i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416

journal homepage: www.intl.elsevierhealth.com/journals/ijmi

Factors influencing health information technology

adoption in Thailand’s community health centers:
Applying the UTAUT model

Boonchai Kijsanayotin a,∗,1 , Supasit Pannarunothai b , Stuart M. Speedie c

a Bureau of Policy and Strategy, Ministry of Public Health, Tiwanon Rd., Nonthaburi 11000, Thailand
b Center for Health Equity Monitoring, Faculty of Medicine, Naresuan University, Phitsanulok, Thailand
c Institute for Health Informatics, University of Minnesota, Minneapolis, MN, USA

a r t i c l e i n f o a b s t r a c t

Article history:
Received 13 May 2008 Background: One of the most important factors for the success of health information tech-
Received in revised form nology (IT) implementation is users’ acceptance and use of that technology. Thailand has
11 November 2008 implemented the national universal healthcare program and has been restructuring the
Accepted 26 December 2008 country’s health IT system to support it. However, there is no national data available regard-
ing the acceptance and use of health IT in many healthcare facilities, including community
health centers (CHCs). This study employed a modified Unified Theory of Acceptance and
Keywords: Use of Technology (UTAUT) structural model, to understand factors that influence health IT
Technology adoption adoption in community health centers in Thailand and to validate this extant IT adoption
Information systems model in a developing country health care context.
Computer systems Methods: An observational research design was employed to study CHCs’ IT adoption and
Attitude of health personnel use. A random sample of 1607 regionally stratified CHC’s from a total of 9806 CHCs was
Diffusion of innovation selected. Data collection was conducted using a cross-sectional survey by means of self-
Thailand administered questionnaire with an 82% response rate. The research model was applied
UTAUT model using the partial least squares (PLS) path modeling.
Results: The data showed that people who worked in CHCs exhibited a high degree of IT
acceptance and use. The research model analyses suggest that IT acceptance is influenced
by performance expectancy, effort expectancy, social influence and voluntariness. Health
IT use is predicted by previous IT experiences, intention to use the system, and facilitating
conditions.
Conclusions: Health IT is pervasive and well adopted by CHCs in Thailand. The study results
have implications for both health IT developmental efforts in Thailand and health infor-
matics research. This study validated the UTAUT model in the field context of a developing
country’s healthcare system and demonstrated that the PLS path modeling works well in a
field study and in exploratory research with a complex model.
© 2008 Elsevier Ireland Ltd. All rights reserved.

∗
Corresponding author.
E-mail addresses: kijs0001@umn.edu, kijs0001@gmail.com (B. Kijsanayotin), supasitp@nu.ac.th (S. Pannarunothai),
speed002@umn.edu (S.M. Speedie).
1
This study has been done as partial fulfillment of the requirements for the doctoral degree in the Health Informatics Graduate Program,
Medical School, University of Minnesota, Minneapolis, MN, USA.
1386-5056/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.ijmedinf.2008.12.005
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
1. Background
More than 40% of information technology (IT) developments
in various sectors including the health sector have failed
or been abandoned [1–5]. One of the major factors leading
to the failure is the inadequate understanding of the socio-
technical aspects of IT, particularly the understanding of
how people and organizations adopt information technology
[6–8].
The Ministry of Public Health (MOPH) of Thailand has
been restructuring the country’s health information system to
support the country’s universal healthcare coverage scheme,
which has been implemented since 2001 and is still evolv-
ing [9–12]. The goal is to have an information system that
integrates health-related information from the village level in
community health centers to hospitals and local administra-
tive organizations up to the central administration at ministry
[9,13]. Multiple levels of health-related information systems
including community health center (CHC) information sys-
tems are being planned and developed using IT to support the
universal healthcare system. Lessons learned from previous
major health IT implementation projects in the country [14,15]
suggest that user acceptance of technology is one of the major
determinants of the project success. The knowledge of how
people who work in the health sector accept and use health
IT, their basic IT knowledge, and factors that influence their IT
acceptance and use not only helps health information system
designers but also enables more efficient implementation and
evaluation processes.
We conducted a national survey to study Thailand CHC’s
Health IT penetration and adoption in 2005. The results show
that health IT is pervasive in Thailand CHCs across the
country and is prevalent in all regions. CHCs have adopted
health IT, but several perceived barriers associated with IT
use were evident. These results have been described else-
where in detail [16]. This paper reports the analyses of
the research model employed in the survey which aims to
understand factors that influence health IT acceptance and
use. Section 2 of paper is the brief review of IT accep-
tance and use model, and describes our research model.
Sections 3 and 4 describe the methodology and the results
of the model analysis. The final two sections discuss the
validity of the model, its implications for health informatics
research and health IT development in developing coun-
tries.
2. IT acceptance and use models
Research on IT users’ acceptance and use has been done
extensively since computer and information technology have
been in wide use and several models have been developed
to explain users’ acceptance and use. The models originated
from different theoretical disciplines such as psychology, soci-
ology and information systems. These are the Theory of
Reasoned Action (TRA), the Technology Acceptance Model
(TAM), the Motivation Model (MM), the Theory of Planned
Behavior (TPB), the Combined TAM and TPB (c-TAM-TPB), the
Model of PC Utilization (MPCU), Innovation Diffusion Theory
405
(IDT) and Social Cognitive Theory (SCT).2 In 2003, Venkatesh
et al. proposed a new IT acceptance and use model which
aimed to unify eight prominent competing IT acceptance and
use models [17]. The model is named the Unified Theory of
Acceptance and Use of Technology (UTAUT). The authors con-
tend that the new model successfully integrates all constructs
in previous models and can explain variance in IT behavioral
intention and use behavior better than the previous models.
The UTAUT model was able to explain 69% of intention to
use IT (technology acceptance) while other previous models
explained approximately 40% of technology acceptance [17].
Our study adapts the UTAUT model to study factors that
predict the intention to use health IT and IT use in community
health centers in Thailand. The UTAUT model was developed
in a western industrial developed country’s (USA) business
context (banking, accounting, entertainment and telecom-
munications services) [17]. Applying the model to a study of
health IT acceptance and use in a developing country’s public
health context, such as Thailand, will expand the understand-
ing of the model’s robustness in explaining acceptance and
use.
The basic UTAUT model (see Fig. 1) consists of several com-
ponents or constructs that are hypothesized to relate to the
intention to use IT. In turn intention to use IT predicts IT use.
Performance expectancy (PE) is defined as the degree to which
an individual believes that using health IT will help him or her
to attain gains in job performance. Venkatesh et al. [17] inte-
grated similar concept from other models, namely, perceived
usefulness [18], outcome expectancy [19], relative advantage
[20], job-fit [21] and extrinsic motivation [22] into this con-
struct. In several previous acceptance studies, performance
expectancy was shown to be a strong predictor of intention to
use IT [17,18,23–25].
Effort expectancy (EE) is defined as the degree of ease of
use associated with health IT. The concept is similar to the
perceived ease of use construct in TAM and the IDT model
and the complexity of technology construct in the MPCU
model. Although many previous studies have shown that
effort expectancy was a significant influence on intention to
use behavior [18,20,21,23,26,27], some did not [28].
Social influence (SI) is defined as the degree to which an
individual perceives that important others believe he or she
should use health IT. The construct contains a notion that an
individual’s behavior is influenced by the way in which one
believes others will view him/her as a result of having used
health IT. Venkatesh et al. integrated subjective norms in TRA,
TAM2, and TPB, social factors in MPCU, and image in IDT to this
construct [17]. The effect of social influence on intention to use
technology has been shown to be significant in several previ-
ous acceptance studies [23,25,29], but some studies exhibited a
non-significant effect, especially studies in professionals with
high autonomy such as physicians [27,28,30]. However, the
people in our study who were working in CHCs were health
workers, nurses and public health specialists. They do not gen-
erally act as autonomously as physicians, so it is possible that
2
For a comprehensive review of information technology adop-
tion theories and models, the interested reader is referred to
Venkatesh et al. [17].
406 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
Fig. 1 – Research model exhibits relationship between constructs (latent variables—LV) and measurements (manifest
variable—MV) and path relationship between constructs.
social influence has a positive influence on their intention to
use IT.
Voluntariness (VO), defined as the degree to which an indi-
vidual perceives that he or she has a choice to use or not
use health IT, is an important concept that also influences
the intention to use information technology and is included
in the explanatory model. Moore’s dissertation (1989) showed
that the degree of perceived voluntariness of use affects
attitudes toward usage and the attitudes toward usage pre-
dicted use. The less voluntary the behavior, the less one’s
attitude toward usage predicts use [29,31]. Perceived volun-
tariness was treated as a moderating dichotomous variable
(voluntary/compulsory) in the original UTAUT model, but it
was treated as a determinant variable with ordinal scale in
Karahanna et al.’s technology innovation adoption study [29].
Moore and Benbasat (1991) and Agarwal and Prasad (1997)
argued that in practice, users may perceive different degrees of
voluntariness in using technology and suggested that it might
be empirically ordinal [20,32].
The construct labeled facilitating conditions (FC) is defined
as the degree to which an individual believes that an orga-
nizational and technical infrastructure exists to support use
of the system. The UTAUT’s facilitating conditions construct
captures the perceived behavior control concept from the
TPB and c-TAM-TPB models, the facilitating conditions con-
cept from the MPCU model and compatibility from the IDT
model [17]. Previous technology adoption studies exhibited
a positive effect of facilitating conditions on innovation use.
[17,20,21,23,24,28,33] They found that facilitating conditions
significantly predicted technology use but did not predict
intention to use IT when both the performance expectancy
and effort expectancy constructs are present in the model.
Furthermore, we anticipate that the CHC personnel’s basic
IT knowledge should have a positive influence on the per-
ception of facilitating conditions because having adequate
knowledge to use a computer is one of the operationalized
items in the facilitating conditions construct. Consequently
we have added a construct of IT knowledge (Knowledge) to the
model as a predictor of facilitating conditions.
We also argue that Experience, defined as the past use of
health IT, is an important predictor of IT use in addition
to a person’s intention to use IT. Thompson and Higgins
found that the direct effect of experience on PC utiliza-
tion was statistically and substantially significant while the
indirect effect was present but less profound [34]. The orig-
inal UTAUT model treats Experience as a moderator between
effort expectancy, social influence, and behavior intention,
and between facilitating conditions and use behavior. In addi-
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
tion, the “Experience” construct in the original UTAUT model
was operationalized by converting aggregating data from three
consecutive time periods [17]. But, our study operationalized
Experience as the length of time each respondent reported
using health IT and tested only the direct effect of experience
on health IT use.
Finally, the intention-behavior relationship is well doc-
umented in many research fields including the technology
acceptance literature and has been found to be strong
[17,24,25,35,36].
Since the inception of the UTAUT model, several works
in the healthcare context have employed and modified the
model to study IT adoption and IT use. The studies show
that the UTAUT model and its modified model are applica-
ble in explaining IT adoption in healthcare settings. However,
all these studies were conducted in developed countries
[23,27,30]. To the best of our knowledge, there is no study
employing the UTAUT model that has been conducted in a
developing countries healthcare context.
The latent variable components in Fig. 1 portray this study’s
proposed structural model adapted from UTAUT. As shown in
the figure, performance expectancy, effort expectancy, social
influence and voluntariness are the determinants of intention
to use IT. Following the path, IT use is predicted by intention to
use, facilitating conditions, and computer experience. Facili-
tating conditions is predicted by basic IT knowledge that is
defined as general knowledge (similar to a college entrant)
related to basic personal computer (PC) components and basic
functions of PC and the Internet [37]. The IT use construct
is hierarchically modeled. Overall IT use is represented by
three dimensions of IT use related to CHC’s activities and
one dimension of use frequency. Health IT in this study is
defined as the information and communication technology
that is used for health information systems.
3. Methods
A cross-sectional national survey was conducted in Thailand
during July–October 2005. We randomly sampled population
weighted provinces. This resulted in the selection of twelve
provinces with 1607 CHCs: three from the north with 337
CHCs, four from the central with 376 CHCs, three from the
northeast with 708 CHCs, and two from the south with 146
CHCs. At each CHC, we asked an officer who was respon-
sible for the CHC’s information management or the CHC’s
administrative officer (the CHC’s head officer) to complete the
survey. Research collaborators at the provincial health office in
each province distributed the self-administered paper-based
survey. Respondents could return the questionnaire either
through the provincial collaborators or by mailing the ques-
tionnaires direct to the research team at Center for Health
Equity Monitoring (CHEM), Naresuan University. The ques-
tionnaires collected by the provincial research collaborators
were also sent directly to CHEM. Each questionnaire included
a pre-paid return envelope. Two to four weeks after ques-
tionnaire distribution, the provincial research collaborators
reminded the CHCs to complete the questionnaire by phone
and/or by an in-person reminder at the monthly meeting of
the provincial health office’s director and CHCs’ head officers.
407
The details of this survey process and the survey instrument
have already been reported elsewhere [16].
The survey contained a number of questions that were
designed to capture information about the constructs in the
research model. These are labeled as manifest variables in
Fig. 1 and form the outer ring of that model. The questions that
measured performance expectancy (PE), effort expectancy
(EE), social influence (SI), facilitating conditions (FC) and vol-
untariness (VO), including intention to use IT (IN) construct
were adapted from Venkatesh et al. [17]. IT use associated with
activities in CHCs, drawn from our previous study [37], was
measured with ten questions using a four-point scale ranging
from one (never use) and four (always use). These ten items
included four for “use for providing care and routine report-
ing” (Care and Report Use), three for “use for management and
administration” (Administration Use) and another three for “use
for information searching and collaboration with colleagues”
(Communication Use). Table 1 contains a description of each
of the items and the constructs that they are intended to
measure. All items were minimally modified and translated
into Thai language so as to be clearly understood by Thai
health center personnel. As previously reported, the study
was approved by the University of Minnesota Institutional
Review Board and the Ethical Review Committee for Research
in Human Subjects, the Ministry of Public Health Thailand.
The survey was piloted tested for its understandability in Thai
prior to its distribution.
SPSS V15.0 statistics package was used to obtain descrip-
tive statistics for the manifest variables. Partial least square
path modeling (PLS-SmartPLS V 2.0 M3) [38,39] was applied to
evaluate the predictive research model.
PLS path modeling is one of the statistical methods for
structural equation modeling (SEM), a modeling procedure
that performs path-analytics modeling with latent variables
(unobservable variables). SEM has been used to model the
complex relationship of multiple endogenous (independent)
and exogenous (dependent) variables. It is considered a sec-
ond generation of multivariate analysis and is originated to
overcome limitations of those first generation ones including
standard regression-based analyses, such as multiple regres-
sion, discriminant analysis, logistic regression, and analysis
of variance [40–42]. SEM is capable of simultaneously assess-
ing the reliability and validity of the measures of theoretical
constructs and estimating the relationships among these
constructs [43]. There are two approaches to SEM analysis.
The first approach is covariance-based analysis which typi-
cally uses maximum likelihood (ML) and was popularized by
LISREL, AMOS and EQs statistics software. The second
approach is variance-based analysis, or component-based
analysis, which uses least square (LS) functions and is called
partial least squares path modeling.
PLS path modeling is sometimes known as “soft model-
ing” [43–45] whereas ML is a covariance-based SEM approach,
known as “hard modeling.” This is because ML “aims at opti-
mality in statistical inference, and is designed for testing
hypotheses that are sharp and pure; accordingly, ML is sen-
sitive to the inaccuracies of real-world models and impurities
of real data. PLS is distribution-free (employing nonparamet-
ric statistics), and aims only at (predictive) consistency, and is
therefore insensitive to impurities in the model and the data”
408 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416

Table 1 – Mean scores and standard deviations of each item in the questionnaire related to the research model constructs
(N = 1323).
Variables Mean SD Construct (LV) definition/Item (MV) in questionnaire
a
Performance Expectancy (PE) = the degree to which an individual believes that using the system will help him or her to attain
gains in job performance
PE1 6.28 1.30 I would find the computer system useful in my job.
PE2 6.34 1.23 Using the computer system enables me to accomplish tasks more quickly.
PE3 6.17 1.27 Using the computer system increases my productivity.
PE4 5.73 1.43 If I use the computer system. I will increase my chances of getting a raise.

Effort Expectancy (EE)a = the degree of ease associate with use of system
EE1 5.66 1.31 My interaction with the computer system would be clear and
understandable.
EE2 5.68 1.29 It would be easy for me to become skillful at using the computer system.
EE3 5.68 1.29 I would find the computer system easy to use.
EE4 5.22 1.40 Learning to operate the computer system is easy for me.

Social Influence (SI)a = the degree to which an individual perceives that important others believe he or she should use the system
SI1 5.47 1.46 People who influence my behavior think that I should use the computer
system.
SI2 5.51 1.42 People who are important to me think that I should use the computer
system.
SI3 5.30 1.63 The senior health administration has been helpful in the use of the
computer system.
SI4 5.71 1.46 In general, provincial health office has supported the use of the computer
system.

Voluntariness (VO)a = the degree to which use of IT is perceived as voluntary or free will
VO1 4.80 1.72 My boss does not require me to use computer.
VO2 4.91 1.75 Although it might be helpful, using a computer system is certainly not
compulsory in my job.

Intention to Use (IN)a = the degree to which an individual intends to use IT

IN1 6.17 1.11 I intend to use the computer system in the next 2 months.
IN2 6.17 1.10 I predict I would use the computer system in the next 2 months.
IN3 6.13 1.16 I plan to use the computer system in the next 2 months.

Facilitating Conditions (FC)a = the degree to which an individual believes that an organizational and technical infrastructure
exists to support use of the system
FC1 5.60 1.45 I have the resource necessary to use the computer system.
FC2 5.26 1.27 I have knowledge necessary to use the computer system.
FC3 5.73 1.22 The computer system is compatible with other systems I use.
FC4 5.13 1.56 Health IT persons in province are available for assistance with system
difficulty.
FC5 5.65 1.23 I think that using computer system fits well with the way I like to work.
FC6 5.21 1.37 I have knowledge sources (e.g. books, documents, consultants) help me
learn about computer system.

Experienceb = IT experience
Experience 6.69 0.94 How long ago did you first start to use a computer?

IT knowledge = general knowledge (similar to a college entrant) related to basic personal computer (PC) components and basic
functions of PC and the Internet
Knowledge 13.20 4.50 There are 20 test statements. The highest score is 20, the lowest score is 0.

Care and Report Usec = use for providing care and routine reporting
C&R Use1 3.35 0.88 Recording people (individual and family) information in catchments area.
C&R Use2 3.49 0.87 Recording patients (clients) information.
C&R Use3 3.24 0.91 Retrieving previously recorded individual information for providing care.
C&R Use4 3.51 0.77 Generating mandatory reports.

Administration Usec = use for management and administration

Adm Use1 3.53 0.76 Writing (e.g. official letters, reports, etc.).
Adm Use2 2.89 1.07 Preparing presentation slides or overheads.
Adm Use3 2.48 0.05 Performing statistical analysis.
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416 409

Table 1 – (Continued )
Variables Mean SD Construct (LV) definition/Item (MV) in questionnaire

Communication Usec = use for information search and collaboration with colleagues
Comm Use1 1.76 1.01 Use for information search and collaboration with colleagues.
Comm Use2 2.28 1.02 Searching for information associating with office tasks (e.g. governmental
documents, products prices).
Comm Use3 2.24 1.06 Searching for information for personal interest such as knowledge
associating with personal continuing education and general knowledge.

Use Frequencyd = frequency of IT use

Use Freq 5.42 1.00 How often do you use computer system hand on?

a
Seven-point scale: 1 = strongly disagree, 2 = quite disagree, 3 = slightly disagree, 4 = neither agree nor disagree, 5 = slightly agree, 6 = quite agree
and 7 = strongly agree.
b
Eight-point scale: 1 = never use, 2 = experience < 1 month, 3 = experience 1–6 months, 4 = experience 7–12 months, 5 = experience > 1–2 year,
6 = experience > 2–5 years, 7 = experience > 5–10 years and 8 = experience > 10 years.
c
Four-point scale: 1 = never use, 2 = sometimes use, 3 = frequent use and 4 = always use.
d
Six-point scale: 1 = don not use at all, 2 = use < once a week, 3 = use about once each week, 4 = use several times each week, 5 = use about once
a day and 6 = use several times each day.

[46]. The PLS method is quite robust against manifest vari-
ables’ skewed distributions, multi-collinearity within blocks
of manifest variables and between latent variables, and mis-
specification of the structural model [47–49].
PLS path modeling simultaneously evaluates the measure-
ment model and the structural model relating the associated
constructs. The measurement model is that part of the
research model which portrays the relationships between a
construct and its associated manifest variables (measurement
items). A PLS path model analyzes and interprets data in two
stages: (1) assessment of the measurement model by exam-
ining the reliability and validity of the composite of items
measuring each construct, and (2) assessment of the structural
model. The interpretation sequence aims to ensure that we
have reliable and valid measures of constructs before drawing
conclusions regarding the relationships among those con-
structs.
We used PLS path modeling for model analysis because
our project attempted to predict factors that influence IT
acceptance and IT use. It leaned more toward a predictive
research model less toward a theory confirmatory model. We
also wanted to simultaneously evaluate the reliability and
validity of the measures of the constructs in the model, and
estimate the relationships among these constructs. Further-
more, many manifest variables in our research data were not
normally distributed. In addition, there was a potential multi-
collinearity between latent variables in the model. Given all
these circumstances, we considered PLS method the better
alternative for analysis than covariance-based SEM in our
research project. Finally PLS path modeling has been com-
monly used by information system researchers and those
investigating technology adoption including those who con-
ducted the study that developed the UTAUT model.
4. Results
4.1. Respondent characteristics
The survey response rate was 82% with 1323 out of 1607 CHCs
responding. Of the 1323 respondents, male and female were
represented in essentially equal proportions (54% and 46%)
and average age was 36 years old (SD = 7, range 17–59). Three
quarters (74%) of the respondents had a bachelor degree and
one-third (32%) held the position of a public health admin-
istration officer. On the whole, respondents were not new
employees but rather had worked in centers for several years,
and had worked their way up through the position classifi-
cation system. They spent approximately 40% of their work
hours providing health services and almost equal proportion
of work hours in data management and report production. The
detailed results of the rest of the survey have been described
elsewhere [16]. For the manifest variables in the research
model, Table 1 provides the item descriptions and descriptive
statistics of the measured items from the survey.
Examination of all 1323 responses revealed that ten percent
(136) were inconsistent cases. They abnormally shifted the
predicting effects of other variables [50]. These responses were
from subjects who declared very low intention but reported a
high degree of IT use, which conflicted with the Theory of Rea-
soned Action (TRA) where intention has a positive effect on
behavior. Comparisons of the characteristics of inconsistent
cases with the rest of the respondents showed no differences
in terms of respondents’ demographic information, CHCs’
characteristics, IT resources or among the other model vari-
ables. Because we could not distinguish the two groups with
using any other characteristics, we concluded that they were
outliers who did not conform to the basic assumptions of the
model. Therefore we excluded them from the analysis and
reported the results for the remaining respondents (1187).
4.2. Measurement model evaluation
The reliability of construct measurement was evaluated by
examining the composite reliability and internal consistency
reliability (Cronbach’s alpha) as determined by PLS for each
construct (see Table 2). All constructs exhibited composite
reliability and Cronbach’s alpha greater than the acceptable
level of 0.70 indicating that the measurement errors were rel-
atively small [51,52]. Convergent validity of a set of items with
respect to their associated construct is assessed by examin-
ing the factor loadings of the items on the model’s constructs.
410 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416

Table 2 – Construct composite reliability (CR), internal consistency reliability (ICR), diagonal elements are the square root
of the shared variance between the constructs and their measures (AVE); off-diagonal elements are correlations between
constructs (N = 1187).

Note: Construct identifiers: PE = Performance expectancy, EE = Effort Expectancy, SI = Social Influence, VO = Voluntariness, IN = Intention,
FC = Facilitating Conditions, Knw = Knowledge, Exp = Experience, Use Freq = Use Frequency, C&R Use = Care and Report Use, Adm
Use = Administration Use, Comm Use = Communication Use.

A high loading of the item on its underlying construct and
lower loadings on unrelated constructs indicates convergent
validity. Barclay et al. suggested that a heuristic criteria for
convergent validity is to accept items with loadings of 0.70
or more [43]. However, Falk and Miller argued that the fac-
tor loading might be acceptable at the value of 0.55 [44]. As
shown in Table 3, factor loadings of all items on their respec-
tive associated constructs are equal or greater than 0.70 while
their loadings on unrelated constructs are lower. For the sec-
ond order construct of IT Use, factor loadings of individual first
order constructs on the second order construct met the same
criteria with the exception of Frequency on Use which was 0.44
which was slightly less than 0.55. Overall, the results demon-
strated reasonable convergent validity of the measurement
model with respect to the model’s constructs.
Discriminant validity indicates the extent to which a given
construct is different from other constructs. In PLS path mod-
eling analysis adequate discriminant validity is demonstrated
when a construct shares more variance with its measures
(indicators) than with other constructs in the model. A mea-
sure of variance shared between a construct and its measures
is the Average Variance Extract (AVE) that should be greater
than the variance shared between the construct and others
constructs in the model [51]. Table 2 also displays the corre-
lation matrix for the constructs. The diagonal of the matrix
contains the square roots of the AVEs which provide a metric
comparable to a correlation (the square root of the variance
shared between two variables). For adequate discriminant
validity, the diagonal elements should be greater than the off-
diagonal elements in the corresponding rows and columns. It
should be noted that in the case of IT Knowledge, Use Frequency,
and Experience the diagonal value is 1.0 indicating that there
was only one measurement item associated with each of these
constructs. The results demonstrated adequate discriminant
validity for all constructs in our research model.
4.3. Structural model evaluation
Structural model evaluation is the assessment of the predic-
tive or causal relationship between constructs in the model.
Fig. 2 exhibits the structural model and the analytical results.
It portrays the path coefficients (ˇ), the explained variance (R2 )
and the effect size (f2 ) for each path segment in the model. All
path coefficients are significant and in the direction proposed.
The constructs of performance expectancy (PE), effort
expectancy (EE), Voluntariness (VO) and social influence (SI)
were predictive of Intention to Use (IN) with an R-square
of 0.54. IN, facilitating conditions (FC) and Experience (Exp)
accounted for a smaller R-square (0.27) in predicting IT Use
(Use). The ability of IT Knowledge (Knowledge) to predict FC
(R-square = 0.02) was low, even though the path coefficient of
Knowledge on FC was statistically significant. Nonetheless, the
predicting constructs in the model each accounted for more
than 1.5% of the variance in a predicted construct. The R-
squares of all predicted constructs in our model were greater
than 0.10 except the R-square for FC. The results indicate
that our structural model is adequate in that it has met Falk
and Miller’s criteria for level of variance explained (R-squared
≥0.10 and predictor variable explaining ≥1.5% of variance) [44].
Examining the individual effect sizes of predictors (f2 )
provides further information about the unique and separate
contributions of each of the constructs. According to Chin, f2
of 0.02, 0.15 and 0.35 can be viewed as a gauge for whether a
predicting construct has a small, medium, or large effect at the
structural level [40]. With respect to IN, PE had the largest effect
size (0.19) classified as “medium” according to the categories
proposed by Chin. EE, SI, and VO all had less of an effect with
sizes in the “small” category. This indicates that the majority
of the variance was accounted for by the combined effect of
these four constructs rather than their separate, independent
contributions.
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416 411

Table 3 – Loadings and cross-loadings of individual items (MVs) on their constructs (LVs) (N = 1187).
PE EE SI VO IN FC Knw Exp Use Freq C&R Use Adm Use Comm Use USE

PE1 0.89 0.59 0.53 0.25 0.63 0.54 0.08 0.10 0.09 0.25 0.23 0.12 0.26
PE2 0.93 0.61 0.53 0.23 0.64 0.56 0.13 0.12 0.11 0.28 0.25 0.13 0.29
PE3 0.93 0.62 0.57 0.22 0.64 0.58 0.11 0.11 0.12 0.24 0.25 0.15 0.28
PE4 0.77 0.54 0.52 0.22 0.5 0.49 0.06 0.07 0.07 0.19 0.19 0.16 0.23
EE1 0.62 0.86 0.52 0.26 0.57 0.60 0.15 0.18 0.11 0.26 0.26 0.22 0.32
EE2 0.61 0.91 0.51 0.25 0.55 0.59 0.15 0.17 0.11 0.24 0.22 0.18 0.28
EE3 0.61 0.90 0.53 0.23 0.55 0.60 0.13 0.19 0.13 0.25 0.25 0.20 0.30
EE4 0.46 0.80 0.43 0.27 0.42 0.62 0.15 0.24 0.13 0.24 0.25 0.24 0.32
SI1 0.52 0.53 0.84 0.19 0.47 0.52 0.01 0.06 0.07 0.17 0.16 0.15 0.20
SI2 0.54 0.50 0.85 0.22 0.52 0.52 0.04 0.07 0.07 0.18 0.15 0.14 0.20
SI3 0.44 0.41 0.80 0.21 0.39 0.50 0.01 0.04 0.04 0.12 0.10 0.07 0.12
SI4 0.49 0.43 0.79 0.24 0.45 0.55 0.02 0.04 0.03 0.16 0.13 0.07 0.16
VO1 0.23 0.28 0.24 0.91 0.27 0.36 0.04 0.14 0.04 0.15 0.16 0.20 0.21
VO2 0.25 0.25 0.24 0.93 0.30 0.38 0.02 0.13 0.05 0.15 0.16 0.21 0.22
IN1 0.67 0.58 0.55 0.31 0.95 0.63 0.15 0.17 0.11 0.3 0.28 0.15 0.32
IN2 0.66 0.59 0.54 0.30 0.97 0.63 0.14 0.19 0.14 0.32 0.27 0.17 0.33
IN3 0.64 0.57 0.53 0.28 0.94 0.6 0.11 0.15 0.13 0.31 0.27 0.16 0.32
FC1 0.44 0.45 0.49 0.29 0.44 0.72 0.10 0.11 0.06 0.20 0.18 0.21 0.25
FC2 0.41 0.58 0.43 0.31 0.45 0.82 0.18 0.22 0.15 0.26 0.26 0.26 0.34
FC3 0.60 0.60 0.56 0.33 0.59 0.84 0.13 0.14 0.13 0.28 0.25 0.21 0.32
FC4 0.38 0.41 0.47 0.28 0.39 0.69 0.01 0.05 0.04 0.15 0.12 0.14 0.17
FC5 0.59 0.61 0.57 0.34 0.65 0.84 0.12 0.20 0.14 0.34 0.29 0.27 0.39
FC6 0.42 0.51 0.46 0.34 0.45 0.77 0.10 0.16 0.12 0.19 0.19 0.26 0.27
Knw 0.11 0.16 0.02 0.03 0.14 0.15 1.00 0.15 0.09 0.11 0.2 0.09 0.17
Use Exp 0.11 0.22 0.07 0.15 0.18 0.2 0.15 1.00 0.38 0.26 0.30 0.29 0.40
Use Freq 0.11 0.14 0.07 0.05 0.13 0.15 0.09 0.38 1.00 0.30 0.26 0.18 0.44
C&R U1 0.22 0.24 0.16 0.12 0.29 0.23 0.07 0.25 0.24 0.80 0.44 0.20 0.65
C&R U2 0.16 0.17 0.09 0.09 0.21 0.20 0.09 0.17 0.22 0.82 0.32 0.14 0.58
C&R U3 0.20 0.23 0.15 0.20 0.23 0.28 0.09 0.19 0.24 0.81 0.38 0.27 0.65
C&R U4 0.26 0.25 0.20 0.10 0.28 0.28 0.09 0.20 0.24 0.67 0.43 0.22 0.59
Adm U1 0.24 0.21 0.14 0.11 0.26 0.19 0.17 0.22 0.20 0.43 0.74 0.21 0.59
Adm U2 0.21 0.24 0.13 0.15 0.22 0.23 0.16 0.25 0.22 0.41 0.84 0.34 0.67
Adm U3 0.18 0.21 0.13 0.16 0.21 0.26 0.14 0.25 0.20 0.37 0.78 0.46 0.67
Comm U1 0.07 0.15 0.06 0.18 0.07 0.21 0.07 0.27 0.15 0.17 0.37 0.82 0.55
Comm U2 0.15 0.22 0.13 0.18 0.15 0.27 0.04 0.21 0.14 0.27 0.37 0.87 0.61
Comm U3 0.17 0.24 0.15 0.20 0.19 0.28 0.13 0.26 0.17 0.24 0.38 0.88 0.61

Note: Construct Identifiers: PE = Performance expectancy, EE = Effort Expectancy, SI = Social Influence, VO = Voluntariness, IN = Intention,
FC = Facilitating Conditions, Knw = Knowledge, Exp = Experience, Use Freq = Use Frequency, C&R Use = Care&Report Use, Adm
Use = Administration Use, Comm Use = Communication Use, USE = IT Use.

IN, FC and previous computer experience (Exp) together
explained 27% of the variance in IT use (Use). Among these,
Exp exerted the strongest individual effect on the explained
variance followed by FC and IN respectively. The effect size of
Exp and FC was “medium” while IN was “small”.
Overall the results of measurement model analysis indi-
cated that measures of all constructs demonstrated values for
reliability and validity meeting existing criteria and thus indi-
cating that they were psychometrically acceptable measures
for this structural model. The structural model analytic results
indicated that more than half of the variance in intention to
use (IT acceptance) and over one quarter of the variance in IT
use was explained by these measures. Given that there are a
large number of factors that could impact both IT acceptance
and IT use, the variance explained by our research model is
substantial.
5. Discussion
This study sought to identify factors that predict the sur-
vey respondents’ intention to use and their use of health
IT in community health centers in Thailand by applying
an extant theoretical model. As a result, the focus of the
study was the prediction of two theoretical constructs—the
intention to use IT and IT use employing an adaptation
of the UTAUT model. The results showed that intention to
use health IT is a function of the perception that health
IT is useful (performance expectancy), that it exhibits ease
of use (effort expectancy), that important others believed
that he/she should use health IT (social influence) and the
perception that one has a choice in the use of IT (volun-
tariness). The predictive power of these four factors was
substantial and accounted for more than half of the vari-
ance in the intention to use IT. Among these four influencing
factors, performance expectancy was by far the strongest
predicting factor. Our results are consistent with a number
of prior studies where performance expectancy has more
of an influence than effort expectancy and social influ-
ence. The influence of voluntariness on the intention to
use is also similar to previous findings that the percep-
tion of freedom of choice has positive effect on intention to
use.
412 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
Fig. 2 – PLS path modeling analytical results: ˇ = path beta coefficient, R2 = R-square, f2 = effect size, all ˇs are significant at a
p < 0.001.
Model analysis also revealed that for Thai CHCs health IT
use, the construct of our ultimate interest, was predicted by
previous IT experiences (Experience), intention to use IT (IN)
and the perception that organizational and technical support
for the use of health IT exists (FC—facilitating conditions).
Twenty seven percent of the variance in IT use was explained
by these three factors. Recognizing that there are a variety of
other factors not include in the study that can influence IT
use, the result provides us a better understanding of the fac-
tors influencing IT use in the context of this study. The positive
effects of facilitating conditions, experience and intention to
use are similar to findings from previous IT adoption research.
In this study, previous IT experience appeared to have stronger
effect on IT use than did facilitating conditions and intention
to use. The smaller than expected effect of intention to use
may be attributable to the nature of our study methods and
population. The Use Frequency question in the survey revealed
that the respondents were already frequent users of health
IT. According to Triandis (1980), frequently perform behaviors
are likely to come under the control of habits, and the impact
of intentions on behavior is thereby reduced [53,54]. The the-
ory of plan behavior (TPB) suggests that intention to perform
behavior predicts future behavior. This study may not be able
to demonstrate the full effect of the intention to use on use
behavior because it was a “snap-shot” observation which cap-
tured attitudes reflecting intention to use IT and reports of use
behavior simultaneously.
The positive relationship between IT knowledge and the
facilitating conditions supported the validity of facilitating
conditions construct since one of the manifest variables that
reflected the FC construct was “I have knowledge necessary
to use the computer system”. The positive relationship indi-
cates that a higher IT knowledge score is reflected in the
perception of having necessary IT knowledge. The perception
of possessing knowledge is an aspect of the perception of “self-
behavior control”. The UTAUT model integrates the concept of
self-behavior control into its facilitating condition construct.
Therefore, the observed positive relationship provides sup-
porting evidence for the validity of the facilitating condition
construct.
The study provides further evidence for the basic validity
of the UTAUT model in that it confirmed the relationships
among the variables in the model proposed by Venkatesh.
The relationships between performance expectancy, effort
expectancy, social influence and intention to use IT were con-
firmed as existing and in the same direction as the UTAUT
model proposes. Furthermore there was a positive though
weak relationship between Intention to Use and reported
IT Use. All this provides evidence that the UTAUT model is
applicable in health care settings and in other cultures as rep-
resented in this study.
In addition, this study demonstrated the direct effects
of effort expectancy, social influence, facilitating condition
which are the core constructs of the UTAUT model whereas the
original UTAUT model demonstrated these three main con-
structs effect through the three or four way interaction terms
with age, gender, experience and voluntariness. Although our
model does not explain IT use to the same extent as the orig-
inal UTAUT model, our expanded UTAUT model provides a
better understanding of the technology adoption in a devel-
oping country’s healthcare system. Moreover, our research
model has demonstrated the adequate construct validity and
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
reliability of a measurement model related to the core con-
structs of the UTAUT model. Rather than measuring IT use in
one dimension as was done in previous studies, we expanded
the measure to represent three dimensions of use associ-
ated with activities and one dimension of use frequency.
Modeling IT use in this multi-dimensional way may pro-
vide a more comprehensive measure of IT use than previous
studies.
Moreover, the study applied the partial least squares
path modeling method which is not commonly used in the
health/biomedical informatics research studies but which
has significant advantages in terms of dealing with complex
models and studies with many variables. This study demon-
strated that the method works well in a field study and in
exploratory research. Based on the fact that PLS works with
a complex model, small sample sizes and violations of the
normal distribution assumption it could be a powerful tool
for many health/biomedical informatics researchers who deal
with complex path models, particularly in studying socio-
technical aspects of IT.
Knowledge acquired from this study can potentially benefit
individuals who currently work with health information sys-
tems and health IT implementation in developing country like
Thailand. The strong influence of performance expectancy
and effort expectancy on the adoption of health IT suggests
that work-related benefits of implemented systems must be
perceivable, identifiable and substantial, and they should be
viewed as relatively easy to use. An awareness of these effects
on the system adoption can help develop and accelerate the
process of the implementation. Adequate facilitating condi-
tions (continuous training and technical support to users) also
play an important role in technology use. Social influence and
perception of voluntariness do affect technology adoption.
Therefore it is important to identify individuals with strong
personal influence (formal and informal) and work with them
to become advocates for technology use in order to facilitate
the implementation process. Fostering an environment, both
from a top down and bottom up perspective, where use of tech-
nology is desirable, and that maintains the perception that use
is a choice, could do much to facilitate the implementation
process.
There are several potential limitations of this study. We
assumed that one respondent in a given CHC represented
all people who worked in the CHC. If the received response
was not representative then the generalizability of our find-
ings would be limited. Nevertheless, given the CHC working
context, the potential selection bias was minimized by a rea-
sonably representation of people who work in CHCs: 28% of
respondents were health workers, 12% nurses, 26% public
health specialists (the higher career for health worker and
nurse working in CHCs) and 31% public health administra-
tor (the highest career in CHCs). Regarding the CHC’s working
context, a CHC is a sub-district (or “Tambon” in Thai) health
service unit. It is a first-line unit, covering a population of
1000–5000, with 3–5 health staff typically a health worker,
a midwife and a technical nurse, without full-time doctors.
Because of the small office, the main job of every employee is
providing healthcare services including information manage-
ment tasks (data input is embedded in the primary healthcare
service tasks). These individuals are highly aware of all situ-
413
ations in their work place and thus the survey results should
be reasonable reflective of all CHC personnel.
Another potential limitation related to the representative-
ness of CHC responses and is related to the sampling method
used. Because of limited budget, the need to depend on provin-
cial officials to distribute our surveys to the CHCs and the lack
of any existing master list of CHCs with contact information
we were unable to use simple random sampling. Rather, we
chose to use a cluster sampling approach [55]. We performed
a first level cluster sampling via random sampling of provinces
in each region. This was followed by random sampling of CHC’s
within a region. In terms of representativeness, the number of
sampled provinces and CHCs were proportionate to the distri-
bution of population and CHCs in each region. Furthermore,
the response rates in all provinces were high and similar. Given
all these circumstance, we maintain that the results were not
compromised by this sampling strategy.
A third potential limitation was due to the cross-sectional
nature of the survey design where causal relationships
between model factors due to time precedence could not be
claimed. Similar follow up studies in the same cohort could
shed the light on the causal relationships. The cross-sectional
design introduced a limitation on interpreting the relationship
between the measures of intention to use and IT use. Since
they were measured at the same time the antecedent rela-
tionship of intention to the use behavior was not captured.
Therefore the causal link between intention to use IT and IT
use is less logically obvious.
A fourth potential limitation has to do with the decision to
exclude the inconsistent cases who reported a low Intention
to Use and a high Usage rate. These persons did not conform
to one of the basic model assumptions that there is a positive
relationship between these two constructs. By excluding this
group of respondents, the generality of the findings is limited
to persons who exhibit a non-inverse relationship which is
the vast majority of the responding group. Since we were not
able to differentiate this group on any other characteristics
or responses we do not know if they represent an important
subgroup requiring further study or simply a group that mis-
understood the Usage rating scale. In any event, the results of
the study do represent 90% of the original respondents.
The fifth potential limitation of the study is associated with
the relationship between past experience and IT use. Previous
IT experience was modeled to influence IT use in the study, but
it may well be a bidirectional relationship. For example, people
who frequently use IT may report a high degree of experience
and vice versa. This possible bidirectional relationship affects
the interpretation of the influence of IT experience on IT use.
Nevertheless, at least, the positive correlation between past
experience and IT use can be claimed.
Finally the results of this study can only apply to health
IT in general, because the focus of the study is not a specific
technology artifact. Therefore, results may vary in adoption
of a specific technology (e.g. applications, services, machines)
because of the differences in characteristics of those artifacts.
As a result, adoption studies of specific technology artifacts
are still needed but should be guided by the overall results of
this study.
Our study can be characterized as one of the studies in the
IT adoption research stream that focuses on individual adop-
414 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
Summary points
What was known before the study?
• At the national level, there are no studies about health
providers’ IT acceptance and use and its predicting fac-
tors in Thailand.
• The UTAUT model had been employed to study user’s
IT acceptance and use in many developed countries
and many industries including health industry, but
there are a few studies employing the UTAUT model
to study IT acceptance and use in healthcare system
in developing countries.
What the study has added to the body of knowledge?
• Community health centers’ personnel in Thailand
demonstrated high level of health IT acceptance and
use. Their intention to use health IT was the function
of performance expectancy, effort expectancy, social
influence and voluntariness. CHC’s health IT use was
influenced by past IT experience, facilitating condi-
tions and intention to use IT.
• The study confirms the validity of the UTAUT model in
the field context of a developing country’s healthcare.
tion of information technology. The other research stream
focuses on adoption of information technology at organization
level. However, the two research streams are usually pur-
sued separately though they are related because successful
implementation of an IT project requires both organizational
commitment and individual utilization. Although, there may
be a communality of factors that influence IT adoption at both
levels, there are number of factors that have been studied
at the organization level and have not been studied at the
individual level. Future research models should address the
relationships and linkages between the two levels of IT adop-
tion to develop a more complete model of adoption. Therefore,
modification of our research model with other predictors at
the organization level to comprehensively understand health
IT adoption could be conducted in the future. Predictors sug-
gested in a comprehensive review of the relationship between
individual and organization predictors of the adoption and
diffusion of IT-base innovation [56] by Jeyaraj et al. could be
reasonable candidates for the model development.
6. Conclusions
This study applied the modified UTAUT model to understand-
ing factors that influence health IT acceptance and use in
Thailand community health centers. The results show that
Intention to use health IT is a function of various concepts
including the perception that (1) health IT is useful, (2) it is not
too difficult to use, (3) important persons/others believed that
he/she should use health IT and (4) the perception of free will
to use IT influence the intention to use. Among these factors,
performance expectancy exerted the strongest effect. CHCs’
IT use was influenced by (1) past IT experience, (2) the per-
sonnel’s intention to use the system, and (3) the belief that
an organization and technical support exists to support the
use of health IT. Past IT experience and facilitating conditions
were prominent in predicting IT use. This study confirms the
validity of the UTAUT model in the field context of a develop-
ing country’s healthcare system. The study has implications
for the development of health IT in developing country and
health/biomedical informatics research. Knowledge gained
from the study is beneficial to both health IT policy makers
and people who work with health IT development and imple-
mentation projects.
Acknowledgements
The authors gratefully acknowledge Dr. Suvat Siasiriwattana,
Mrs. Aree Wadwongtham, Dr. Narong Kasitipradith and the
staffs at the Bureau of Health Policy and Strategy, Ministry of
Public Health, Thailand for their great support. The authors
thank Ms. Parunee Yimsabai and staffs at the Center for Health
Equity Monitoring (CHEM), Faculty of Medicine, Naresuan Uni-
versity, Thailand for helping the survey process. The primary
funding source of this project was from WHO Thailand, project
activities proposal number 040199.
references
[1] P. Beynon-Davies, Human error and information systems
failure: the case of the London ambulance service
computer-aided despatch system project, Interacting with
Computers 11 (1999) 699–720.
[2] R. Heeks, Information systems and developing countries:
failure, success, and local improvisations, The Information
Society: An International Journal 18 (2) (2002) 101–112.
[3] P. Littlejohns, J.C. Wyatt, L. Garvican, Evaluating
computerised health information systems: hard lessons still
to be learnt, BMJ 326 (7394) (2003) 860–863.
[4] B. Kaplan, N.T. Shaw, Future directions in evaluation
research: people, organizational, and social issues, Methods
of Information in Medicine 43 (3) (2004) 215–231 (see
comment).
[5] S. Whetton, Health Informatics: A Socio-Technical
Perspective, South Melbourne, Vic, Oxford, 2005.
[6] J. Aarts, P. Gorman, IT in health care: sociotechnical
approaches “to err is system”, International Journal of
Medical Informatics 76 (Supplement 1) (2007) S1–S3.
[7] M. Berg, J. Aarts, J. van der Lei, ICT in health care:
sociotechnical approaches, Methods of Information in
Medicine 4 (42) (2003) 297–301.
[8] D.A. Giuse, K.A. Kuhn, Health information systems
challenges: the Heidelberg conference and the future,
International Journal of Medical Informatics 69 (2–3) (2003)
105–114.
[9] Health Policy in Thailand 2006 (Internet resource) 2006 (cited
11 June 2006), Available from:
http://www.searo.who.int/LinkFiles/On-going projects HP-
Tha-2006.pdf.
[10] S. Pannarunothai, D. Patmasiriwat, S. Srithamrongsawat,
Universal health coverage in Thailand: ideas for reform and
policy struggling, Health Policy 68 (1) (2004) 17–30.
[11] V. Tangcharoensathien, S. Pitayarangsarit, C. Vasavid,
Universal Health Care Coverage and Medium Term
 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
Financing Implications, International Health Policy
Program-Thailand, Nonthaburi, 2002.
[12] A. Towse, A. Mills, V. Tangcharoensathien, Learning from
Thailand’s health reforms, BMJ 328 (January (7431)) (2004)
103–105.
[13] Information and Communication Technology Center,
Ministry of Public Health’s Information and Communication
Technology Master Plan 2004–2006 (Thai Language) (Web
page) 2004 (cited 2004 April 25, 2004), Available from:
www.moph.go.th/ict/data/MOPH ICTPlan.zip.
[14] N. Kasitipradith, The Ministry of Public Health telemedicine
network of Thailand, International Journal of Medical
Informatics 61 (2–3) (2001) 113–116.
[15] K. Raungarrerat, Development Guideline for Public Health
Information Technology in Thailand, Health Systems
Research Institute, Ministry of Public Health of Thailand,
Bangkok, Report No.: 974-291-702-7, 1998.
[16] B. Kijsanayotin, S. Pannarunothai, S. Speedie, Penetration
and adoption of health information technology (IT) in
Thailand’s community health centers (CHCs): a national
survey, Medinfo 12 (Pt 2) (2007) 1154–1158.
[17] V. Venkatesh, M. Morris, G. Davis, F. Davis, User acceptance
of information technology: toward a unified view, MISQ 27
(3) (2003) 425–478.
[18] F.D. Davis, Perceived usefulness, perceived ease of use, and
user acceptance of information technology, MISQ 13 (3)
(1989) 319–339.
[19] D.R. Compeau, C. Higgins, Social cognitive theory and
individual reactions to computing technology: a longitudinal
study, MISQ 23 (2) (1999) 145–158.
[20] G.C. Moore, I. Benbasat, Development of an instrument to
measure the perceptions of adopting an information
technology innovation, ISR 2 (3) (1991) 192–222.
[21] R. Thompson, C. Higgins, J. Howell, Personal computing:
toward a conceptual model of utilization, MISQ 15 (1) (1991)
125–143.
[22] F.D. Davis, R. Bagozzi, P. Warshaw, Extrinsic and intrinsic
motivation to use computers in the workplace, Journal of
Applied Social Psychology 22 (14) (1992) 1111–1132.
[23] I.C. Chang, H.-G. Hwang, W.-F. Hung, Y.-C. Li, Physicians’
acceptance of pharmacokinetics-based clinical decision
support systems, Expert Systems with Applications 33 (2)
(2007) 296–303.
[24] S. Taylor, P. Todd, Understanding information technology
usage: a test of competing models, ISR 6 (2) (1995)
144–176.
[25] V. Venkatesh, F. Davis, A theoretical extension of the
technology acceptance model: four longitudinal field
studies, Management Science 46 (2) (2000) 186–204.
[26] R. Agarwal, J. Prasad, A field study of the adoption of
software process innovations by information systems
professionals, IEEE Transactions on Engineering
Management 47 (3) (2000) 295–308.
[27] L.K. Schaper, G.P. Pervan, ICT and OTs: a model of
information and communication technology acceptance and
utilisation by occupational therapists, International Journal
of Medical Informatics 76 (Supplement 1) (2007) S212–
S221.
[28] P.Y.K. Chau, P.J.-H. Hu, Investigating healthcare
professionals’ decisions to accept telemedicine technology:
an empirical test of competing theories, Information &
Management 39 (4) (2002) 297–311.
[29] E. Karahanna, D.W. Straub, N.L. Chervany, Information
technology adoption across time: a cross-sectional
comparison of pre-adoption and post-adoption beliefs, MIS
Quarterly 23 (2) (1999) 183–213.
[30] S. Han, P. Mustonen, M. Seppanen, M. Kallio, Physicians’
behavior intentions regarding the use of mobile technology:
415
an exploratory study, in: The Eighth Pacific-Asia Conference
on Information Systems 2004 July 8–11, 2004, Shanghai,
China, 2004.
[31] G.C. Moore, An examination of the implementation of
technology by users: A diffusion of innovations perspective,
Unpublished doctoral dissertation, Vancouver, University of
British Columbia, 1989.
[32] R. Agarwal, J. Prasad, The role of innovation characteristics
and perceived voluntariness in the acceptance information
technology, Decision Science 28 (3) (1997) 557–582.
[33] V. Venkatesh, C. Speier, Computer technology training in the
workplace: a longitudinal investigation of the effect of
mood, Organizational Behavior and Human Decision
Processes 79 (1) (1999) 1–28.
[34] R.L. Thompson, C.A. Higgins, Influence of experience on
personal computer utilization: testing a conceptual model,
Journal of Management Information Systems 11 (Summer
(1)) (1994) 167.
[35] I. Ajzen, The theory of planned behavior, Organizational
Behavior and Human Decision Processes 50 (2) (1991)
179–211.
[36] B.H. Sheppard, J. Hartwick, P.R. Warshaw, The theory of
reasoned action: a meta-analysis of past research with
recommendations for modifications and future research,
The Journal of Consumer Research 15 (3) (1988)
325–343.
[37] B. Kijsanayotin, S. Pannarunothai, S. Speedie, Development
of an instrument to measure health center (HC) personnel’s
computer use, knowledge and functionality demand for HC
computerized information system in Thailand, AMIA
Annual Symposium Proceedings (2005) 1007.
[38] T. Dirk, K. Henning, H. Lutz, PLS Path Modeling A Software
Review, 2006 December (cited 10 June 2007), Available from:
http://ideas.repec.org/p/hum/wpaper/sfb649dp2006-
084.html.
[39] C.M. Ringle, S. Wende, A. Will, SmartPLS 2.0 M3. 2.0 M3 (beta)
ed., University of Hamburg, Hamburg, Germany, 2005.
[40] W. Chin, The partial least squares approach to structural
equation modeling, in: G.A. Marcoulides (Ed.), Modern
Methods for Business Research: Quantitative Methodology
Series, Lawrence Erlbaum, Mahwah, NJ, 1998, pp. 295–336.
[41] W. Chin, P. Newsted, Structural equation modeling analysis
with small samples using partial least squares, in: R.H.
Hoyle (Ed.), Statistical Strategies for Small Sample Research,
Sage Publications, Thousand Oaks, CA, 1999, pp. 307–339.
[42] C. Fornell, A Second Generation of Multivariate Analysis,
Praeger, New York, NY, 1982.
[43] D. Barclay, C. Higgins, R. Thompson, The partial least
squares (PLS) approach to causal modeling: personal
computer adoption and use as an illustration, Technology
Studies 2 (2) (1995) 285–323.
[44] R.F. Falk, B. Miller Nancy, A Primer for Soft Modeling, 1st ed.,
University of Akron Press, Akron, Ohio, 1992.
[45] J.-B. Lohmöller, Latent Variable Path Modeling with Partial
Least Squares, Physica-Verlag, Heidelberg, 1989.
[46] H. Wold, Soft modeling: the basic design and some
extensions, in: K.G. Jöreskog, H. Wold (Eds.), Systems Under
Indirect Observation: Causality, Structure, Prediction Part II,
Amsterdam North-Holland, 1982, pp. 1–54.
[47] C. Cassel, P. Hackl, A.H. Westlund, Robustness of partial
least-squares method for estimating latent variable quality
structures, Journal of Applied Statistics 26 (4) (1999) 435–446.
[48] C.M. Cassel, P. Hackl, A.H. Westlund, On measurement of
intangible assets: a study of robustness of partial least
squares, Total Quality Management 11 (7) (2000) 897–907.
[49] Jürgen K-UB, J.Z. Yu, Organizational use of the internet: scale
development and validation, Internet Research 15 (1) (2005)
67–87.
416 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 8 ( 2 0 0 9 ) 404–416
[50] Anonymous, What are outliers in the data? in: C.M.
Croarkin, J.J. Filliben, W.F. Guthrie, A.N. Heckert, B. Hembree,
J. Prinz, et al. (Eds.), NIST/SEMATECH e-Handbook of
Statistical Methods, The National Institute of Standards and
Technology/Semiconductor Manufacturing Technology,
2003.
[51] C. Fornell, D.F. Larcker, Evaluating structural equation
models with unobservable variables and measurement error
Journal of Marketing Research (JMR) 18 (1) (1981) 39–50.
[52] J.C. Nunnally, Psychometric Theory, 2nd ed., McGraw-Hill,
New York, 1978.
[53] H.C. Triandis, Values, sttitudes, and interpersonal behavior,
in: H.E. Howe, Jr., M. Page (Eds.), Nebraska Symposium of
Motivation, Lincoln, 1980, pp. 195–259.
[54] L.T. Webb, P. Sheeran, Does changing behavioral intentions
engender behavior change? A meta-analysis of the
experimental evidence, Psychological Bulletin 132 (March
(2)) (2006) 249–268.
[55] S. Bennett, T. Woods, W.M. Liyanage, D.L. Smith, A simplified
general method for cluster-sample surveys of health in
developing countries, World Health Statistics Quarterly 44
(3) (1991) 98–106.
[56] A. Jeyaraj, J.W. Rottman, M.C. Lacity, A review of the
predictors, linkages, and biases in IT innovation adoption
research, Journal of Information Technology 21 (2006)
1–23.