Managing skills

Why do people use information technology? A critical review of the technology acceptance model

Paul Legris a,*

, John Ingham b , Pierre Collerette

c

a Université du Québec à Hull, Office of the President, Pavillon Lucien-Brault, Piece B-2076,

Case postale 1250, Succursale B, Hull, Que., Canada J8X 3X7 b Faculty of Business, Université de Sherbrooke, Sherbrooke, Que., Canada J1H 5N4

c Department of Administrative Science, Québec University in Hull, Hull, Que., Canada J8X 3X7

Received 19 June 2001; received in revised form 17 August 2001; accepted 15 December 2001

Abstract

Information systems (IS) implementation is costly and has a relatively low success rate. Since the seventies, IS research has

contributed to a better understanding of this process and its outcomes. The early efforts concentrated on the identification of

factors that facilitated IS use. This produced a long list of items that proved to be of little practical value. It became obvious that,

for practical reasons, the factors had to be grouped into a model in a way that would facilitate analysis of IS use.

In 1985, Fred Davis suggested the technology acceptance model (TAM). It examines the mediating role of perceived ease of

use and perceived usefulness in their relation between systems characteristics (external variables) and the probability of system

use (an indicator of system success). More recently, Davis proposed a new version of his model: TAM2. It includes subjective

norms, and was tested with longitudinal research designs. Overall the two explain about 40% of system’s use. Analysis of

empirical research using TAM shows that results are not totally consistent or clear. This suggests that significant factors are not

included in the models.

We conclude that TAM is a useful model, but has to be integrated into a broader one which would include variables related to

both human and social change processes, and to the adoption of the innovation model.

# 2002 Elsevier Science B.V. All rights reserved.

Keywords: Technology acceptance model; Information technology; Ease of use; Usefulness; IS use; Change management; Innovation

1. Introduction

1.1. Problem statement

Enterprises decide to invest in information systems

(IS) for many reasons; among these are: pressures to

cut costs, pressures to produce more without increasing

costs, and simply to improve the quality of services or

products in order to stay in business.

Despite considerable investments in IS, research

reports mixed results. A study performed in 1998

by the Standish Group 1

once again found that only

26% of all MIS projects, and less than 23.6% of large

company projects, are completed on time and within

budget, with all requirements fulfilled. In excess of

Information & Management 40 (2003) 191–204

* Corresponding author. Tel.: þ1-819-595-3900;

fax: þ1-819-773-1699. E-mail addresses: [email protected] (P. Legris),

[email protected] (J. Ingham),

[email protected] (P. Collerette).

1 Chaos: charting the seas of information technology, a special

compass report, The Standish Group International.

0378-7206/02/$ – see front matter # 2002 Elsevier Science B.V. All rights reserved.

PII: S 0 3 7 8 - 7 2 0 6 ( 0 1 ) 0 0 1 4 3 - 4

46% of projects were over budget, late, and with

fewer features and functions than originally speci-

fied. Almost one third of the projects (28%) were

cancelled.

Since the seventies, researchers have concentrated

their efforts on identifying the conditions or factors

that could facilitate the integration of IS into busi-

ness. Their search has produced a long list of factors

that seem to influence the use of technology [4].

From the mid-eighties, IS researchers [6,7] have

concentrated their efforts in developing and testing

models that could help in predicting system use. One

of them, technology acceptance model (TAM) was

proposed by Davis in 1986 in his doctoral thesis.

Since then, it has been tested and extended by many

researchers. Overall, TAM was empirically proven

successful in predicting about 40% of a system’s use

[3,14].

1.2. Research objectives

This article first discusses research using TAM,

with three main objectives in mind: (1) to provide a

critical analysis of the research methods; (2) to

highlight the convergence or divergence in results;

and (3) to bring out the added value of TAM in

explaining system use. This is done by studying the

various parts of the model and discussing the results

of a meta-analysis of empirical research done with

TAM.

1.3. Background: origin and overview of TAM

In their effort to explain system use, researchers

first developed tools for measuring and analysing

computer user satisfaction. As indicated by Bailey

and Pearson, it was natural to turn to the efforts of

psychologists, who study satisfaction in a larger sense.

In general terms, satisfaction is considered as the sum

of one’s feelings or attitudes toward a variety of

factors affecting the situation. Therefore, it is defined

as the sum of m user’s weighted reactions to a set of n

factors.

Satisfaction ¼ X

WijRij ðj ¼ 1;...; n; i ¼ 1;...; mÞ

where Rij is the reaction to factor j by individual i and

Wij is the importance of factor j to individual i.

Bailey and Pearson identified 39 factors (see

Appendix A) that can influence user satisfaction.

Faced with such a long list of factors, Bailey and

Pearson and others, worked to abbreviate it and thus

make it more practical. Cheney et al. grouped factors

into three categories of variables: (1) uncontrollable

(task technology and organisational time frame); (2)

partially controllable (psychological climate and sys-

tems development backlog); and (3) fully controllable

(end-user computing (EUC) training, rank of EUC

executive, and EUC policies).

Davis [8] and Davis et al. [10] proposed TAM to

address why users accept or reject information tech-

nology. Their model is an adaptation of the theory

of reasoned action (TRA, see Fig. 1) proposed by

Fishbein and Ajzen [12] to explain and predict the

behaviours of people in a specific situation. Fig. 2

present original version of TAM [8].

A key purpose of TAM is to provide a basis for

tracing the impact of external variables on internal

beliefs, attitudes, and intentions. It suggests that

perceived ease of use (PEOU), and perceived use-

fulness (PU) are the two most important factors in

explaining system use.

TRA and TAM propose that external variables

intervene indirectly, influencing attitude, subjective

Fig. 1. Theory of reasoned action.

192 P. Legris et al. / Information & Management 40 (2003) 191–204

norms, or their relative weight in the case of TRA, or

influencing PEOU and PU in the case of TAM.

Attitude towards using (AT) and behavioural inten-

tion to use (BI) are common to TRA and TAM, and

Davis used Fishbein and Ajzen’s method to measure

them. Davis chose not to keep the variable subjective

norms, because he estimated that it had negligible

effect on BI. In TAM2, Venkatesh and Davis recon-

sidered this choice [29].

2. Methodology

We reviewed the articles published from 1980 to the

first part of 2001 in periodicals known to include this

type of study. They were in:

� MIS Quarterly; � Decision Sciences; � Management Science; � Journal of Management Information Systems; � Information Systems Research and � Information and Management.

The bibliographical references of the articles

initially selected also allowed us to trace a number

of research findings. We also explored specialised

databases and other information sources available

on the WEB. Out of more than 80 articles consulted,

we kept 22 (covering 28 measurements) for analysis.

They were selected using the following criteria:

1. TAM is used in an empirical study;

2. the integrity of TAM is respected;

3. the research methodology is well described and

4. the research results are available and complete.

3. Findings

3.1. Diversified settings

In our review we examined the type of software

introduced, the size of the sample, and the presence of

the models tested. Table 1 presents the overall results.

To analyse this data, we grouped the studies under

three software tool categories: (1) office automation;

(2) software development; and (3) business applica-

tion (Table 2).

TAM was compared four times to either the TRA or

the theory of planned behaviour (TPB). Five times

subjective norm was added to the model.

3.2. Versions of TAM

In its original version, TAM had the following

components: PU, PEOU, AT, BI, and actual use (U).

Thus on the basis of the five components present and

taking into account the structure of the model, 10

relations could potentially be examined: (1) PEOU–PU;

(2) PU–AT; (3) PEOU–AT; (4) PU–BI; (5) PEOU–BI;

(6) AT–BI; (7) AT–U; (8) BI–U; (9) PEOU–U; and (10)

PU–U.

As shown in Table 3, no single study incorporated

all these relations, but they are all measured in at least

one study. Results of this analysis are summarised in

Table 4. This shows a high proportion of positive

results for all relations, but with a number of incon-

sistencies. These favourable results highlight variables

that are related to IT adoption intention, but it does not

mean that these variables are sufficient to predict IT

adoption.

Fig. 2. Original technology acceptance model.

P. Legris et al. / Information & Management 40 (2003) 191–204 193

Table 1

Methodological details

Author Software Sample size Model used (usually TAM)

Davis et al. [8] Text-editor 107 full time MBA students TAM þ TRA Davis [9] E-mail, text-editor 112 professionals and managers TAM

Mathieson [20] Spreadsheet 262 students course intro-management TAM þ TPB Davis et al. [10] Writeone, chartmaster 200 and 40 MBA students TAM, TAM

Subramanian [24] Voice mail system, customer dial-up

system

75 and 104 subjects TAM

Taylor and Todd [26] University computing, resource centre,

business school student

786 students TAM þ subjective norm þ perceived behavioural control

Taylor and Todd [27] University computing, resource centre,

business school student

786 students TAM þ TPB þ decomposed TPB

Keil et al. [18] Configuration software 118 salespersons TAM

Szajna [25] Electronic mail 61 graduate students TAM

Chau [6] Case 2500 IT professionals TAM modified for long- and short-term usefulness

Davis et al. [28] Three experiences with six software Total of 108 students TAM model of antecedents of perceived ease of use

Jackson et al. [16] Spreadsheet, database, word processor,

graphics

244, 156, 292, 210 students TAM validation of perceived usefulness and ease of

use instruments (each six items tools)

Igbaria and Craig [15] Personal computing 596 PC users TAM in small firms

Bajaj et al. [5] Debugging tool 25 students TAM þ loop back adjustments Gefen and Keil [13] Configuration software 307 salesman TAM testing for effect of perceived developers

responsiveness

Agarwal and Prasad [2] Word processing spreadsheet graphics 205 users of a Fortune100 company TAM testing for individual differences

Lucas and Spitler [19] Multifunctional workstation 54 brokers, 81 sales assistant of

financial company

TAM þ social norms and perceived system quality

Straub et al. [17] Microsoft windows 3.1 77 potential adopters, 153 users in a

corporation

Adaptation of TAM þ subjective norms

Hu et al. [14] Telemedicine software 407 physicians TAM

Dishaw and Strong [11] Software maintenance tools 60 maintenance projects in three

Fortune50 firms, no indications

of the number of subjects

TAM and task technology fit

Venkatesh and Davis [29] Four different systems in four

organisations

48 floor supervisors, 50 members

of personal financial services,

51 employees small accounting firm, 51

employees of small investment banking

Extension of TAM including subjective norms

and task technology fit

Venkatesh and Morris [30] Data and information retrieval 342 workers TAM þ subjective norms, gender and experience

1 9

4 P .

L eg

ris e t

a l./

In fo

rm a

tio n

& M

a n

a g

e m

e n

t 4

0 (2

0 0

3 )

1 9

1 –

2 0

4

Table 2

Software categories

Office automation tool [11] Software development tool [6] Business application tool [5]

Software used in the operation of an

office environment

Software used in application

development

Software used in the core business

process

Text-editor Programming tools

Spreadsheet Case tools MIS software (ERP)

Graphics presentation Debugging tools Production control tools

Electronic mail Software maintenance tools

Voice mail

Table 3

Type of relations found a

Author PEOU–PU PU–AT PEOU–AT PU–BI PEOU–BI AT–BI AT–U BI–U PEOU–U PU–U

Davis et al. [10]

Post training Yes Yes No Yes Yes Yes Yes

End semester Yes Yes Yes Yes Yes Yes Yes

Davis [8,9] Yes Yes Yes Yes Yes

Mathieson [20] Yes Yes Yes Yes Yes

Davis et al. [10]

Writeone Yes Yes Yes Yes Yes Yes

Chartmaster Yes Yes Yes Yes Yes Yes

Subramanian [24]

Voice mail No Yes No

Customer dial-up No Yes No

Taylor and Todd [26,27]

With experience Yes Yes Yes Yes No Yes

No experience Yes Yes Yes Yes No Yes

Taylor and Todd [26,27] Yes Yes Yes Yes No Yes

Keil et al. [18] Yes No Yes

Szajna [25]

Pre-implementation Yes Yes Yes Yes No No

Post-implementation Yes Yes Yes Yes No No

Chau [6] Yes Yes Yes

Davis et al. [10] Yes Yes Yes

Jackson et al. [16] No No Yes No Yes No

Igbaria et al. [15] Yes Yes Yes

Bajaj and Nidumolu [5] No Reverse Yes Yes No

Gefen and Keil [13] Yes No Yes

Agarwal and Prasad [1,2] Yes Yes Yes Yes Yes

Lucas and Spitler [19] Yes No No No No

Karahanna et al. [17]

Potential adopters Yes Yes Yes

Actual users Yes Yes Yes

Hu et al. [14] No Yes No Yes Yes

Dishaw and Strong [11] Yes Yes No Yes No No

Venkatesh and Davis [28,29] Yes Yes Yes Yes

Venkatesh and Morris [30] Yes Yes Yes

a Yes indicates that the relation was found to be significant and positive, blank the relation was not measured, no the relation was found to

be non-significant and reverse indicates that the relation was found to be significant but negative.

3.2.1. Attitude towards using and behavioural

intention to use

In its original form (Fig. 2), TAM included both AT

and BI as in TRA. Out of the 22 studies, only seven

included both AT and BI. Three included only AT,

while eight included only BI. This leaves four studies

that ignored both AT and BI, measuring only the direct

effect of PU and PEOU on use.

3.2.2. Use

The ultimate objective of TAM was to predict use,

and for this, a linear regression model was most often

used. To build the model, use has to be measured. In

eleven of the 22 studies, use was measured through

self-reporting. The method used normally consisted of

two or three questions about the frequency of use and

the amount of time spent using the system. In one

study, use was measured by an automatic measuring

tools. In 10 other studies, use was not measured: it was

either mandatory or this variable was ignored.

3.3. External variables

TAM postulates that external variables intervene

indirectly by influencing PEU and PU. Table 5 pre-

sents the external variables considered. We note that

there is no clear pattern with respect to the choice of

the external variables considered.

Table 4

Counting of relations

PEOU–PU PU–AT PEOU–AT PU–BI PEOU–BI AT–BI AT–U BI–U PEOU–U PU–U

Positive relation 21 12 10 16 10 7 3 10 4 8

Non-significant relation 5 1 3 3 3 4 0 1 5 5

Negative relation 0 1 0 0 0 0 0 0 0 0

Not tested 2 14 15 9 15 17 25 17 19 15

Table 5

External variables

Author External variable

Jackson et al. [16] Situational involvement, intrinsic involvement, prior use, argument of change

Igbaria et al. [15] Internal computing support, internal computing training, management support, external computing

support, external computing training

Bajaj and Nidumolu [5] No external variable

Gefen and Keil [13] Perceived developer responsiveness

Agarwal and Prasad [1,2] Role with regard to technology, tenure in workforce, level of education, prior similar experiences,

participation in training

Lucas and Spitler [19] Quality perceived subjectiveness

Karahanna et al. [17] Compatibility, trainability, visibility, result demonstrability

Hu et al. [14] No external varibale

Dishaw and Strong [11] Tool functionality, tool experinece, task technology fit, task characteristics

Venkatesh and Davis [28,29] Subjective norms, voluntariness, image, job relevance, output quality, result demonstrability

Venkateshand Morris [30] Gender, experience

Davis [8,9] No external variable

Davis et al. [10] No external variable

Mathieson [20] No external variable

Davis et al. [10] Output quality

Subramanian [24] No external variable

Taylor and Todd [26,27] Affect of experience

Taylor and Todd [26,27] No external variable

Keil et al. [18] No external variable

Szajna [25] No external variable

Chau [6] Implementation gap, transitional support

Davis et al. [10] Computer self efficacy, objective usability, direct experience

196 P. Legris et al. / Information & Management 40 (2003) 191–204

Research results provided by these studies confirm

that external variables are fully mediated by PEOU

and PU and that the addition of such variables con-

tributes marginally to the explanation of the variance

in system use. Actually, external variables provide a

better understanding of what influences PU and

PEOU, their presence guide the actions required to

influence a greater use.

3.4. Measures of PU and PEOU

3.4.1. Perceived usefulness

Table 6 presents the items used for measuring PU

and, when available, the reported internal consistency

of the resulting constructs. Davis, in his study of

PU, proposed a six items measurement tool. The

six items include the four items most commonly used:

(1) using (application) increases my productivity;

(2) using (application) increases my job performance;

(3) using (application) enhances my effectiveness on

the job; and (4) overall, I find the (application) useful

in my job. All measures of PU are found to lead to an

acceptable level of internal consistency (greater or

equal to 0.83).

3.4.2. Perceived ease of use

Table 7 presents the constructs present in the tools

for measuring PEOU. We observe that four items are

more frequently used: (1) learning to operate (the

application) is easy for me; (2) I find it easy to get

the (application) to do what I want to do; (3) the

(application) is rigid and inflexible to interact with;

and (4) overall, I find the (application) easy to use.

These are found to lead to a reasonable degree of

internal consistency (with alpha most of the time

greater than 0.79) in 12 articles or more.

Davis gave great attention to building a solid tool to

measure PEOU. His 1989 study proposes a six items

measurement tool, which includes the four most

commonly used items.

3.4.3. Attitude towards using and behavioural

intention to use

Where required, the studies use the constructs

suggested by Fishbein and Azjen. These constructs

demonstrated a good degree of reliability with all

alphas greater than 0.8 in 16 out of the 18 studies.

4. A meta-analysis

Research results with TAM have been, over the

years, generally consistent. When analysing the type

of relation between the different components of TAM,

we observed that in one case, the research findings

were contradictory. The relation between PU and AT is

found to be significant and positive in all but one study

[5].

Meta-analysis aims at integrating a large number of

results to determine if they are homogeneous. Statis-

tical methods are applied to summary statistics. The

focus is not on statistical significance but on the size of

treatment effects. The objective is a detailed review

that supports making a sound judgement on the aver-

age of the findings computed and on the reasons for

inconsistencies.

It would be helpful to investigate the homogeneity

of the relations between the components used in

TAM across the different studies. In order to do this,

the correlation coefficients between the components

observed must be available. Unfortunately, the coef-

ficient correlation matrices were present in only three

of the 22 studies examined. In most studies, a

measure of the strength of the relation was given

through the result of the computed linear regression.

In models that account for most of the factors,

measuring the total effect (direct and indirect) will

compare favourably to the results of the coefficient

correlation matrix.

To validate the possibility of using the data avail-

able (from the model results) we undertook a com-

parison of the results from the three studies where the

coefficient matrices were provided. In two, the

results were the same. In the third, the coefficient

correlations were slightly different from the total

effect between the components measured in the

models. For these reasons, we must be cautious,

because we cannot rely on strong statistical evidence.

Nevertheless, we conducted the meta-analysis to see

if trends appear.

We proceeded first by grouping the studies by type

of samples (students and non-students) and second

by grouping with software categories. We used the

meta-analysis procedure and software programs pro-

vided by Ralf Schwarzer on his WEB site (http://

www.fu-berlin.de/gesund/gesu_engl/meta_e.htm) to

process the data.

P. Legris et al. / Information & Management 40 (2003) 191–204 197

Table 6

Measuring PU a

Perceived

usefulness

Davis

[8,9]

Davis

et al. [10]

Mathieson

[20]

Davis

et al. [10]

Subramanian

[24]

Taylor and

Todd [26,27]

Taylor and

Todd [26,27]

Keil et al.

[18]

Szajna

[25]

Chau

[6]

Davis

et al. [10]

Alpha 0.902 0.91 0.963 0.94 pre,

0.95 post

0.95 0.9 short-term, 0.928

long-term

0.90

Using (application) improves the quality

of the work I do

X X

Using (application) gives me greater control

over my work

X

Application enables me to accomplish tasks

more quickly

X X X X

Application supports critical aspects of my job X

Using (application) increases my productivity X X X X X X X X X

Using (application) increase my job performance X X X X X X X X X X

Using (application) allows me to accomplish

more work than would otherwise be possible

X

Using (application) enhances my effectiveness

on the job

X X X X X X X X

Using (application) makes it easier to do my job X X X X X

Overall, I find the (application) useful in my job X X X X X X X X X X

Jackson

et al.

[16]

Igbaria

et al.

[15]

Bajaj and

Nidumolu

[5]

Gefen and

Keil

[13]

Agarwal and

Prasad [1,2]

Lucas and

Spitler [19]

Karahanna

et al. [17]

Hu et al.

[14]

Dishaw

and Strong

[11]

Venkatesh

and Davis

[28,29]

Venkatesh

and Morris

[30]

Alpha 0.83 0.94 0.96 0.93 0.95 0.91 0.90 0.89 0.98 0.86–0.98 0.93

Using (application) improves the quality of the

work I do

X X X

Using (application) gives me greater control over

my work

X

Application enables me to accomplish tasks

more quickly a

X X X X X

Application supports critical aspects of my job

Using (application) increases my productivity a

X X X X X X X X X X

Using (application) increases my job performance a

X X X X X X X

Using (application) allows me to accomplish

more work than would otherwise be possible

Using (application) enhance my effectiveness

on the job a

X X X X X X X X X X

Using (application) makesk it easier to do my job a

X X X X X

Overall, I find the (application) useful in my job a

X X X X X X X X X

a Iitems proposed by Davis.

Table 7

Measuring PEOU

Perceived ease

of use

Davis

[8,9]

Davis

et al.

[10]

Mathieson

[20]

Davis

et al.

[10]

Subramanian

[24]

Taylor

and Todd

[26,27]

Taylor

and Todd

[26,27]

Keil

et al.

[18]

Szajna

[25]

Chau

[6]

Davis

et al.

[10]

Jackson

et al.

[16]

Igbaria

et al.

[15]

Bajaj and

Nidumolu

[5]

Gefen

and Keil

[13]

Agarwal

and Prasad

[1,2]

Lucas

and Spitler

[19]

Karahanna

et al.

[17]

Hu et al.

[14]

Dishaw

and Strong

[11]

Venkatesh

and Davis

[28,29]

Venkatesh

and Morris

[30]

Alpha 0.91 0.93 0.938 0.88 0.903 0.82 0.96 0.900 >0.90 0.91 0.94 0.87 0.89 0.87 0.87 0.90 0.79 0.97 0.86–o.98 0.92

I find (application)

cumbersome to use

X X X1

Learning to operate (application)

is easy for me

X X X X X X1 X1 X X X X X X X1

Interacting with the (application)

is often frustrating

X X

I find it easy to get the (application)

to do what I want to do

X X X X X X X X X X X X X X X X

The (application) is rigid and

inflexible to interact with

X X1 X1 X1 X1 X1 X X1 X1 X1

It is easy for me to remember

how to perform tasks using the

(application)

X X

Interacting with the (application)

requires a lot of mental effort

X X X X X X1 X1 X1

My interaction with the (application)

is clear and understable

X X X X X

I find it takes a lot of effort to become

skilful at using the (application)

X X1 X1 X1 X1 X1 X X

Overall, I find the (application)

easy to use

X X X X X X X X X X X X X X X X X, X1 X X X X

Fig. 3. TAM2.

Table 8

Summary of research findings

Article Finding

Davis [8,9] TAM fully mediated the effects of system characteristics on use behaviour, accounting for 36% of the variance

in use

Perceived usefulness was 50% more influential than ease in determining use

Davis et al. [10] Perceived usefulness predicts intentions to use whereas perceived ease of use is secondary and acts through

perceived usefulness

Attitudes have little impact mediating between perceptions and intention to use

Relatively simple models can predict acceptance

Mathieson [20] Both models (TAM and TRA) predict intentions to use well

TAM is easier to apply, but provides only general information

TPB provides more specific information for developers

Davis et al. [10] Together, usefulness and enjoyment explained 62% (study 1) and 75% (study 2) of the variance in use

intentions

Usefulness and enjoyment were found to mediate fully the effects on use intentions of perceived output quality

and perceived ease of use

A measure of task importance moderated the effects of ease of use and output quality on usefulness but not

enjoyment

Subramanian [24] Perceived usefulness and not ease of use is a determinant of predicted future use

Taylor and Todd [26,27] Modified TAM explains use for both experienced and inexperienced users

Stronger link between behavioural intention and behaviour for experienced users

Antecedent variables predict inexperienced user’s intentions better

200 P. Legris et al. / Information & Management 40 (2003) 191–204

Table 8 (Continued )

Article Finding

Taylor and Todd [26,27] All models performed well based on fit and explanation of behaviour

TPB provides a fuller understanding of intentions to use

In TAM attitudes are not significant predictors of intention to use

Keil et al. [18] Usefulness is a more important factor than ease of use in determining system use

Ask/tool fit plays a role in shaping perceptions of whether or not a system is easy to use

Szajna [25] Questions self-report measures vs. actual measurement of use

Experience component may be important in TAM

Chau [6] Findings indicate that ease of use has the largest influence on software acceptance

Davis et al. [10] Individual’s perception of a particular system’s ease of use is anchored to her or his general computer self-

efficacy at all times

Objective usability has an impacton ease of use perceptions about a specific system only after direct

experience with the system

Jackson et al. [16] Direct effect of situational involvement on behavioural intention as well as attitude is significant in the

negative direction

Attitude seems to play a mediating role

Intrinsic involvement plays a significant role in shaping perceptions

Igbaria et al. [15] Perceived ease of use is a dominant factor in explaining perceived usefulness and system use, and PU has a

strong effect on use

Exogenous variables influence both PEOU and PU particularly management support and external support

Relatively little support was found for the influence of both internal support and internal training

Bajaj and Nidumolu [5] Past use apparently influences the ease of use of the system and is a key factor in determining future use

Gefen and Keil [13] Proposes that IS managesrs can influence both the perceived usefulness and the perceived ease of use of an

IS through a constructive social exchange with the user

Agarwal and Prasad [1,2] It appears that there may be nothing inherent in individual differences that strongly determines acceptance

(use)

Identifies several individual difference variables (level of education, extent of prior experiences, participation

in training) that have significant effects on TAM’s beliefs

Lucas and Spitler [19] Field setting, organizational variables such as social norms and the nature of the job are more important in

predicting use of the technology than are user’s perceptions of the technology

Karahanna et al. [17] Pre-adoption attitude is based on perceptions of usefulness, ease of use, result demonstrability, visibility and

triability

Post-adoption attitude is only based on instrumental beliefs of usefulness and perceptions of image

enhancements

Hu et al. [14] TAM was able to provide a reasonable depiction of user’s intention to use technology

Perceived usefulness was found to be a significant determinnt of attitude and intention

Perceived ease of use was not a significant determinant

Dishaw and Strong [11] Suggests an integration of TAM and task-technology fit constructs

Integrated model leads to a better understanding of choices about using IT

Venkatesh and Davis [28,29] The extended model accounted for 40–60% of the variance in usefulness perceptions and 34–52% of the

variance in use intentions

Both social influence process (subjective norm, voluntariness, and image) and cognitive instrumental

processes (job relevance, output quality, result demonstrability, and perceived ease of use) significantly

influenced user acceptance

Venkatesh and Morris [30] Compared to women, men’s technology use was more strongly influenced by their perceptions of usefulness

Women were more strongly influenced by perceptions of ease of use and subjective norms, although the

effect of subjective norms diminished over time

P. Legris et al. / Information & Management 40 (2003) 191–204 201

Because of the statistical shortfall, we limit the pre-

sentation of the findings to the general conclusion: in

all grouping except one, the research findings were

found to be heterogeneous.

The only situation where the results of the research

findings were found to be homogeneous was when

studies were grouped by type of software and only for

students. The results from the meta-analysis tend

to support the view of some researchers [1,19,25]

who suggest that TAM should be modified to include

other components in order to explain consistently

more than 40% of system use. As for the homogeneity

of the results with TAM when dealing with students,

we believe that it reveals the robustness of TAM

when not exposed to all the factors (structure, roles

and responsibilities) of the real world environment,

since students function in a simpler environment.

5. Conclusion

TAM has proven to be a useful theoretical model in

helping to understand and explain use behaviour in IS

implementation. It has been tested in many empirical

researches and the tools used with the model have

proven to be of quality and to yield statistically

reliable results.

From its original model, TAM has evolved over

time. This is the model used by Venkatesh and Morris

[30] (Fig. 3). The notion of time has been included in

the analysis of the factors that influence use. Research

has shown that the influence of some factors on

intention to use IS, varies at different stages in the

IS implementation process. Rogers’ work [23] on

innovation introduced also: triability; relative advan-

tage; complexity; compatibility; and observability.

The following Table 8, inspired by Lucas, presents

the summary of our findings. Here we notice that,

although the results are mostly convergent, there are

situations where they are conflicting. A closer analysis

of these situations [15–17] points out that the original

model of TAM needed to be improved and that the

latest model goes a long way in that direction. How-

ever, even if established versions include additional

variables, the model hardly explains more than 40% of

the variance in use.

In conclusion, we underline three limits of TAM

research to date.

1. Involving students

Nine of the studies involved students. Although

this minimised the costs, we think that research

would be more better if it was performed in a

business environment.

2. Type of applications

We also notice that most studies examined the

introduction of office automation software or

systems development applications. We think that

research would benefit from examining the

introduction of business process applications.

3. Self-reported use

Since most of the studies do not measure system

use, what TAM actually measures is the variance

in self reported use. Obviously this [9,24] is not a

precise measure. Not only is it difficult to measure

rigorously, but it also involves problems. At best,

self reported use should serve as a relative

indicator.

The following is an example of the difficulty with

self reported use (La Presse Montréal, Tuesday 17

October 2000).

Observers in public washrooms in New Orleans,

New York, Atlanta, Chicago and San Francisco

noted that only 67% of the persons washed their

hands after visiting the toilet cabinet. When 1201

Americans, in a telephone survey, were asked if

they washed their hands after going to the bath-

room, 95% answered yes.

Since our objective is to explain use, greater empha-

sis should be put on measurement. Another important

limitation of TAM is in considering IS to be an

independent issue in organisational dynamics.

Research in the field of innovation and change man-

agement suggests that technological implementation

is related to organisational dynamics, which will have

a strong impact on the outcomes. Orlikowski and

Hofman [21] acknowledge that the effectiveness of

any change process relies on the interdependence

between the technology, the organisational context,

and the change model used to manage the change. This

support the suggestion that it may be difficult to

increase the predictive capacity of TAM if it is not

integrated into a broader model that includes organi-

sational and social factors. Orlikowski and Tyre [22]

found that effective IS implementation tend to follow a

pattern where the management proceeds with disjoint

202 P. Legris et al. / Information & Management 40 (2003) 191–204

periods of intensive implementation, rather than with

continuous improvement. This information is particu-

larly useful for managers who have to make decisions

about implementation strategies.

Appendix A. Factors affecting information system satisfaction (Bailey and Pearson)

Following are the factors affecting IS satisfaction:

1. Top management involvement

2. Organisational competition with the EDP unit

3. Priorities determination

4. Charge-back method of payment for services

5. Relationship with the EDP staff

6. Communication with the EDP staff

7. Technical competence of the EDP staff

8. Attitude of the EDP staff

9. Schedule of products and services

10. Time required for new development

11. Processing of change requests

12. Vendor support

13. Response/turnaround time

14. Means of input/output with EDP centre

15. Convenience of access

16. Accuracy

17. Timeliness

18. Precision

19. Reliability

20. Currency

21. Completeness

22. Format of input

23. Language

24. Volume of output

25. Relevancy

26. Error recovery

27. Security of data

28. Documentation

29. Expectations

30. Understanding of systems

31. Perceived utility

32. Confidence in systems

33. Feeling of participation

34. Feeling of control

35. Degree of training

36. Job effects

37. Organisational position of the EDP function

38. Flexibility of systems

39. Integration of systems

References

[1] R. Agarwal, J. Prasad, The role of innovation characteristics

and perceived voluntariness in the acceptance of informa-

tion technologies, Decision Sciences 28 (3), 1997, pp. 557–

582.

[2] R. Agarwal, J. Prasad, Are individual differences germane to

the acceptance of new information technologies? Decision

Sciences 30 (2), 1999, pp. 361–391.

[3] I. Ajzen, M. Fishbein, Understanding Attitudes and predict-

ing Social Behavior, Prentice-Hall, Englewood Cliffs, NJ,

1980.

[4] J.E. Bailey, S.W. Pearson, Development of a tool for

measuring and analysing computer user satisfaction, Manage-

ment Sciences 29 (5), 1983, pp. 530–545.

[5] A. Bajaj, S.R. Nidumolu, A feedback model to understand

information system usage, Information and Management 33,

1998, pp. 213–224.

[6] P.Y.K. Chau, An empirical investigation on factors affecting

the acceptance of CASE by systems developers, Information

and Management 30, 1996, pp. 269–280.

[7] P.H. Cheney, R.I. Mann, D.L. Amoroso, Organizational

factors affecting the success of end-user computing, Journal

of Management Information Systems III (1), 1986, pp. 65–

80.

[8] F.D. Davis, Perceived usefulness, perceived ease of use, and

user acceptance of information technologies, MIS Quarterly

13 (3), 1989, pp. 319–340.

[9] F.D. Davis, User acceptance of information technology:

system characteristics, user perceptions, and behavioral

impacts, International Journal of Man Machine Studies 38,

1993, pp. 475–487.

[10] F.D. Davis, R. Bagozzi, P.R. Warshaw, User acceptance of

computer technology: a comparison of two theoretical

models, Management Science 35 (8), 1989, pp. 982–1003.

[11] M.T. Dishaw, D.M. Strong, Extending the technology

acceptance model with task-technology fit constructs, In-

formation and Management 36, 1999, pp. 9–21.

[12] M. Fishbein, I. Ajzen, Belief, Attitude, Intention, and

Behavior: An Introduction to Theory and Research, Addi-

son-Wesley, Reading, MA, 1975.

[13] D. Gefen, M. Keil, The impact of developer responsiveness

on perceptions of usefulness and ease of use: an extension

of the technology acceptance model, The DATA BASE

for Advances in Information Systems 29 (2), 1998, pp.

35–49.

[14] P.J. Hu, P.Y.K. Chau, O.R. Liu Sheng, K. Yan Tam,

Examining the technology acceptance model using physician

acceptance of telemedicine technology, Journal of Manage-

ment Information Systems 16 (2), 1999, pp. 91–112.

[15] M. Igbaria, N. Zinatelli, P. Cragg, A. Cavaye, Personal

computing acceptance factors in small firms: a structural

equation model, MIS Quarterly, September (1997) 279–

302.

[16] C.M. Jackson, S. Chow, R.A. Leitch, Toward an under-

standing of the behavioral intention to use an information

system, Decision Sciences 28 (2), 1997, pp. 357–389.

P. Legris et al. / Information & Management 40 (2003) 191–204 203

[17] E. Karahanna, D.W. Straub, N.L. Chervany, Information

technology adoption across time: a cross-sectional compar-

ison of pre-adoption and post-adoption beliefs, MIS Quarterly

23 (2), 1999, pp. 183–213.

[18] M. Keil, P.M. Beranek, B.R. Konsynski, Usefulness and ease

of use: field study evidence regarding task considerations,

Decision Support Systems 13, 1995, pp. 75–91.

[19] H.C. Lucas, V.K. Spitler, Technology use and performance: a

field study of broker workstations, Decisions Sciences 30 (2),

1999, pp. 291–311.

[20] K. Mathieson, Predicting user intentions: comparing the tech-

nology acceptance model with the theory of planned behavior,

Information Systems Research 2 (3), 1991, pp. 173–191.

[21] W.J. Orlikowski, J.D. Hofman, An improvisational model for

change management: the case of groupware technologies,

Sloan Management Review Winter (1997) 11–21.

[22] W.J. Orlikowski, M.J. Tyre, Exploiting opportunities for

technological improvement in organizations, Sloan Manage-

ment Review (1993) 13–26.

[23] E.V. Rogers, Diffusion of Innovations, 4th Edition, The Free

Press, New York, 1995.

[24] G.H. Subramanian, A replication of perceived usefulness and

perceived ease of use measurement, Decision Sciences 25 (5/

6), 1994, pp. 863–874.

[25] B. Szajna, Empirical evaluation of the revised technology

acceptance model, Management Science 42 (1), 1996, pp. 85–

92.

[26] S. Taylor, P. Todd, Understanding information technology

usage: a test of competing models, Information Systems

Research 6 (2), 1995, pp. 144–176.

[27] S. Taylor, P. Todd, Assessing IT usage: the role of prior

experience, MIS Quarterly, December (1995) 561–570.

[28] V. Venkatesh, F.D. Davis, A model of the antecedents of

perceived ease of use: development and test, Decision

Sciences 27 (3), 1996, pp. 451–481.

[29] V. Venkatesh, F.D. Davis, A theoretical extension of the

technology acceptance model: four longitudinal field studies,

Management Science 46 (2), 2000, pp. 186–204.

[30] V. Venkatesh, M.G. Morris, Why do not men ever stop to ask

for directions? Gender, social influence, and their role in

technology acceptance and usage behavior, MIS Quarterly 24

(1), 2000, pp. 115–139.

Paul Legris, Counsellor to the President,

is an expert in computer science and

public administration. He has accumu-

lated more than 20 years experience in

management positions in the fields of

information technology and administra-

tion in general. He is particularly inter-

ested in improving the success rate of

the integration of technology into com-

pany business processes and is pursuing

research in that field.

John Ingham is a Professor and

Researcher of Information Systems

Management at the University Sher-

brooke (Canada). He has been for several

years Associate Dean and Dean of the

Faculty of Business. His current re-

search interests include: virtual teams,

the management of electronic commerce,

information systems strategic planning

and enterprise resource planning in a

global context.

Pierre Collerette is a Professor and

Researcher in management at Québec

University in Hull (Canada). He has

published a number of works in the field

of organisational change and manage-

ment structures. Aside from his aca-

demic activities, he holds a management

position and has been a consultant in

many projects in Canada and in Europe.

204 P. Legris et al. / Information & Management 40 (2003) 191–204