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

glassbbc@muohio.edu (B. Glassberg)8 ykim@binghamton.edu(Y.J. Kim)8 mgtsand@mgt.buffalo.edu (G.L. Sanders)8

Decision Support Systems 39 (shin@uri.edu (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: garrity@canisius.edu (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