The theory of user acceptance and use of technology (UTAUT): A meta-analytic review of empirical findings

Abstract

The unified theory of acceptance and use of technology (UTAUT) proposes that performance expectancy, effort expectancy, and social influence predict behavioral intention towards the acceptance of information technology. The theory further proposes that facilitating conditions and behavioural intention predicts use behavior in the acceptance of information technology. Ever since its inception, the theory has been assessed using different applications, and it has become a dè factor model of measuring user acceptance. Nonetheless, in terms of statistical significant magnitude and direction, reports on the model are diverse. Therefore, in this study, based on 37 selected empirical studies, a meta-analysis was conducted in order to harmonize the empirical evidence. The outcome of the study suggests that only the relationship between performance expectancy and behavioural intention is strong, while the relationships between effort expectation, social influence and behavioural intention are weak. Similarly, the relationship between facilitating condition, behavioural intention and use behaviour is also weak. Furthermore, the significance of the relationship between facilitating condition and use behaviour does not pass the fail safe test while the significance of the relationship between behavioural intention and use behaviour does not pass the fail safe test satisfactorily. Implications for further studies are also discussed.

Full text

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
48
THE THEORY OF USER ACCEPTANCE AND USE OF
TECHNOLOGY (UTAUT): A META-ANALYTIC REVIEW OF
EMPIRICAL FINDINGS
1
AYANKUNLE ADEGBITE TAIWO,
2
ALAN G. DOWNE
1
Department of Computer and Information science, Universiti Technologi PETRONAS,Malaysia.
2
Assoc. Prof., Department of Marketing and Management, Curtin University Sarawak, Malaysia
E-mail:
1
kunletaiwong@gmail.com,
2
alan.downe@curtin.edu.my
ABSTRACT
The unified theory of acceptance and use of technology (UTAUT) proposes that performance expectancy,
effort expectancy, and social influence predict behavioral intention towards the acceptance of information
technology. The theory further proposes that facilitating conditions and behavioural intention predicts use
behavior in the acceptance of information technology. Ever since its inception, the theory has been assessed
using different applications, and it has become a dè factor model of measuring user acceptance.
Nonetheless, in terms of statistical significant magnitude and direction, reports on the model are diverse.
Therefore, in this study, based on 37 selected empirical studies, a meta-analysis was conducted in order to
harmonize the empirical evidence.
The outcome of the study suggests that only the relationship between performance expectancy and
behavioural intention is strong, while the relationships between effort expectation, social influence and
behavioural intention are weak. Similarly, the relationship between facilitating condition, behavioural
intention and use behaviour is also weak. Furthermore, the significance of the relationship between
facilitating condition and use behaviour does not pass the fail safe test while the significance of the
relationship between behavioural intention and use behaviour does not pass the fail safe test satisfactorily.
Implications for further studies are also discussed.
Keywords: Information Systems (IS), Adoption, UTAUT, Meta-Analytic Review.
1. INTRODUCTION
Information technology pervades the
international community from programmable home
appliances to organization applications. Increase in
technological innovation and application with
awesome advantages brought changes to human life
and work endeavours. As people, organizations and
governments moved towards the use of Information
Technology. Such move of change has increased
the human computer interaction, which is the sole
aim of performing a task (Card, Moran and Newell
,1983). Interaction between humans and computers
is affected by quite a number of human factors and
its characteristics (Whitley, 1997), to which studies
have come up with theories and models to
investigate factors that influences humans to use
computers and its applications.
The design, development and acceptance of
information technologies have received substantial
attention in the past few decades. Many theoretical
models have been proposed to give explanations to
end users acceptance behaviour. The newest
amongst them is the Unified theory of adoption and
use of technology (UTAUT) by Venkatesh et al.
(2003), which has been applied and empirically
tested in different domains. Since its inception
many empirical studies have been conducted using
UTAUT. The model is believed to be more robust
than other Technology acceptance model in
evaluating and predicting technology acceptance
(Venkatesh et al., 2003). Although, the model has
been widely used, tested and validated, the outcome
of empirical studies has been inconclusive in
respect to the magnitude, direction and significance
of the relationships amongst the model. In social
sciences the issue of variety in statistical
significance is common because of complexity in
human behaviour. Therefore, mixed outcomes in
different studies are not uncommon, but it does
undermine the accuracy of the models, UTAUT
inclusive. Consequently, identifying users’ history
towards technology acceptance is difficult and

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
49
complicated for the academia and information
technologist.
Information Systems (IS) researchers,
Information Technology (IT) managers and e-
commerce decision makers can benefit from the
importance of meta-analysis on UTAUT as a
knowledge cumulating tool (Hwang and
Schmidt,2011) by having better understanding of
concrete pre-cursors to users acceptance towards a
technology and its applications. Armed with this
knowledge IT managers and other decision makers
can take more successful steps in attaining increase
in technological patronage and usage. Studies have
shown that to achieve a top level IT management
success, accurate IT prescription is of paramount
importance (Benbasat & Zmud, 1999).
The objective of this study is to investigate the
validity of UTAUT and reveal how much this
validity is substantiated in present literature. In
order to achieve this we harmonized existing results
on UTAUT through a meta-analysis. Integrating
empirical results of the theory can assist in
understanding the application of UTAUT to variety
of technology in general. Meta-analysis also fosters
examination of relationship between the dimensions
of a model as a whole. Thus, analyzing
relationships between the constructs of UTAUT
with a larger sample of subjects becomes feasible
than any individual study.
The outline of this study is as follows, a revision
on UTAUT with discrepancies and consistent
results in the existing literature, methods of study
selection and coding of empirical findings based on
37 carefully selected studies. We concluded by
discussing the outcome of the study, limitation of
the study and implications for future studies. We
anticipate that the outcome of this study can be
relatively used as a point of reference while testing
UTAUT in the nearest future.
2. LITERATURE REVIEW
The UTAUT is a unified model that was
developed by Vankatesh et al (2003) based on
social cognitive theory with a combination of eight
prominent information technology (IT) acceptance
research models. The authors examined the
predictive validity of eight models in determining
the behavioural intention and usage to allow fair
comparison of the models. The eight models are;
The Theory of reasoned action (TRA), The theory
of Planned behavior (TPB), The technology
acceptance model (TAM), The motivational Model
(MM), A model combining the technology
acceptance model and the theory of planned
behavior (C-TAM-TPB) ,The model of PC
Utilization (MPCU),The innovation diffusion
theory (ID) and Socio Cognitive Theory (SCT).
The unified model outperformed the eight
individual models (adjusted variance (R
2
) of 70
percent). The UTAUT model uses four core
determinants of usage and intention (performance
expectancy, effort expectancy, social influence, and
facilitating conditions) alongside with four
moderators (gender, age, experience and
voluntariness of use) of key relationships. Previous
works that have used UTAUT are briefly discussed
in the following paragraphs.
AlAwadhi and Morris (2008) investigated the
adoption of e-government services using UTAUT,
the survey was carried out on 880 students revealed
that performance expectancy, effort expectancy and
peer influence determine students’ behavioural
intention. Similarly facilitating conditions and
behavioural intentions determine students’ use of e-
government services. Also, Biemans, Swaak,
Hettinga & Schuurman (2005) used the UTAUT
model to examine nurses behavioural intentions
towards the use of Medical Teleconferencing
Application, the study revealed that performance
expectancy and effort expectation are high
predictors of behavioural intention but social
influence prediction power is low. In a cross
cultural study of IT adoption, Oshlyansky,Cairns
and Thimbleby (2007) found that performance
expectancy, effort expectancy and social influence
predicts use intention. Furthermore, Šumak,
Polančič and Heričko (2010) found that social
influence have a significant impact on students
behavioural intention to use moodle and students’
behavioural intentions is a powerful predictor of the
use of the e-learning system. Cheng, Liu, Song and
Qian (2008) investigated the validity of UTAUT
using 313 intended users of Internet banking in
China, the result suggest that performance
expectancy and social influence are strong
predictors of behavioural intention. In a similar
study, Cheng, Liu, and Qian (2008) found
performance expectancy and social influence of the
UTAUT constructs as predictors of users
behavioural intention towards internet banking.
In addition, an empirical study by Fang, Li, and Liu
(2008) suggests that performance expectancy, effort
expectancy and social influence significantly
predicts managers intention to engage in knowledge
sharing using web2.0. Maldonado, Khan, Moon and
Rho (2009) examined the acceptance of an e-
learning technology in secondary school in Peru,

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
50
240 Students took part in the survey. Result from
their study suggests that social influence
significantly predicts behavioural intention. In the
same study, Maldonado et al. (2009) found
behavioural intention to significantly predict use
behaviour. Carlsson, Carlsson and Hyvönen (2006)
examined the acceptance of mobile telephone and
found that performance expectancy, effort
expectancy and social influence are predictors of
behavioural intention.
Also, Wu, Tao and Yang (2007) investigated the
acceptance of 3G services in Taiwan and found
performance expectancy and social influence as
predictors of behavioural intention. Interestingly,
the authors also found performance expectancy,
effort expectation, social influence and facilitating
conditions as predictors of use behaviour. He and
Lu (2007) further suggest that performance
expectancy and social influence are predictors of
behavioural intention towards consumer’s
acceptances of mobile advertising. The authors also
found that facilitating condition and behavioural
intention predicts use behaviour. Cheng, Liu, Qian
& Song (2008) examined the acceptance of internet
banking, results suggest that performance
expectancy and social influence predicts intention.
Figure1. Unified Theory Of Acceptance And Use Of Technology (UTAUT) (Source: Venkatesh Et Al., 2003).
As much as some studies have supported that the
four predictive factors of UTAUT predicts intention
and use behavior, results from some other studies
suggest otherwise. Li & Kishore (2006) studied the
Use of Online Community Weblog Systems, the
results indicated that scales for the four constructs
in UTAUT including performance expectancy,
effort expectancy, social influence, and facilitating
conditions have invariant true scores across most
but not all subgroups. The authors expressed need
for caution when interpreting UTAUT. In a
structured PLS-Graph Conceptual Model,
Tibenderana and Ogao (2008) found performance
expectancy and social influence to be non-
significant in predicting behavioural intention to
use electronic Library services in Ugandan
Universities. Performance expectancy, effort
expectancy and social influence were found to be
non-significant in predicting intention in a study
investigating the acceptance of an interface robot
and a screen agent by elderly users
(Heerink,Kröse,Wielinga and Evers, 2009).
Similarly, Šumak, Polančič and Heričko (2010)
suggested that performance expectancy and effort
expectancy are non-significant predictor of
behavioural intention. In a related study, Cheng,
Liu, Song and Qian (2008) discovered that effort
expectancy does not significantly predict
behavioural intention. Similar studies have also
found effort expectancy to be non-significant in
predicting behavioural intention (See Cheng, Liu
and Qian 2008; He and Lu 2007;Wu, Tao and Yang
2007). In a study to investigate the role played by
motivation in e-learning technology adoption,
Maldonado, Khan, Moon and Rho (2009) found
facilitating condition to be non- significant in
predicting use behaviour. Cheng, Liu, Qian, Song
(2008) also examined the acceptance of internet
banking and found that effort expectancy does not
predict customers intention to use internet banking.
In the context of eGovernment, Schaupp, Carter
and Hobbs (2009) investigated the acceptance of E-
Filing by the American tax payers. Results from the
study suggest that performance expectancy and

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
51
social influence predicts behavioural intention.
Interestingly, the study revealed that effort
expectancy is not a predictor of behavioural
intention. The inconsistency in the outcomes of the
studies on UTAUT leaves the output of the
relationships in the model inconclusive.
3. METHODOLOGY
Glass (1976) defined meta-analysis as statistical
analysis of a large collection of analysis results for
the purpose of integrating the findings. Shaughnessy
et al. (2006) further defined meta-analysis as the
approach in which data is summarized and reported.
Lastly, DeCoster (2004) described meta-analysis as
a method used to provide information to support a
particular theoretical statement concerning strength
or consistency of a specific relationship. Following
the approach of Lipsey & Wilson, (2001) and Ma
and Liu (2004) we employed their approach for data
collection, selection criteria, articles usage
permission, coding and subsequent data analysis.
3.1 Selection of Studies
A detailed literature search was performed using
Science Direct, Emerald, ProQuest, EBSCO,
PsycArticles, and Dissertation Abstracts Online,
IEEE, GoogleScholar and ACM. A compendium of
studies on technology acceptance that used UTAUT
was gathered using search words such as "UTAUT,
User Acceptance, and Technology Adoption".
Furthermore, we identified the use of the UTAUT
model using references in other articles (Lipsey &
Wilson, 2001). In order to be included in the meta-
analysis, the paper has to meet the following criteria
adapted from Ma and Liu (2004):
a. Be a behavioural study.
b. Involve technology investigation.
c. Involves empirical testing of UTAUT directly or
indirectly.
d. Reported correlation co-efficient between the
constructs of UTAUT or other values that can be
converted to correlations.
e. Reports a sample size.
f. It must be published or dated after 2003 when
UTAUT was first published to 2011.
In addition, effect sizes (r) from various samples
vary, therefore, studies that fail to report enough
statistical data to calculate effect sizes of the study
were not included in the meta-analysis. A general
concern while conducting a meta-analysis is
publication bias. Owing to the fact that only studies
with statistical significant results are published in
academic journals, the sizes of the effect sizes
reported in articles are larger than unpublished
studies (Ma and Liu, 2004). Therefore, reporting
only the published papers will result into file
drawers problem since an important method in meta-
analysis is the calculation of the average effect size
of individual studies. There is every likelihood that
meta-analytic results may be inflated due to the file-
drawers problem. In order to avoid this, we included
academic conference papers from the Association of
computer machinery, IEEE digital library and
Proquest Academic database for online thesis,
besides the papers from journals.
Furthermore, we took caution in assuming that
individual study is independent (independent effect
sizes), which is a general assumption in meta-
analysis. Because some authors violate this by
reporting two or more correlation based on a single
sample. Thus, we cross checked to see if the
correlations were not based on the same study before
final selection. In this study, a total of 96 empirical
studies were discovered, seventy one (71) studies
were without correlation coefficient or other
statistical metrics that can be used. We corresponded
with 20 authors with contact details in their paper
and 12 of the authors provided us with inter-item
correlations between the constructs in their study.
Amongst the 37 selected information systems (IS)
studies, there are 24 Journal Papers, 6 conference
papers, 7 PhD thesis papers. A total of 153
correlation coefficients were obtained from the
studies.
3.2 Variables Recorded
The six (6) variables examined in each study were
coded as performance expectation (PE), effort
expectancy (EE), social influence (SI), facilitating
condition (FC) behavioural intention (BI) and Use
Behaviour (UB). The calculation of effect sizes and
other derivatives of meta-analysis were based on
these important variables of the UTAUT model.
Other variables investigated by Venkatesh et al.
(2003) but without significant effect were not
included in this study.
3.3 Computation of Effect Sizes
Correlation co-efficient (r) was the effect size metric
chosen for this study because of its wide
accessibility and availability in technology accepted
literature, also because of its ease of interpretation
and availability of formulae to convert other test
statistics to correlation coefficient. In many studies,
it is not uncommon to have researchers reporting
different statistical value such as correlation
coefficient, F,t-value, p-value and chi square.

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
52
Nonetheless, achieving insight into the depth of the
strength of the relationships between factors in a
study becomes cumbersome without a uniform
metric. Therefore, converting metrics into a uniform
format becomes pertinent before a meta-analysis can
be conducted. In this study, we adopted the
Pearson’s correlation co-efficient as effect size index
representing the empirical strength of the
relationship between each pair of the UTAUT
construct. We followed the approach described by
Lipsey & Wilson (2001) and Ma & Liu (2004), for
each of the pair of the UTAUT construct:
performance expectancy (PE), effort expectancy
(EE), social influence (SI), faciliatating condition
(FC), behavioural intention (BI) and use behaviour
(UB) the effect size was computed such that it is
simply a correlation coefficient (r) if reported,
otherwise a conversion is made using equation (1) if
other metrics such as t-value was reported. This
procedure by Rosenthal (1984) and has been widely
adopted by several studies (See Szymansky&
Henard, 2001; Ma and Liu 2004). The effect sizes of
variables in each study were computed to access
prediction effect towards behavioural intention and
use behaviour. Effect sizes reported by authors were
not recalculated but were used directly. The
computed outcome of effect values were computed
into excel spreadsheet. Generally, about 96% of the
effect sizes were calculated using the means and the
mean and standard deviation spreadsheet. 4% of the
effect sizes were calculated using the F or t test
spreadsheet.
…..equation (1)
4. DATA ANALYSIS
The analysis of the data computed was reported in
two phases. The first phase described the range,
direction, statistical significance and the sample size
of the correlation. The multiplicities of the existing
findings were described. The univariate analysis of
the correlation was investigated in the second phase.
Therefore, statistical significance and essential
tendencies of the findings were deducted.
4.1 The descriptive Statistics
Using the 153 correlation coefficients obtained from
the studies, Table 1 shows that some studies did not
report all the five correlation or their equivalents.
Out of the 153 correlation coefficients, (43) PE-BI
correlations were obtained from (37) studies, (42)
PE-BI correlations were reported from 36 studies,
(36) SI-BI correlations were reported from 31
studies, 16 FC-BI correlations were reported from
13 studies and 16 BI-UB correlations were reported
from 13 studies. The number of studies for PE-BI
and EE-BI were approximately the same, the
number of studies for SI-BI is a little lower than the
first two. While the number of studies for FC-BI and
BI-UB were low and reporting the same low 16
correlations from 13 studies respectively.
Concerning the strength of individual correlation
coefficient, Table1 shows that the range of the effect
sizes moves from insignificant to strongly
significant. Although, most studies reported
significant results, it could be noted that some of the
studies reported insignificant results. All the five
relationships reported high positive significance but
PE-BI reports the highest positive significant
correlations. Moreover, FC-BI and BI-UB have
equal but highest negative non-significant
correlation. Furthermore, the sample size varies
from one study to another, as the sample size was as
low as 41 in a study, while it is as high as 1607 in a
related study. The average sample size shows that
number of subjects in the PE-BI, EE-BI and SI-BI
are very close while the sample size for both FC-BI
and BI-UB are the same and lower than the first
three relationships.
4.2 Direct Effect Analysis
Using the simple means, sample size adjusted mean
and the Fisher r to Z transformation method (Fisher
1932; Lipsey and Wilson (2001). The average of all
the individual effect sizes is regarded as a simple
mean. A sample size-weighted average of the
individual effect size is considered as sample size
adjusted mean (equation 2). Correlations were
transformed to Fisher's Z (using equation 3) for
analysis and later back-transformed in the
correlation metrics for result presentation.
..equation (2)
..equation (3)

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
53
Table1:Summary Of Selected Correlation
Link No of
Studies No of
Correlation
Coefficients
Range of
correlation Positive
significant
correlation
Negative
significant
correlation
Range of Sample
size Cumulative
sample size
from To # % # % from To Ave
PE-BI 37 43 0.10 0.70 42 97.7 1 2.3 41 1607
257 11057
EE-BI 36 42 0.07 0.70 40 95.2 2 4.8 41 1607
262 10995
SI-BI 31 36 0.12 0.89 35 94.6 2 5.4 41 1607
252 9304
FC-UB 13 16 0.11 0.79 13 81.3 3 18.7 55 722 191 3048
BI-UB 13 16 0.12 0.84 13 81.3 3 18.7 55 722 191 3048
Table2: Summary Of Means, Variances, Correlations
Link Sample
Adjusted r Simple
mean r Standard
Deviation Correlation
from Zr Standard
Error Sample
Variance
Fail safe Confidence
Interval
@99%
PE-BI 0.4982 0.4919 0.1486 0.5361 0.1581 0.0221 167.03 (.43,.55)
EE-BI 0.4224 0.4131 0.1509 0.4356 0.1601 0.0228 76.67 (.35,.47)
SI-BI 0.4235 0.4019 0.1658 0.4236 0.1715 0.0275 47.15 (.33,.47)
FC-UB 0.3566 0.3556 0.2047 0.3769 0.2774 0.0419 -1.05 (.22,.50)
BI-UB 0.4104 0.4125 0.0221 0.4356 0.2774 0.0490 5.57 (.40,.42)
In meta-analysis, some argue that there is no much
difference between the simple mean and the Fisher’s
r to Z transformation (See Szymanski & Henard
(2001); Mo et al 2004), but we decided to engage
both methods is our study (equation 2 and 3) for
clarity. Reliability adjusted mean was not computed
in this study but sample size adjusted mean.
The Fisher’s r to Z transformation results is quite
larger than the simple mean and the sample adjusted
mean except for some of the values SI-BI are almost
the same. All in all the three values from the three
methods are approximately the same. Therefore we
can interpreted the result from these study based on
both the Fisher’s r to Z transformation method and
sample size adjusted method. The magnitude of an
effect size is regarded as small when it is close to
0.10, medium when it is close to 0.30 and large
when it is close to 0.50 (Cohen 1977;Mo et al 2004).
Following this rule, our meta-analysis implies a
medium size effect for FC-BI and large size effect
for all other relationship. Furthermore, our study
suggests that the PE-BI relationship is the strongest
while FC-UB is the weakest of the relationships.
Also, EE-BI relationship is stronger than SI-BI
relationship (See Table 2). For more confidence that
intervals include means (µ) a wider confidence
interval is recommended (Cumming and Finch
2005). Therefore, a 99% confidence interval was
computed for each mean estimate to depict its
statistical significance. With the existence of
variances and error, the confidence provides a range
of effects that may exist in a true population. None
of the intervals in Table2 is 0, thus we can conclude
that all the mean effect is significantly different from
zero. The significance of the effect sizes were
further tested by computing the failsafe test. Using
equation 3, the failsafe test (N) shows the amount in
numbers of additional studies that is needed to
confirm the null hypothesis (r=0) required to annul
the conclusion that there exist a significant
relationship amongst the pair of variables.
Consequently, Table 2 shows that the mean effect
sizes of PE-BI, EE-BI and SI-BI are significantly
different from zero to the extent that 47-167 null
effect size is needed to revert mean effect sizes to a
level considered as non – significant statistically, BI-
UB requires 6 of null effect to make the relationship
non-significance(Many study suggest significant
relationship with behavioural intention).
Nevertheless, the mean effect size of FC-UB did not
pass the fail safe test with the negative (Nfs.05).
Since individual level analysis (use of individual
correlation) was employed in this study against the
study level analysis (use of average correlation). We
computed the sampling error variance and the
standard deviation for each of the relationship to
avoid under estimation of the sampling error
sampling and generalization (Mo et al., 2004). Our
result shows that the variances of sampling error are

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
54
close to each other. Therefore, individual level
analysis is satisfactory for this meta-analytic study.
5. DISCUSSIONS
The summary of the results computed in the meta-
analysis is depicted by Table 3. Using Cohen (1992)
criteria of categorizing mean effect sizes as non-
significant, small, medium or large. As depicted by
the table, most of the effect sizes are small. This
result however is consistent with the result of
Venkatesh et al.(2003), except the results of BI-USE
which is lower. This could be as a result of inability
of authors to measure actual usage of systems being
investigated.
Table3: Summary Of Effect Sizes By Dependent Variables
Link Correlation
from Zr
Size/weight
PE-BI 0.5361 Medium
EE-BI 0.4356 Small
SI-BI 0.4236 Small
FC-UB 0.3769 Small
BI-UB 0.4356 Small
Note: Effect sizes classification is based on Cohen’s
1992 s = small (.2 <d <.5); M = Medium (.5 <d <.8);
L = Large (d >.8);
5.1 Performance Expectancy and behavioural
intention (PE-BI)
The effect size of performance expectancy can be
classified as medium with a Zr of 0.5361. This is the
largest effect size in the study, and it is consistent
with previous literature showing that amongst the
four major construct of UTAUT, performance
expectancy has the highest co-efficient path weight
(Venkatesh et al 2003, Wnag & Shih 2009; Dijk et
al 2008). Users of Information Systems gives high
regard to the level at which the system is
advantageous to them in their daily routine. The
ability of the system to assist users to achieve task
quickly will motivate users to adopt the system.
The fail safe test shows that 167 null effect sizes are
expected to make the computed effect size non-
significant. This however is not possible considering
that fourty three (43) coefficients were used to
compute the effect size.
However, the derived effect size of PE-BI in this
study is further tested with it statistical significance
and we found that the out-put is statistically different
from zero with a confidence interval of (.43,55) and
thus we can say that the difference between the
theorized parameter and the observed estimate in
this study is not statistically significant at the 1%
level using a 99% confidence Interval.
5.2 Effort Expectancy and behavioural intention
(EE-BI)
Table 3 shows that the effect size (Zr) of effort
expectancy is small with a weight of 0.4356.
However this value is consistent with previous
literature (Venkatesh et al 2003; Dijk et al 2008).
Users of Information Systems are concerned with
the ease that is associated with the use of the
information system. A complex system or a web
interface that is difficult to navigate can make users
uninterested in adopting the system or website
(Byun & Finnie, 2011). The issue regarding the level
of computer literacy amongst the population can
alter the perception of respondents to the ease
associated with using an information system,
because computer savvy users may be indifferent.
The fail safe test shows that 76 null effect sizes are
expected to make the computed effect size non-
significant. However this is hard to realize
considering that fourty two (42) coefficients were
used to compute the effort expectancy effect size.
Furthermore, the derived effect size in this study is
further tested with it statistical significance and we
found that the out-put is statistically different from
zero with a confidence interval of (.35,.47). Thus the
difference between the theorized parameter and the
observed estimate in this study is not statistically
significant at 1% level using a 99% confidence
interval.
5.3 Social Influence and behavioural intention
(SI-BI)
The effect size of social influence can be classified
as small (See Table 3) with a Zr of 0.4236. This
result is consistent with previous literature showing
the effect of social influence on intention to adopt a
technology (Venkatesh et al 3003; Wang &Shih
2009).
Besides an effective and easy to use information
system, end-users might not be obliged to use the
system until they are motivated by important others
(people) that can influence their attitude and
behaviour. With the way people's life are molded
round role models, public figures, sportsmen and
celebrities, an encouragement by such important
figures to use the system can motivate users to adopt
the use of an information system (Taiwo et al.,
2012).
The fail safe test shows that 47 null effect sizes are
expected to make the computed effect size non-

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
55
significant. This however is hard to obtain
considering that thirty six (36) coefficients were
used to compute the effect size.
However, the derived effect size in this study is
further tested with it statistical significance and we
found that the out-put is statistically different from
zero with a confidence interval of (.33, .47) and thus
we can say that the difference between the theorized
parameter and the observed estimate in this study is
not statistically significant at the 1% level using a
99% confidence Interval.
5.4 Facilitating condition and behavioural
intention (FC-BI)
Table 3 depicts a small effect size (Zr) of facilitating
condition with a weight of 0.3769.The FC-BI
relationship accounts for the lowest effect size in the
study. This could be as a result of inability of most
studies to measure the actual usage of the
information systems being invested. Therefore, few
studies actually reports the outcome of the effect of
facilitating conditions on use behaviour, rather many
authors reports the effect of facilitating condition on
intention which Venkatesh et al had hypothesized
and found non-significant. In some studies,
facilitating condition has been found to be
significant in predicting intention (Foon and Fah,
2011; Venkatesh et al 2011a; Venkatesh et al
2011b). Thus, the outcome on the relationship
between the facilitation condition and use behaviour
can be said to be inconclusive. Although
empirically, there might be inconclusive argument
on the effect of FC on UB or BI, it is important to
note that qualitative research have shown that the
contribution of provision of organizational and
technical infrastructure for users towards the
acceptance of a technology cannot be
overemphasized (Alawadhi and Morris 2009).
Furthermore, the derived effect size of the FC-UB
relationship is further tested with it statistical
significance and we found that the out-put is
statistically different from zero with a confidence
interval of (.22,.50). Thus the difference between the
theorized parameter and the observed estimate in
this study is not statistically significant at the 1%
level using a 99% confidence Interval. Nevertheless,
the effect of facilitating condition on intention do
not pass the fail safe test showing a negative value,
thus the consistency in the outcome of FC-BI
relationship is questionable.
5.5 Behavioural intention and use behaviour (BI-
UB)
The effect size of BI-USE can be classified as small
with a Zr of 0.4356(See table 3). This however
could be as a result of inability of many studies that
employed UTAUT in their investigation of
acceptance and adoption of technologies to measure
the actual use behaviour of the new information
systems being investigated. Therefore, few studies
actually investigate the effect of intention on use
behaviour rather many authors relied on the premise
that there exist a strong relationship between
intention and usage which Venkatesh et al had
hypothesized and found significant.
The derived effect size in this study is further tested
with it’s statistical significance and we found that
the out-put is statistically different from zero with a
confidence interval of (.40, .42). Thus, the difference
between the theorized parameter (Venkatesh
outcome) and the observed estimate in this study is
not statistically significant at the 1% level using a
99% confidence Interval. Users that show positive
intention towards a technology actually exhibit that
specific behaviour at a later time. However, our
study found the BI-UB relationship had a less than
desired fail safe test value of 5.56, thus about 6 null
effect sizes are expected to make the computed
effect size non-significant.
6. LIMITATIONS AND FUTURE-WORKS
Besides the selection criteria surrounding studies
involved in this study, one of the limitations of this
study is that the UTAUT theory is merely cited in
many articles but not actually used. This led to the
relative small sample size of studies that were
employed in this study. The inability of a
standardized and generally accepted effect size
statistics would enhance meta-analytic outcomes
(Lipsey and Wiley 2000;Ma and Liu,2004).
Secondly, meta-analysis has the ability of
indentifying whether the variation of correlation is
due to chance, dimensions or methods employed.
We were unable to achieve this because of the small
size of the selected studies, although we earlier
planned to examine the effects of the moderators and
methods used by running multiple regressions.
Thirdly, inability of the studies to measure use
behaviour and the integration of some dimensions of
UTAUT led to uneven number of correlations
between the constructs. Therefore, some caution is
advised when interpreting the results in the study.
Future works could to attempts to resolve the above
limitations. Besides the use of coefficient paths
magnitude, reports on future studies of UTAUT
should include correlations, T-test and other
statistical measures that can be used to compute a
meta-analysis. This shall enhance measuring the

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
56
consistency of findings and the parsimonious nature
of UTAUT.
7. CONCLUSIONS
Through the meta-analysis, this study has succeeded
in combining and investigating existing empirical
literatures on Unified theory of acceptance and use
of technology (UTAUT). The relationship between
UTAUT was examined using a larger sample size of
over 11,000 which could have been difficult to
achieve in a single study. On the basis of meta-
analysis reported in this article, generally, our
findings confirms Venkatesh et al. initial findings
amongst the five constructs of UTAUT, only the
relationship between PE and BI is strong while
others are slightly weak but significant.
Also the relationship between BI and UB is also
reliable while the relationship between FC and UB is
found to be fairly less than desired. Furthermore, all
the five mean effects are statistically positive at
α=0.01 to attest to their statistical significance. The
fail safe test further asserts the significance of the
relationships. We discovered that 47-167 null effect
size is needed to be hidden in file drawer for the
mean correlation between the trio of PE,EE,SI and
BI to be non significant, this seems unlikely.
However the mean effect size of BI-UB has a weak
fail safe value, suggesting that six (6) reports with
null effects can make the effect non-significant.
While the mean effect size of FC-UB failed the fail
safe test, suggesting that addition of just one report
with null effect can make the effect non-significant.
In conclusion, we discovered that majority of
researchers cited UTAUT in their articles in order to
support an argument rather than using it. Others that
reported the use of UTAUT actually used it partially
while only a few have reported the use of the
actually theory. This paper contributes to the area of
IS/IT adoption and diffusion research by showing
the inadequacy and inconsistency in the use and
output of a theory.
REFRENCES:
[1] Alawadhi, A and Morris,A (2008). “The use of
the UTAUT model in the adoption of e-
government services in Kuwait”. Proceedings of
the 41st Hawaii International Conference on
System Sciences, Hawaii, 2008.
[2] Alawadhi,A and Morris,A (2009). “Factors
Influencing the Adoption of E-government
Services”. Journal of Software, Vol. 4, NO. 6,
AUGUST 2009
[3] Benbasat,I and Zmud,R. (1999). “Empirical
research in information systems”. The practices
of relevance.MIS Quarterly,23(1),3-16.
[4] Biemans,M, Swaak,J, Hettinga,M and
Schuurman,J.G (2005). “Involvement Matters:
The Proper Involvement of Users and
Behavioural Theories in the Design of a
Medical Teleconferencing Application”.
Proceeds of GROUP’05, November 6–9, 2005,
Sanibel Island, Florida, USA.
[5] Byun,D-H and Finnie,G.2011. “Evaluating
usability, user satisfaction and intention to
revisit for successful e-government websites”.
Electronic Government, An International
Journal, Vol. 8, No. 1, 2011,pp.1–19.
[6] Card, S K, Moran, T P and Newell, A (1983).
“The psychology of human-computer
interaction”. Hillsdale, NJ: Erlbaum.
[7] Cheng,D,Liu,G.,Song,Y-F and Qian,C. “(2008)
Adoption of Internet Banking: An Integrated
Model”. Proceeds of the 4th International
Conference on Wireless Communications,
Networking and Mobile Computing,12-14 Oct.
2008,Dalian.
[8] Cheng,D, Liu,G and Qian,C. “(2008) On
Determinants of User Acceptance of Internet
Banking: A Theoretical Framework and
Empirical Study”. Advanced Management of
Information for Globalized Enterprises, 2008.
AMIGE 2008.
[9] Carlsson,C, Carlsson,J., Hyvönen,K.,
Puhakainen,J and Walden. (2006). “Adoption of
Mobile Devices/Services – Searching for
Answers with the UTAUT”. Proceedings of the
39th Hawaii International Conference on
System Sciences.
[10] Cohen, J (1992). “A power Primer”.
Psychological Bulletin, 112,155-159
[11] Cumming G and Finch S. (2005) “Inference by
Eye Confidence Intervals and How to Read
Pictures of Data”. American Psychological
Association. Vol. 60, No. 2, 170–180 DOI:
10.1037/0003-066X.60.2.170
[13] DeCoster, J. (2004). “Meta-analysis Notes".
Retrieved on 20th May, 2011, from
http://www.stat-help.com/notes.html.
[14] Dijk,J.A.G.M.V, Peters,O and Ebbers,W.
(2008). “Explaining the acceptance and use
government Internet services: A multivariate
analysis of 2006 survey data in Netherlands”.
Government information Quarterly. 25,375-
399.

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
57
[15] Fang,W.C., Li,M.W and Liu,C.W (2008).
Measurement of the Knowledge-Sharing
Efficacy of Web2.0 Site Constructed on the
Basis of Knowledge-based Systems by
Applying the Model of UTAUT: Evidence of
the early adopters”. Proceeds of the 3rd
Intetnational Conference on Innovative
Computing Information and Control
(ICICIC'08).
[16] Foon, Y.S and Fah, B.C.Y. (2011). “Internet
banking adoption in Kuala Lumpur: An
application of UTAUT model. International
Journal of Business and Management.
Vol.6,No.4, pp161-167.
[17] Glass, G V (1976). “Primary, secondary, and
meta-analysis of research”. Educational
Researcher, 5, 3-8.
[18] He,D and Lu,Y (2007). “Consumers
Perceptions and Acceptances towards Mobile
Advertising: An Empirical Study in China”. In
proceedings of International Conference
Wireless Communications, Networking and
Mobile Computing.
[19] Heerink,M., Kröse,B, Wielinga,B and Evers,V
(2009). “Measuring the influence of social
abilities on acceptance of an interface robot and
a screen agent by elderly users”. People and
Computers XXIII(HCI) – Celebrating people
and technology.
[20] Hwang M I and Schmidt F L (2011).
“Assessing moderating effect in meta-analysis:
a re-analysis of top management support studies
and suggestions for researchers”. European
Journal of Information Systems online
publication. doi:10.1057/ejis.2011.12
[21] Li,J.P and Kishore,R (2006). “How Robust is
the UTAUT Instrument? A Multigroup
Invariance Analysis in the Context of
Acceptance and Use of Online Community
Weblog Systems”. Proceeds of SIGMIS-
CPR’06, April 13–15, 2006, Claremont,
California, USA.
[22] Lipsey, M. And Wilson, D. (2001). “Practical
meta-analysis”. CA: Sage Publications.
[23] Ma,Q and Liu,L (2004). “The technology
Acceptance Model:A meta-Analysis of
Empirical Findings”. Journal of Organizational
and End User Computing.16(1),59-72.
[24] Maldonado,U.P.T., Khan,G.F., Moon,J And
Rho,J.J. (2009). “E-learning motivation,
Students’ Acceptance/Use of Educational Portal
in Developing Countries”. Proceeds of the 4th
International Conference on Computer Sciences
and Convergence Information Technology.
[25] Oshlyansky,L.,Cairns,P. and Thimbleby,H
(2007). “Validating the Unified Theory of
Acceptance and Use of Technology (UTAUT)
Tool Cross-Culturally”. Proceedings of the 21st
BCS HCI Group Conference.3-7 September
2007, Lancaster University, UK.
[26] Rosenthal, R., and Dimatteo, M. R. (2001)
Meta-analysis: Recent developments in
quantitative methods for literature reviews.
Annual Review of Psychology, 52, 59- 82.
[27] Schaupp, LC, Carter, L and Hobbs, J (2009). “
E-File Adoption: A Study of U.S. Taxpayers
Intentions”. In proceedings of the 42nd Hawaii
International Conference on System Sciences –
2009.
[28] Shaughnessy, J. J., Zechmeister, E. B and
Zechmeister, J. S (2006). “Research methods in
psychology”. Seventh edition. New York:
McGraw-Hill.
[29] Šumak,B, Polančič,G, and Heričko,M. (2010)
“An Empirical Study Of Virtual Learning
Environment Adoption Using UTAUT”. In
Proceedings of 2010 Second International
Conference on Mobile, Hybrid, and On-Line
Learning.
[30] Szymantki, D.M and Henard, DH (2001)
Customer Satisfaction: A Meta-Analysis of
empirical evidence. Journal of the academy of
Marketing Science,29(1),16-35.
[31] Taiwo, A.A., Downe,A.G And Mahmood, A.K
(2012). “User Acceptance of Egovernment:
Integrating Risk and Trust Dimensions with
UTAUT Model”. In proceedings at the
International Conference on Computer and
Information sciences. Kuala Lumpur,
[23] Tibenderana,P.K and Ogao,P.J (2008)
“Acceptance and Use of Electronic Library
Services in Ugandan Universities”. In
Proceedings of JCDL 2008, Pittsburgh,
Pennsylvania, USA.
[33] Venkatesh,V, Morris, M. G, Davis, G. B. And
Davis,F. D (2003) “User acceptance of
information technology: toward a unified view”.
MIS Quarterly, vol. 27, no. 3, pp. 425-478,
2003.
[34] Venkatesh V, Zhang X and Sykes T A (2011a)
“Doctors do too little technology : a
longitudinal field study of an electronic
healthcare”. Journal Information Systems
Research. Volume: 22 Issue: 3 pp.523-546
[35] Venkatesh,V, Sykes T A And Zhang X (2011b)
“Just What the Doctor Ordered’: A Revised
UTAUT for EMR System Adoption and Use by
Doctors”. Proceedings of the 44th Hawaii

Journal of Theoretical and Applied Information Technology
10
th
March 2013. Vol. 49 No.1
© 2005 - 2013 JATIT & LLS. All rights reserved.
ISSN:
1992-8645
www.jatit.org E-ISSN:
1817-3195
58
International Conference on System Sciences -
2011
[36] Wang, Y-S and Shih,Y-W (2009). “Why do
people use information Kiosks?A validation of
the Unified theory of Acceptance and use of
Technology”. Government Information
Quaterly,26,158-165.
[37] Whitley,B.E (1997). “Gender Differences in
Computer-Related Attitudes and Behavior: A
Meta-Analysis”. Computers in Human
Behavior, Vol. 13, No. l, pp. 1-22.
[38] Wu, Y, Tao, Y and Yang, P (2007). “Using
UTAUT to explore the behavior of 3G mobile
communication users”. In Proceedings of The
International Conference on Industrial
Engineering and Engineering Management
(IEEM).
References included in the Meta-Analysis
[39] Al-Gahtani,S.S,Hubona,G.S and Wang,J. (2007).
“Information technology (IT) in Saudi
Arabia:Culture and the acceptance and use of IT”.
Journal of Information and management.
(44),pp.140-154
[40] Basaglia,S, Caporarello,L, Magni,M and
Pennarola, F (2010) Individual adoption of
convergent mobile phone in Italy. Review of
Managerial Science.3,pp1-18
[41] Chea, S (2006). “Understanding the roles of
emotion in technology acceptance. University of
Hawaii at Manoa)”. ProQuest Dissertations and
Theses, 146 p. ( UMI No 3216051).Retrieved from
http://search.proquest.com/docview/305319501?ac
countid=47520 on Retrieved June 20, 2011
[42] Chiu,C-M and Wang, E.T.G. (2008)
“Understanding web-based learning continuance
intention: The role of subjective task value”.
Information & Management,45,pp194-201.
[43] Chiu,Y-T,H.,Fang, S-C and Tseng,C-C (2010)
“Early versus potential adopters Exploring the
antecedents of use information in the context of
retail service innovations”. International journal
of retail and distribution management.
Vol.38,No.6, pp443-459.
[44] Clarke, D. C (2001) “Differences in student
dropout rate, attendance rate, and grade point
average after the implementation of the North
Carolina Dropout Prevention/Drivers License
Law”. Dissertation Abstract International, 62(3),
853A. (UMI No. AAT 3010522) Retrieved August
20, 2004, from Dissertations and Theses database
[45] Dijk,J.A.G.M.V., Peters,O and Ebbers,W (2008).
“Explaining the acceptance and use government
Internet services: A multivariate analysis of 2006
survey data in Netherlands”. Government
information Quarterly. 25,375-399.
[46] Duyck, P, Pynoo B, Devolder P, Voet T, Adang L
and Vercruysse J (2008a). “User acceptance of a
Picture Archiving and Communication System -
Applying the unified theory of acceptance and use
of technology in a radiological setting”. Methods
of Information in Medicine 47(2): 149-156.
[47] Duyck, P, Pynoo B, Devolder P, Voet T, Adang L
and Vercruysse J (2008b). “Do hospital physicians
really want to go digital? Acceptance of a picture
archiving and communication system in a
university hospital”. Röfo-Fortschritte Auf dem
Gebiet der Röntgenstrahlen und der Bildgebenden
Verfahren 180(7): 631-638.
[48] Duyck P, B Pynoo, Devolder P, Voet,T,Adang.,
Ovaere, D and Vercruysse, J (2010). “Monitoring
the PACS implementation process in a large
University Hospital – Discepancies between
radiologist and physicians”. Journal of Digital
Imaging.Vol.23,No1,pp73-80.
[49] Foon YS and Fah, B.C.Y (2011). “Internet
banking adoption in Kuala Lumpur: An
application of UTAUT model”. International
Journal of Business and Management. Vol.6,No.4,
pp161-167.
[50] Goh, S and Yoon, T. (2011) “If you build it will
they come?An empirical investigation of
facilitators and inhibitors of hedonic Virtual world
acceptance”. Proceedings of the 44
th
Hawaii
International conference on system sciences.
[51] Gupta B, Dasgupta,S and Gupta,A (2008)
“Adoption of ICT in a government organization in
a developing country: An empirical study”.
Strategic Information systems,17,pp140-154.
[52] He D and Lu Y (2007). “Consumers perception
and acceptances towards Mobile advertising: An
empirical study in China”. International
Conference on Wireless Communications
Networking and Mobile Computing Vols 115.
[53] Johnston A.C.(2006). “An empirical investigation
of the influence of fear appeals on attitudes and
behavioral intentions associated with
recommended individual computer security
actions”. Mississippi State University). ProQuest
Dissertations and Theses, , 188 p. UMI No
3213966 Retrieved from
http://search.proquest.com/docview/305307755?ac
countid=47520 on June 20, 2011.