Information & Management 38 (2000) 119128

Implementation in a world of workstations and networks$

Henry C. Lucas Jr.*, Valerie Spitler
Stern School of Business, New York University, Department of Information Science,
44 West 4th Street, Suite 9-67, New York, NY 10012-1126, USA
Accepted 26 May 2000

Abstract
This paper explores the relevance of research on implementation and user acceptance given the pervasiveness of technology
in modern organizations. The paper presents the results of applying an extended model of technology acceptance to the use of
broker workstations. We argue that implementation success is important in obtaining a return from the rm's investment in
technology. Data are collected at two points in time to assess user acceptance of the workstations. The results provide some
support for the models and the unanticipated nding that perceptions of system quality and system ease of use decreased over
time. The paper suggests that a combination of research methodologies may be appropriate for assessing user acceptance of
the complex technology typically found in today's rms. # 2000 Elsevier Science B.V. All rights reserved.
Keywords: Implementation; Technology acceptance model; Workstations; Task-technology t

1. Introduction
The pace of change and user adoption increases
with each technological innovation. It took 5 years for
personal computers to become a major factor in
business. Once the Internet could be used for prot,
the number of service providers and users grew exponentially. With such rapid change and widespread
adoption, does the organization need to be concerned
with acceptance and implementation issues that have
characterized earlier technology innovations? Are
users nally ready to adopt technology without some
of the implementation problems witnessed in the
past?
This paper addresses two questions: (1) is implementation research relevant in today's high-technology business environment; and (2) how well do
$

The Chairman of the Editorial Board processed this paper.

Corresponding author. Present address: Robert H. Smith School
of Business, University of Maryland, USA.
*

models of implementation predict user acceptance

of technology?
Interest in the implementation of technology began
in the 1960s and 70s as many information systems
failed to achieve their potential [10]. There has been a
continuing stream of research on implementation, and
more recently, the acceptance of technology [3,16].
With large numbers of people having computer-based
workstations on their desks, and some organizations
having more computers than employees, is implementation still relevant?
1.1. Implementation research
Implementation research seeks to determine what
motivates people to use and accept technology. Is
this an interesting and relevant question today? In
the era before networked computing, managers frequently complained that they did not receive a return
from their investment in technology. It seemed that
many senior managers were unaware of their rm's

0378-7206/00/$ see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 7 2 0 6 ( 0 0 ) 0 0 0 5 9 - 8

120

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

technology initiatives. With the exponential growth of

the Internet, the impact of technology stocks on the
market, and the frequent articles about information
technology, electronic commerce and the Web, it is
hard to believe that any U.S. executive lacks awareness of the pervasiveness of today's technology. Does
this awareness translate into acceptance and use? Can
research on implementation help answer this question?
Past research on implementation has been categorized as falling into `factor' and `process' studies
[11,17]. The setting for this research generally
involved mainframe or minicomputers with batch
and on-line applications; there are few studies involving PCs, and only one that we are aware of includes
networked, multifunctional workstations [13]. Factor
studies are variance research; the investigator develops a model, instruments to measure the value of
variables in the model, collects data and analyzes it
using statistical techniques like correlation and regression analysis. The name comes from the analysis in
which the variance in one variable or construct is
correlated with the variance in another. Process
research focuses on the actual process, generally over
time, of implementing a system. This researcher is
generally looking for a combination of events and
actions that are necessary for success. The process
researcher is likely to be a participant observer of the
development process and adopt a qualitative design.
Researchers who favor factor studies site the rigor
of the research designs and the fact that they generally
collect data from a large number of users. Statistical
techniques make it possible to generalize to other
settings, though the vast majority of factor studies
is cross-sectional. Process researchers argue that
implementation occurs over time and that one has
to observe the process to understand all of its nuances.
Most studies, whatever their approach, assume that
system use is voluntary, or that the researchers can
distinguish voluntary from required use. Requiring
one to use a system reduces variance in use and makes
it difcult to adopt usage as a dependent variable
measuring implementation success.

some functions may be mandatory, for example, a

stock broker may only be able to obtain a real-time
quote through his or her workstation. However, workstation usage also may have a highly voluntary component; the stock broker chooses whether to use
analytic functions or not. The actual usage of a workstation is very hard to monitor when the device is
turned on all day on the user's desk. While a program
could count keystrokes and identify functions being
used, the results may be misleading. A broker may
have stock tickers running across the screen all day. Is
the broker looking at it or is it background?
In this environment with everyone having a workstation, why should management be concerned about
implementation? First, current information technology represent a very large investment, more so than
systems in earlier eras. The Department of Commerce
has been estimated that 45% of capital investment in
the U.S. is for information technology. A few years
ago NationsBank had a $500 million software budget
and an overall IT expenditure of $1.9 billion a year.1
When it and Bank of America merged, the combined
bank had an initial annual technology budget of $4
billion.
If management makes this kind of investment, it
must believe the systems that result will be of value to
the rm when users work with them. Systems like the
DSS at Frito-Lay, communications-oriented technology at VeriFone, and groupware at a number of rms
suggests that there is a large component of voluntary
use for some of the functions found on a typical
workstation. Successful implementation and user
acceptance of this technology should be a question
of great importance to senior management and to IT
management in the rm. We feel that the answer to the
rst question in the paper is that implementation is an
important issue in today's technological environment.
2.1. Models of implementation and acceptance

2. The contemporary environment

Early research on implementation can be characterized as empirically-driven. Factor researchers proposed descriptive models of implementation,
including variables other researchers had found to
be related to success, and usually adding new variables

The IT environment today is characterized by networked, multifunctional workstations. The use of

1
Comments by Hugh McColl, NationsBank's chairman at an
NYU CEO Forum, 4 July 1997.

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

as well [12,17]. Process researchers often employed a

model of the consultation process from the organization development literature as a framework for their
observations and analysis. None of these models had a
strong theoretical base.
Davis [4] proposed the technology acceptance
model (TAM) for implementation research, which is
based on the theory of reasoned action [5]. This model
is a variance or factor model and it contains many of
the variables in past implementation research. However, the model is more grounded in theory than the
past descriptive models of implementation. Davis'
original model includes two perceptions, ease of
use and usefulness. These perceptions predict attitudes
toward using a system. Attitudes predict intentions to
use, and intentions predict actual usage. In various
tests of the model, Davis and others have recommended using a simplied version of the model in
which the two perceptions predict either intentions or
use, dropping attitudes, which have proven to be weak
predictors of either intentions or usage. In most cases,
researchers have been able to measure only intentions
or usage, so one or the other has been the dependent
variable.
Davis' results supported his model (1989); several
other studies also provide evidence for TAM. Davis,
Bagozzi and Warshaw compared a model based on the
theory of reasoned action with TAM and found mixed
results for both models, though there was support for
the perceptions and their positive relationship with
intentions to use a system (1989). Mathieson [14] also
compared TAM with the theory of planned behavior
(TPB) and found that both models predicted intention
to use well, but that TAM was slightly better from an
empirical view.
Taylor and Todd [20] looked at TAM and the TPB in
a longitudinal study of a resource center; they concluded that a decomposed TPB provided more insights
than TAM, though TAM received support from their
data. In another study [21], these same authors found
that TAM, modied to include subjective norms and
perceived behavioral control, performed well in predicting acceptance for various users. Straub et al. [18]
used TAM to compare self-report and computer monitored voice mail usage in a eld setting. Szajna [19]
found that a revised TAM, dropping attitudes from the
model and making a slight change for pre versus postimplementation, predicted usage. She recommended

121

adding an experience component as an extension of

the model and suggested that measures of actual usage
may work better than self-report measures, at least
when looking at the acceptance of e-mail.
TAM has primarily been tested in experimental or
quasi-experimental studies with students as the subjects rather than in the eld with business professionals. The technology involved has generally
included a single application or a few packages like
a word processor and presentation application. There
is one study of broker workstations running many
applications simultaneously using TAM in a crosssectional design [13]. However, for the most part,
there has been little research on implementation and
acceptance featuring modern technology.
2.2. Extending TAM
Researchers have tested TAM in laboratory settings
more often than the eld. We believe that the greater
richness and complexity of a eld setting requires
consideration of more variables than are included in
TAM. What theoretical justication is there for making additions to TAM, and what variables might we
use to extend the model to be used in eld studies of
modern technology? Davis has indicated that a class of
`external' variables inuences perceived usefulness
and perceived ease of use. He sees additional variables
acting through these perceptions to inuence usage.
Past research on implementation [12], especially
related to technology acceptance by brokers [11] also
offers guidelines for extending the model.
This reasoning suggests that the following variables
might be added to TAM in a eld setting involving
stock brokers using a multifunctional workstation, as
shown in Fig. 1: social norms; broker strategy; perceived system quality; broker performance.
2.2.1. Broker strategy
While the tasks of all brokers are similar, there are
different ways to approach the job (Lucas, 1979). A
broker may feel she can best succeed by doing
research for her clients to encourage them to trade.
This broker would be likely to use the features of a
workstation that provided access to research information. Another strategy is to develop social ties; most of
the time of a broker following this strategy would be
spent at meals, visiting clients and at social functions.

122

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

Fig. 1. Research model.

The workstation would be less relevant for this broker.

The broker's strategy affects the degree to which the
technology is relevant, and should be a determinant of
perceptions of the workstation. The inclusion of broker strategy is consistent with the ideas expressed in
the task-technology t literature [7].
2.2.2. System quality
The quality of a system is related to whether a user
can accomplish his or her tasks with a workstation. A
system that omits important functions or which offers
irrelevant services will not satisfy the user's needs.
System quality is a variable that IS managers should
be able to inuence through design and operating
practices.
2.2.3. Performance
There is some evidence that technology may best t
the needs of the high performing user, or in some
instances, the needs of a poorly performing worker
[10]. If the system supports the worker in solving
problems, its use should be associated with high levels
of performance. For example, a broker might use the
workstation with a client to develop recommendations
on stocks to buy and sell. On the other hand, a person
with poor performance may use an information system
to diagnose problems and develop a strategy for
improving performance. This broker might use analytic tools provided by a workstation to understand
better the performance of different client portfolios.
2.2.4. Social norms
The theory of reasoned action includes social norms
in predicting behavior. ``According to our theory, the

more a person perceives that others who are important

to him think he should perform a behavior, the more he
will intend to do so [6], p. 57.'' Participants are not
subject to the inuence of senior management in a
laboratory setting, and may not perceive much inuence from their peers. It should be noted that in one
empirical study of participation, Hartwick and Barki
[8] incorporated social norms and found only weak
associations with other variables. In the current study,
we hope to show that norms are important in a eld
setting in which TAM provides part of the underlying
research model.
2.3. The research model
The model in Fig. 1 follows the guidelines of Davis
and Fishbein and Azjen. In particular, two external
variables are hypothesized to inuence perceived
usefulness and perceived ease of use, strategy and
system quality. Why are we interested in predicting
these perceptions? The items in the scales for perceived ease of use and perceived usefulness are difcult to inuence directly, for example, a typical item in
these scales is `the workstation enhances my effectiveness'. If one believes in the causal implications of
TAM, then a way to increase acceptance is to improve
the user's perceptions of ease of use and usefulness.
We believe that broker strategy and system quality
inuence these two perceptions.
As described above, individual brokers have
different approaches to their tasks; technology
which ts a broker's tasks should be perceived as
more useful and easier to use than technology
which exhibits a poor t. A broker whose strategy

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

is supported and reinforced by the workstation

should have favorable perceptions of it. The broker
for whom the workstation is largely irrelevant is
likely to have no or possibly unfavorable perceptions
of the technology.
System quality is a rating of how easy it is to
retrieve data using the workstation, its response time,
accuracy, etc. It could be argued that these items are
perceptions just as are ease of use and usefulness.
However, management does control some of the characteristics of the technology which lead to these
perceptions. Assuming causal linkages, management
might be able to inuence perceptions of ease of use
and usefulness, and then user acceptance, by increasing the quality of the workstation's features, for
example, by improving response times.
Following the logic of Fishbein and Azjen, we have
added social norms to the TAM as a direct inuence on
usage. Employees in a company setting are inuenced
by superiors, colleagues and clients. While norms may
be relatively unimportant in research involving students, they are likely to have a signicant inuence on
technology acceptance in an organization like a
brokerage rm.
Based on earlier ndings on performance [10], the
research model hypothesizes a positive relationship
between usage and performance when the workstation
supports the successful broker. If a broker with low
performance uses the system to help nd problems,
one would nd a negative relationship between use
and performance.
The research model in Fig. 1 extends TAM to t the
setting of an ongoing rm. It includes two external
variables of broker strategy and system quality. We
feel these variables inuence perceived ease of use and
perceived usefulness, and that they are important in
designing technology. For example, the broker strategy variable suggests to designers that they consider
the t between the user's task and the technology.
Designers and managers have some control over the
quality of a system, and the model proposes that
quality affects acceptance through its inuence on
perceptions. The inclusion of social norms is
important in a work environment where a variety of
actors create norms that inuence potential users.
Finally, past research has shown that system usage,
one measure of acceptance, is related to user performance.

123

3. The study
We collected data in the fall of 1993 and again in the
fall of 1995 to test the model in Fig. 1 from the
institutional brokerage group of a major stock brokerage rm. The 1993 data (time 1) was collected within
6 months of the installation of a new workstation for
all brokers in the group. The workstation provided a
number of functions including the ability to obtain
stock quotations, to communicate via e-mail, and to
analyze securities data. We developed a survey instrument based on past research, and three research assistants administered the survey in a 2030 minute phone
interview. A total of 41 out of 59 brokers in four
locations completed the interviews. Two years later
(time 2) we again collected data from the brokers who
had responded at time 1. We were surprised to nd that
only 25 brokers from the original 41 remained in their
same position at the time 2 survey; this is a high
mortality rate, particularly for a job described as being
very attractive with low turnover.
3.1. Data and variables
Table 1 provides the denition of the data included
in the study. We used scales from past research or new
scales which we developed after interviewing brokerage rm managers and brokers. Because brokers' time
is very valuable, the length of the phone call was
limited. Therefore, we dropped some items from past
scales in order to t interview time constraints. The
individual items in the scales may be found in the
Appendix A.
The four strategy variables came from these interviews and are single-item questions. There was virtually no correlation among the strategies and we
consider them to be independent of each other.
The system quality scale of ve items is based on a
number of past studies [12]. The reliability of these
scales is in the 0.60.75 range, which is somewhat
lower than the scale reliabilities in similar studies in
the past.
Perceived usefulness and ease of use have been
adapted from Davis' scales used in his studies of TAM
[3]. We developed an instrument to measure social
norms based on Ref. [5] and pretested it on a group of
MBA students. Based on these results, we constructed
the individual items for social norms shown in the

124

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

Table 1
Description of variables in the study
Variable name

Description

Number
responses

a-reliability

Mean

Standard
deviation

STGMEET
STGMTCH
STGPORT
STGRES
SYSQUAL
PERUSE
PEREOU
NORMS

Strategy: meet with clients

Strategy: find securities to match clients' strategy
Strategy: broker acts as portfolio manager for clients
Strategy: broker does research for clients
System quality evaluated by five items
Perceived usefulness based on two items
Perceived ease of use based on three items
Support in using workstation from others,
managers, senior managers and clients
Intent to use new features
Use from two items
Broker sales for 1992a
Broker sales for 1993a

41
41
41
41
39
41
41
39

na
na
na
na
0.63
0.69
0.63
0.82

4.44
4.90
4.12
3.85
3.71
4.44
3.92
3.81

0.92
0.30
0.90
0.94
0.62
0.60
0.60
0.79

38
41
41
41

0.75
0.73
na
na

2.91
4.50
17373
29519

0.98
0.58
23809
22645

INTENT
USAGE
SALES92
SALES93
a

Disguised data from firm records.

Appendix A; a scale of these items forms the variable,

Norms, in the analysis.
Intent is measured by a three-item scale asking
about the user's intentions to use features that were
planned for the system in 1993. We did not include
intent in 1995 because the workstation had been in use
for so long at this point. Usage is a two-item selfreport measure. Because it is difcult to observe
usage, and because monitoring usage may not be very
accurate, we used the self-report measure.
The brokerage rm provided a listing of revenues
by account by broker. We compiled these data into
broker sales revenue for 1992 and 1993. Unfortunately, the rm changed its performance reporting
system between time 1 and time 2, so that we were
unable to obtain comparable performance data for the
1995 survey (We also could not restate the 1992 and
1993 data to t the new performance measurement
algorithm.). In the analysis, we use the natural log of
sales to provide a better statistical distribution.

change in the dependent variable in standard deviation

units from a change of one standard deviation unit in
the independent variable. Eqs. (1) and (2) present the
results of predicting perceived usefulness and perceived ease of use with the four broker strategy
variables and system quality. In these equations, we
had too many independent variables for the number of
observations, so we used a stepwise algorithm to select
independent variables to enter. System quality is a
predictor of both perceptions. A strategy of meeting
with the client is negatively related to perceived ease
of use; the broker who follows a strategy of meeting
with clients nds the workstation less easy to use than
his or her colleagues. It is likely that the workstation
provides less support for meeting with clients than for
other broker strategies; it is not easy to use or easily
available in a client setting. It is also possible that
brokers who prefer to meet with clients are more
oriented toward social contact than working with
technology.

3.2. Results time 1

PERUSE 0:49 SYSQUAL

Following the practice of the majority of past

studies of TAM, we used regression analysis to analyze the results (The limited size of the data set
prevented us from using path anlaysis or structural
equation modeling.). Each of the equations is in
standardized form without the intercept and with beta
weights for coefcients; the beta weight represents the

3:42

F 11:72 n 39

Stepwise



R2 0:22

p  0:10;  p  0:05;

p  0:01

(1)

0:34 SYSQUAL
PEREOU 0:36 STGMEET

2:48

R 0:25 F 7:30



Stepwise

2:35

n 39

(2)

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

125

Eq. (3) predicts usage for the original TAM. Here only
perceived usefulness is significant, and the two variables explain 15% of the variance in usage. The
relationships posited in TAM are supported by the
data, but the results are weak.

1992 performance and usage of the workstation. Both

variables are positively related to 1993 performance,
and workstation usage is surprisingly significant
(Frequently, lagged performance dominates other
variables in predicting current performance.).

0:21 PEREOU
USAGETAM 0:31 PEURSE


LNSALES93 0:87 LNSALES92

0:16 USAGE



1:34

2:01

R 0:15 F 4:50



n 39

14:12

(3)

Eq. (4) predicts usage for the extended model of Fig. 1;

it adds social norms and the log of 1992 sales. Only the
log of 1992 sales is significant statistically, and at the
0.10 level; higher performers use the system more.
The four variables in this equation are weakly dominated by sales. The results explain a little more of the
variance than Eq. (3), 20 versus 15%.
USAGE 0:14 NORMS 0:25 PERUSE
0:77

1:32

0:16 PEREOU 0:29 LNSALES92


0:97

1:92

R 0:20 F 3:34



n 39

(4)

Eqs. (5) and (6) repeat the analysis above for intentions to use the system rather than self-reported usage.
In Eq. (5) for the original TAM, we see the same
pattern as Eq. (3); perceived usefulness is significant
and both perceptions are positively related to Intent.
The amount of variance explained is only 7%, which
means the model does not shed much light on intentions.
0:07 PEREOU
INTENTTAM 0:31 PERUSE

0:43

1:83

R 0:07 F 2:34 n 38

(5)

Eq. (6) predicts intentions to use the system for the

extended model. Social norms are the dominant independent variable; they are significant at the 0.05 level.
Norms do not enter into reports of current usage, but
they are associated with future intentions. Norms may
play more of a role in shaping future behavior than
current work practices.
0:7 PERUSE
INTENT 0:44 NORMS

2:23

0:37

0:01 PEREOU 0:05 LNSALES92

0:05

R2 1:5 F 2:58 n 36

0:31

(6)

Eq. (7) predicts 1993 performance as a function of

R 0:86 F 124:66



n 41

2:60

(7)

3.3. Results time 2

Table 2 presents the results of a matched samples
t-test of the 25 `survivors' for the variables we were
able to measure at both time 1 and time 2. All results
but two are remarkably similar and changed little in
the 2-year period between surveys. The consistent
values hold in spite of the high sample mortality rate
of 39% from time 1 to time 2.
User ratings of systems quality declined signicantly as did perceptions of ease of use. Company
management indicated that they were making continual improvements and upgrades to the system,
which should have resulted in improvements in
quality. It is possible that at time 1, brokers rated
the new system favorably in comparison to the
old system it had just replaced. By time 2 the old
system was long forgotten and the relevant reference
system was the broker workstation, itself. Management may nd that providing the newest technology raises user expectations and that users do
not remain satised long with current technology.
Professionals have colleagues in other organizations
and may use their systems as a point of comparison.
Software vendors do their best to obsolete their
products with new releases and versions in order
to maintain sales. We may nd that users expect
the same improvements from internal, proprietary
systems.
4. Discussion
Considering the rst question raised in the paper, is
implementation research relevant today, we believe
that the implementation of information technology
and its acceptance is even more important today
than in the past. Organizations rely on technology

126

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

Table 2
Time 2 vs. Time 1
Variable
name

Description

t2

Time 1
mean

Time 2
mean

Difference

Matched
t-statistic

SYSQUAL
STGMEET
STGMTCH
STGPORT
STGRES
PERUSE
PEREOU
NORMS

System quality evaluated by five items

Strategy: meet with clients
Strategy: find securities to match clients' strategy
Strategy: broker acts as portfolio manager for clients
Strategy: broker does research for clients
Perceived usefulness based on two items
Perceived ease of use based on three items
Support in using workstation from others, managers,
senior managers and clients
Use from two items

24
25
25
25
25
25
25
24

0.75
na
na
na
na
0.68
0.42a
0.75

3.6
4.4
4.9
4.0
3.9
4.4
3.9
3.8

3.3
4.4
4.8
4.1
3.8
4.4
3.5b
3.7

0.3
0
0.1
0.1
0.1
0
0.4
0.1

2.30**
0
1.44
0.57
0.57
0
3.36**
0.94

25

0.85

4.5

4.4

0.1

0.77

USAGE
a

The decrease in reliability from t1 (0.63) to t2 (0.42), for PEREOU is troubling. While some of the decrease might be attributed to a
smaller sample size at time 2, reliabilites for other variables at time 2 did not show similar declines. The standard deviation for PEREOU is
very similar for time 1 vs. time 2. The Pearson correlation for PEREOU at time 1 with PEREOU at time 2 is 0.63 which is a measure of test
retest reliability. This number is comparable to other time 12 correlations for individual variables. When we examined the items in the scale,
we found that by eliminating `received sufficient training' the a rose to 0.7. Please see the discussion of time 2 results in the text for further
information.
b
Using the more reliable version of this scale, eliminating the item `received sufficient training', results in a time 2 mean of 3.6, a
difference of 0.3 and a t-value of 2.51**. PEREOU and SYSQUAL still remain the only two variables which are significantly different
between time 1 and time 2, and both of them are less favorable in 1995 than 1993.

to operate their businesses, provide reservoirs of

knowledge and organizational intelligence, and structure their rms. Organizations are rapidly adopting
workstation and network technology platforms. Management has to be concerned with the acceptance of
this technology if it is to obtain a return from the rm's
investment in IT.
The second question in our research was how well
models of implementation predict acceptance in
today's workstation and networked environment.
The answer to this question is mixed. TAM performs
well in quasi-experimental research on students. In
this eld setting, its variables tend to be dominated by
other variables like social norms and performance.
However, this dominance is not strong, and neither
TAM in its original or extended versions explains a
large amount of variance in workstation use.
Our results, however, have to be treated with caution due to the limitations of the sample size. Testing
models like TAM in a eld study is difcult because
one needs to locate a signicant number of users who
work with the same system. As we learned, a longitudinal study is also very difcult due to subject
mortality, even with a job that is believed to have
low turnover.

A problem with eld studies may be in separating

mandatory from voluntary use; almost all models of
implementation attempt to predict voluntary system
usage. Mandatory use makes all of the other variables
in a model moot. Some functions of the workstation
have to be used in order to do one's job. It is the
broker's only source of stock market quotations, for
example. However, brokers have been getting these
quotations from some kind of computer device for
decades, and it is not clear that they even think about
this function when evaluating the capabilities of the
workstation.
Intentions to use the system refers to new features
whose usage is voluntary; here social norms in the
extended TAM are highly signicant, suggesting that
extending the model is important in a eld setting.
System quality is associated with perceived ease of use
and perceived usefulness. If one believes the causal
implications of the model, these results suggest that
management should concentrate on the quality of the
system and on establishing norms that favor system
use in order to encourage acceptance.
While TAM and its extensions do not predict
large amounts of variance in usage, they are relatively
simple models that can be applied to predict accep-

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

tance and to intervene if necessary. The variables in

these models appear to be quite stable, and the changes
from time 1 to time 2 suggest that they can be used in
an ongoing evaluation of a system. For example, a rm
implementing technology might track user acceptance
after adding new functions to a system, offering
advanced training courses, or releasing an entirely
new version of the application. In periods of rapidly
advancing technology, continued assessment of systems like the multifunctional workstation in the study
are very important in taking advantage of technology
investments.
5. Conclusions
The results of this study show that TAM and its
extensions are weak predictors of the acceptance of a
broker workstation. In future work, we recommend
including the variables in these models because they
have shown an association with acceptance in both
laboratory and eld studies. An important question is
what other models and variables are needed to better
explain the phenomenon of IT implementation and
acceptance? Possibilities include ethnographic studies
such as those by Barley [1] on roles and networks and
Orlikowski [15], which offers a situated change perspective on organization change over time. Researchers might also want to examine managerial tasks
related to adoption [2] and the role of technological
champions in implementation [9].
Based on this study, we feel that variance models
like TAM, combined with qualitative research, offer
the best opportunity for understanding the implementation of modern information technology. These two
approaches to research compliment each other and
their combination will provide the most insights possible into the complexities of implementation in an
environment of workstations, LANs, intranets and the
Internet.
It may be necessary to develop variables to extend
TAM in each setting. For example, there may be no
analogy to broker strategy in a manufacturing company. Our suggestion is to look for important external
variables, include social norms, and seek additional
explanatory variables for technology acceptance in
building models. Combining variance and ethnographic research designs should also contribute a great

127

deal to our understanding of acceptance. The implementation and acceptance of information technology
is and will remain a key determinant of the return the
organization receives from its IT investments, and
models of acceptance and implementation should lead
to more successful IT initiatives.
Appendix A. Survey items
All responses are 15, strongly agree to strongly
disagree.
STGMEET The successful broker has to spend considerable time meeting with clients
STGMTCH The successful broker finds stocks and
opportunities that match the client's
strategy
STGPORT The successful broker functions as an
unpaid portfolio manager
STGRES
The successful broker spends significant
time each day doing research for clients
SYSQUAL It is very easy to retrieve and organize
data using the workstation
The workstation's response time is very
fast
The workstation's data is very accurate
(the data are correct)
The workstation is very reliable (does
not go down)
Using the workstation is much better
than using the previous [system's name]
PERUSE
Using the workstation improves my
performance
Using the workstation enhances my
effectiveness
PEREOU
I find the workstation easy to use
I have received sufficient training to use
the workstation effectively
I find it easy to get the workstation to do
what I want it to do
NORMS
Other institutional brokers whom I
respect strongly support my using the
Institutional Broker Workstation
My immediate manager strongly supports my using the workstation
[Firm's name] senior management
strongly supports my using the workstation

128

INTENT

USAGE

H.C. Lucas Jr., V. Spitler / Information & Management 38 (2000) 119128

My most important clients strongly support my using the workstation

I intend to make frequent use of electronic mail on the workstation
I intend to make frequent use of the
office automation tools like spreadsheets and word processing on the
workstation
I intend to make frequent use of the
______functions like data retrieval and
transactions status reports on the workstation
At the present time, I consider myself to
be an extremely frequent user of the
workstation
I currently use the workstation continuously throughout the day

[14]
[15]
[16]

[17]

[18]
[19]
[20]

References
[1] S. Barley, The alignment of technology and structure through
roles and networks, Administrative Science Quarterly 35 (1),
1990, pp. 61103.
[2] R. Cooper, R. Zmud, Information technology implementation
research: a technological diffusion approach, Management
Science 36 (2), 1990, pp. 123139.
[3] F. Davis, Perceived usefulness, perceived ease of use, and
user acceptance of information technology, MIS Quarterly 13
(3), 1989, pp. 319340.
[4] F. Davis, R. Bagozzi, P. Warshaw, User acceptance of
computer technology: a comparison of two theoretical
models, Management Science 35 (8), 1989, pp. 9821003.
[5] M. Fishbein, I. Ajzen, Belief, Attitude, Intention and
Behavior: An Introduction to Theory and Research, Addison-Wesley, Reading, MA, 1975.
[6] M. Fishbein, I. Ajzen, Understanding Attitude and Predicting
Social Behavior, Prentice-Hall, Englewod Cliffs, NJ, 1980.
[7] D. Goodhue, Understanding user evaluations of information
systems, Management Science 41 (12), 1995, pp. 18271844.
[8] J. Hartwick, H. Barki, Explaining the role of user participation in information system use, Management Science 40 (4),
1994, pp. 440465.
[9] J. Howell, C. Higgins, Champions of technological innovations,
Administrative Science Quarterly 35 (2), 1990, pp. 317341.
[10] H. Lucas, Why Information Systems Fail, Columbia University Press, New York, 1975.
[11] H.C. Lucas, Jr., The implementation of an Operations
Research Model in the Brokerage Industry, TIMS Studies in
the Management Sciences, No. 13 (1979), pp. 139154.
[12] H. Lucas, Jr., R. Schultz, M. Ginzberg, Information Systems
Implementation: Testing a Structural Model, Ablex, 1990.
[13] H. Lucas, V. Spitler, Technology use and performance: a field

[21]

study of broker workstations, Decision Sciences 30 (2), 1999,

pp. 291311.
K. Mathieson, Predicting user intentions: comparing the technology acceptance model with the theory of planned behavior,
Information Systems Research 2 (3), 1991, pp. 173191.
W. Orlikowski, Improvising organizational transformation
over time: a situated change perspective, Information Systems
Research 7 (1), 1996, pp. 6391.
R. Sabherwal, D. Robey, An empirical taxonomy of
implemention processes based on sequences of events in
information system development, Organization Science 4 (4),
1993, pp. 548576.
T. Shaw, S. Jarvenpaa, Process models in information
systems, In: A.S. Lee, J. Liebenau, J. DeGross, (Eds.),
Information Systems and Qualitative Research, Chapman and
Hall, London, pp. 70100.
D. Straub, M. Limayem, E. Karahanna-Evaristo, Measuring
system usage: implications for IS theory testing, Management
Science 41 (8), 1995, pp. 13281342.
B. Szajna, Empirical evaluation of the revised technology acceptance model, Management Science 42 (1), 1996, pp. 8592.
S. Taylor, P. Todd, Understanding information technology
usage: a test of competing models, Information Systems
Research 6 (2), 1995, pp. 144176.
S. Taylor, P. Todd, Assessing it usage: the role of prior
experience, MIS Quarterly 19 (4), 1995a, pp. 561570.

Henry C. Lucas, Jr. has been on the

faculty of the Stern School of Business at
New York University. Beginning in
August of 2000, he will be the Robert
H. Smith Professor of Information Systems at the Robert H. Smith School of
Business at the University of Maryland.
Professor Lucas has written several
books and articles about information
systems, publishing articles in Management Science, Information Systems Research, MIS Quarterly and
other journals. He is the editor-in-chief of the Association for
Information Systems Electronic Journals.
Valerie Spitler is a candidate for the Ph.D.
degree in the Information Systems Department at New York University's Stern
School of Business. In addition to research
on implementation, she conducts research
on individual use of information technologies in practice, using situated learning
theory and interpretive research methods.
She has been published in Decision
Sciences, and in several conference proceedings, including the Proceedings of the International Federation
for Information Processing, Working Group 8.2 and the MIS Quarterly
workshop proceedings on new organizational roles of information
technology in the Information Age. Ms. Spitler holds an M.B.A. from
Insead (France) and a B.S. from the Wharton School and has worked
for firms in financial services, software development and advertising.