A Theoretical Extension of the Technology

Acceptance Model: Four Longitudinal

Field Studies
Viswanath Venkatesh Fred D. Davis
Robert H. Smith School of Business, Van Munching Hall, University of Maryland, College Park, Maryland 20742
Sam M. Walton College of Business Administration, University of Arkansas, Fayetteville, Arkansas 72701
vvenkate@mbs.umd.edu fdavis@walton.uark.edu

he present research develops and tests a theoretical extension of the Technology

Acceptance Model (TAM) that explains perceived usefulness and usage intentions in
terms of social influence and cognitive instrumental processes. The extended model, referred
to as TAM2, was tested using longitudinal data collected regarding four different systems at
four organizations (N 156), two involving voluntary usage and two involving mandatory
usage. Model constructs were measured at three points in time at each organization:
preimplementation, one month postimplementation, and three months postimplementation.
The extended model was strongly supported for all four organizations at all three points of
measurement, accounting for 40% 60% of the variance in usefulness perceptions and
34%52% of the variance in usage intentions. Both social influence processes (subjective norm,
voluntariness, and image) and cognitive instrumental processes (job relevance, output quality,
result demonstrability, and perceived ease of use) significantly influenced user acceptance.
These findings advance theory and contribute to the foundation for future research aimed at
improving our understanding of user adoption behavior.
(Adoption of Information Technology; Technology Acceptance Model; Social Influence; Perceived
Usefulness)

Introduction
Information technology adoption and use in the workplace remains a central concern of information systems research and practice. Despite impressive advances in hardware and software capabilities, the
troubling problem of underutilized systems continue.
Low usage of installed systems has been identified as
a major factor underlying the productivity paradox
surrounding lackluster returns from organizational
investments in information technology (Sichel 1997).
Understanding and creating the conditions under
which information systems will be embraced by the
human organization remains a high-priority research
issue.
Management Science 2000 INFORMS
Vol. 46, No. 2, February 2000 pp. 186 204

Significant progress has been made over the last

decade in explaining and predicting user acceptance
of information technology at work. In particular,
substantial theoretical and empirical support has
accumulated in favor of the Technology Acceptance
Model (TAM) (Davis 1989, Davis et al. 1989). Numerous empirical studies have found that TAM
consistently explains a substantial proportion of the
variance (typically about 40%) in usage intentions
and behavior, and that TAM compares favorably
with alternative models such as the Theory of
Reasoned Action (TRA) and the Theory of Planned
Behavior (TPB) (see Venkatesh 1999 for recent review). TAM theorizes that an individuals behav0025-1909/00/4602/0186$05.00
1526-5501 electronic ISSN

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

ioral intention to use a system is determined by two

beliefs: perceived usefulness, defined as the extent
to which a person believes that using the system
will enhance his or her job performance, and perceived ease of use, defined as the extent to which a
person believes that using the system will be free of
effort. TAM theorizes that the effects of external
variables (e.g., system characteristics, development
process, training) on intention to use are mediated
by perceived usefulness and perceived ease of use.
According to TAM, perceived usefulness is also
influenced by perceived ease of use because, other
things being equal, the easier the system is to use
the more useful it can be. As of January 2000, the
Institute for Scientific Informations Social Science
Citation Index listed 424 journal citations to the two
journal articles that introduced TAM (i.e., Davis
1989, Davis et al. 1989). In 10 years, TAM has
become well-established as a robust, powerful, and
parsimonious model for predicting user acceptance.
Across the many empirical tests of TAM, perceived
usefulness has consistently been a strong determinant
of usage intentions, with standardized regression coefficients typically around 0.6. Since perceived usefulness is such a fundamental driver of usage intentions,
it is important to understand the determinants of this
construct and how their influence changes over time
with increasing experience using the system. Perceived ease of use, TAMs other direct determinant of
intention, has exhibited a less consistent effect on
intention across studies. Whereas some research has
been done to model the determinants of perceived
ease of use (Venkatesh and Davis 1996), the determinants of perceived usefulness have been relatively
overlooked. A better understanding of the determinants of perceived usefulness would enable us to
design organizational interventions that would increase user acceptance and usage of new systems.
Therefore, the goal of the present research is to extend
TAM to include additional key determinants of TAMs
perceived usefulness and usage intention constructs,
and to understand how the effects of these determinants change with increasing user experience over
time with the target system.

Management Science/Vol. 46, No. 2, February 2000

Theory and Hypotheses

Figure 1 shows the proposed model, referred to as
TAM2. Using TAM as the starting point, TAM2 incorporates additional theoretical constructs spanning social influence processes (subjective norm, voluntariness, and image) and cognitive instrumental processes
(job relevance, output quality, result demonstrability,
and perceived ease of use). Below we define each of
these constructs and develop the theoretical rationale
for the causal relationships of the model.
Social Influence Processes
TAM2 reflects the impacts of three interrelated social
forces impinging on an individual facing the opportunity to adopt or reject a new system: subjective
norm, voluntariness, and image.
Subjective Norm. Consistent with TRA, which
was a key theoretical underpinning for the original
development of TAM, we tap into social influences via
subjective norm, defined as a persons perception that
most people who are important to him think he
should or should not perform the behavior in question (Fishbein and Ajzen 1975, p. 302). Subjective
norm is included as a direct determinant of behavioral
intention in TRA (Fishbein and Ajzen 1975) and the
subsequent TPB (Ajzen 1991). The rationale for a
direct effect of subjective norm on intention is that
people may choose to perform a behavior, even if they
are not themselves favorable toward the behavior or
its consequences, if they believe one or more important referents think they should, and they are sufficiently motivated to comply with the referents. User
acceptance research examining the direct effect of
subjective norm on intention has yielded mixed results. Mathieson (1991) found no significant effect of
subjective norm on intention, whereas Taylor and
Todd (1995) did find a significant effect. In their
empirical comparison of TAM and TRA, Davis et al.
(1989) found that subjective norm had no significant
effect on intentions over and above perceived usefulness and ease of use, and they omitted it from the
original TAM, but they did acknowledge the need for
additional research to investigate the conditions and
mechanisms governing the impact of social influences
on usage behavior (p. 999).

187

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Figure 1

Proposed TAM2Extension of the Technology Acceptance Model

Voluntariness and Compliance with Social Influence. A contingency underlying the mixed findings
regarding subjective norm was identified by Hartwick
and Barki (1994): After separating their respondents
into mandatory and voluntary usage contexts, they
found that subjective norm had a significant effect on
intention in mandatory settings but not in voluntary
settings. We refer to the causal mechanism underlying
this effect as compliance. In general, the direct compliance effect of subjective norm on intention is theorized to operate whenever an individual perceives
that a social actor wants him or her to perform a
specific behavior, and the social actor has the ability to
reward the behavior or punish nonbehavior (French
and Raven 1959, Kelman 1958, Warshaw 1980). TAM2
theorizes that, in a computer usage context, the direct
compliance-based effect of subjective norm on intention over and above perceived usefulness and perceived ease of use will occur in mandatory, but not
voluntary, system usage settings. To distinguish between mandatory and voluntary usage settings, our
model posits voluntariness as a moderating variable,

188

defined as the extent to which potential adopters

perceive the adoption decision to be non-mandatory
(Agarwal and Prasad 1997, Hartwick and Barki 1994,
Moore and Benbasat 1991). As Hartwick and Barki
(1994) found, even when users perceive system use to
be organizationally mandated, usage intentions vary
because some users are unwilling to comply with such
mandates.
Hypothesis 1a. Subjective norm will have a positive
direct effect on intention to use when system use is
perceived to be mandatory.
Hypothesis 1b. Subjective norm will have no significant direct effect on intention to use when system use is
perceived to be voluntary.
Hypothesis 1c. Voluntariness will moderate the effect
of subjective norm on intention to use.
Internalization of Social Influence. Whereas the
direct relationship between subjective norm and intention in TRA and TPB is based on compliance, TAM2
encompasses two additional theoretical mechanisms

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

by which subjective norm can influence intention

indirectly through perceived usefulness: internalization and identification. Internalization (Kelman 1958,
Warshaw 1980) refers to the process by which, when
one perceives that an important referent thinks one
should use a system, one incorporates the referents
belief into ones own belief structure. Internalization is
equivalent to what Deutsch and Gerard (1955) refer to
as informational (in contrast to normative) social influences, defined as influence to accept information
from another as evidence about reality (p. 629). In the
present context, if a superior or co-worker suggests
that a particular system might be useful, a person may
come to believe that it actually is useful, and in turn
form an intention to use it. In French and Ravens
(1959) taxonomy, the basis of internalization is expert
power, where the target individual attributes expertise and credibility to the influencing agent (Kelman
1958). In the case of internalization, subjective norm
has an indirect effect on intention through perceived
usefulness, as opposed to a direct compliance effect on
intention. Research based on Salancik and Pfeffers
(1978) social information processing model is consistent with the proposed internalization effect (e.g., Fulk
et al. 1987, Rice and Aydin 1991). TAM2 theorizes that
internalization, unlike compliance, will occur whether
the context of system use is voluntary or mandatory.
That is, even when system use is organizationally
mandated, users perceptions about usefulness may
still increase in response to persuasive social information.
Hypothesis 2. Subjective norm will have a positive
direct effect on perceived usefulness.
Image and Social Influence. Individuals often respond to social normative influences to establish or
maintain a favorable image within a reference group
(Kelman 1958). Drawing from research on diffusion of
innovations, Moore and Benbasat (1991, p. 195) define
image as the degree to which use of an innovation is
perceived to enhance ones . . . status in ones social
system. TAM2 theorizes that subjective norm will
positively influence image because, if important members of a persons social group at work believe that he
or she should perform a behavior (e.g., using a sys-

Management Science/Vol. 46, No. 2, February 2000

tem), then performing it will tend to elevate his or her

standing within the group (Blau 1964, Kiesler and
Kiesler 1969, Pfeffer 1982). Kelman (1958) refers to this
source of social influence as identification and distinguishes it from compliance and internalization.
Viewed from the perspective of French and Ravens
(1959) taxonomy, the basis of identification is referent
power. In the typical work environment, with a high
degree of interdependence with other social actors in
carrying out ones duties, increased status within the
group is a basis of power and influence via processes
such as social exchange, coalition formation, and resource allocation (Blau 1964, Pfeffer 1981, 1982). As
Pfeffer (1982, p. 85) argues, by performing behaviors
that are consistent with group norms, an individual
achieves membership and the social support that
such membership affords as well as possible goal
attainment which can occur only through group action
or group membership. The increased power and
influence resulting from elevated status provides a
general basis for greater productivity. An individual
may thus perceive that using a system will lead to
improvements in his or her job performance (which is
the definition of perceived usefulness) indirectly due
to image enhancement, over and above any performance benefits directly attributable to system use.
This identification effect is captured in TAM2 by the
effect of subjective norm on image, coupled with the
effect of image on perceived usefulness. TAM2 theorizes that identification, like internalization but unlike
compliance, will occur whether the context of system
use is voluntary or mandatory.
Hypothesis 3a. Subjective norm will have a positive
effect on image.
Hypothesis 3b. Image will have a positive effect on
perceived usefulness.
Changes in Social Influence with Experience.
How might the effects of these social influence processes change with increasing experience using a target
system? There is theory and evidence to suggest that
the direct effect of subjective norm on intentions may
subside over time with increased system experience.
Hartwick and Barki (1994) found that, although subjective norm had a significant effect on intentions prior

189

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

to system development, the effect became nonsignificant three months after system implementation. Their
interpretation of this pattern is that, before a system is
developed, users knowledge and beliefs about a system are vague and ill-formed, and they must therefore rely more on the opinions of others as a basis for
their intentions (Hartwick and Barki 1994, pp. 458
459). After implementation, when more about the
systems strengths and weaknesses are known
through direct experience, the normative influence
subsides. A similar pattern of results was obtained by
Agarwal and Prasad (1997), who found that mandating the use of a system can increase initial system
utilization, enabling users to overcome the hurdle of
first-time use, but that such pressure seems to erode
over time (p. 575). Research outside the domain of
information systems also indicates that normative
pressure attenuates over time (Ram and Jung 1991).
Therefore, TAM2 theorizes that the direct effect of
subjective norm on intentions for mandatory usage
contexts will be strong prior to implementation and
during early usage, but will weaken over time as
increasing direct experience with a system provides a
growing basis for intentions toward ongoing use.
Similarly, we expect the effect of subjective norm on
perceived usefulness (internalization) to weaken over
time, since greater direct experience will furnish concrete sensory information (Doll and Ajzen 1992, Fazio
and Zanna 1981, Tybout and Scott 1983), supplanting
reliance on social cues as a basis for usefulness perceptions. In contrast, we do not expect the influence of
image on perceived usefulness (identification) to
weaken over time since status gains from system use
will continue as long as group norms continue to favor
usage of the target system.
Hypothesis 4a. The positive direct effect of subjective
norm on intention for mandatory systems will attenuate
with increased experience.
Hypothesis 4b. The positive direct effect of subjective
norm on perceived usefulness will attenuate with increased
experience for both mandatory and voluntary systems.
Cognitive Instrumental Processes
Beyond the social influence processes affecting perceived usefulness and usage intention discussed

190

above, we theorize four cognitive instrumental determinants of perceived usefulness: job relevance, output
quality, result demonstrability, and perceived ease of
use. As we argue below, people form perceived usefulness judgments in part by cognitively comparing
what a system is capable of doing with what they need
to get done in their job. To develop the theoretical
basis for these instrumental determinants, we draw
from recent developments in the reference paradigms
upon which TAMs perceived usefulness construct
was originally formulated. These theoretical underpinnings come from three main areas: work motivation theory (e.g., Vroom 1964), action theory from
social psychology (e.g., Fishbein and Ajzen 1975), and
task-contingent decision making from behavioral decision theory (e.g., Beach and Mitchell 1978). Recent
work in these areas has converged on a view of
behavior being driven by a mental representation that
links higher-level goals to specific actions that are
instrumental for achieving those goals.
Within work motivation theory, Locke and Latham
(1990) discuss task-specific plans, which are cognitive
mechanisms by which acts are selected, combined,
and sequenced in order to achieve goals. Task-specific
plans guide behavior through a conception-matching
process (see also Bandura 1986) linking instrumental
acts to goals. Within social psychology, action identification theory (Vallacher and Wegner 1987, p. 4)
posits an organized cognitive representation of action,
called the identity structure, which links lower-level
identities regarding specific actions to higher-level
identities indicating why the action is done or what its
effects or implications are. This identity structure is
seen as a fundamental mechanism by which people
cognitively regulate their behaviors in the furtherance
of higher-level goals (Vallacher and Kaufman 1996).
Within behavioral decision theory, recent work in
image theory (Beach and Mitchell 1996, 1998) has
embraced the concepts of a trajectory image, which is
a mental representation of adopted goals and the ideal
future state, and a strategic image, which is a mental
representation of possible action sequences that can
guide behavior toward the goal states of the trajectory
image. When deciding among alternative instrumental action sequences (called the adoption decision),

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

image theory posits two distinct decision stages. In the

first stage, a compatibility test is used to screen from
further consideration any options that are incompatible with ones decision standards. In the second stage,
a profitability test is performed to compare the acceptable options directly with one another to determine
the best option. For example, in searching for and
hiring a new employee, an employer would first
eliminate any candidates whose attributes violate
more than a threshold number of standards. If more
than one acceptable candidate remains in the choice
set after the compatibility test, then the profitability
test is used to select the best candidate on a compensatory basis. Although different with respect to the
pattern of information processing required, both the
compatibility and profitability tests are based on cognitively assessing the match between the characteristics of the trajectory image and the perceived consequences of alternative action plans.
These recent theoretical developments cutting
across action theory, work motivation theory, and
behavioral decision theory share the view that the
impetus for engaging in specific behaviors stems from
a mental representation linking instrumental behaviors to higher-level goals or purposes. Following this
line of reasoning, TAM2 theorizes that people use a
mental representation for assessing the match between important work goals and the consequences of
performing the act of using a system as a basis for
forming judgments about the use-performance contingency (i.e., perceived usefulness).
Job Relevance. One key component of the matching process discussed above is a potential users
judgment of job relevance, which we define as an
individuals perception regarding the degree to which
the target system is applicable to his or her job. In
other words, job relevance is a function of the importance within ones job of the set of tasks the system is
capable of supporting. Research in human-computer
interaction (Black et al. 1987, Norman 1987) has postulated similar goal-hierarchy models, though operating at more micro levels of analysis wherein higherlevel goals include tasks such as writing a document,
and lower-level actions are at the level of keystrokes
and mouse clicks. Kieras and Polson (1985) and Polson

Management Science/Vol. 46, No. 2, February 2000

(1987) argued that users possess distinct knowledge

about their job situation, which they can use as a basis
for determining what tasks can be performed with a
given system. The existence of well-defined knowledge structures regarding important job goals is supported by research from personnel psychology (e.g.,
Roberson 1989). We regard job relevance as a cognitive
judgment that exerts a direct effect on perceived
usefulness, distinct from social influence processes.
Empirically, user acceptance has been linked in other
research to variables similar to job relevance, including job-determined importance (Leonard-Barton and
Deschamps 1988), involvement (defined by Hartwick
and Barki (1994) as personal importance and relevance), task-technology fit (Goodhue 1995), and cognitive fit (Vessey 1991). We conceptualize perceptions
of job relevance to be part of a compatibility test
within the context of Beach and Mitchells (1996, 1998)
image theory since systems below a minimal threshold value of perceived job relevance would be
screened from further adoption consideration.
Hypothesis 5. Job relevance will have a positive effect
on perceived usefulness.
Output Quality. TAM2 posits that, over and above
considerations of what tasks a system is capable of
performing and the degree to which those tasks match
their job goals (job relevance), people will take into
consideration how well the system performs those
tasks, which we refer to as perceptions of output
quality. Empirically, the relationship between perceived output quality and perceived usefulness has
been shown before (Davis et al. 1992). We expect
output quality to be empirically distinct from, and to
explain significant unique variance in, perceived usefulness over and above job relevance because a different underlying judgmental process is involved. In the
context of image theory (Beach and Mitchell 1996,
1998), judgments of job relevance are more apt to take
the form of a compatibility test whereby systems that
are judged not to be job-relevant are eliminated from
ones choice set for further consideration. Judgments
of output quality, in contrast, are less likely to be used
for excluding options from consideration. Instead,
they are more apt to take the form of a profitability test

191

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

in which, given a choice set containing multiple relevant systems, one would be inclined to choose a
system that delivers the highest output quality.
Hypothesis 6. Output quality will have a positive
effect on perceived usefulness.
Result Demonstrability. Even effective systems
can fail to garner user acceptance if people have
difficulty attributing gains in their job performance
specifically to their use of the system. Therefore,
TAM2 theorizes that result demonstrability, defined by
Moore and Benbasat (1991, p. 203) as the tangibility
of the results of using the innovation, will directly
influence perceived usefulness. This implies that individuals can be expected to form more positive perceptions of the usefulness of a system if the covariation
between usage and positive results is readily discernable. Conversely, if a system produces effective jobrelevant results desired by a user, but does so in an
obscure fashion, users of the system are unlikely to
understand how useful such a system really is. Empirically, Agarwal and Prasad (1997) found a significant correlation between usage intentions and result
demonstrability. The relationship between result demonstrability and perceived usefulness is also consistent with the job characteristics model, which emphasizes knowledge of the actual results of work activities
as a key psychological state underlying work motivation (Hackman and Oldham 1976, Loher et al. 1985).
Hypothesis 7. Result demonstrability will have a positive effect on perceived usefulness.
Perceived Ease of Use. TAM2 retains perceived ease
of use from TAM as a direct determinant of perceived
usefulness (Davis et al. 1989), since, all else being
equal, the less effortful a system is to use, the more
using it can increase job performance. There is extensive empirical evidence accumulated over a decade
that perceived ease of use is significantly linked to
intention, both directly and indirectly via its impact on
perceived usefulness (e.g., Davis et al. 1989, Venkatesh
1999). Although beyond the scope of the present
extension of TAM, other research has begun to model
the antecedents of perceived ease of use. For example,
Venkatesh and Davis (1996) model perceived ease of

192

use as being anchored on ones general computer

self-efficacy and adjusted to account for a systems
objective usability via direct behavioral experience
using the target system. We note that the mechanisms
by which ease of use perceptions are theorized to form
and evolve are conceptually distinct from, and complementary with, the social influence and cognitive
instrumental processes within TAM2.
Hypothesis 8. Perceived ease of use will have a positive effect on perceived usefulness.
Changes in Cognitive Instrumental Influences
with Experience. Consistent with the theories of
mental representation discussed above, we expect
that, even over time, people will continue to rely on
the match between their job goals and the consequences of system usage (job relevance) as a basis for
their ongoing usefulness perceptions. Just as the role
of what a system does remains influential, we expect
that how well a system does what it does (output
quality) will remain a significant determinant of perceived usefulness over time. Similarly, we have no
basis to expect the effect of result demonstrability on
perceived usefulness to become either stronger or
weaker over time. Although there is no theoretical
account of such temporal shifts, the effect of perceived
ease of use on perceived usefulness has sometimes
been found to increase over time (e.g., Davis et al.
1989, Venkatesh and Davis 1996), whereas the direct
effect of perceived ease of use on usage intention has
been observed in some research to decrease over time
(e.g., Davis et al. 1989), and in other research to
increase over time (e.g., Venkatesh and Davis 1996).
Presently, we lack sufficient theoretical rationale to
hypothesize specific temporal shifts in the strength of
any of the effects of cognitive instrumental determinants (job relevance, output quality, result demonstrability, or ease of use) on either perceived usefulness or
intention to use.
In summary, the proposed TAM2 encompasses social influence processes (subjective norm, voluntariness, and image) and cognitive instrumental processes
(job relevance, output quality, result demonstrability,
and perceived ease of use) as determinants of perceived usefulness and usage intentions. We also hy-

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

pothesize a decrease in the strength with which social

influence processes affect perceived usefulness and
intention to use with increasing experience over time.

on-the-job training, to familiarize them with the use of

the system. Thirty-eight subjects completed the study
and provided usable responses at all points of measurement.

Method

Study 2. Subjects were 50 members of the personal

financial services department of a large financial services firm, representing various levels of the organizational hierarchy. The specific system project was to
move all current mainframe operations to a Windowsbased environment. Given the magnitude of the
change, subjects were also given continued access to
the old system and the discretion to use either system,
thus rendering use of the new system to be voluntary.
A one-and-a-half-day on-site training program was
conducted to educate the users on the new environment. Thirty-nine subjects provided usable responses
at all points of measurement.

Four longitudinal field studies were conducted to test

TAM2. In the interest of ecological validity, we sought
naturalistic research sites that closely mirrored the
target situation to which we would like our results to
generalize: where a new system is about to be introduced in the workplace. The four field sites spanned a
range of industries, organizational contexts, functional
areas, and types of system being introduced. We chose
two sites where usage of the system was voluntary,
and two where usage was mandatory to explicitly
examine the theorized moderating role of voluntariness. To assess whether system use was mandatory or
voluntary, the researchers initially interviewed senior
managers in each of the organizations prior to the
actual system introduction. In the actual questionnaire
we measured user perceptions of voluntariness using
a three-item scale (discussed below) since, although
we expected these measures to coincide with senior
manager reports, our theoretical model posits perceived rather than actual voluntariness as a moderator
of the subjective norm-intention relationship. Questionnaires were administered to potential users at
three points in time: after initial training (T1), one
month after implementation (T2), and three months
after implementation (T3). Self-reported usage behavior was measured at T2 and T3, and also five months
after implementation (T4).
Data Sample
Study 1. Subjects were 48 floor supervisors in a
medium-sized manufacturing firm who were being
introduced to a proprietary system for their day-today activities, which included floor and machine
scheduling and personnel assignment. Before the introduction of the system, most floor supervisors used
a manual system, though there was software on multiple stand-alone Macintosh machines. Use of the new
system was voluntary, and none of the subjects had
any prior knowledge of the system. The subjects
received two days of training, including one day of

Management Science/Vol. 46, No. 2, February 2000

Study 3. Subjects were 51 employees from various

levels of the organizational hierarchy of a small accounting services firm. The subjects performed various activities related to customer accounts and telephone customer service. The new system introduced
was a Windows-based customer account management
system, which replaced a combination of a paperbased and a DOS-based system. A one-day on-site
training program was conducted to educate all employees on the use of the new system and software
environment. Unlike the first two studies, where use
was voluntary, the old system was phased out over a
one-week period, so use of the new system became
mandatory. Forty-three subjects provided usable responses at all points of measurement.
Study 4. Subjects were 51 employees of a small
international investment banking firm. The subjects,
who represented various levels of the organization,
performed a variety of functions related to domestic
and international stock management. The new system
introduced was designed to assist in analyzing and
creating financially sound international stock portfolios. Compared with the existing system, the new
system introduced new functionality such as the ability to quantitatively weight and incorporate subjective
information (e.g., political stability) into risk assessment. The new system (a custom-built DOS-based

193

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

system) was developed by a different vendor and had

a very different look and feel from the current system
(which was also DOS-based). The subjects participated
in a four-hour training program on the new system.
Similar to Study 3, system use was mandatory, as the
old system was immediately phased out. Thirty-six
subjects provided usable responses at all points of
measurement.
Measurement Scales
Theoretical constructs were operationalized using validated items from prior research (see Appendix 1). The
TAM scales of perceived usefulness, perceived ease of
use, and behavioral intention were measured using
items adapted from Davis (1989) and Davis et al.
(1989). The measurement of subjective norm was
adapted from Taylor and Todd (1995). Measures of
result demonstrability and image were adapted from
Moore and Benbasat (1991). Measures of job relevance
and output quality were adapted from Davis et al.
(1992). Following Davis (1989), we measured usage
behavior by asking subjects: On average, how much
time do you spend on the system every day? _ hours
and _ minutes. Though some research suggests that
self-reported usage measures are biased (e.g., Straub
et al. 1995), other research suggests that self-report
usage measures correlate well with actual usage measures (e.g., Taylor and Todd 1995). Our design
avoided a major limitation associated with self-reported usage: common-method variance resulting
from measuring both self-reported usage and its determinants (intention, perceived usefulness, etc.) on a
single questionnaire. We examined the relationship
between intention and other usage determinants measured at one point in time with self-reported usage
measured at the subsequent time period, such that the
measurement of usage and its determinants were
separated by one month (T1T2), three months (T2
T3), and two months (T3T4), respectively. A threeitem scale of voluntariness developed by Moore and
Benbasat (1991) was included. To ensure that the
measurement scales were adapted appropriately to
the current context, we conducted a focus group
pretest with five business professionals. Minor suggested wording changes were performed.

194

Procedure
Although the specific initial training in each organization was different, the process of questionnaire administration followed was very similar. Following the
training (T1), the subjects filled out a questionnaire
with the above measurement scales regarding the new
system for which they just received training. The
questionnaire was administered online and each subjects log-in ID was captured and used to create a
barcode (for subsequent administrations in T2, T3, and
T4) to help track individual responses over time. To
preserve privacy and confidentiality, only the researchers had access to specific respondent information and only summarized responses were shared
with the organization. One of the authors observed the
process of questionnaire administration at T1 to ensure no biases were introduced. At T2 and T3, in
addition to user reactions, self-reported usage was
also measured. In each of these cases, a paper version
of the questionnaire was delivered to subjects mailboxes with a request that it be returned directly to the
researchers within a one-week time window. At T4,
only self-reported usage was measured using a procedure similar to T2 and T3. Intention measured in T1
was used to predict usage in T2, intention measured in
T2 was used to predict usage in T3, and intention
measured in T3 was used to predict usage in T4,
respectively.

Results
Psychometric Properties of Measures
The measurement scales exhibited strong psychometric properties. All measurement scales showed high
reliability, with Cronbach alpha coefficients for all
four studies and three time periods exceeding 0.80 (see
Appendix 1). Construct validity was strongly supported both by principal components analysis with
direct oblimin rotation (Appendix 2), in which all
cross-loadings were lower than 0.30, and by an analysis of the multitrait-multimethod matrix. This pattern
of high reliability and validity is consistent with much
prior research (e.g., Davis 1989, Davis and Venkatesh
1996, Mathieson 1991, Taylor and Todd 1995). Appendix 3 presents cross-sectional correlations among these
constructs at each of the three points of measurement.

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Based on measures of voluntariness (where 1 mandatory and 7 voluntary), we confirmed that employees in Studies 1 and 2 perceived system use to be
voluntary (across companies and time periods, mean
voluntariness ratings ranged from 6.2 to 6.7, and
standard deviations ranged from 0.4 to 0.6), whereas
employees in Studies 3 and 4 perceived system use to
be mandatory (across companies and time periods,
mean voluntariness ratings ranged from 1.2 to 1.5, and
standard deviations ranged from 0.3 to 0.6). Further,
correlations between measures of voluntariness across
the three time periods exceeded 0.80 in all four
studies.
Explaining Intention and Usage
Table 1 shows the effects of perceived usefulness,
perceived ease of use, and subjective norm on intentions. Consistent with much prior research, perceived
usefulness was a strong determinant of intention to
use, and perceived ease of use was a significant
secondary determinant. The effect of subjective norm
on intention (compliance) was consistent with our
expectations. That is, when usage was mandatory
(Studies 3 and 4), subjective norm did have a direct
effect on intention at T1 and T2 (supporting HypothTable 1

esis 1a), and this effect weakened to the point of

nonsignificance by T3 (supporting Hypothesis 4a). In
contrast, in cases where usage was voluntary (Studies
1 and 2), subjective norm had no direct effect on
intention over and above what was explained by
perceived usefulness and perceived ease of use (consistent with Hypothesis 1b). Unlike subjective norm,
perceived usefulness and perceived ease of use remained consistent significant determinants of intention across all time periods in all four studies. TAM2
thus explained between 37% and 52% of the variance
in usage intentions. The correlations between intention and usage behavior were found to be in the range
of 0.44 to 0.57 for all studies at all points of measurement. For all studies and time periods, intention fully
mediated the effects of perceived usefulness, perceived ease of use, and subjective norm on usage
behavior.
Explaining Perceived Usefulness
TAM2 theorizes main effects for all determinants of
perceived usefulness but does not explicitly posit any
interactions. To address the possibility that some of
the constructs combined multiplicatively rather than
additively in their effect on perceived usefulness, we

TAM2 Regression Results Explaining Intention to Use

Voluntary Settings
Study 1 (n 38)

Time of Measurement

R2

Preimplementation (T1)
Perceived Usefulness
Perceived Ease of Use
Subjective Norm

0.39

One month postimpl. (T2)

Perceived Usefulness
Perceived Ease of Use
Subjective Norm

0.44

Three month postimpl. (T3)

Perceived Usefulness
Perceived Ease of Use
Subjective Norm

0.42

Mandatory Settings

Study 2 (n 39)
R2

0.37
0.58***
0.18*
0.11

R2

0.44
0.51***
0.27**
0.10

0.34
0.55***
0.17*
0.06

Study 4 (n 36)
R2

0.47

0.52***
0.18*
0.28**
0.42

0.54***
0.15*
0.26**
0.39

0.64***
0.16*
0.02

0.52
0.48***
0.13*
0.31**

0.50***
0.21*
0.08
0.42

0.63***
0.14*
0.11

Study 3 (n 43)

0.44***
0.11*
0.24**
0.39

0.57***
0.17*
0.10

0.50***
0.22*
0.08

* p 0.05, ** p 0.01, *** p 0.001. Adjusted R 2 s are shown. : standardized regression coefficients.

Management Science/Vol. 46, No. 2, February 2000

195

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Table 2

TAM2 Regression Results Explaining Perceived Usefulness

Voluntary Settings
Study 1

Time of Measurement

R2

Preimplementation (T1)
Subjective Norm
Image
Job Relev. Quality
Result Demonstrability
Perceived Ease of Use
One month postimpl. (T2)
Subjective Norm
Image
Job Relev. Quality
Result Demonstrability
Perceived Ease of Use
Three month postimpl. (T3)
Subjective Norm
Image
Job Relev. Quality
Result Demonstrability
Perceived Ease of Use

0.60

Mandatory Settings
Study 2

R2

Study 3

0.60
0.50***
0.19*
0.40***
0.27**
0.23**

0.48

0.55

0.31**
0.36***
0.33**
0.22**
0.20**
0.46

0.27**
0.28**
0.30**
0.30**
0.38***
0.43

0.08
0.19*
0.36***
0.26**
0.34***

0.50

0.51

0.40

R2

0.38***
0.31**
0.32**
0.30**
0.23**

0.34***
0.18*
0.38***
0.26**
0.35***

0.20*
0.17*
0.38***
0.34**
0.28**

0.51
0.47***
0.21*
0.38***
0.30**
0.24**

0.39***
0.22*
0.32**
0.24*
0.26**
0.44

R2

Study 4

0.21*
0.27**
0.33***
0.30**
0.29**
0.40

0.16*
0.27**
0.36***
0.21*
0.35***

0.10
0.25**
0.32**
0.30**
0.35***

Note: Main effects of task importance and output quality were included in regressions but omitted from table due to noninterpretability in the presence of the
interaction term.
* p 0.05; ** p 0.01; *** p 0.001. Adjusted R 2 s are shown. : standardized regression coefficients.

conducted stepwise regression analyses. Specifically,

following the test of the main-effects model, we used
stepwise regression to enter into the model any significant two-way interactions. The only two-way interaction term that was significant was between job
relevance and output quality. Moreover, this interaction was significant at p 0.01 or p 0.001 for all
four studies at all three points of measurement. We
presume this interaction effect to be robust, since the
possibility of it occurring across four independent
samples by chance, under the null hypothesis of no
significant effect (type I error), is less than 0.00001.
Therefore, we added this interaction term to the model
and reestimated all parameters.
As Table 2 shows, TAM2 explained up to 60% of the
variance in perceived usefulness. The effect of subjective norm on perceived usefulness (internalization)
was significant at T1 and T2 (supporting Hypothesis
2), but weakened by T3 (supporting Hypothesis 4b).

196

The influence of image on perceived usefulness (identification) was significant at all three points of measurement (supporting Hypothesis 3a). Also, as hypothesized, the effect of subjective norm on image was
significant at all points of measurement (supporting
Hypothesis 3b). As discussed earlier, the interaction of
job relevance and output quality was significant in all
four studies at all points of measurement (supporting
Hypotheses 5 and 6, respectively). Note that the main
effects of job relevance and output quality were found
to be significant before the interaction term was introduced into the regression model. After the interaction
term was included, the main effects were included in
the model for parameter estimation, but they were
omitted from Table 2 because main effects should not
be interpreted in the presence of the interaction term
under interval scale measurement. As theorized, result demonstrability and perceived ease of use were
significant across all four studies and three time peri-

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

ods (supporting Hypotheses 7 and 8, respectively). To

further substantiate the assertion that perceived usefulness fully mediates the effects of these determinants
on intention, we performed hierarchical regression
analyses with intention as the dependent variable and
all the other variables as possible predictors. As theorized, perceived usefulness fully mediated the effects
of all of its determinants on usage intentions, except
for the expected direct effects of perceived ease of use
in all studies and subjective norm for the mandatory
usage studies (3 and 4).
Summary of Results Pooled Across Studies and
Time Periods
Figure 2 provides a summary presentation of the
results for TAM2 when data are pooled across studies
and time periods. Pooling across the four studies
yielded a sample of 156 per time period. Pooling
across the three time periods yielded a sample of 468,

Figure 2

which was used to estimate the summary model.

Voluntariness was coded as a dummy variable with
values of 0 for voluntary use (Studies 1 and 2) and 1
for mandatory use (Studies 3 and 4). Experience was
coded as an indicator variable with values of 0 for T1,
1 for T2, and 2 for T3. Regression analyses with tests
for moderation and mediation were done to test the
summary model, which yielded results that were
highly consistent with the 12 individual model tests
for each combination of four organizations and three
time periods (Tables 1 and 2). Consistent with theory,
the basic TAM relationships (i.e., intention-usage, usefulness-intention, ease of use-intention, and ease of
use-usefulness) were well supported, with full mediation by intention and no moderation by either voluntariness or experience. As theorized, the subjective
norm-intention relationship was significantly moderated by both experience (supporting Hypothesis 4a)

TAM2 Results Pooled Across Studies and Time Periods (n 468)

Management Science/Vol. 46, No. 2, February 2000

197

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

and voluntariness (supporting Hypothesis 1c), mirroring the more detailed findings reported in Table 1 that
subjective norm significantly affects intention directly
only when usage is mandatory and experience is in
the early stages. Also consistent with TAM2, the
subjective norm-usefulness relationship (internalization) was significantly moderated by experience (supporting Hypothesis 4b), whereas the image-usefulness
relationship (identification) was not. Consistent with
Table 2, the effect of job relevance and output quality
on perceived usefulness was interactive. Thus, all
hypotheses were supported by all studies at all points
of measurement.
Cross-Temporal Correlations of Intention
Determinants
In a post hoc analysis performed to better understand
how preimplementation user evaluations of a system
relate to postimplementation evaluations, we examined the cross-temporal correlations of usage intention
and its direct determinants, perceived usefulness, perceived ease of use, and subjective norm (Table 3).
Perceived usefulness was the most stable of the three

Table 3

Cross-Temporal Stability Correlations for Perceived

Usefulness, Perceived Ease of Use, Subjective Norm, and
Intention to Use
Voluntary Settings
Time Span

Time 1Time 2 (1 month)

Perceived Usefulness
Perceived Ease of Use
Subjective Norm
Intention to Use
Time 2Time 3 (2 months)
Perceived Usefulness
Perceived Ease of Use
Subjective Norm
Intention to Use
Time 1Time 3 (3 months)
Perceived Usefulness
Perceived Ease of Use
Subjective Norm
Intention to Use

Study 1

Study 2

Study 3

Study 4

0.68***
0.26*
0.56***
0.68***

0.56***
0.23*
0.54***
0.78***

0.76***
0.12
0.59***
0.70***

0.60***
0.28*
0.50***
0.66***

0.71***
0.30*
0.61***
0.76***

0.66***
0.17
0.60***
0.82***

0.79***
0.26*
0.57***
0.71***

0.62***
0.37**
0.65***
0.61***

0.66***
0.14
0.58***
0.69***

0.62***
0.19
0.57***
0.71***

0.71***
0.28*
0.51***
0.63***

0.69***
0.36***
0.52***
0.68***

* p 0.05. ** p 0.01. *** p 0.001.

198

Mandatory Settings

determinants, with correlations ranging from 0.56 to

0.79. Similarly, the stability correlations for subjective
norm were relatively high (0.51 0.65). In contrast, the
stability correlations were systematically lower for
perceived ease of use (0.12 0.37). Reflecting the comparatively strong effect of perceived usefulness on
intentions, as found in the present studies (see Table 2)
as well as in previous studies, intention exhibited
cross-temporal stability correlations comparable to
those of usefulness, ranging from 0.63 to 0.82. Using
distributed lag models, we confirmed that intentions
fully mediated the effects of perceived usefulness,
perceived ease of use, and subjective norm measured
in one time period (T1 or T2) on intention measured in
a later time period (T2 or T3).

Discussion
TAM2 was strongly supported across four organizations and three points of measurement (preimplementation, one month postimplementation, and three
months postimplementation). Encompassing both social influence processes (subjective norm, voluntariness, and image) and cognitive instrumental processes
(job relevance, output quality, result demonstrability,
and perceived ease of use), TAM2 provides a detailed
account of the key forces underlying judgments of
perceived usefulness, explaining up to 60% of the
variance in this important driver of usage intentions.
Moreover, TAM2 extends TAM by showing that subjective norm exerts a significant direct effect on usage
intentions over and above perceived usefulness and
perceived ease of use for mandatory (but not voluntary) systems.
The effects of social influence processes were consistent with TAM2. Subjective norm significantly influenced perceived usefulness via both internalization,
in which people incorporate social influences into
their own usefulness perceptions, and identification,
in which people use a system to gain status and
influence within the work group and thereby improve
their job performance. Beyond these two indirect
effects via perceived usefulness, subjective norm had a
direct effect on intentions for mandatory, but not
voluntary, usage contexts. This may explain previous
research that found a nonsignificant role for social

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

influences in voluntary contexts (e.g., Davis et al. 1989,

Mathieson 1991). We found that as individuals gained
direct experience with a system over time, they relied
less on social information in forming perceived usefulness and intention but continued to judge a systems usefulness on the basis of potential status benefits resulting from use.
The effects of cognitive instrumental processes
were also consistent with TAM2. An important and
interesting finding that emerged was the interactive
effect between job relevance and output quality in
determining perceived usefulness. This implies that
judgments about a systems usefulness are affected
by an individuals cognitive matching of their job
goals with the consequences of system use (job
relevance), and that output quality takes on greater
importance in proportion to a systems job relevance. Although the interaction was not explicitly
hypothesized, similar interaction effects between
task characteristics and technology characteristics
have been observed in other research (Davis et al.
1992, Goodhue 1995). As hypothesized, user perceptions of result demonstrability and ease of use were
significant. Unlike social influence processes, the
effects of cognitive instrumental processes remained
significant over time.
Limitations
The present research has several limitations that
should be noted. First, the sample sizes were less
than 50 for each of the four longitudinal samples,
which could reduce the power of significance tests.
Fortunately, few of the findings were nonsignificant,
and the small sample sizes give assurance that the
significance we did observe indicates meaningful
effect sizes. The consistency of the findings across
studies further reduces the likelihood that the major
findings are spurious. Second, several of the constructs were measured with only two items. The
two-item scales exhibited adequate reliability
(Cronbach alpha was over 0.80 in all cases) and
factorial validity, mitigating concerns about measure reliability or validity. Also, the data analysis
did not use structural equation modeling, in which
two-item scales in a measurement model can introduce problems of underidentification resulting in

Management Science/Vol. 46, No. 2, February 2000

instability of parameter estimates. Third, the field

studies were longitudinal observational designs and
hence did not involve experimental manipulation of
theoretical constructs. Experimental replication of
these findings would enhance causal interpretations
of key relationships. Fourth, we employed selfreported as opposed to objectively measured usage.
Since self-reported usage was measured in the wave
of measurement subsequent to that in which intentions were measured (at minimum of one month
later), common-method variance was minimized.
Nevertheless, the interchangeability of self-report
and objective usage measures remains a controversial point in IS research (Straub et al. 1995). Fifth,
although we employed a longitudinal design spanning five months in each organization, we employed
a variance theory approach. Additional insights into
the underlying dynamics of the causal mechanisms
addressed by the model might emerge by using a
process approach that analyzes the detailed events
and action sequences that unfold over time.
Contributions and Implications
There are several practical implications of our findings. Mandatory, compliance-based approaches to introducing new systems appear to be less effective over
time than the use of social influence to target positive
changes in perceived usefulness. Practical alternatives
to usage mandates based on social information should
be developed and tested, such as increasing the source
credibility of social information to increase internalization or designing communication campaigns that
raise the prestige associated with system use to increase identification. On the instrumental side, in
addition to designing systems to better match jobrelevant needs, improving the quality of their output,
or making them easier to use, our research suggests
that practical interventions for increasing result demonstrability, such as empirically demonstrating to
users the comparative effectiveness of a new system
relative to the status quo (e.g., Davis and Kottemann
1995), may provide important leverage for increasing
user acceptance.
The current research represents an important contribution to theory by extending TAM to address
causal antecedents of one of its two belief constructs,

199

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

perceived usefulness. Other research has begun to

model the determinants of TAMs other major belief,
perceived ease of use (Venkatesh and Davis 1996).
Further research on TAM, in addition to refining the
models of the determinants of perceived usefulness
and perceived ease of use, should address the role of
other direct determinants of usage intentions and
behavior and continue to map out the major contingency factors moderating the effects of perceived
usefulness, perceived ease of use, subjective norm,
and other constructs on intention.
Another important avenue for future research
concerns the temporal dynamics of the determinants
of user acceptance. The present research found that
TAM2 held up well at three points of measurement
spanning from preimplementation to three months
postimplementation. Perceived usefulness, subjective norm, and intention were quite stable over this
three-month horizon. Perceived ease of use, in contrast, was less stable over time, which we attribute
to the known role of direct hands-on experience in
forming this belief (Venkatesh and Davis 1996).
Future research might profitably seek to establish
how early in a system development process, for
example, even before a working prototype is built,
we can measure key user reactions, such as perceived usefulness and intention, and still rely on
them as indicators of postimplementation success of
the system concept.
Additional research is needed to elucidate the processes involved in cognitively matching important job
goals to the consequences of system use. Beach and
Mitchells (1996, 1998) compatibility and profitability
tests offer theoretical structure that might be brought
to bear for fleshing out this important process.
Progress might also come from adapting profile-comparison processes from research on person-job fit
(Caldwell and OReilly 1990) to the assessment of
system-job fit. In any case, understanding this matching process is a vital research direction since it explicitly links the functional design characteristics of a
system to perceived usefulness and ultimately user
acceptance.
More broadly, future research should seek to
further extend models of technology acceptance to

200

encompass other important theoretical constructs,

such as the choice sets of available alternative
technologies, learning and training (e.g., Venkatesh
and Speier 1999), misperceptions of usefulness or
ease of use and how to correct them (e.g., Davis and
Kottemann 1995), changes in work content or job
goals, and changing social environments. Future
research should compare our findings regarding the
role of subjective norm with other recent findings
(e.g., Venkatesh and Morris 2000). Of particular
interest would be a comparison of TAM2 with
recent approaches from the media choice paradigm
that attempt to integrate normative and utilitarian
determinants (e.g., Kraut et al. 1998, Webster and
Trevino 1995). The continuing trend in organizations away from hierarchical, command-and-control
structures toward networks of empowered, autonomous teams underscores our finding regarding the
limits of organizational mandate as a lever for
increasing usage. To adapt user acceptance theory
to this trend, the conceptualization of perceived
usefulness will need to be expanded from its current
focus on expected individual-level performance
gains to encompass team-based structures and incentives. As the adoption decision becomes more of
a team- rather than individual-level decision, the
nature and role of social influence processes (both
within teams and across teams) will need to be
elaborated beyond TAM2.
In conclusion, user acceptance of information technology in the workplace remains a complex, elusive,
yet extremely important phenomenon. Research over
the past decade has made significant progress toward
unraveling some of its mysteries. The development
and test of TAM2 reported here advances theory and
research on this important issue. 1

1
The authors thank Professors Gordon Davis and Shawn Curley at
the University of Minnesota and Ruth Kanfer and Phillip Ackerman
at Georgia Institute of Technology for their many useful comments
and suggestions on earlier versions of the paper. They also thank
Tracy Ann Sykes for her excellent editorial work and assistance in
data collection. Part of this research was done while the second
author was on the visiting faculty of Salisbury State University. The
authors contributed equally to this research; order of authorship
was determined by a coin flip.

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Appendix 1. TAM2 Measurement Scales and Reliabilities

Intention to Use
Assuming I have access to the system, I intend to use it.
Given that I have access to the system, I predict that I would use it.
Perceived Usefulness
Using the system improves my performance in my job.
Using the system in my job increases my productivity.
Using the system enhances my effectiveness in my job.
I find the system to be useful in my job.
Perceived Ease of Use
My interaction with the system is clear and understandable.
Interacting with the system does not require a lot of my mental effort.
I find the system to be easy to use.
I find it easy to get the system to do what I want it to do.
Subjective Norm
People who influence my behavior think that I should use the system.
People who are important to me think that I should use the system.
Voluntariness
My use of the system is voluntary.
My supervisor does not require me to use the system.
Although it might be helpful, using the system is certainly not compulsory
in my job.
Image
People in my organization who use the system have more prestige than
those who do not.
People in my organization who use the system have a high profile.
Having the system is a status symbol in my organization.
Job Relevance
In my job, usage of the system is important.
In my job, usage of the system is relevant.
Output Quality
The quality of the output I get from the system is high.
I have no problem with the quality of the systems output.
Result Demonstrability
I have no difficulty telling others about the results of using the system.
I believe I could communicate to others the consequences of using the
system.
The results of using the system are apparent to me.
I would have difficulty explaining why using the system may or may not be
beneficial.

(Cronbachs ranged from 0.82 to 0.97 across studies and time periods)

(Cronbachs ranged from 0.87 to 0.98 across studies and time periods)

(Cronbachs ranged from 0.86 to 0.98 across studies and time periods)

(Cronbachs ranged from 0.81 to 0.94 across studies and time periods)

(Cronbachs ranged from 0.82 to 0.91 across studies and time periods)

(Cronbachs ranged from 0.80 to 0.93 across studies and time periods)

(Cronbachs ranged from 0.80 to 0.95 across studies and time periods)

(Cronbachs ranged from 0.82 to 0.98 across studies and time periods)

(Cronbachs ranged from 0.80 to 0.97 across studies and time periods)

Note. All items were measured on a 7-point Likert scale, where 1 strongly disagree, 2 moderately disagree, 3 somewhat disagree, 4 neutral (neither
disagree nor agree), 5 somewhat agree, 6 moderately agree, and 7 strongly agree.

Management Science/Vol. 46, No. 2, February 2000

201

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Appendix 2. Principal Components Analysis with Oblimin Rotation: Pooled Across Studies and Time
Periods (n 468)
Construct Loadings
Item

Intent to Use

Perc Usef

Perc EOU

Subj Norm

Image

Reslt Demon

Output Qual

Job Rel

Intention to Use 1
Intention to Use 2

0.91
0.93

0.03
0.06

0.09
0.12

0.04
0.28

0.03
0.07

0.03
0.05

0.03
0.01

0.11
0.02

Perc.
Perc.
Perc.
Perc.

Usefulness
Usefulness
Usefulness
Usefulness

0.11
0.13
0.05
0.22

0.89
0.88
0.93
0.91

0.07
0.02
0.11
0.04

0.09
0.22
0.04
0.00

0.01
0.04
0.01
0.11

0.00
0.18
0.12
0.06

0.06
0.22
0.03
0.08

0.06
0.02
0.05
0.12

Perc.
Perc.
Perc.
Perc.

Ease
Ease
Ease
Ease

0.10
0.11
0.15
0.09

0.09
0.11
0.09
0.04

0.91
0.93
0.96
0.82

0.03
0.11
0.12
0.02

0.08
0.19
0.12
0.17

0.03
0.09
0.02
0.08

0.11
0.17
0.10
0.02

0.12
0.05
0.07
0.02

Subjective Norm 1
Subjective Norm 2
Image 1
Image 2
Image 3

0.27
0.22
0.14
0.04
0.11

0.03
0.20
0.05
0.06
0.11

0.11
0.18
0.21
0.13
0.10

0.90
0.91
0.02
0.04
0.01

0.09
0.06
0.86
0.89
0.91

0.00
0.04
0.09
0.11
0.16

0.05
0.08
0.08
0.13
0.21

0.07
0.02
0.21
0.17
0.12

Result
Result
Result
Result

0.01
0.10
0.16
0.12

0.15
0.17
0.09
0.01

0.08
0.16
0.11
0.10

0.00
0.11
0.14
0.19

0.05
0.02
0.08
0.09

0.90
0.91
0.88
0.84

0.07
0.18
0.00
0.12

0.05
0.12
0.17
0.12

Output Quality 1
Output Quality 2

0.19
0.02

0.00
0.03

0.06
0.05

0.00
0.01

0.13
0.14

0.13
0.12

0.88
0.85

0.09
0.14

Job Relevance 1
Job Relevance 2

0.13
0.15

0.04
0.13

0.12
0.17

0.07
0.08

0.16
0.12

0.11
0.04

0.07
0.02

0.91
0.92

Cronbachs

0.91

0.93

0.93

0.90

0.91

0.89

0.88

0.90

of
of
of
of

1
2
3
4

Use
Use
Use
Use

Demonst.
Demonst.
Demonst.
Demonst.

1
2
3
4

1
2
3
4

Note. The same simple factor structure was found within each time period in each of the four studies and when data were separately pooled across time and across
mandatory versus voluntary settings.

Appendix 3. Cross-Sectional Correlation Matrices of TAM2 Constructs

Voluntary Setting (Studies 1 & 2 Pooled)

Mandatory Setting (Studies 3 & 4 Pooled)

(a) Preimplementation (T1)

U
U
EOU
SN
IMG
JR
QUAL
RD
BI

202

0.27***
0.21*
0.26**
0.30***
0.28***
0.28**
0.45***

EOU

0.12
0.13
0.21*
0.22**
0.13
0.25**

(b) Preimplementation (T1)

SN

0.45***
0.27**
0.12
0.22*
0.13

IMG

0.11
0.20*
0.13
0.24**

JR

0.31***
0.16*
0.27***

QUAL

0.24**
0.21**

RD

0.20*

BI
U
EOU
SN
IMG
JR
QUAL
RD
BI

EOU

SN

IMG

JR

QUAL

RD

0.28***
0.40***
0.25**
0.32***
0.29**
0.30***
0.47***

0.21*
0.16
0.23*
0.25**
0.18
0.23**

0.43***
0.20*
0.19*
0.28***
0.20*

0.14
0.19*
0.20**
0.27***

0.29**
0.15
0.25**

0.30***
0.31***

0.29***

BI

Management Science/Vol. 46, No. 2, February 2000

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

Appendix 3

(Continued)

(c) One month postimplementation (T2)

U
U
EOU
SN
IMG
JR
QUAL
RD
BI

EOU

0.27***
0.16*
0.30**
0.29***
0.22*
0.29**
0.50***

0.15
0.11
0.20*
0.21*
0.11
0.25**

SN

(d) One month postimplementation (T2)

IMG

0.40***
0.11
0.13
0.23**
0.09

JR

0.13
0.22*
0.04
0.26**

QUAL

0.30***
0.12
0.20*

0.22*
0.23**

RD

BI
U
EOU
SN
IMG
JR
QUAL
RD
BI

0.21*

(e) Three month postimplementation (T3)

U

U
EOU
SN
IMG
JR
QUAL
RD
BI

0.31***
0.19*
0.27***
0.30***
0.21*
0.23*
0.46***

EOU

0.23*
0.13
0.20*
0.23**
0.14
0.20*

SN

0.40***
0.12
0.21*
0.12
0.18*

EOU

0.30***
0.33***
0.26**
0.28***
0.24**
0.33***
0.44***

SN

0.15
0.19*
0.20*
0.29**
0.9
0.20*

IMG

0.40***
0.22*
0.21*
0.22*
0.13

JR

0.19*
0.22**
0.16
0.23*

QUAL

0.22*
0.13
0.26***

0.28***
0.21*

RD

BI

0.24**

(f) Three month postimplementation (T3)

IMG

0.13
0.14
0.19*
0.31***

JR

0.33***
0.06
0.20*

QUAL

0.23*
0.23**

RD

BI

0.20*

U
EOU
SN
IMG
JR
QUAL
RD
BI

EOU

SN

IMG

JR

QUAL

RD

0.33***
0.27**
0.21**
0.31***
0.29***
0.36***
0.48***

0.15
0.19*
0.20*
0.29**
0.9
0.28**

0.40***
0.22*
0.21*
0.22*
0.07

0.19*
0.22**
0.16
0.26**

0.22*
0.13
0.21*

0.28***
0.22*

0.24***

BI

Note. U perceived usefulness; EOU perceived ease of use; SN subjective norm; IMG image; JR job relevance; QUAL output quality; RD result
demonstrability; BI behavioral intention to use; * p 0.05; ** p 0.01; *** p 0.001.

References
Agarwal, R., J. Prasad. 1997. The role of innovation characteristics
and perceived voluntariness in the acceptance of information
technologies. Decision Sci. 28 557582.
Ajzen, I. 1991. The theory of planned behavior. Organ. Behavior and
Human Decision Processes 50 179 211.
Bandura, A. 1986. Social Foundations of Thought and Action: A Social
Cognitive Theory. Prentice-Hall, Englewood Cliffs, NJ.
Beach, L. R., T. R. Mitchell. 1978. A contingency model for the
selection of decision strategies. Acad. Management Rev. 3 439
449.
,
. 1996. Image theory, the unifying perspective. L. R. Beach,
ed. Decision Making in the Workplace: A Unified Perspective.
Lawrence Erlbaum Associates, Mahwah, NJ. 120.
,
. 1998. The basics of image theory. L. R. Beach, ed. Image
Theory: Theoretical and Empirical Foundations. Lawrence Erlbaum
Associates, Mahwah, NJ. 318.
Black, J. B., D. S. Kay, E. M. Soloway. 1987. Goal and plan
knowledge representations: From stories to text editors and

Management Science/Vol. 46, No. 2, February 2000

programs. J. M. Carroll, ed. Interfacing Thought. The MIT Press,

Cambridge, MA. 36 60.
Blau, P. M. 1964. Exchange and Power in Social Life. John Wiley, New
York.
Caldwell, D. F., C. A. OReilly. 1990. Measuring person-job fit within
a profile-comparison process. J. Appl. Psych. 75 648 657.
Davis, F. D. 1989. Perceived usefulness, perceived ease of use, and
user acceptance of information technology. MIS Quart. 13
319 339.
, R. P. Bagozzi, P. R. Warshaw. 1989. User acceptance of
computer technology: A comparison of two theoretical models.
Management Sci. 35 9821002.
,
,
. 1992. Extrinsic and intrinsic motivation to use
computers in the workplace. J. Appl. Social Psych. 22 1111
1132.
, J. E. Kottemann. 1995. Determinants of decision rule use in
production planning task. Organ. Behavior and Human Decision
Processes 63 145157.
, V. Venkatesh. 1996. A critical assessment of potential mea-

203

VENKATESH AND DAVIS

A Theoretical Extension of the Technology Acceptance Model

surement biases in the technology acceptance model: Three

experiments. Internat. J. Human-Comput. Stud. 45 19 45.
Deutsch, M., H. B. Gerard. 1955. A study of normative and informational social influences upon individual judgment. J. Abnormal and Social Psych. 51 629 636.
Doll, J., I. Ajzen. 1992. Accessibility and stability of predictors in the theory
of planned behavior. J. Personality and Social Psych. 63 754765.
Fazio, R. H., M. Zanna. 1981. Direct experience and attitude-behavior
consistency. L. Berkowitz, ed. Advances in Experimental Social
Psychology, Vol. 14. Academic Press, San Diego, CA. 161202.
Fishbein, M., I. Ajzen. 1975. Belief, Attitude, Intention and Behavior: An
Introduction to Theory and Research. Addison-Wesley, Reading, MA.
French, J. R. P., B. Raven. 1959. The bases of social power. D.
Cartwright, ed. Studies in Social Power. Institute for Social
Research, Ann Arbor, MI. 150 167.
Fulk, J., C. W. Steinfield, J. Schmitz, J. G. Power. 1987. A social
information processing model of media use in organizations.
Comm. Res. 14 529 552.
Goodhue, D. L. 1995. Understanding the linkage between user
evaluations of systems and the underlying systems. Management Sci. 41 18271844.
Hackman, J. R., G. R. Oldham. 1976. Motivation through the design
of work: Test of a theory. Organ. Behavior and Human Performance 16 250 279.
Hartwick, J., H. Barki. 1994. Explaining the role of user participation
in information system use. Management Sci. 40 440 465.
Kelman, H. C. 1958. Compliance, identification, and internalization:
Three processes of attitude change. J. Conflict Resolution 2 51 60.
Kieras, D. E., P. G. Polson. 1985. An approach to the formal analysis
of user complexity. Internat. J. of Man-Machine Stud. 22 365394.
Kiesler, C. A., S. B. Kiesler. 1969. Conformity. Addison-Wesley,
Reading, MA.
Kraut, R. E., R. E. Rice, C. Cool, R. S. Fish. 1998. Varieties of social
influence: The role of utility and norms in the success of a new
communication medium. Organ. Sci. 9 437 453.
Leonard-Barton, D., I. Deschamps. 1988. Managerial influence in the
implementation of new technology. Management Sci. 34 12521265.
Locke, E. A., G. P. Latham. 1990. A Theory of Goal Setting and Task
Performance. Prentice-Hall, Englewood Cliffs, NJ.
Loher, B. T., R. A. Noe, N. L. Moeller, M. P. Fitzgerald. 1985. A
meta-analysis of the relation of job characteristics to job satisfaction. J. Appl. Psych. 70 280 289.
Mathieson, K. 1991. Predicting user intentions: Comparing the
technology acceptance model with the theory of planned behavior. Inform. Systems Res. 2 173191.
Moore, G. C., I. Benbasat. 1991. Development of an instrument to
measure the perceptions of adopting an information technology innovation. Inform. Systems Res. 2 192222.
Norman, D. A. 1987. Cognitive engineeringCognitive science.
J. M. Carroll, ed. Interfacing Thought. The MIT Press, Cambridge,
MA. 325336.
Pfeffer, J. 1981. Power in Organizations. Ballinger, Cambridge, MA.

. 1982. Organizations and Organization Theory. Pitman, Marshfield, MA.

Polson, P. G. 1987. A quantitative theory of human-computer
interaction. J. M. Carroll, ed. Interfacing Thought. MIT Press,
Cambridge, MA. 184 235.
Ram, S., H. Jung. 1991. Forced adoption of innovation in organizations: Consequences and implications. J. Product Innovation
Management 8 117126.
Rice, R. E., C. Aydin. 1991. Attitudes towards using new organizational technology: Network proximity as a mechanism for
social information processing. Admin. Sci. Quart. 36 219 244.
Roberson, L. 1989. Assessing personal work goals in the organizational
setting: Development and evaluation of the work concerns inventory. Organ. Behavior and Human Decision Processes 44 345367.
Salancik, G. R., J. Pfeffer. 1978. Social information processing approach
to job attitudes and task design. Admin. Sci. Quart. 23 224 253.
Sichel, D. E. 1997. The Computer Revolution: An Economic Perspective.
The Brookings Institution, Washington, DC.
Straub, D., M. Limayem, E. Karahanna-Evaristo. 1995. Measuring
system usage: Implications for IS theory testing. Management
Sci. 41 1328 1342.
Taylor, S., P. A. Todd. 1995. Understanding information technology
usage: A test of competing models. Inform. Systems Res. 6 144 176.
Tybout, A. M., C. A. Scott. 1983. Availability of well-defined internal
knowledge and the attitude formation process: Information
aggregation versus self-perception. J. Personality and Social
Psych. 44 474 491.
Vallacher, R. R., J. Kaufman. 1996. Dynamics of action identification:
Volatility and structure in the mental representation of behavior. P. M. Gollwitzer and J. A. Bargh, eds. The Psychology of
Action: Linking Cognition and Motivation to Behavior. The Guilford Press, New York. 260 282.
, D. M. Wegner. 1987. What do people think theyre doing?
Action identification and human behavior. Psych. Rev. 94 315.
Venkatesh, V. 1999. Creation of favorable user perceptions: Exploring the role of intrinsic motivation. MIS Quart. 23 239 260.
, F. D. Davis. 1996. A model of the antecedents of perceived ease
of use: Development and test. Decision Sci. 27 451 481.
, M. G. Morris. 2000. Why dont men ever stop to ask for
directions? Gender, social influence, and their role in technology acceptance and usage behavior. MIS Quart. Forthcoming.
, C. Speier. 1999. Computer technology training in the workplace: A longitudinal investigation of the effect of mood. Organ.
Behavior and Human Decision Processes 79 128.
Vessey, I. 1991. Cognitive fit: A theory-based analysis of the graphs
versus tables literature. Decision Sci. 22 219 240.
Vroom, V. H. 1964. Work and Motivation. Wiley, New York.
Warshaw, P. R. 1980. A new model for predicting behavioral intentions: An alternative to Fishbein. J. Marketing Res. 17 153172.
Webster, J., L. K. Trevino. 1995. Rational and social theories as
complementary explanations of communication media choices:
Two policy-capturing studies. Acad. Management J. 38 15441572.

Accepted by John C. Henderson; received April 24, 1997. This paper has been with the author 15 months for 2 revisions.

204

Management Science/Vol. 46, No. 2, February 2000