (TSS), Decision Support Satisfaction (DSS), and Interface Satisfaction. The
model explains approximately 50% of the variance in users intention to use Web-based information systems. It is concluded
that Decision Support Satisfaction plays an important role in Web-based information systems success. In light of these findings,
implications for theory and practice are discussed.
D 2004 Published by Elsevier B.V.
Keywords: Web information systems; Electronic commerce success metrics; Information systems success; Technology acceptance model; User
satisfaction; Partial least squares
1. Introduction
The emergence of the World Wide Web as an
* Corresponding author. Tel.: +1 716 645 2373; fax: +1 716
645 6117.
E-mail addresses: [email protected] (E.J. Garrity)8success in the context of electronic commerce
Edward J. Garritya,1, Bonnie Glassbergb,2, Yong Jin Kimc,3, G. Lawrence Sandersd,*,
Seung Kyoon Shine,4
aInformation Systems Department, Canisius College, Buffalo, New York 14208, United StatesbDepartment of Decision Sciences and MIS, Miami University, Oxford, Ohio 45056, United States
cSchool of Management, Room 240, State University of New York at Binghamton, Binghamton, New York 13902, United StatesdDepartment of Management Science and Systems, 310A Jacobs Management Center, State University of New York at Buffalo, Buffalo,
New York 14260, United StateseCollege of Business Administration, University of Rhode Island, 210 Flagg Road, Kingston, RI 02881-0802, United States
Received 1 July 2003; accepted 1 June 2004
Available onlnie 2 October 20040167-9236/$ - see front matter D 2004 Published by Elsevier B.V.
doi:10.1016/j.dss.2004.06.015
2 Tel.: +1 53 Tel.: +1 607 777 6638; fax: +1 607 777 4422.4 Tel.: +1 401 874 5543; fax: +1 401 874 4312.eb-based information systems
2005) 485503
www.elsevier.com/locate/dswexpanding rapidlyUS$500 million and have been31 529 4826; fax: +1 531 529 9689.since [64]. The total electronic commerce spending
estimates for 2004 range from US$963 billion to
-
organizational setting, investigating the causal path
from User Satisfaction to individual impacts is
uppoUS$4 trillion [20]. Many firms positioned and
deployed systems to compete in the electronic
marketplace. The systems, in general, are referred to
as Web-based information systems (WIS) [76]. Web-
based information systems [76] represent a new
frontier for businesses trying to establish an on-line
presence where consumers are free to shop in more
efficient bfriction free markets.QIncorporating the radical changes in global market-
places created by this new application of information
technology, Web-based systems are considerably
different from the traditional systems in terms of their
scope and focus [91]. WIS are usually built to
facilitate consumer oriented tasks and focus on
enhancing the consumer decision making process.
As the Web environment is characterized as non-linear
in nature, presentation and delivery of information and
frictionless work support are very critical for attract-
ing consumers and enhancing the shopping experi-
ence. Lederer et al.s [46] study found that the TAM
model did not adequately explain Web use, and this
implies that traditional information systems have
different characteristics than WIS.
Given the importance of electronic commerce and
WIS, in terms of the magnitude of its impact on
business process and organization structure, there is
little research addressing fundamental issues such as
how electronic commerce systems success can be
measured, whether existing measures of success can
be applied, and how consumers react to the online
shopping experience. Without a clear understanding
of the dynamics of Web-based system success to
guide firms, proper strategies and system designs are
mere speculation. A central activity for researchers
will be to define and operationalize the constructs for
understanding the success of electronic commerce
systems. This paper, while drawing on the traditional
information systems success literature, extends that
literature to the development of a Web-based model
for electronic commerce success in the context of a
consumer purchase decision.
The contributions of the current study are three-
fold. First, this research addresses the issue of how to
predict users intent to use a particular Web site
utilizing traditional information systems success
theory. The results should be of great interest to
E.J. Garrity et al. / Decision S486Web designers, IS staff and researchers. Second, by
explaining the dynamic relationship between Userfruitless [68]. Finally and most importantly, the model
does not explain clearly and fully the relationship
between User Satisfaction and individual/organiza-
tional impacts.
2.2. Technology Acceptance ModelSatisfaction components which influence WIS suc-
cess, the current research can aid researchers in further
refinement of information systems success models in
general. Third, the current study provides a reasonable
background for applying existing measures of infor-
mation systems success to the electronic commerce
environment.
2. Review of information systems success literature
There is a rich tradition of information systems
success research including user information satisfac-
tion, tasktechnology fit, user involvement, and
participation. In this paper, three important, compre-
hensive studies are further examined in terms of
providing theoretical background. They are the
DeLone and McLean Model of IS Success [17], the
Technology Acceptance Model (TAM) [1315], and
the Garrity and Sanders Model of IS Success [25].
2.1. DeLone and McLean Model of IS Success
DeLone and McLean [17] suggested a model of IS
success that is comprised of six multi-level constructs:
Information Quality, System Quality, User Satisfac-
tion, System Use, Individual Impact, and Organiza-
tional Impact. While the model integrates the
comprehensive dependent variables used by IS
researchers, there exist several criticisms (although
they do not reduce the importance of the model). First,
IS Use in the DeLone and McLean model contains too
many meanings to be appropriately examined [68]. IS
Use is also argued to play a problematic and
controversial role in modeling system success (cf.
Refs. [15,49,68,69]). Second, because User Satisfac-
tion represents individual impacts of IS in an
rt Systems 39 (2005) 485503A second stream of IS success research is focused
on the use of information technology as conceptual-
-
ers m
upport Systems 39 (2005) 485503 487ized by the Technology Acceptance Model [1315].
This model of IS success asserts that Ease of Use
and Perceived Usefulness are primary determinants
of System Use. The model specifically postulates
that technology Usage is determined by Behavioral
Intention to Use the technology, which is itself
determined by both Perceived Usefulness and
Attitude.
TAM has been widely studied in a number of
contexts using different information systems. More
recently, it has been applied to examining the use of
Web-based information systems [27,46,54,77]. How-
Fig. 1. Garrity and Sand
E.J. Garrity et al. / Decision Sever, this model does not capture the complex
psychological and social influences on technology
adoption [50] and long-term use [9,50]. In particular,
when applied to the context of Web information
systems, the model appears to account for only a small
portion of variance for Usage [46]. Lederer et al. [46]
called for researchers to develop better instruments for
measuring antecedents of Perceived Ease of Use and
Perceived Usefulness for assessing Web sites. In
addition, they urged a reexamination of factors or
latent variables related to Web site Usage via study
replication, and additional use of factor analysis to
uncover important subdimensions of the Technology
Acceptance Model.
2.3. Garrity and Sanders Model of IS Success
Garrity and Sanders [25] extended the DeLone and
McLean model and proposed an alternative model inthe context of organizational and socio-technical
systems (Fig. 1). The model identifies four subdi-
mensions of User Satisfaction: Interface Satisfaction,
Decision Support Satisfaction (DSS), Task Support
Satisfaction (TSS), and Quality of Work Life Sat-
isfaction. These factors were derived from an exten-
sive review of IS success research and reasoning from
basic principles of systems and general systems
theory. The four factors correspond with three view-
points of information systems: the organizational
viewpoint (that views IS as a component of the larger
organization system), the humanmachine viewpoint
odel of IS success [25].(which focuses on the computer interface and the user
as components of a work system), and the socio-
technical viewpoint (that considers humans as also
having goals that are separate from the organization
and whereby the IT or technical artifact impacts the
human component in this realm).5
Both Task Support Satisfaction and Decision
Support Satisfaction attempt to assess the effective-
ness of an IS within the context of the organizational
5 The fourth dimension of the Garrity and Sanders model is
quality of work life satisfaction. This dimension addresses the fit
between an IS and the socio-technical work world of the respective
users. Quality of work life satisfaction is an important dimension of
success in organizational settings. However, in the context of a
consumer-oriented web information system, the users will not be
impacted in terms of work-life or job satisfaction, and thus, this
dimension will not be included in this study.
-
viewpoint of systems. The Task Support Satisfaction
dimension captures the overall set of tasks associated
with job activities while Decision Support Satisfaction
is more focused on decision support (i.e., structuring,
analyzing, and implementing a decision).
Interface Satisfaction assesses IS success from the
humanmachine viewpoint. Interface Satisfaction
measures the quality of the interface in terms of
presentation, format, and processing efficiency. The
quality of the interface is related to both Task Support
information systems success [17,26,35,36,52]. Fig. 2
shows the proposed research model. The next sections
3.1. Theoretical background of the research model
In this research study, we are concerned with
assessing User Satisfaction with Web-based informa-
tion systems in the context of consumer purchasing
activities. The nature of a consumer purchasing Web-
based system is closely tied to providing decision
support capabilities to aid consumers in product and
service purchase decisions [57].
A consumer-oriented WIS can be decomposed
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503488elaborate a theoretical background of the proposed
research model and detailed hypotheses developed to
examine the relationships among the constructs of
interest.and Decision Support Satisfaction. An improperly
designed interface can cause users difficulty with task
completion or it can impair their ability to make
decisions.
3. Research model and hypothesis
The model used in this study draws on the DeLone
and McLean Model [17], TAM, and the Garrity and
Sanders Model of IS Success [25] to examine Web-
based information systems success. As outlined by
Lederer et al. [46], we seek to discover important
underlying factors that will help to explain Web-based
information systems use and adoption. In essence, the
research model, Web-based Information Systems
Success Model (hereafter, referred to as WISSM),
deals with the reciprocal relationship between User
Satisfaction and systems use. This relationship has
been touted as one of the key constructs representingFig. 2. Web-based Information Systeinto numerous tasks, subtasks and decisions. For
example, a product purchasing WIS may involve a
number of cognitive tasks, such as product search-
ing, advice gathering, decisions on information
presentation, rank ordering of alternatives, criteria
discovery, learning about the domain of product use,
deciding on the relevance of criteria, determining
tradeoffs among criteria, prices, and products, actual
purchase tasks such as badding to shopping carts,Qnavigating from one task context to the next,
checkout, payment, confirmation, and shipping
methods. Many of these cognitive tasks can be
categorized as intellectual or decision support based,
while others may be more mundane tasks such as
filling out self-report information.
In sum, Decision Support Satisfaction assesses the
support provided by the WIS in the context of
decision making. Task Support Satisfaction, on the
other hand, assesses the overall support for the
purchasing task. Since this is a general assessment
of the overall task, it takes into account the overall
design of the software and its usefulness. In addition
to these variables, Trust in a Web site is regarded as an
important determinant of EC success [27,82]. Gefen
and Straub [27] included a Trust item as an indicator
variable for the behavioral intention to purchase.ms Success Model (WISSM).
-
However, the current study does not directly incor-
porate Trust as an indicator variable reflecting
E.J. Garrity et al. / Decision SupporBehavioral Intention to Use an EC web site.
The dependent variable, Behavioral Intention to
Use an EC Web site, was taken from the Technology
Acceptance Model [1315] and Seddons [68] respe-
cification of the DeLone and McLean Model [17].
Their work provides the theoretical basis for the
relationship between User Satisfaction and the Behav-
ioral Intention to Use an information system.
System Use was not selected as the dependent
variable because of the well-documented problems in
measuring system success (cf. Seddon [68]).6 This
paper posits that Behavioral Intention to Use a system
represents an individuals perception of the benefit
from System Use [68]. In essence, Behavioral
Intention to Use a system effectively mediates the
effect of User Satisfaction on actual systems use. The
positive relationship between User Satisfaction and
Behavioral Intention has been frequently reported in
the marketing literature (e.g., Refs. [2,53,83]). The
innovation literature also reports that dissatisfaction
with an innovation leads to the termination of the
innovation [60].
It should be noted that the constructs used in this
model are more context based than the System Quality
and Information Quality measures from the DeLone
and McLean model. Information Quality and System
Quality were developed from a product focus as
opposed to a system, tool, or goal directed focus. Our
approach, using a more goal-oriented perspective,
asserts that a system of high quality should support
users in performing their task related responsibilities.
This is the approach taken by Davis [1315] in the
development of the Perceived Usefulness dimension
of IS Success. Recall that Perceived Usefulness is
defined as the users subjective assessment that using
a system will increase his or her job performance. The
dimensions of Task Support and Decision Support
Satisfaction are simply more specific measures of the
type of job support provided by the system.
Finally, the Interface Satisfaction dimension may
be thought of in several ways. It provides an assess-
ment of the design of the hardware and software
6 According to Seddon [68], information systems use in Deloneand McLeans model has three distinct meanings: a proxy for the
benefits from use, future IS Use, and an event in a process model.interface and is closely related to Daviss Ease of Use
dimension. If systems are tools to help workers
accomplish tasks, then a well-designed tool should
be made in such a way that it is easy and efficient to
use. In addition, the interface is also the focal point of
interaction between humans and information systems.
Thus, Information Quality can also be explained by
Interface Satisfaction, in the sense that the vehicle for
presenting the information (e.g., a textbox, graph,
listbox, or form) cannot be separated from the
information itself [56]. A bchunkQ of information isbeneficial to a user if it is presented in an effective and
efficient manner, helping users accomplish task
objectives. bProviding high-quality data along thedimensions of value and usefulness relative to the
consumers task contexts places a premium on
designing flexible systems with data that can be easily
aggregated and manipulatedQ [73]. Thus, we believe aseparate assessment of the interface is especially
critical for WIS because users must successfully
navigate the Web site to obtain informational support
for tasks and decisions.
Thus, the focus of the Web-based Information
Systems Success Model presented in Fig. 2 is on: (1)
How well the system supports workers in the
accomplishment of their goals (an effectiveness
measure via Task Support and Decision Support
Satisfaction), and (2) How well designed and imple-
mented the interface is so that workers may accom-
plish their goals more easily (an efficiency measure
via Interface Satisfaction).
3.2. Research question and hypotheses
Fig. 2 illustrates hypothesized relationships between
constructs in the study. The major determinant of
Behavioral Intention to Use an electronic commerce
Web site is hypothesized to be Task Support Satisfac-
tion which is influenced by both Decision Support
Satisfaction and Interface Satisfaction. Note that there
is no direct relationship between Interface Satisfaction
and Behavioral Intention to Use an EC Web site. The
reasons are twofold. First, as discussed earlier, the
current study assumes that Task Support Satisfaction is
a focal construct that mediates the effects of Interface
and Decision Support Satisfaction on Behavioral
t Systems 39 (2005) 485503 489Intention. The second reason relates to the way that
Interface Satisfaction affects technology acceptance
-
uppo[43]. Over time, the impact of Ease of Use on
Behavioral Intention fades away [86].
3.2.1. Interface Satisfaction
Lohse and Spiller [47] showed that designing online
stores with user-friendly interfaces is critical in traffic
and sales, while Szymanski and Hise [75] found site
design important in creating a satisfying customer
experience. The Interface Satisfaction dimension cap-
tures the users overall impression of the interface in
terms of presentation, navigability, format, ease of use
and efficiency of interaction with the Web site.
Shneiderman [70] argues that an appropriate design
of user interfaces includes presentation, and interactiv-
ity. Zhang and von Dran [90] report that ease of
navigation was the most important Web site quality
across six electronic commerce domains. Poorly
designed screens and interfaces make task support
and decision support activities more difficult. In
particular, high interface quality in hypermedia systems
like the Web is critical in reducing cognitive overhead
[11,48] and disorientation [11,78,81].
According to the theory of symbolic representation
in problem solving [32,40,41], there are three types of
symbolic representation: linguistic representation,
visual imagery representation, and exploratory rea-
soning. These symbolic representations can be
referred to as representational information quality
[73], which is, in turn, interpreted as interface quality.
In most problem solving situations, various combina-
tions of these three representations are needed to
provide dominant or controlling focus in problem
solving. Two critical stages in the problem solving
process are information search and the evaluation of
alternatives [19,34,55], and they are fundamental for
decision making and task performance.
The importance of the user interface has a rich
research tradition. Assimilation Theory [3] posits that
meaningful learning of computer systems begins with
the humancomputer interface. Davis and Bostrom
[16] and Mayer [51] found that the interface can have
a positive impact on task performance and decision
making within the context of Assimilation Theory.
High levels of usability are related to improved task
performance [80]. Ausubel [3] emphasized the impor-
tance of external aids in the form of an badvance
E.J. Garrity et al. / Decision S490organizerQ that facilitates the development of newknowledge structures. The user interface as anbadvance organizerQ should enhance task performanceand decision making performance. According to
Nielsen [56], user-centered navigability significantly
enhances users performance with various Web tasks.
Poor interface design has been cited as a key element
in a number of high profile Web site failures [58]. It is
hypothesized, therefore, that Interface Satisfaction
positively influences both dimensions of satisfaction
(TSS, DSS) with the system.
H1a. Interface Satisfaction has a positive impact on
Decision Support Satisfaction.
H1b. Interface Satisfaction has a positive impact on
Task Support Satisfaction.
3.2.2. Decision Support Satisfaction
Decision Support Satisfaction is defined in terms of
the capability of an information system to assist in
decision-making and better performance of the users
jobs [65], even when those decision-making activities
are ill-structured, situation-specific, or involve choos-
ing from a number of alternatives [44].
Humans tend to adapt their decision making
strategies to specific situations and environments
[61]. Individuals can be described as bcognitivemisersQ who strive to reduce the amount of cognitiveeffort associated with decision making [71]. The
notion that people are typically willing to settle for
imperfect accuracy in their decisions, in return for a
reduction in effort, is well supported [8,37], and is
consistent with the idea of bounded rationality and
the concept of satisficing [72]. Due to this tradeoff
between effort and accuracy, decision makers often
satisfice when alternatives are numerous and/or
difficult to compare, or when the complexity of the
decision environment is high [62]. In the context of
Web-based consumer purchasing decision systems, it
is imperative not only to assess the Perceived
Usefulness of these systems (the purchasing task
itself) but also to include an assessment of the degree
to which the system supports or reduces the decision
making efforts of the user.
OKeefe and McEachern [57] delineate five critical
task areas that need to be modeled in a Web-based
consumer decision support system: need recognition,
rt Systems 39 (2005) 485503information search, evaluation, purchase, and after-
purchase evaluation. Decision support features that
-
The two consequences of dissatisfaction negatively
affect repeat purchasing behavior. The positive
upporcan be provided by a Web-based consumer decision
support system include virtual catalogs, internal
search capabilities, structured interactions, and eval-
uative decision models. While a number of these
features fall outside the realm of simply supporting
the decision process, they are vital to supporting the
entire task process (i.e., a purchase task). Examples of
actions to support the task include placing an order,
arranging a payment, or providing detailed informa-
tion on the purchase task. Additionally, this would
encompass after sale support and feedback, and
response about the process itself.
Unlike traditional organizational decision support
systems, Web-based consumer purchasing systems
must be designed to battractQ consumers to use thesystem [57]. The non-linear structure of the Web
environment entails more cognitive effort than the
linear, sequential structure of the traditional system
environment and without sufficient support for deci-
sion making and task performance, consumers tend to
simply click-away to competitors.
Hence, decision support features or actions in WIS
should not only support the decision process but also
support the entire task process [57]. This reasoning
leads to the hypothesis of a positive relationship
between Decision Support Satisfaction and Task
Support Satisfaction as in H2.
H2. Decision Support Satisfaction is positively
associated with Task Support Satisfaction.
3.2.3. Task Support Satisfaction
Antecedents for e-commerce success examined by
previous studies highlight the importance of support-
ing online consumers at each step of their activities
with an e-commerce system. The most common
online tasks include product search, comparison
shopping, negotiation of terms, order placement,
payment authorization, and receipt of product [38].
The essential question surrounding the success of the
online experience relates to the net value of the
benefits and costs of finding, ordering, and receiving a
product [42]. User satisfaction with such task support
can be referred to as the net benefits of a system [68].
The main behaviors of satisfied or dissatisfied
customers with product or service offerings include
E.J. Garrity et al. / Decision Scomplaining behavior, negative word of mouth, and
repeat purchasing [74]. Complaining behavior isrelationship between the level of satisfaction and the
repeat purchasing behavior has been reported to be
very strong (correlation 0.882) [74]. In the informa-
tion systems context, the repeat purchasing behavior
is interpreted as the repeated Use of an information
system, which is primarily determined by Behavioral
Intention to Use the information system [14,15,68,85].
Hence, the relationship between Task Support
Satisfaction, as a global evaluation of systems
support, and Behavioral Intention to Use an electronic
commerce Web site as a precursor for future Use of a
system, can be hypothesized as H3.
H3. Task Support Satisfaction positively influences
Behavioral Intention to Use an e-commerce Web site.
Hypotheses one through three relate to the specific
paths of WISSM. Task Support Satisfaction is
hypothesized to be the primary determining factor of
Behavioral Intent to Use an EC Web site. Task
Support Satisfaction is partially derived from Per-
ceived Usefulness [25]. A separate but closely related
dimension, Decision Support Satisfaction, is hypothe-
sized to positively influence Task Support Satisfaction
(see H2). Decision support capabilities are important
and vital task considerations, and this is particularly
true for systems used in electronic markets [6,7,57].
4. Research method
The research model was tested in the context of an
Internet buying simulation. The subjects of this study
were given a Web-based purchasing task. The task was
to buy a digital camera under given limitation in terms
of budget, time, and retail Web sites. The budget limit
given to the subjects was US$600, which is the medianexplained as consumers tendency to complain to
suppliers and this works as a mechanism for dissat-
isfied consumers to relieve cognitive dissonance,
while negative word of mouth behavior to other
consumers stands for another type of complaining
behavior of dissatisfied consumers. Repeat purchasing
behavior represents the phenomenon that satisfied
consumers will likely buy the offerings repeatedly.
t Systems 39 (2005) 485503 491price of digital cameras. This budget limit was required
so that subjects would have a greater cognitive burden
-
on the decision task. It was expected that few of the
respondents would have already purchased the target
product, while most would be familiar with the general
features of a camera.
E.J. Garrity et al. / Decision Suppo4924.1. Sampling method
Each subject was given two identical sets of survey
instruments to be used in assessing the quality of two
Web sites (CameraWorld.com and Advandig.com).
Each set contained the four constructs (Decision
Support Satisfaction, Task Support Satisfaction, Inter-
face Satisfaction, and Behavioral Intent to Use) under
study.
4.1.1. Data collection procedure
The data collection activity for this study consisted
of three main phases. First, an expert survey, for
identifying two diverse Web sites, was conducted.
Second, a pilot study, involving 97 subjects and 194
observations, was performed. It was used to refine the
items and constructs used in the study. It also enabled
the researchers to clarify the wording, content, and
general layout of the survey instrument. Finally, a
main survey was administered, involving 163 subjects
and 326 observations.
In the first phase of the data collection, ten experts
who had practical and academic experiences with
electronic commerce evaluated six Web sites.7 The
experts were asked to rate the Web sites in terms of
Decision Support, Task Support, Interface, and Over-
all Satisfaction level, by responding to a 7-point
Likert scale questionnaire. The scores for Decision
Support, Task Support, and Interface Satisfaction were
summed. Both the sum of the scores and Overall
Satisfaction level consistently indicate that all ten
experts nominated CameraWorld.com as the superior
commercial Web site. Nine of the experts voted
Advandig.com as the least desirable (inferior) of the
six Web sites.
The second phase of the data collection consisted
of a pilot study. The objective of this phase was to
7 We selected six most popular commercial web sites special-
izing digital cameras and accessories, which commonly appeared in
the lists of three Internet portal sites. The six sites include Advandig,Digital Camera, Digital Solution, Camera World, Digital Camera
Club, and Digital Camera Store.focus on measurement issues and to refine the
research instruments prior to the final data collection.
Pilot survey questionnaires were distributed to 97
junior and senior undergraduate business majors.
The main study consisted of 163 students that were
not involved in the pilot study. Two hundred fifty
surveys were given out and 163 were returned
(response rate: 65.2%). As noted earlier, the survey
instrument consisted of two identical questionnaires
that were used by subjects to evaluate the Web sites
according to the research constructs. The order of the
appearance of the Web site addresses on the ques-
tionnaire was randomized.
4.1.2. Sample frame
Identifying the sample frame for empirical research
is an important step in ensuring that the population of
interest has been correctly identified. The ideal
candidate for this study was computer literate, had
previous exposure to Web technology and had current
access to a Web-enabled computer. The Web environ-
ment is highly graphical, requiring knowledge of the
use of icons, frames, and scrolling. There may be
multiple browser windows open at the same time, so
general navigation and mouse skills were needed. As
the purpose of this study was to predict purchasing
behavior by individuals, in-depth knowledge of Web
technology was neither assumed nor expected in the
target population.
How much exposure to the technology is suffi-
cient? Empirical studies based on specific technolo-
gies, where the individual is the unit of analysis, have
employed a 1 hour hands-on task [14,84]. Other
studies used exposure without a required hands-on
task [39,50] and some asked subjects for pure
speculation about a technology which did not exist
[12]. Required courses at the institution used to
conduct the study typically use the Web in a variety
of ways including distributing course syllabi, presen-
tation materials and handouts, online cases, and a
variety of Web-based assignments. Subjects would
thus have satisfied the exposure constraint. In
addition, respondents were asked to rate themselves
on their computer skills and Web experience.
The final requirement for the population of interest
was current availability of a computer with Web
rt Systems 39 (2005) 485503access. All subjects had access to the Internet via
school or work and many had their home computer
-
the average age was 21.5 years. Respondents were
asked to rate the extent of their experience with theWeb
and with computers in general on a scale of 1 to 7, with
1 being a novice and 7 being an expert. This group of
subjects considered themselves to be relatively expe-
rienced users of the Web (4.90) and slightly above
average on their use of computers (5.01). Most reported
they logged onto the Web about 20 times a week,
spending just under 16 h on the Internet (weekly). A
majority of the subjects have one to ten buying
experiences over the Internet, while 24% of partic-
ipants have never bought over the Internet. Addition-
ally, subjects reported they spend about 24 h per week
on the computer itself. The majority of respondents
(78%) had completed 3 years of college.
4.2. Partial Least Squares (PLS)
Assessment of the research model was conducted
using PLS. PLS is a structured equation modeling
technique that can analyze structural equation models
(SEMs) involving multiple-item constructs, with
upporlinked to the Web through an Internet Service
Provider.
4.1.3. Subject pool
Human subjects identified for the pilot test and final
data collection were a sample of upper level under-
graduate students enrolled in management information
systems courses at a state research institution in the
northeast. All subjects voluntarily participated in the
survey, and were advised that they could withdraw
from participation at any time without adverse con-
sequence. The survey was conducted in an unsuper-
vised environment. All the participants were given
extra credit for participating in the study. The final data
collection utilized all undergraduates enrolled in
introductory MIS classes taught by an instructor. This
was a required course for all business majors.
Although the reliance on college students has been
criticized for many studies in applied research, there is
justification for using them under specific conditions
[31]. In a meta-analysis comparing studies using both
undergraduate students and non-students, Gordon et
al. [31] noted that between-groups differences are less
pronounced when no prior knowledge or familiarity
with the task is needed to make judgments, or when
the objective is to perform simple reasoning tasks. A
majority of undergraduates taking courses are exposed
to Web technology as part of the regular curriculum of
their coursework. Any complexity or skill in perform-
ing experimental tasks is reduced by the graphical
nature of the technology. Thus, only general knowl-
edge of computers and English were required.
The use of undergraduate students made sense in
this study because a primary focus of this research is on
understanding Web-based decision behavior and col-
lege students are the present and future consumers of
Web technology. It should be pointed out that major
users of the Web are individuals who have a college
degree and are 1644 years old [33], which is similar to
the subject pool used in this study (age range 1850
years old).
4.1.4. Demographics
One hundred sixty-three upper-level undergraduates
participated in the final survey. The demographics for
this sample are shown in Table 1. Approximately 60%
E.J. Garrity et al. / Decision Sof the subjects were male and 40% were female. The
age of respondents ranged from 18 to 50 years old, andTable 1
Demographics
Age Average:
21.5 years
Range:
1850 years
Gender Male 60.5%
Female 39.5%
Education level: years
completed in
college
2 30.6%
3 54.4%
4 or more 15.0%
Experience with web
technology
Average self-rating
(1=Novice, 7=Expert)
4.90
Experience with
computers
Average self-rating
(1=Novice, 7=Expert)
5.01
Online buying
experience
Never 24%
15 50%
610 15%
More than 10 times 11%
Current web usage Average accesses per
week
20
Average hours per
week
16
Years using web Less than 1 year 6%
Between 1 and
1.5 years
31%
More than 2 years 13%
More than 3 years 78%
t Systems 39 (2005) 485503 493direct and indirect paths. PLS works by extracting
successive linear combinations of the predictors and is
-
Table 2
Operationalization of latent variables
Decision Support Satisfaction
DBET Use of the web site
enables me to make
better purchasing
decisions
[65]
DEFF This web site assists me
in making a decision
more effectively
Adapted from
Ref. [14]
DPRI Use of the web site
enables me to set my
priorities in making the
purchase decision
[65]
Task Support Satisfaction
TUSE This web site is more
useful than I had
expected
[24]
TEXT This web site is
extremely useful
[65]
TQIK Using this web site
enables me to
accomplish tasks more
quickly
[14]
TESY This web site makes it
easier to do my task
[14]
Interface Satisfaction
ICLR The information
provided by this web
site is clear and
understandable
[18,2830]
ILEN Learning to use this web
site was easy for me
[14]
IFRN This web site is
user-friendly
[14,18,2830]
IESY This web site is easy to
use
[14,18,2830]
IWNT I found it easy to get this
web site to do what I
want it to do
[14]
IINT My interaction with this
web site was clear and
understandable
[14]
ISKL It would be easy for me
to become skillful at using
this web site
[14]
INAV This web site is easy to
navigate
[46]
Behavioral Intent to Use
UINF I predict I will use this
web site to get the
information I need
Adapted from
Ref. [86]
E.J. Garrity et al. / Decision Suppo494effective in explaining both response and predictor
variation [10]. PLS is a powerful approach for
analyzing models because of the minimal demands
on measurement scales, sample size, and residual
distributions [89]. In addition, the component-based
PLS avoids two serious problems, inadmissible
solutions and factor indeterminancy [22].
SEM approaches, such as LISREL and AMOS, are
not able to deal with non-normal distributions [21],
and they can yield non-unique or otherwise improper
solutions in some cases [22]. PLS is not as susceptible
to these limitations [88,89]. The emphasis of PLS is
on predicting the responses as well as in under-
standing the underlying relationship between the
Behavioral Intent to Use
UCRT The use of this web site
would be critical in my
future decisions
Adapted from
Ref. [24]
UCRE I would use my credit
card to purchase from
this web site
[27]
USIM I would use this web site
again for a similar task
Adapted from
Ref. [1]
UINC I intend to increase my
use of this web site in
the future
[1]
UFUT In the future, I plan to
use this web site often
Adapted from
Ref. [86]
All questionnaire items use a seven-point Likert scale, varying from
Strongly disagree to Strongly agree.
Table 2 (continued)
rt Systems 39 (2005) 485503variables [79]. For example, PLS is very useful in
screening out factors that have a negligible effect on
the dependent variables.
PLS has one additional property that is useful in
studying the nature of the linkages between constructs
and their measures. The nature of the links is referred
to as epistemic relationships, or drules of corre-spondenceT [4,21]. Two basic types of epistemicrelationships are relevant to causal modeling: reflec-
tive indicators and formative indicators. In this study,
we use reflective indicators for measuring the
unobserved, underlying constructs.
4.3. Operationalization of research variables
As shown in Table 2, the variables of the study
were operationalized using a variety of sources. All of
the measures were taken directly or adapted from
-
previous studies on IS success such as Doll and
Torkzadeh [18], Franz and Robey [24], and Sanders
[65] and on technology acceptance such as Davis [14]
and Lederer et al. [46].
To keep the length of the instrument reasonable,
four and five items were selected from the Task
Support Satisfaction and the Decision Support Sat-
isfaction categories, respectively, and eight items were
selected for Interface Satisfaction. For Behavioral
Intention to Use an EC Web site, the study draws on
three previous studies including Agarwal and Prasad
Satisfaction (This Web site assists me in performing
my task better).
5. Analysis and results
PLS analysis involves two stages: (1) the assess-
ment of the measurement model, including the
reliability and discriminant validity of the measures,
and (2) the assessment of the structural model. For the
assessment of the measurement model, individual
I
I
I
I
IFRN 0.9014 0.1585 UCRE 0.7757 0.1879
I
I
I
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503 495[1], Gefen and Straub [27], and Venkatesh and Morris
[86]. The study follows the criteria recommended by
Agarwal and Prasad [1] for choosing survey items.
They recommend removing items that are not relevant
to the specific innovation examined in the study, and
also deleting items that are very similar to other items.
By using these criteria, the items selected ensure
complete coverage of the constructs at hand. Behav-
ioral Intention to Use the EC Web site was measured
using six items.
Among them, five items involve the likelihood of
future use, while one item bUsing my credit card topurchase from this Web siteQ is used to assess levelof Trust. Trust is considered a critical aspect of
electronic commerce success [67], and is an
important determinant of Behavioral Intention to
Use the Web site [82]. This item was also used in
Gefen and Straub [27] to measure Intention to
Purchase.
Two items were subsequently excluded from the
final analysis because of their high correlation
coefficient: one item from Decision Support Satisfac-
tion (This Web site improves the quality of my
purchase decision making) and one from Task Support
Table 3
Measures, loadings, and weights
Decision Support Satisfaction Task Support Satisfaction
Item Loadings Weight Item Loadings Weight
DBET 0.9228 0.3579 TUSE 0.8471 0.2844
DEFF 0.9202 0.3862 TEXT 0.9079 0.2931
DPRI 0.8952 0.3511 TQIK 0.9104 0.2743
TESY 0.8848 0.2749I
IESY 0.9271 0.1503 USIM 0.8831 0.2196
WNT 0.8777 0.1582 UINC 0.8784 0.1761
INT 0.8782 0.1439 UFUT 0.8475 0.1850item loadings and internal consistency were examined
as a test of reliability. Individual item loadings and
internal consistencies greater than 0.7 are considered
adequate [23].
As shown in Table 3, loadings for all measurement
items are above 0.7, which indicates there is sound
internal reliability and all the weights are statistically
significant at pb0.001. The almost uniformly distrib-uted weights show each item contributes to each
construct equivalently. We used PLS-Graph Version
2.91.03.04 to perform the analysis.
5.1. Reliability and validity tests
In assessing the internal consistency for a given
block of indicators, the composite reliability (CR),
also referred to as convergent validity (see Werts et
al. [87]), was calculated. All the CR values are over
0.9, which suggests that the parameter estimates are
sound (Table 4).
The average variance extracted (AVE) was also
calculated. AVE measures the amount of variance that
a construct captures from its indicators relative to the
variance contained in measurement error. This statistic
nterface Satisfaction BI to use EC Web site
tem Loadings Weight Item Loadings Weight
CLR 0.8623 0.1625 UINF 0.8575 0.2112
LEN 0.8695 0.1256 UCRT 0.8724 0.1923SKL 0.8368 0.1244
NAV 0.7939 0.1248
-
compared the proposed four-factor model to all
possible three-factor models (there were four such
comparisons). In all cases the change in the v2
comparing the proposed four-factor model to each of
the three-factor models was significant at pb0.001. Thebest three-factor model involved treating DSS and TSS
as one construct, which was accomplished by setting
the correlation between the two constructs to one. The
v2 difference between the four-factor model and thisthree-factor model was v2=105 (df=1, pb0.001).
The analyses with both PLS and AMOS indicate
that there exists reasonable discriminant validity
among all of the constructs. It is important to note
that this study is in fact examining an information
systems success model consisting of dependent
Table 4
Composite reliability (CR) and average variance extracted (AVE)
Constructs CR AVE Formula
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503496can be interpreted as a measure of reliability for the
construct and as a means of evaluating discriminant
validity [23]. AVE values should be greater than 0.50.
All AVEs for the constructs used in this study are
greater than .70. This indicates that more than 70% of
the variance of the indicators can be accounted for by
the latent variables.
The AVE can also be used to assess discriminant
validity. The AVEs should be greater than the square
of the correlations among the constructs. That is, the
amount of variance shared between a latent variable
and its block of indicators should be greater than
shared variance between the latent variables. In this
study, the square roots of each AVE value is greater
than the off-diagonal elements (Table 5, refer to
Appendix A which contains the cross loadings
between the measurement items and the constructs).
This indicates that there exists reasonable discriminant
validity among all of the constructs.8
The correlation between Decision Support Satis-
faction and Task Support Satisfaction appears to be a
little high although valid in terms of discriminant
validity criteria. Such a close but distinctive relation-
Decision Support
Satisfaction
0.9374 0.8332
Task Support
Satisfaction
0.9371 0.7884 CR=(P
k i)2/[(P
k i)2
+P
ivar(e i)]Interface Satisfaction 0.9610 0.7554 AVE=
Pk i2/[P
k i2
+P
ivar(e i)]Behavioral Intention
to Use EC web site
0.9413 0.7280
k i is the component loading to an indicator and var(e i)=1ki2.ship between Decision Support Satisfaction and Task
Support Satisfaction was expected (see Garrity and
Sanders [25]). Hence, Decision Support Satisfaction
together with Task Support Satisfaction can be used to
provide more insight into the features of the Web site.
To further evaluate/establish the discriminant val-
idity of the constructs, we conducted discriminant
analysis tests [5] using AMOS. These analyses
8 In personal correspondence with Wynne Chin, he indicated
that the difference between the off-diagonal (correlation) and the
diagonal (square root of AVE) should be greater than 0.1 in order to
secure reasonable discriminant validity.variable constructs (see Refs. [17,25] for a discussion
of dependent variable models). Accordingly, the
correlations between the research variables are
expected to be high. However, they should not be so
high as to preclude them from being considered
separate constructs. As noted above, the constructs in
the model do indeed satisfy discriminate validity
criteria detailed by Fornell and Larcker [29] and
Bagozzi and Phillips [5].
5.2. Assessment of the structural model
The path coefficients in the PLS model represent
standardized regression coefficients. The suggested
lower limit of substantive significance for regression
coefficients is 0.05 [63]. In a more conservative
Table 5
Correlations of latent variables
Decision
Support
Satisfaction
Interface
Satisfaction
Task
Support
Satisfaction
BI to use
EC Web
site
Decision
Support
Satisfaction
(0.913)
Interface
Satisfaction
0.501*** (0.869)
Task Support
Satisfaction
0.778*** 0.636*** (0.888)
Behavioral
Intention
to Use EC
0.612*** 0.482*** 0.697*** (0.853)web site
The number parentheses is the square root of AVE, and *** p b 0.001.
-
ructur
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503 497position, path coefficients of 0.10 and above are
preferable. As shown in Fig. 3, all path coefficients
are over 0.1 satisfying both conservative criteria and
the suggested lower limit. They are also statistically
significant at p=0.001, which indicates that all
hypotheses (H1a, H1b, H2, and H3) are supported
by the data. Moreover, a high R2 for each endogenous
variable in the structural model shows that this model
can be used to predict the Behavioral Intention to Use
a Web site.
5.3. Order effects and treatment effects
The survey questionnaires were designed in two
different formats with identical measures (Format A
listed the inferior site first, and Format B listed the
superior site first). Order effects during the experi-
ments were investigated. The results of a two-way
ANOVA (Table 6a) showed that there was no
statistically significant difference in Behavioral Inten-
Fig. 3. WISSM sttion to Use caused by the order of the presentation of
the Web sites on the questionnaire.
The treatment effects were explored next. Table 6a
and 6b present that Behavioral Intention to Use the
Table 6a
Tests of between-subjects effects
Source Sum of squares df M
Superior/inferior sites 74.34 1 74
Formats 3.603 1 3.6
Superior/inferiorFormats 3.244e03 1 3.2Error 464.122 271 1.7
Total 5340.028 275
Corrected total 543.474 274
Dependent variable: behavioral intention to use an EC web site.superior Web site (Mean=4.699, S.D.=1.192) was
significantly higher than the inferior Web site
(Mean=3.650, S.D.=1.417). As expected, most sub-
jects (112 out of 156 respondents) decided to purchase
a digital camera from the superior site. Additionally,
there are no interaction effects between the order of
presentation and the experimental treatment.
5.4. Effects of online buying experience
Experience entails general learning by trial and
error, and an increasing awareness and familiarity
with the types of domain problems and their structure
[45]. Through the accumulation of online buying
experience, the online buyers can develop a repertoire
of Internet skills, knowledge, and meta-knowledge
that leads to effective performance in a specific
domain [66]. Thus, it is clear that online buying
experience plays a role in individuals Web site
assessment process and influences the selection of a
al model results.Web site that a buyer would purchase from.
The current study investigated the impacts of
subjects online buying experience on their Intention
to Use for the four experience groups. Based on the
ean square F Significance Eta squared
.347 43.411 0.000 0.138
03 2.104 0.148 0.008
44e03 0.002 0.965 0.00013
-
the difference between group 1 and the other groups.
The analysis offers evidence that a buyers online
to the Interface Satisfaction concept, will lead to
high bIntention to Use the Web siteQ via an indirectpath through Decision Support and Task Support
Satisfaction.
Table 6b
Mean (S.D.) statistics
Table 7a
Group means and standard deviations
Group
no.
Number of
experience
Sample
size
Mean score
differencea
(Mn)
S.D. Contrast
estimate
(MnM1)1 0 30 0.344 1.362
2 15 64 0.914 1.676 0.569
3 610 18 1.555 1.679 1.211b
4 10b 14 2.119 1.878 1.774c
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503498experience has significant impacts on Intention to
Use. The results show that more experienced online
buyers tend to more clearly observe the difference
between two sites than less experienced buyers
( pb0.005). In order to investigate the divergence ofsubjects intention in different groups, mean score
differences9 were calculated. While subjects in group
1 who have never experienced online buying do not
show much difference in their buying intention with
superior and inferior Web sites (M1=0.344), ones in
group 4 who have more than ten online-buying
experiences better recognize the difference
(M4=2.119). In essence, the more an individual
experiences online purchasing, the better they eval-
uate the quality of Web sites, and the more likely they
intend to use a superior site.
6. Discussion and implications
This study has found support for the Web-based
Information Systems Success Model introduced here.
Overall, the results indicate strong support for the
model. Table 8 summarizes the results of the hypothesismean score of the seven-scale items of Behavioral
Intention to Use an EC Web site, the contrasts are
developed by first performing an ANOVA test (Table
7b) and then conducting a post hoc test where the
ANOVA indicates a difference among treatments
(Table 7a). The post hoc tests involve three t tests on
Format A Format B Total
Superior site 4.821 (1.316) 4.585 (1.059) 4.699 (1.192)
Inferior site 3.772 (1.466) 3.549 (1.377) 3.650 (1.417)
Total 4.317 (1.481) 4.053 (1.334) 4.176 (1.408)testing. WISSM explains almost 50% of the variation
in Behavioral Intent to Use theWeb information system
(R2=0.486). All of the paths in the model are statisti-
cally significant (at p=0.001 level). Compared to prior
studies, WISSM effectively predicts the intention to
9 Mean score difference is the difference between the mean
score of a subjects Behavioral Intention to Use the superior site and
the mean score of Behavioral Intention to Use the inferior site.use a Web-based information system. Behavioral
Intention to Use is primarily explained via Task
Support Satisfaction. The results of this model also
show, as in Davis et al.s [15] findings, that Task
Support Satisfaction (corresponding to Perceived Use-
fulness) directly influences Behavioral Intention to Use
an ECWeb site and Interface Satisfaction (correspond-
ing to Perceived Ease of Use) has an indirect impact on
Behavioral Intention to Use an EC Web site through
Task Support Satisfaction.
Lederer et al. [46] have urged IS researchers to
explore and develop different factors or latent
variables that can better explain information System
Use and lead to models that explain greater variance.
In this research, important antecedents of Task
Support Satisfaction, the Interface Satisfaction and
Decision Support Satisfaction dimensions, were
validated in the context of a Web-based purchasing
information system. The results are consistent with
researchers who study Web usability such as Nielsen
[56]. Web systems that are well developed according
a Mean score differences (Mn) are the difference between the
mean scores of superior and inferior web sites.b Total mean difference=1.004, pb0.08.c Total mean difference=1.004, pb0.06.Table 7b
Tests of between-subjects effects
Source Sum of
squares
df Mean
square
F Significance
Intercept 74.34 1 138.316 51.972 0.000
Experience
group
3.603 3 12.150 4.565 0.005
Error 464.122 122 2.661
Total 5340.028 126
-
Satisfaction, for this study, is the users assessment of
Decision Support Satisfaction, which involves activ-
upporities such as comparing important product features
and prices, and examining the myriad of tradeoffs in
bundled attributes offered at a Web site. Given the
new demands of electronic commerce and markets,
such specialized task characteristics warrant further
consideration and measurement by researchers.
7. ConclusionThe most important determinant of Task Support
Table 8
Summary of hypothesis testing
Hypothesis Support Significance
level
H1a Interface Satisfaction has a
positive impact on Decision
Support Satisfaction.
Yes pb0.001
H1b Interface Satisfaction has a
positive impact on Task
Support Satisfaction.
Yes pb0.001
H2 Decision Support
Satisfaction is positively
associated with Task upport
Satisfaction.
Yes pb0.001
H3 Task Support Satisfaction
positively influences
Behavioral Intention to Use
an EC Web site.
Yes pb0.001
E.J. Garrity et al. / Decision SThis study attempts to address several of the issues
in electronic commerce research: How should Web-
based information systems success be measured in the
context of User Satisfaction, whether existing meas-
ures of success can be applied, and how consumers
react to commercial Web sites. The results of this
study, which draw on the DeLone and McLean Model
[17], the Garrity and Sanders [25] model of IS
success, and TAM, support the use of WISSM for
examining Web-based information systems success.
The results also confirm three fundamental WIS User
Satisfaction components (Task Support Satisfaction,
Decision Support Satisfaction, and Interface Satisfac-
tion) that explain approximately 50% of the variance
in users Intention to Use a WIS. The amount of
variance explained in Behavioral Intention to Use,
however, needs to be carefully interpreted with the
understanding that the experiment was done within acontrived situation. The respondents were not actually
making a purchase.
An important facet of conducting WIS research is
the availability of a dependent variable that is able to
reflect the concept of system success. A set of
questions has been presented that can be used to
measure WIS success. This is, of course, not the only
available scale nor should it be considered the final
instrument. The brevity of the scale along with its
reasonably interpretable and clear dimensions desig-
nates this measure as a useful addition to the current
literature.
This research is not without limitations. First, this
experiment utilized a Web-based information sys-
tem, focusing on a fundamental but core purchasing
task. The subjects in this study were asked to
determine a product to purchase, given a constrained
budget. A complete assessment of consumer pur-
chasing behavior in an electronic market should,
however, include after sale support, order tracking,
and should be extended to the areas of bneedassessment,Q advertising, and promotion (see Ref.[57]). While the current study examined those
respondents who were considered a major user
group, further experiments could utilize new referent
groups and control the experimental environment
more closely. That is, future study needs to include
various user groups (i.e., individuals in high school,
those who do not go to college, and users with
disposable income) and investigate differences in
their behavior. Notwithstanding these limitations, the
study supports the basic research question that
Decision Support Satisfaction and Task Support
Satisfaction do matter in WIS success. This follows
recommendations by Lederer et al. [46] to develop
more robust constructs to explain Web-based infor-
mation systems success.
In the future, research on Web-based information
systems assessment could be extended to consider
variations in Web support required for different types
of products. For example, the experimental design
could manipulate differences in product information
intensity and determine the effect on purchase
behavior. Palmer and Griffith [59] call for research
into Web-based product purchase decisions that
manipulate the product type to account for differences
t Systems 39 (2005) 485503 499in consumer binvolvementQ in the purchase. Web sitedesign for high involvement products may require
-
different types of decision and task support. In
addition, most research in the Web success area has
found a strong link between Trust and purchase
behavior (e.g., Ref. [82]). For future work, it would
be worthwhile to consider the effect of Trust on
Behavioral Intention and different User Satisfaction
dimensions. Future research might also examine the
potential of the Web-based Information Systems
Success Model for different Web-based applications.
Models of success may need to be modified as
technology changes and new models of business and
computing evolve.
Acknowledgement
The authors thank two anonymous reviewers for
their constructive comments and suggestions on
earlier drafts of this article. The authors also thank
Appendix A. Cross loadings between measurement
items and constructs
TUSE 0.692 0.847 0.563 0.632
E.J. Garrity et al. / Decision Suppo500TEXT 0.730 0.908 0.548 0.651
TQIK 0.666 0.910 0.556 0.604
TESY 0.671 0.885 0.592 0.586
ICLR 0.484 0.623 0.862 0.477
ILEN 0.371 0.485 0.869 0.334
IFRN 0.479 0.603 0.901 0.478
IESY 0.444 0.579 0.927 0.428
IWNT 0.495 0.588 0.878 0.462
IINT 0.420 0.558 0.878 0.419
ISKL 0.361 0.485 0.837 0.360
INAV 0.399 0.458 0.794 0.357
UINF 0.568 0.639 0.491 0.858
UCRT 0.571 0.582 0.366 0.872
UCRE 0.482 0.569 0.458 0.776
USIM 0.534 0.665 0.521 0.883Item Decision
Support
Satisfaction
Task
Support
Satisfaction
Interface
Satisfaction
BI to use
EC Web
site
DBET 0.923 0.691 0.455 0.527
DEFF 0.920 0.732 0.511 0.580
DPRI 0.895 0.706 0.402 0.568Dr. Wynne Chin for his correspondence related to the
discriminant validity issues.UINC 0.492 0.533 0.303 0.878
UFUT 0.475 0.560 0.293 0.848References
[1] R. Agarwal, J. Prasad, Are individual differences germane to
the acceptance of new information technologies? Decision
Sciences 30 (2) (1999) 361391.
[2] E.W. Anderson, M.W. Sullivan, The antecedents and con-
sequences of customer satisfaction for firms, Marketing
Science 12 (2) (1993) 125143.
[3] D.P. Ausubel, Educational Psychology: A Cognitive View,
Holt, Rinehart and Winston, New York, NY, 1968.
[4] R.P. Bagozzi, A prospectus for theory construction in market-
ing, Journal of Marketing 48 (1984 Winter) 1129.
[5] R.P. Bagozzi, L.W. Phillips, Representing and testing organ-
izational theories: a holistic construal, Administrative Science
Quarterly 27 (3) (1982) 459489.
[6] J.Y. Bakos, Reducing buyer search costs: implications for
electronic marketplaces, Management Science 43 (12) (1997)
16761692.
[7] Y. Bakos, The emerging role of electronic marketplaces on the
Internet, Communications of the ACM 41 (8) (1998) 3542.
[8] J.R. Bettman, E.J. Johnson, J.W. Payne, A componential
analysis of cognitive effort in choice, Organizational Behavior
and Human Decision Processes 45 (1990) 111139.
[9] R.O. Briggs, A technology transition model derived from field
investigation of GSS use aboard the U.S.S. CORONADO,
Journal of Management Information Systems 15 (3) (1998/
1999) 151195.
[10] W.W. Chin, The Partial Least Squares approach to structural
equation modeling, in: G.A. Marcoulides (Ed.), Modern
Methods for Business Research, Lawrence Erlbaum Associ-
ates, Mahwah, NJ, 1998, pp. 295336.
[11] J. Conklin, Hypertext: an introduction and survey, IEEE
Computing Survey 20 (9) (1987) 1741.
[12] P.A. Dabholkar, Incorporating choice into an attitudinal frame-
work: analyzing models of mental comparison processes,
Journal of Consumer Research 21 (1) (1994) 100118.
[13] F.D. Davis, A Technology Acceptance Model for Empirically
Testing New End-User Information Systems: Theory and
Results. PhD dissertation, MIT, 1986.
[14] F.D. Davis, Perceived usefulness, perceived ease of use, and
user acceptance of information technology, MIS Quarterly 13
(3) (1989) 319340.
[15] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance of
computer technology: a comparison of two theoretical models,
Management Science 35 (8) (1989) 9821003.
[16] S.A. Davis, R.P. Bostrom, Training end users: an experimental
investigation of the roles of the computer interface and training
methods, MIS Quarterly 17 (1) (1993) 6185.
[17] W.H. DeLone, E.R. McLean, Information systems success: the
quest for the dependent variable, Information Systems
Research 3 (1) (1992) 6195.
[18] W.J. Doll, G. Torkzadeh, The measurement of end-user
computing satisfaction, MIS Quarterly 12 (2) (1988) 259274.
[19] J.F. Engel, D.T. Kollat, R.D. Blackwell, Consumer Behavior,
21 ed.., Holt, Rinehart and Winston, New York, 1973.
rt Systems 39 (2005) 485503[20] L. Enos, Report: B2B Still Driving e-Commerce, http://
www.ecommercetimes.com, Dec. 11, 2000.
-
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503 501[21] C. Fornell, A Second Generation of Multivariate Analysis,
Praeger, New York, NY, 1982.
[22] C. Fornell, F.L. Bookstein, Two structural equation models:
LISREL and PLS applied to consumer exit-voice theory,
Journal of Marketing Research 19 (4) (1982) 440452.
[23] C. Fornell, D. Larcker, Evaluating structural equation models
with unobservable variables and measurement error, Journal of
Marketing Research 18 (1981) 3950.
[24] C.R. Franz, D. Robey, Organizational context, user involve-
ment, and the usefulness of information systems, Decision
Sciences 17 (3) (1986) 329356.
[25] E.J. Garrity, G.L. Sanders, Dimensions of information systems
success, in: E.J. Garrity, G.L. Sanders (Eds.), Information
Systems Success Measurement, Idea Group Publishing,
Hershey, PA, 1998, pp. 1345.
[26] E.J. Garrity, G.L. Sanders, Introduction to information systems
success measurement, in: E.J. Garrity, G.L. Sanders (Eds.),
Information Systems Success Measurement, Idea Group
Publishing, Hershey, PA, 1998, pp. 112.
[27] D. Gefen, D. Straub, The relative importance of perceived
ease-of-use in IS adoption: a study of e-commerce adoption,
Journal of the Association for Information Systems 1 (8)
(2000) 121.
[28] D.L. Goodhue, Developing a Theory-Based Measure of User
Satisfaction: The Task-Systems Fit Questionnaire, in Working
Paper, 1990 Information and Decision Sciences, University of
Minnesota.
[29] D.L. Goodhue, Development and measurement validity of a
tasktechnology fit instrument for user evaluations of infor-
mation systems, Decision Sciences 29 (1) (1998) 105138.
[30] D.L. Goodhue, R.L. Thompson, Tasktechnology fit and
individual performance, MISQuarterly 19 (2) (1995) 213236.
[31] M.E. Gordon, L.A. Slade, N. Schmitt, The bscience of thesophomoreQ revisited: from conjecture to empiricism, Acad-emy of Management Review 11 (1) (1986) 191207.
[32] T. Helstrup, One, two, or three memories? A problem-solving
approach to memory from performed acts, Acta Psychologica
66 (1987) 3768.
[33] D.L. Hoffman, W.D. Kalsbeek, T.P. Novak, Internet and Web
use in the U.S., Communications of the ACM 39 (12) (1996)
3646.
[34] J. Howard, J. Sheth, The Theory of Buyer Behavior, Free
Press, New York, NY, 1972.
[35] M. Igbaria, M. Tan, The consequences of information
technology acceptance on subsequent individual performance,
Information & Management 32 (3) (1997) 113121.
[36] B. Ives, M.H. Olson, J.J. Baroudi, The measurement of user
information satisfaction, Communications of the ACM 26 (10)
(1983) 785793.
[37] E.J. Johnson, J.W. Payne, Effort and accuracy in choice,
Management Science 31 (4) (1985) 395414.
[38] R. Kalakota, A.B. Whinston, Electronic Commerce: A
Managers Guide, Addison-Wesley, Reading, MA, 1997.
[39] E. Karahanna, D.W. Straub, N.L. Chervany, Information
technology adoption across time: a cross-sectional comparison
of pre-adoption and post-adoption beliefs, MIS Quarterly 23(2) (1999) 183213.[40] G. Kaufmann, Imagery, language and cognition: toward a
theory of symbolic activity in human problem solving,
Columbia University Press, NY, 1980.
[41] G. Kaufmann, A theory of symbolic representation in problem
solving, Journal of Mental Imagery 9 (2) (1985) 5170.
[42] R.L. Keeney, The value of Internet commerce to the customer,
Management Science 45 (4) (1999) 533542.
[43] M. Keil, P.M. Beranek, B.R. Konsynski, Usefulness and ease
of use: field study evidence regarding task considerations,
Decision Support Systems 13 (1) (1995) 7591.
[44] P.J. Kirs, G.L. Sanders, R.P. Cerveny, D. Robey, An
experimental validation of the Gorry and Scott Morton
framework, MIS Quarterly 13 (2) (1989) 183197.
[45] J.L. Kolodner, Towards an understanding of the role of
experience in the evolution from novice to expert, Journal of
ManMachine Studies 19 (1983) 497518.
[46] A.L. Lederer, D.J. Maupin, M.P. Sena, Y. Zhuang, The
Technology Acceptance Model and the world wide web,
Decision Support Systems 29 (3) (2000) 269282.
[47] G.L. Lohse, P. Spiller, Electronic shopping, Communications
of the ACM 41 (7) (1998) 8187.
[48] D. Lucarella, A. Zanzi, A visual retrieval environment for
hypermedia information systems, ACM Transactions on
Information Systems 14 (1) (1996) 329.
[49] M.A. Mahmood, End user computing productivity: a new
approach to help resolve information technology productivity
paradox, Journal of End User Computing 7 (2) (1995) 25.
[50] K. Mathieson, Predicting user intentions: comparing the
Technology Acceptance Model with the theory of planned
behavior, Information Systems Research 3 (3) (1991)
173191.
[51] R.E. Mayer, The psychology of how novices learn computer
programming, Computing Surveys 13 (1981) 121141.
[52] N.P. Melone, A theoretical assessment of the user-satisfaction
construct, Management Science 36 (1) (1990) 7691.
[53] V. Mittal, P. Kumar, M. Tsiros, Attribute-level performance
satisfaction, and behavioral intentions over time: a consump-
tion-system approach, Journal of Marketing 63 (2) (1999)
88101.
[54] M.G. Morris, A. Dillon, How user perceptions influence
software use, Decision Support Systems, IEEE Software (1997
JulyAugust) 5865.
[55] F.N. Nicosia, Consumer decision processes: a futuristic view,
Advances in Consumer Research 9 (1982) 1719.
[56] J. Nielsen, Designing Web Usability: The Practice of
Simplicity, New Riders Publishing, Indianapolis, IN, 2000.
[57] R.M. OKeefe, T. McEachern, Web-based customer Decision
Support Systems, Communications of the ACM 41 (3) (1998)
7178.
[58] J.W. Palmer, Web site usability, design, and performance
metrics, Information Systems Research 13 (2) (2002) 151167.
[59] J.W. Palmer, D.A. Griffith, An emerging model of web site
design for marketing, Communications of the ACM 41 (3)
(1998) 4451.
[60] M. Parthasarathy, A. Bhattacherjee, Understanding post-
adoption behavior in the context of online services, Informa-tion Systems Research 9 (4) (1998) 362379.
-
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503502[61] J.W. Payne, Contingent decision behavior, Psychological
Bulletin 92 (2) (1982) 382402.
[62] J.W. Payne, J.R. Bettman, E.J. Johnson, The Adaptive
Decision Maker, Cambridge, New York, NY, 1993.
[63] E.J. Pedhazur, Multiple regression in behavioral research:
explanation and prediction, 3rd edn., Harcourt Brace College
Publishers, Fort Worth, 1997.
[64] H.R. Rao, A.F. Salem, B. DosSantos, Introduction: marketing
and the internet, Communications of ACM 41 (3) (1998)
3234.
[65] G.L. Sanders, MIS/DSS success measure, Systems, Objec-
tives, Solutions 4 (1984) 2934.
[66] K.D. Schenk, N.P. Vitalari, K.S. Davis, Differences between
novice and expert systems analysts: what do we know and
what do we do? Journal of Management Information Systems
15 (1) (1998) 950.
[67] E. Schonberg, T. Cofino, R. Hoch, M. Podlaseck, S.L.
Spraragen, Measuring success, Communications of the ACM
43 (8) (2000) 5357.
[68] P.B. Seddon, A respecification and extension of the DeLone
and McLean Model of IS Success, Information Systems
Research 8 (3) (1997) 240253.
[69] P.B. Seddon, M. Kiew, A partial test and development of the
DeLone and McLean Model of IS Success, Proceedings of
Fifteenth International Conference on Information Systems
(ICIS 1994), 1994, pp. 99110.
[70] B. Shneiderman, Designing the User Interface: Strategies for
Effective HumanComputer Interaction, 3rd edn., Addison-
Wesley Publishers, Reading, MA, 1998.
[71] S.M. Shugan, The cost of thinking, Journal of Consumer
Research 7 (2) (1980) 99111.
[72] H.A. Simon, A behavioral model of rational choice, Quarterly
Journal of Economics 69 (1955) 99118.
[73] D.M. Strong, Y.W. Lee, R.Y. Wang, Data quality in context,
Communications of the ACM 40 (5) (1997) 103110.
[74] D.M. Szymanski, D.H. Henard, Customer satisfaction: a meta-
analysis of the empirical evidence, Journal of the Academy of
Marketing Science 29 (1) (2001) 1635.
[75] D.M. Szymanski, R.T. Hise, e-Satisfaction: an initial exami-
nation, Journal of Retailing 76 (3) (2000) 309322.
[76] J.M. Tenenbaum, WISs and electronic commerce, Communi-
cations of the ACM 41 (7) (1998) 8990.
[77] T.S.H. Teo, V.K.G. Lim, R.Y.C. Lai, Intrinsic and extrinsic
motivation in internet usage, Omega 27 (1) (1999) 2537.
[78] M. Thqring, J. Hannemann, J.M. Haake, Designing forcomprehension: a cognitive approach to hypermedia deve-
lopment, Communications of the ACM 38 (8) (1995)
5766.
[79] R.D. Tobias, An Introduction to Partial Least Squares
Regression, SAS Institute, 1999, pp. 18.
[80] R. Took, Putting design into practice: formal specification and
the user interface, in: H. Harrison, H. Thimbleby (Eds.),
Formal Methods in HumanComputer Interaction, Cambridge
University Press, Cambridge, UK, 1990.
[81] K. Utting, N. Yankelovich, Context and orientation in hyper-
media networks, ACM Transactions on Information Systems 7(1) (1989) 5884.[82] C. Van Slyke, F. Belanger, C.L. Comunale, Factors influen-
cing the adoption of web-based shopping: the impact of trust,
Database for Advances in Information Systems 35 (2) (2004)
3249.
[83] S. Varki, M. Colgate, The role of price perceptions in an
integrated model of behavioral intentions, Journal of Service
Research 3 (3) (2001) 232240.
[84] V. Venkatesh, F.D. Davis, A model of the antecedents of
perceived ease of use: development and test, Decision
Sciences 27 (10) (1996) 451481.
[85] V. Venkatesh, F.D. Davis, A theoretical extension of the
Technology Acceptance Model: four longitudinal field studies,
Management Science 46 (2) (2000) 186204.
[86] V. Venkatesh, M.G. Morris, Why dont men ever to stop to ask
for directions? Gender, social influence, and their role in
technology acceptance and usage behavior, MIS Quarterly 24
(1) (2000) 115139.
[87] C.E. Werts, R.L. Linn, K.G. Joreskog, Intraclass reliability
estimates: testing structural assumptions, Educational and
Psychological Measurement 34 (1) (1974) 2533.
[88] H. Wold, Causal flows with latent variables, European
Economic Review 5 (1974) 6786.
[89] H. Wold, Partial Least Squares, in: S. Kotz, N.L. Johnson
(Eds.), Encyclopedia of Statistical Sciences, vol. 6, Wiley,
New York, 1985, pp. 581591.
[90] P. Zhang, G.M. von Dran, User expectations and rankings of
quality factors in different web site domains, International
Journal of Electronic Commerce 6 (2) (2001/2002) 9.
[91] V. Zwass, Electronic commerce: structures and issues, Interna-
tional Journal of Electronic Commerce 1 (1) (1996) 323.
Edward J. Garrity is Associate Professor of Information Systems
in the Richard J. Wehle School of Business at Canisius College in
Buffalo, New York. He holds a PhD in MIS and an MBA from the
State University of New York at Buffalo. He has published research
in Datamation, DATA BASE, Encyclopedia of Computer Science,
Encyclopedia of Information Systems, Expert Systems with Appli-
cations, Heuristics, Information Resources Management Journal,
the Journal of Management Information Systems, the Journal of
Management Systems, and several book chapters. He has also co-
authored a book of readings entitled bInformation Systems SuccessMeasurement,Q published by Idea Group Publishing. His currentresearch interests are information systems development issues,
information systems success measurement, and IT evaluation
management.
Bonnie C. Glassberg is currently an Assistant Professor at Miami
University/Ohio. She has a PhD from the University of South
Carolina, an MBA from Indiana University, and a BSc degree from
Cornell. As an Assistant Professor at the University at Buffalo
(19972001), she also taught information systems at the graduate
and undergraduate level. Bonnie has 10 years of industry experience
as a Senior Analyst with a Fortune 100 company. She has been
involved in the planning, design, implementation, and management
of major corporate systems. She has served as a reviewer for JMIS,CACM, Database, and the Mid-America Journal of Business.
-
Yong Jin Kim is an Assistant Professor of Management Information
Systems at the School of Management at the State University of New
York at Binghamton. He holds a PhD in MIS from the State
University of New York at Buffalo. He participated in large IT
projects including a BPR implementation for Samsung Life
Insurance, an online accounting systems development project for
the Bank of Korea, and a business recovery planning project for the
Korea Financial Telecommunication and Clearing Institute. His
research interests are in IS success, e-commerce, and information
technology valuation, and knowledgemanagement. He has published
papers in outlets such as Communications of the ACM, Decision
Support Systems, and Knowledge and Process Management. He has
also published book chapters on IS Success and e-learning.
Seung Kyoon Shin is an Assistant Professor of Information
Systems in the College of Business Administration at the University
of Rhode Island. Prior to joining academia, he worked in industry as
a software specialist and system integration consultant. His research
has appeared in Communications of the ACM, and the International
Conference on Information Systems, and Hawaii International
Conference on System Sciences (HICSS). His current research
interests include strategic database design for data warehousing and
data mining, Web-based information systems success, and economic
and cultural issues related to intellectual property rights.
G. Lawrence Sanders, PhD, is Professor and Chair of the
Department of Management Science and Systems in the School of
Management at the State University of New York at Buffalo. He has
taught MBA courses in the Peoples Republic of China and
Singapore. His research interests are in the ethics and economics
of digital piracy, systems success measurement, cross-cultural
implementation research, and systems development. He has
published papers in outlets such as The Journal of Management
Information Systems, The Journal of Business, MIS Quarterly,
Information Systems Research, the Journal of Strategic Information
Systems, the Journal of Management Systems, Decision Support
Systems , Database Programming and Design , and Decision
Sciences. He has also published a book on database design and
co-edited two other books.
E.J. Garrity et al. / Decision Support Systems 39 (2005) 485503 503
An experimental investigation of web-based information systems success in the context of electronic commerceIntroductionReview of information systems success literatureDeLone and McLean Model of IS SuccessTechnology Acceptance ModelGarrity and Sanders' Model of IS Success
Research model and hypothesisTheoretical background of the research modelResearch question and hypothesesInterface SatisfactionDecision Support SatisfactionTask Support Satisfaction
Research methodSampling methodData collection procedureSample frameSubject poolDemographics
Partial Least Squares (PLS)Operationalization of research variables
Analysis and resultsReliability and validity testsAssessment of the structural modelOrder effects and treatment effectsEffects of online buying experience
Discussion and implicationsConclusionAcknowledgementCross loadings between measurement items and constructsReferences