Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in
The Confirmatory Factor Analysis
(CFA) on GST Compliance Research
Model in Malaysia
Zubaidah Harun1, Pakhriazad Hassan Zaki1, Mohd Hasmadi Ismail1 &
Khairil Wahidin Awang2
1
Faculty of Forestry, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
2
Faculty of Economic and Management, Universiti Putra Malaysia, 43400 UPM Serdang,
Selangor, Malaysia
Abstract: The questionnaire was constructed
based on four constructs which are knowledge
attitude and communication channels as exogenous
variables while GST compliance is endogenous
variable. The questionnaires were distributed to the
wood-based manufacturers throughout seven states
in Peninsular Malaysia. 352 valid data was
evaluated for the validity and reliability. The
ultimate objective of this article is to acquire the
best fit of a research instrument for the GST
compliance study using structural equation model
(SEM) that enable to taking into account the
unreliable factors (items) between exogenous and
endogenous constructs. The items of the constructs
undergo the confirmatory factor analysis (CFA)
procedure involve of unidimensionality test,
convergent validity, construct validity and
discriminant validity. Followed by the
measurement of reliability on all items using
Cronbach alpha using SPSS and construct
reliability (CR) and average variance extracted
(AVE) using AMOS. The result revealed the
constructs of the research model achieved the
validity and reliability for other further analysis in
acquiring high accuracy on the prediction
outcomes.
Keywords: AMOS, Attitude, CFA, GST
Compliance, Knowledge, Reliability and Validity.
1.0 Introduction
Replacement of Goods and Services Tax (GST)
from Sales and Services Tax (SST) is a national tax
system transformation program. As GST is a broad-
based system, it will affect everybody, government
agencies, statutory bodies, companies, social
organizations, politics, associations, clubs, NGOs,
cooperation, welfare and people. Thus, the
Government has given ample time to the parties
involved with kinds of effort and measures have
Imperial Journal of Interdisciplinary Research (IJIR)
been undertaken. The types of educational
programs were provided by the Malaysian Ministry
of Finance and the Royal Malaysian Customs
Department (RMCD) in educating the business
players as the direct person involve in the
implementation of GST.
Generally, the main objective of this study is to
explore the wood-based industry players’
knowledge and attitude toward GST compliance. In
addition, to identify the effectiveness of the
communication channels used to disseminate GST
information. This study also aims to identify the
correlation between knowledge, attitude and
communication channels on GST compliance.
Therefore, Knowledge, Attitude and Practice
(KAP) model as an underpinning theory for
research model is significant in achieving the
objectives of the study. Knowledge, attitude and
communication channels as exogenous constructs
while, GST compliance as an endogenous construct
of the study.
Since this study observes on latent variables such
as the understanding, perception, behaviour and
feeling, Structural Equation Model (SEM) will use
to multiple correlated the latent variables
concurrently in one measurement model that enable
to taking into account the unreliable factors (items)
between exogenous and endogenous constructs.
Therefore, this paper is to emphasize on the
validity and reliability of constructs involve in this
study using confirmatory factor analysis (CFA).
Generally, CFA offers more parsimonious
clarifications and greater modelling flexibility. To
achieve the fitness of the measurement model in
SEM, five models should be applied which are
identification model, specification model,
estimation model, evaluation model and
modification verification model (Zainudin, 2012b).
The researchers deployed the Analysis of Moments
Structures (AMOS) version 20 to confirm the
Page 758
Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in

validity and reliability of the measurement model Fit Index latent

(research model). (CFI) variables are
uncorrelated
and
First of all, the items were tested for compares the
unidimensionality reliability before validation the sample
constructs. The convergent validity and covariance
matrix with a
discriminant validity were undertaken to validate null model
all construct to ensure the consistency of the
measurement model. The unidimensional, validity Tucker Lewis TLI > 0.90 combining
and reliability of measurement model construct are Index (TLI) parsimony
size
used to measure the constructs that could not be
composition
measured directly (Jöreskog & Sörbom, 1993). To of the index
evaluate the fitness of measurement and structural proposed
model (Holmes, Coote & Cunningham, 2006) and model and
null model
Hair, Back, Babin and Rolph (2010) have
suggested using, at least, three fit indexes, which Normed Fit NFI > 0.90 assesses the
are absolute fit, incremental fit and parsimonious Index (NFI) model by
fit for construct validity. At least one index should comparing
hit the threshold of acceptance from each category the χ2 value
of the model
of model fit as in Table 1 below. to the χ2 of
the null
Table 1: Threshold of Acceptance of Indexes model
Name of Name of index Threshold Description Parsimoniou Chi- Chi-Square the chi-
category of s fit Square/Degree / df <5.0 square
acceptance of Freedom divided by
Absolute fit Chi-square P>0.05 evaluate (Chisq/df) degree of
(chisq) (X2) overall freedom
model fit by Source: (Hooper, Coughlan, & Mullen,2008;
assessing the Zainudin,2012b)
degree of
inconsistenc
y among 2.0 Methodology
sample and 2.1 Population and Samples of Study
fitted
covariance The target respondents are among the wood-based
matrices manufacturers located in four regions in Peninsular
Malaysia, namely southern (Johor and Melaka),
Root Mean RMSEA < test for how
Square Error 0.08 well the
central (Kuala Lumpur, Selangor and Negeri
of model with Sembilan), northern (Perlis, Kedah, Penang and
Approximatio chosen Perak) and east coast (Kelantan, Terengganu and
n (RMSEA) parameter Pahang). However, seven states were randomly
estimate
would fit the
selected from eleven these regions. The samples
populations were selected using the propositional stratified
covariance sampling technique since the target population is
matrix heterogeneous (various types of wood-based mills).
Goodness of GFI >0.90 calculate the
They have to be stratified into homogenous groups
Fit Index fraction of to become similar characteristics (Zainudin, 2012).
(GFI) variance that To the date, a total 2,527 of mills became a GST-
is accounted registered person in Peninsular Malaysia. A
for by the
expected
number of samples are 331(n) at confidence level
population 95% and alpha (α) = 0.05 (margin error 5%)
covariance (Krejcie & Morgan, 1970). Next, the simple
Incremental Adjusted AGFI>0.9 calculate the random sampling was employed in the selection of
fit Goodness of 0 ratio of the respondents from four regions.
Fit (AGFI) degree of
freedom for
the model 2.2 The Measurement Instrument
proposed The questionnaire is used as a primary survey
with a degree
of freedom instrument in collecting quantitative data in
for the null numerical form. The structure of the questionnaire
model was developed based on KAP Model. The
questions were designed purely based on the
Comparative CFI > 0.90 assumes all

Imperial Journal of Interdisciplinary Research (IJIR) Page 759

Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in

researcher’s knowledge and working experience in Knowledge require

(CK) d level
the RMCD. The total of 352 questionnaires was is
answered and returned by the respondents who are achieve
d
representative of their company from selected Legal 0.078 0.990 0.996 3.145 The
states of Peninsular Malaysia. These amounts of Knowledge require
data are valid to be analysed. (LK) d level
is
achieve
d
3.0 Data Analysis Technology 0.084 0.988 0.995 3.468 The
Before proceeds to the validity and reliability of the Knowledge require
measurement model, the items firstly should be (IT) d level
is
confirmed the unidimensional items of the achieve
measurement model. What CFA for every latent d
Technical 0.077 0.953 0.989 3.079 The
construct involves in the study was executed to Knowledge require
confirm the first order unidimensional items in the (TK) d level
measurement model. The threshold of the factor is
achieve
loading should be 0.6 and above (Zainudin, 2012b). d
The items with factor loading low than threshold Cognitive 0.00 1.000 1.000 0.053 The
Attitude (C) require
were deleted in order to achieve unidimensionality. d level
The result of unidimensionality as showed in Table is
achieve
2 shows the number of items in the second order d
was deleted where the factor loading below than Affective 0.032 0.998 1.000 1.358 The
threshold (0.6) in the first order. Attitude (A) require
d level
is
Table 2: Unidimensionality Result achieve
Construct Number of Number of d
items items GST 0.039 0.998 0.999 1.54 The
(1st Order) (2nd order) Compliance require
(B) d level
is
Communication Channels (IS) 5 5 achieve
Concept Knowledge (CK) 8 8 d
Legal Knowledge (LK) 7 5
Technology Knowledge (IT) 5 5
Technical Knowledge (TK) 17 9
Cognitive Attitude (C) 6 4 3.1Validity
Affective Attitude (A) 5 4 Both, endogenous and exogenous constructs were
GST Compliance (B) 6 4
validated by three methods, which are convergent,
construct and discriminant validity.
Refer to Table 2 above, the number of items for
3.2 Convergent validity
communication channels, concept knowledge and
technology knowledge are maintained after second The convergent validity is the first method of
order. 2 items of legal knowledge, cognitive validation processes on measurement model.
attitude and GST compliance were dropped to be 5, According to Kline (2011), convergent validity is a
4 and 4 respectively in second order due to less set of items in one construct are inter-correlation, at
than 0.6 factors loading. 8 items were deleted from least, moderate in magnitude and is measured
17 items of technical knowledge became 9 items through average variance extracted (AVE) where
and 1 item was dropped from affective attitude to the threshold is above >0.5 indicates a high
become 4 in the second order measurement model. convergent validity (Fornell & Larcker, 1981).
The goodness of fit indexes was obtained for each Factor loading of each item at ≥ 0.6 considered
construct as shown in Table 3 below. high convergent validity (Hair et al., 2010).
3.3 Construct Validity
Table 3: Fit Indexes for the Constructs The construct validity on all construct of the
Construct Absolut Increment Parsimoniou Comments measurement model as presented in Figure 1 below
e fit al fit s fit
RMSE GFI CFI Chi
achieved the good fitness index with RMSEA =
A (>0.90) (>0.9 Squar 0.080, CFI = 0.883, TLI = 0.876 and Chi-Square/df
(<0.08) 0) e /df
(<5.0)
Communicati 0.00 1.00 1.00 3.27 The
on Channels require
(IS) d level
is
achieve
d
Concept 0.066 0.972 0.988 2.32 The

Imperial Journal of Interdisciplinary Research (IJIR) Page 760

Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in
= 3.270.
Figure 1: The Measurement Model of
Research Model
All constructs statistically achieved the AVE
threshold (>0.5) with high convergent validity (see
column 5) as presented in Appendix A below. AVE
for communication channels is 0.742, concept
knowledge is 0.629, legal knowledge is 0.869,
technology knowledge and technical knowledge is
0.717 and 0.816 respectively. AVE for cognitive
attitude and affective is 0.676 and latter 0.653,
AVE for GST compliance is 0.687.
3.5 Discriminant Validity
The discriminant validity is to avoid any redundant
items in the measurement model (Zainudin, 2012b).
The items should not be related are in reality not
related. It involves the relationship between a latent
construct and other constructs of a similar nature.
Discriminant validity can be identified by
comparing the variance shared by the average AVE
between these two constructs (Bove, Pervan,
Beatty, & Shiu, 2009). The estimated correlations
between constructs should not be greater than 0.85
(Kline, 2011).
Table 4: Discriminant Validity Result
Construct Knowledge Attitude Comm GST
Channels Compliance
Knowledge
0.431
Attitude 0.140 0.814
Comm Channels 0.115 0.246 0.841
GST Compliance 0.301 0.839 0.510 0.829
The result of discriminant validity as revealed in
Table 5 above for all 4 constructs which are
knowledge, attitude, communication channels and
GST compliance statistically achieved the value
Imperial Journal of Interdisciplinary Research (IJIR)
(≤0.85) and correlated between respective
constructs.
3.6 Reliability
Reliability will assess through three criteria
namely, internal reliability using Cronbach alpha
with threshold 0.600 and above (Nunnally and
Bernstein, 1994) calculated using SPSS. While
construct reliability (CR) should be 0.6 and above
and average variance extracted (AVE) should be
greater than 0.5 using AMOS application. The
result as in Table 4 above revealed Cronbach alpha
is greater than 0.600, CR is 0.6 and above and AVE
greater than 0.5.
4.0 Discussion
As a research instrument, the questionnaire should
go through the validation process to ensure its
validity and reliability of the items involve for the
accurate and reliable findings of the study. The
validity and reliability of constructs of the study
which are knowledge, attitude, communication
channels and GST compliance were measured
using CFA with AMOS 20. Only the items of the
constructs with factor loading >0.60 remain in the
measurement model after the unidimensionality
process.Afterward, the AVE of the remain
constructs was calculated with the threshold above
>0.5 achieve the convergent validity (Fornell &
Larcker, 1981). Together with a factor loading of
all item ≥ 0.6 are considered high convergent
validity (Hair et al., 2010). Later, the construct
validity are measured with good fitness index on
the measurement model with RMSEA = 0.080, CFI
= 0.883, TLI = 0.876 and Chi-Square/df = 3.270.
Furthermore, the measurement model was run for
the discriminant validity to confirm no redundancy
of the constructs. The measurement of this study
achieves the discriminant validity where the
correlations between constructs are < 0.85 (Kline,
2011). In addition, Cronbach alpha for the
constructs reliability has achieved the threshold
0.600, met the CR with more than 0.60 and AVE
greater than the threshold at 0.5 above. All the
construct of the study achieved threshold’s validity
and reliability for further correlation measurement
of the research model.
5.0 Conclusion
The constructs of the study must undergo the
validity and reliability process to confirm the
unidimensional of its items in the measurement
model as the research model of the study. All
construct was correlated to each other in the form
of structural equation model (SEM) as a
measurement model to test factor loading of the
Page 761
Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in

items using CFA procedure with software AMOS

20. CFA procedures consist of unidimensionality,
convergent validity, construct validity and
discriminant validity. Next step is the process of
measurement the reliability of all items of the
constructs using Cronbach alpha using SPSS and
construct reliability (CR) and average variance
extracted (AVE) using AMOS.

It is important to ensure all the construct involve

achieved the validity and reliability before
proceeding to the next measurement of
relationship, moderation and mediation. Fail to
achieve the validity and reliability will lead to error
and inaccurate statistical results. Thus, the findings
of the study will totally insignificant.

6.0 References

Bove, L. ., Pervan, S. ., Beatty, S. ., & Shiu, E.

(2009). Service worker role in encouraging
customer organizational citizenship
behaviors. Business Research Journal, 62(7),
698–705.
Fornell, C., & Larcker, D. F. (1981). Evaluating
structural equation models with unobservable
variables and measurement error. Journal of
Marketing Research, 39–50.
Hair, J. F., Back, W. C., Babin, B. J., & Rolph, E.
A. (2010). Multivariate Data Analysis
(Seventh.). Prentice Hall.
Holmes-Smith, H., Coote, L., & Cunningham, E.
(2006). Structural equation modeling: from
the fundamentals to advanced topics.
Melbourne: School Research, Evaluation and
Measurement Services.
Hooper, D., Coughlan, J., & Mullen, M. (2008).
Structural Equation Modelling: Guidelines
for Determining Model Fit. Dublin Institute
of Technology Articles, 6(1), 53–60.
JC Nunnally, I. B. (1994). Psychometric Theory
(3rd ed.). New York: Mc Graw Hill Inc.
Jöreskog, K., & Sörbom, D. (1993). LISREL 8:
Structural equation modeling with the
SIMPLIS command language. Chicago:
Scientific Software International Inc.
Krejcie, R. V, & Morgan, D. (1970). Determining
sample size for research activities.
Educational and Psychological
Measurements Journal, 30(3), 607–610.
Rex B. Kline. (2011). Principles and practice of
structural equation modeling (3rd ed.). New
York: Guilford Press.
Zainudin Hj Awang. (2012a). Research
Methodology and data analysis (2nd ed.).
Kuala Lumpur: UiTM Press.
Zainudin Hj Awang. (2012b). Structural Equation
Modelling Using AMOS Graphic. Kuala
Lumpur: UiTM Press.

Imperial Journal of Interdisciplinary Research (IJIR) Page 762

Imperial Journal of Interdisciplinary Research (IJIR)
Vol-2, Issue-4, 2016
ISSN: 2454-1362, http://www.onlinejournal.in

Appendix A: Result of Validity and Reliability

1 2 3 4 5 6
Average
Cronbach 2nd order CFA Composite
Variance
Constructs/Dimension Items Alpha (Factor Reliability
Extracted
(> 0.6) loading ≥ 0.5) (CR) (> 0.6)
(AVE) (>0.5)
COMMUNICATION CHANNELS 0.881 0.742 0.934
IS1 0.962
IS2 0.693
IS3 0.86
IS4 0.867
IS5 0.901
KNOWLEDGE
Concept Knowledge (CK) 0.934 0.629 0.933
CK1 0.844
CK2 0.822
CK3 0.829
CK4 0.763
CK5 0.787
CK6 0.711
CK7 0.811

Legal Knowledge (LK) 0.945 0.869 0.943

LK2 0.826
LK3 0.905
LK4 0.934
LK6 0.877
LK7 0.839
Technology Knowledge (IT) 0.926 0.717 0.926
IT1 0.817
IT2 0.94
IT3 0.934
IT4 0.827
IT5 0.692
Technical Knowledge (TK) 0.979 0.815 0.975
TK3 0.814
TK7 0.909
TK8 0.913
TK9 0.920
TK10 0.927
TK11 0.945
TK12 0.939
TK13 0.923
ATTITUDE 0.667 0.676 0.859
Cognitive (C)
C3 0.604
C4 0.888

Affective (A) 0.900 0.653 0.832

A1 0.863
A2 0.871
A3 0.683
A5 0.801
GST COMPLIANCE (B) 0.850 0.687 0.898
B1 0.878
B2 0.792
B5 0.827
B6 0.817

Imperial Journal of Interdisciplinary Research (IJIR) Page 763