See discussions, stats, and author profiles for this publication at: [Link]

net/publication/359822943

Multi-layer hierarchical DEMATEL method: Analysis of soft factors in TQM

practice

Article in Journal of Quality · February 2022

DOI: 10.6220/joq.202202_29(1).0005

CITATIONS READS

4 264

1 author:

Jih-Kuang Chen
Zhaoqing University
26 PUBLICATIONS 339 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jih-Kuang Chen on 08 April 2022.

The user has requested enhancement of the downloaded file.

Journal of Quality Vol. 29, No. 1 (2022)
DOI:10.6220/joq.202202_29(1).0005

MULTI-LAYER HIERARCHICAL DEMATEL METHOD: ANALYSIS

OF SOFT FACTORS IN TQM PRACTICE
*
Jih-Kuang Chen
Economics & Management College, Zhaoqing University

ABSTRACT
This paper proposes an innovative multi-layer hierarchical decision making trial and
evaluation laboratory (DEMATEL) method to analyze the hierarchical structure of soft
factors in total quality management (TQM) practice. Ten soft factors of successful TQM
practice were identified from previous studies. Three green manufacturing enterprises
located in the Guangdong-Hong Kong-Macao Greater Bay Area that have implemented
TQM in China were analyzed. The finding is that “strategic quality planning”, “top
management commitment”, “leadership”, “reward and recognition”, and “employee
training” are the five most critical soft factors in successful TQM. The proposed method
contains a new and highly scientific determination of threshold value, which provides
a highly accurate and easily interpretable hierarchical structure reflecting inter-factor
relationships. The results are limited by the narrow range of research objects and the small
number of experts visited, however, they have notable reference significance.

Keywords: soft factor, total quality management (TQM), decision making trial and
evaluation laboratory (DEMATEL), multi-layer hierarchical decision making trial and
evaluation laboratory

1. INTRODUCTION

Total quality management (TQM) was first popularized in Japan as it enhanced the performance
and international competitiveness of Japanese enterprises. It is now widely used in enterprises around
the world. Many scholars have explored a wide array of tools and methods for TQM application
and have effectively defined its abstract management concepts. However, there is no ideal technique
available for concentrating valuable (and potentially very limited) resources owned by the enterprise in
areas that require prioritized management.
Critical success factors (CSFs), when appropriately defined, can guide enterprises in the process of
resource prioritization. Previous studies have revealed many different CSFs of TQM. CSFs may differ

Received: 10/2020; Revised: 12/2020; Accepted: 02/2021

*
Correspondence: Economics & Management College, Zhaoqing University; No. 1, Zhaoqing St., Duanzhou Dist.,
Zhaoqing 526061, Guangdong, China.
E-mail: jkchen1146@[Link]
 Journal of Quality Vol. 29, No. 1 (2022) 85

in number, object, or methodology in different contexts. Researchers tend to classify the importance
of CSFs as-deployed in TQM practices based on the frequency of text analysis and/or their place
in the literature. There is a lack of empirical research on CSFs. Few previous studies have centered
on those factors selected for TQM-based analysis of other variables (e.g., performance, innovation,
customer satisfaction) and they tend to be assessed similarly. Modern factor decomposition analysis
methods have indicated that the effects of individual TQM dimensions on other variables may differ
(Idris and Zairi, 2006; Prajogo and McDermott, 2005).
Wilkinson (1992) first pointed out that TQM has both “hard” and “soft” sides corresponding to
different factors. “Hard TQM” is generally linked to the tools and techniques necessary for quality
improvement, which encompasses a variety of production technologies including statistical process
control, organizational programming, just-in-time, and basic quality control tools (Thiagaragan et al.,
2001). Hard TQM analysis is more systematic and quantitative than soft TQM analysis, so previous
researchers have tended to focus on hard factors (Habtoor, 2016). Soft factors are fundamental,
however, and require time-consuming inputs and management. Previous scholars have also explored
the relationship between soft TQM and hard TQM. Powell (1995), for example, proposed that soft
TQM is more useful than hard TQM for the successful implementation of quality management
strategies. By contrast, Lewis et al. (2006) applied an analytic hierarchy process (AHP) to find that
hard TQM practices (e.g., system development and continuous improvement) are more popular in
modern enterprises. Rahman and Bullock (2005) found that both soft TQM and hard TQM have
a positive impact on organizational performance. Thiagarajan and Zairi (1997) suggested that hard
TQM can provide support for soft TQM. Despite these valuable contributions, there is a general lack
of knowledge regarding the hierarchy in the soft factor system itself and the importance of its various
components.
Decision making trial and evaluation laboratory (DEMATEL) is an algorithmic tool for the
factor analysis of complex systems. It was first established by Gabus and Fontela (1973), who used the
principles of graph theory and matrix theory to analyze the causal relationships in elements in certain
structures and to identify key elements in complex systems. DEMATEL has strong universality, its
mechanisms are straightforward, and its results are easily interpreted; it has thus garnered a great deal
of research attention and has been deployed in many research fields (Chen, 2018; Hidayanto et al.
2015; Khanam et al. 2016; Seleem et al., 2016; Sun, 2015). However, the DEMATEL method has
some persistent drawbacks. It is difficult to properly set threshold values. It is also not possible to
effectively stratify the elements into hierarchical structure. This study centers on the two problems.
The threshold determination methods used by previous scholars are generally too subjective or
lack a sound theoretical basis (as discussed in detail in the next section). The traditional DEMATEL
simply divides the factors into two levels, cause and effect, after only one operation. In actuality, the
relationships among factors may exist on multiple levels. The relationships between elements and
the order of priority are determined only according to the relation (cause degree) and prominence
(centrality) of each factor as obtained in a single calculation process. It is thus easy to confuse the
causal relationships between elements in the system, thus limiting the exertion of their functions.
Researchers have incorporated interactive structural modeling methods to define hierarchical
structures, but such integrated methods require overly complex calculations and can create other
problems. For example, the total-relation matrix of DEMATEL, when transformed into an interactive
structural modeling reachability matrix, has no transitivity check. This may render the stratification
result inaccurate (as also discussed in detail in the next section).
The goal of this study is to screen and identify the key success factors of possible soft factors in
TQM practices from past studies. An innovative multi-layer hierarchical DEMATEL method, which
86 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

includes a new and highly scientific threshold value determination, is established which can intuitively
reveal the hierarchical structure relationship of factors. The proposed model may be highly beneficial
in designing effective TQM practices.

2. LITERATURE REVIEW

2.1 Extensions of DEMATEL and controversy

DEMATEL has been widely used in various fields, but the application methods and algorithms
associated with it are still controversial. Researchers have developed various extended applications and
improved methods, as summarized below.

2.1.1 Cross fusion with other methods

Many scholars have integrated DEMATEL with AHP (L. Abdullah and Zulkifli, 2015), analytic
network process (ANP) (Büyüközkan and Güleryüz, 2016), or technique for order performance by
similarity to ideal solution (TOPSIS) (Baykasoğlu and Gölcük, 2017) in recent years. These techniques
have a certain theoretical value from the perspective of innovation and fusion, but there is still a lack
of scientific and rigorous argumentation supporting them; more importantly, they are impractical
in terms of real-world operation. W.-W. Wu (2008) similarly pointed out that the scales used by
DEMATEL and ANP do not match, which may give experts substantial psychological perception
differences in using the two methods to provide judgments. The rationality and effectiveness of the
cross-fusion of these methods still need further research.
Other scholars have integrated DEMATEL with methods such as failure mode and effect analysis
(FMEA) (Tsai et al., 2017), importance-performance analysis (IPA) (Ho et al., 2012), and balanced
scorecard (BSC) (Falatoonitoosi et al., 2012), which have extended its application fields and achieved
good results.

2.1.2 Expression of expert information

Because of the uncertainty and ambiguity of expert preference judgment, some researchers have
added language variables such as fuzzy numbers directly to DEMATEL algorithms. At present, there
are three kinds of expert judgment information expressions: fuzzy numbers (Jeng, 2015), gray theory
(Bai and Sarkis, 2013), and rough sets (Song and Cao, 2017). Scholars have attempted to pursue the
complexity of the model mechanism with complex methods. However, it is not yet known whether
the complex method can reflect the internal mechanism of the complex object scientifically. Methods
such as blurring or graying may be superfluous in terms of the expression of expert information.

2.1.3 Multi-theme issues

W.-W. Wu et al. (2011) pointed out that the traditional DEMATEL law places all the constituent
elements under the same topic and cannot manage two or multiple topics at the same time (e.g., the
pairs satisfaction/dissatisfaction, positive/negative emotions, or interests/risks). To enable the traditional
DEMATEL to deal with both positive and negative topics, a duo-theme DEMATEL method was first
proposed for key factor analysis by the Software as a Service (SaaS) organization. Research teams such as
Lee and Wu (2014) have since used this approach to analyze the driver factors of online luxury sales.

2.2 Factor of hierarchical structure

The conventional DEMATEL simply divides factors into two levels, cause and effect, without
considering that the influence relationships among factors may exist on multiple levels. The
 Journal of Quality Vol. 29, No. 1 (2022) 87

prominence and relation of each factor are obtained under only one calculation step to determine the
priority of improvement, which easily confounds the system and limits the accuracy of its analysis.
Michnik (2013) proposed a DEMATEL deformation method, weighted influence non-linear gauge
system, which not only associates the scheme with the criterion but also accounts for the self-
dependence of the criterion itself when stratifying it. However, this method has no universal model
and its criteria-stratification mechanism is not explicitly clear.
Many scholars have used interpretive structural modeling to analyze the hierarchical structure
of factors after DEMATEL analysis (e.g., Kumar and Dixit, 2018). However, Zhou et al. (2006)
proposed integrating them into the DEMATEL-ISM method. Others have considered the total-
relation matrix of DEMATEL to contain more information than the reachability matrix of ISM; the
reachability matrix is more difficult to calculate. The hierarchical structure of the complex system can
be divided by the total-relation matrix of DEMATEL and then transformed into an ISM reachability
matrix (e.g., Shakeri and Khalilzadeh, 2020; Zhang and Luo, 2017). This method, however, involves
simply adding a total-relation matrix T to identify matrix I to become total influence matrix H. After
setting the threshold and comparing the information, the system is directly transformed into the
reachability matrix. Although this makes the reachability matrix simple to obtain and easy to operate,
it is doubtful whether it has the proper transitive characteristics. The lack of scientific standard for
setting the threshold value is also problematic.

2.3 Threshold value determination

There are three primary approaches available for threshold determination. The first is to set the
threshold based on recommendations by decision-makers or to determine it jointly with multiple
experts. This generally emphasizes the experience of decision-makers or the joint investigating
parties (Hsu, 2012; Wang, 2012) and does not explain the specific process of how determinations
are made. The second approach centers on the statistical mean (Cebi, 2013; Horng et al. 2013) or
median (Bai and Sarkis, 2013) of the elements in a total-relation matrix. The mean is easily affected
by extreme values and lacks a theoretical basis. The median is not affected by extreme values, but still
lacks sensitivity to the representativeness of skewed data. These statistical approaches also artificially
delete half of the factors affecting the relationships; the cognition of causality in system factors is too
subjective and arbitrary. The third method is based on the statistical distribution, where the threshold
value is calculated using µ + σ or µ + 1.5σ (Rajesh and Ravi, 2015; Tzeng and Huang, 2012); µ
and σ are the mean value and standard deviation (SD) of the elements of the total-relation matrix,
respectively. In essence, the total-relation matrix data are not necessarily in a normal distribution, so
this may not be consistent with the actual situation at hand.
In addition, L. Wu et al. (2013) assert that a lower threshold can be set as the average value of the
upper quartile and the median while the upper threshold is the average value of the lower quartile and
maximum number. However, the choice still depends on the subjective judgment of decision-makers,
so excessive subjectivity is still problematic. Li et al. (2012) set a judgment equation for relation and
prominence values, eliminating the elements that meet the above judgment equation in the system
while deleting the rows and columns containing the unimportant elements in the total-relation matrix
and renumbering them to obtain a new direct-relation matrix. However, their theoretical basis and
derivation are unreliable.

2.4 Soft factors in TQM

Many researchers have explored soft TQM factors. We screened out the more commonly used
factors from the literature and organized them as shown in Table 1 to facilitate subsequent analysis.
88 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

Table 1. Soft factors in TQM

Authors (Year) Soft factors
Fotopoulos and Psomas Top management commitment, strategic quality planning, employee involvement,
(2009) supplier management, customer focus, process orientation, continuous improvement,
facts-based decision making, human resource development
Garcia (2011) Employee involvement, employee commitment, customer focus, continues improvement
M. M. B. Abdullah and Management commitment, customer focus, employee involvement, training and
Tarí (2012) education, reward and recognition, supplier relationship
Abdallah (2013) Customer focus, training, top management, leadership, workforce management, supplier
relationship
Al-Khalili and Subari Leadership, vision and plan statement, evaluation, employee participation, recognition
(2013) and reward, education and training, customer focus
Psomas et al. (2014) Continuous improvement, top management commitment, customer focus, human
resource development, fact-based decision making, strategic quality planning, process
focus, employee involvement, supplier involvement
Imeri et al. (2014) Continues improvement, strategies quality planning, continuous training, teamwork,
empowerment, customer satisfaction, information and analysis, supplier management,
top management commitment and support, democratic management style, culture
change
Saleh and Sweis (2017) Customer focus, education and training, top management leadership, supplier
relationship
Modgil and Sharma Top management support, customer focus, training and development, employee
(2017) empowerment
Saleh et al. (2018) Customer focus, education and training, top management leadership, supplier
relationship
Source: This study.

3. METHOD

The multi-layer hierarchical DEMATEL method is described below. The first half is similar to the
traditional DEAMEL method, where influencing factors of the system are determined and the degree
of direct influence among factors is scored by experts. The scale can be set as 0–3 quadrilateral or 0–4
quadrilateral from no impact to great impact. The evaluation results are placed into the initial direct-
relation matrix X:

0 x12 ⋯ x1n
x21 0 ⋯ x2n
X= (1)
⋮ ⋮ 0 ⋮
xn1 xn2 ⋯ 0

where xij is the direct influence degree of factor i on factor j. When i = j, xij = 0.
The initial direct-relation matrix X is then normalized by the following equation:

1
N= × X(2)
max ∑ n xij
0≤i≤n j=1
 Journal of Quality Vol. 29, No. 1 (2022) 89

and the total-relation matrix T is built with the following equation:

-1
T = N × (I – N) (3)

where I is the identity matrix and -1 is the inverse matrix.

The threshold value (λ) is then calculated and compared to obtain the final total-relation matrix Ť.
To minimize the cost, time, and complexity of the experiment, it is very important to allocate factors
beforehand and to analyze the effect without excluding any important factors afterward. Box and
Meyer’s (1986) method has been adopted by many scholars for this purpose. Lenth (1989) proposed an
effect-sparsity assumption based on work by Box and Meyer, where the main effect and some low-order
interactions have specific effects on the response variables while many high-order interactions can be
ignored. This is consistent with the DEMATEL threshold. The effect-sparse hypothesis concept was used
in this study to design the determination method of threshold. Its step-wise implementation is as follows.
(1) Bottom value (S0) calculation:

S0 = 1.1 × Mt(4)

where Mt is the median of all element values in the total-relation matrix.

(2) Pseudo-standard error (PSE) calculation:

PSE = 1.1 × Ms(5)

where Ms is the median of the element value ≥ S0 in the total-relation matrix T.

(3) Margin of error (ME) calculation:

PSE
ME = (6)
t1 – ∝; m3

where t1 – ∝; m3 is the t distribution value under the confidence level of 1 – ∝ with degrees of freedom
m . When m is not an integer, the nearest integer value is taken instead.
3 3
The ME is regarded as the threshold value (λ). The elements in the total-relation matrix T are
compared with the threshold value (λ) according to the following equation. The final total-relation
matrix Ť can then be obtained after conversion:

tij = {tij | tij ≥ λ}

(7)
{0 | tij < λ}

The influence degree affected degree of each factor are able to calculated based on the final total-
relation matrix. The influence degree Di indicates the comprehensive influence of various factors on
other factors in the system; the affected degree Rj indicates the degree to which each factor is affected
by other factors.
90 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

Influence degree: Di = ∑ j = 1tij

n

(8)
Affected degree: Rj = ∑ i = 1tij
n

At last the prominence and relation of each factor are able to obtain. The prominence is defined as
Di + Rj, which indicates that the importance of factors in the system. The relation is defined as Di – Rj.
When the relation value is positive, the factor belongs to the cause factor category; when negative, it
belongs to the effect factor category.

Prominence: Pi = {Di + Rj | i = j}
(9)
Relation: REi = {Di – Rj | i = j}

The factor with relation (REi) > 0 (cause factor) is the first-layer factor, which has the strongest
driving effect on other factors and should be given priority attention. The rows and columns of
these factors are deleted from Ť, then the prominence and relation of residual factors are calculated.
The residual factor REi > 0 is the second-layer factor. This process is repeated until there is no
factor belonging to the cause factor set (or only one effect factor remains). The calculations are then
terminated and the analysis of the hierarchical structure factor is complete.

4. ANALYSIS

Three green manufacturing enterprises located in the Guangdong-Hong Kong-Macao Greater

Bay Area that have implemented TQM were analyzed in this study as research objects. A total of 10
soft factors were screened out from those implemented in the enterprises. Each was confirmed to be
suitable for China’s national conditions and industrial characteristics based on previously published
research (Table 1), as shown in Table 2.
A total of 13 core personnel, including five experts from academia in the field of quality
management and eight responsible managers and major executors from three enterprises engaged
in TQM, were invited to assign 4, 3, 2, 1, or 0 scores to the 10 soft factors corresponding to great

Table 2. Soft factors assessed

Code Soft factors
S1 Strategic quality planning
S2 Human resource development
S3 Customer focus
S4 Supplier relationship
S5 Top management commitment
S6 Leadership
S7 Reward and recognition
S8 Employee empowerment
S9 Employee involvement
S10 Employee training
Source: This study.
 Journal of Quality Vol. 29, No. 1 (2022) 91

impact, strong impact, middle impact, weak impact, and no impact, respectively. The initial direct-
relation matrix X was obtained after all the evaluation values are averaged to represent the consensus
of these experts (Table 3).
After normalization of Equation (2) and calculation of Equation (3), the total-relation matrix T
was obtained as shown in Table 4.
The threshold value was calculated according to Equations (4)–(6). The obtained S0 is 0.093 and the
PSE is 0.140. The t0.95,33 value was determined by the look-up table from the statistics textbook as 1.692,
so ME was calculated to be 0.083. The elements of the total-relation matrix T were converted to a final
total-relation matrix Ť after comparison with the threshold according to Equation (7) (Table 5).
According to Equations (8) and (9), the influence degree, affected degree, prominence, and
relation of each factor were calculated as shown in Table 6.
According to the data in Table 6, the cause effect among factors was drawn as shown in Figure 1.
Next, we performed a hierarchy structure analysis. We found that S1, S5, S6, S7, and S10 are cause
factors while S2, S3, S4, S8, and S9 are effect factors. The 5 cause factors can be regarded as the first-

Table 3. Initial direct-relation matrix X

Code S1 S2 S3 S4 S5 S6 S7 S8 S9 S10
S1 0.0 2.1 3.7 3.8 3.1 2.2 1.8 2.3 3.2 1.9
S2 2.2 0.0 1.1 1.3 0.0 0.0 0.0 0.0 1.7 1.8
S3 2.1 1.7 0.0 2.3 0.0 2.8 0.0 0.0 0.0 0.8
S4 1.9 1.8 2.2 0.0 0.0 0.0 0.0 0.0 0.0 1.2
S5 3.8 3.7 3.9 2.8 0.0 3.2 2.9 1.9 2.8 0.0
S6 2.1 0.0 0.0 0.0 2.3 0.0 1.8 1.7 2.2 0.0
S7 0.0 1.8 1.7 2.2 0.0 0.0 0.0 2.1 3.7 0.0
S8 0.0 0.0 2.1 1.8 0.0 0.0 0.0 0.0 1.9 1.1
S9 2.1 0.0 2.3 1.8 0.0 0.0 0.0 0.0 0.0 0.0
S10 2.2 2.1 0.8 1.2 0.0 0.0 0.0 0.0 1.8 0.0
Source: This study.

Table 4. Total-relation matrix

Code S1 S2 S3 S4 S5 S6 S7 S8 S9 S10
S1 0.122 0.172 0.264 0.262 0.153 0.148 0.109 0.134 0.221 0.125
S2 0.127 0.035 0.090 0.097 0.018 0.024 0.013 0.020 0.101 0.092
S3 0.132 0.103 0.050 0.135 0.029 0.133 0.022 0.025 0.048 0.059
S4 0.112 0.102 0.123 0.043 0.016 0.026 0.012 0.014 0.033 0.071
S5 0.261 0.229 0.277 0.236 0.050 0.188 0.154 0.130 0.220 0.062
S6 0.134 0.047 0.075 0.070 0.112 0.034 0.097 0.099 0.148 0.024
S7 0.048 0.097 0.117 0.133 0.008 0.018 0.006 0.091 0.173 0.025
S8 0.034 0.023 0.110 0.099 0.006 0.016 0.004 0.005 0.089 0.057
S9 0.115 0.031 0.128 0.110 0.017 0.027 0.012 0.015 0.025 0.021
S10 0.127 0.113 0.079 0.094 0.018 0.022 0.013 0.016 0.105 0.026
Source: This study.
92 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

Table 5. Final total-relation matrix

Code S1 S2 S3 S4 S5 S6 S7 S8 S9 S10
S1 0.122 0.172 0.264 0.262 0.153 0.148 0.109 0.134 0.221 0.125
S2 0.127 0.000 0.000 0.097 0.000 0.000 0.000 0.000 0.101 0.092
S3 0.132 0.103 0.000 0.135 0.000 0.133 0.000 0.000 0.000 0.000
S4 0.112 0.102 0.123 0.000 0.000 0.000 0.000 0.000 0.000 0.000
S5 0.261 0.229 0.277 0.236 0.000 0.188 0.154 0.130 0.220 0.000
S6 0.134 0.000 0.000 0.000 0.112 0.000 0.097 0.099 0.148 0.000
S7 0.000 0.097 0.117 0.133 0.000 0.000 0.000 0.091 0.173 0.000
S8 0.000 0.000 0.110 0.000 0.000 0.000 0.000 0.000 0.089 0.000
S9 0.115 0.000 0.128 0.110 0.000 0.000 0.000 0.000 0.000 0.000
S10 0.127 0.113 0.000 0.094 0.000 0.000 0.000 0.000 0.105 0.000
Source: This study.

Table 6. Prominence, relation, influence degree, affected degree

Code Di Ri Di + Ri Di – Ri
S1 1.709 1.212 2.921 0.497
S2 0.523 0.953 1.476 -0.430
S3 0.738 1.223 1.960 -0.485
S4 0.509 1.137 1.646 -0.628
S5 1.758 0.375 2.133 1.383
S6 0.839 0.636 1.475 0.203
S7 0.714 0.442 1.156 0.272
S8 0.344 0.544 0.888 -0.201
S9 0.499 1.163 1.662 -0.663
S10 0.612 0.560 1.172 0.052
Source: This study.

Figure 1. Prominence-causal diagram

Source: This study.
 Journal of Quality Vol. 29, No. 1 (2022) 93

layer factors; their rows and columns were deleted from the final total-relation matrix Ť to obtain the
matrix shown in Table 7.
After repeated calculations, we found that S8 and S9 are the second-layer factors. The rows and
columns of S2, S3 and S4 were retained to repeat the calculation again and to determine that S3 is a
third-layer factor. Then S2 and S4 were retained to repeat the calculation again until only S4 was an
effect factor, at which point the calculation was stopped. The whole hierarchy structure and causes/
effects among factors were obtained as shown in Figure 2.

5. RESULTS

According to our results, enterprises should strengthen management from the factors in first
layer and follow the order of transitivity of cause in order to achieve maximum effects. We found
that “strategic quality planning”, “top management commitment”, “leadership”, “reward and
recognition”, and “employee training” are the five key soft factors for successful implementation of
TQM as per their maximal driving forces. The second-most important factors appear to be “employee
empowerment” and “employee involvement”. “Customer focus”, “human resource development”,
and “supplier relationship” are the latter-layer factors in the hierarchy. It is conceivable that the key
to successful TQM implementation lies in the determination, commitment, and leadership of senior

Table 7. 2nd final total-relation matrix

Code S2 S3 S4 S8 S9
S2 0.000 0.000 0.097 0.000 0.101
S3 0.103 0.000 0.135 0.000 0.000
S4 0.102 0.123 0.000 0.000 0.000
S8 0.000 0.110 0.000 0.000 0.089
S9 0.000 0.128 0.110 0.000 0.000
Source: This study.

Figure 2. Hierarchy structure

Source: This study.
94 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

executives as well as superior quality planning, complete staff training, and an appropriate incentive
system. The degree of employee authorization, employee participation, strong human resource
development, and an effective management system are important as well, though to lesser extent.
Although the three external related factors of “customer focus”, “human resource development”, and
“supplier relationship” were not as important in this case, the relative weakness of their influence may
be due to the characteristics of the industry investigated here or to the fact that the enterprises already
have strong external relationships. In short, these results should be treated with caution and may not
be universal.

6. CONCLUSIONS

The DEMATEL method is widely used because it reveals causal correlations and key elements in
the structural relationship of elements in complex systems. However, the threshold determination is
too subjective and may lack a theoretical basis, which can degrade the hierarchical relationships among
factors. Although it can be used separately or in combination with ISM to define these relationships,
the calculation process is overly complex and not widely applicable. The effect-sparse principle
concept was introduced in this study to establish a novel threshold value determination method. The
hierarchical structure of factors was defined in a case study to make the DEMATEL more scientific
while expanding its application range.
Soft factors that are specifically compatible with China’s national conditions and industrial
characteristics were screened out in this study from the soft factors used in previous studies. Three
green manufacturing enterprises located in the Guangdong-Hong Kong-Macao Greater Bay Area
that have implemented TQM were selected as cases to validate the rationality and effectiveness of the
proposed method. “Strategic quality planning”, “top management commitment”, “leadership”, “reward
and recognition”, and “employee training” emerged as the five most critical soft factors for successful
TQM.
The main purpose of this study was to verify the proposed method, which includes a new and
highly scientific threshold value determination. The proposed method can intuitively reveal the
hierarchical structure relationship among factors. The results are limited by the narrow range of
research objects and the small number of experts who participated, however, they still have noteworthy
reference significance.

ACKNOWLEDGMENT

This work was supported by Zhaoqing University Research Fund (202002).

REFERENCES
Abdallah, A. B., 2013, The influence of “soft” and “hard” total quality management (TQM) practices on total
productive maintenance (TPM) in Jordanian manufacturing companies, International Journal of Business
and Management, 8(21), 1-13. doi:10.5539/ijbm.v8n21p1
Abdullah, L. and Zulkifli, N., 2015, Integration of fuzzy AHP and interval type-2 fuzzy DEMATEL: an
application to human resource management, Expert Systems With Applications, 42(9), 4397-4409.
doi:10.1016/[Link].2015.01.021
Abdullah, M. M. B. and Tarí, J. J., 2012, The influence of soft and hard quality management practices on
 Journal of Quality Vol. 29, No. 1 (2022) 95

performance, Asia Pacific Management Review, 17(2), 177-193. doi:10.6126/APMR.2012.17.2.04

Al-Khalili, A. and Subari, K., 2013, Understanding the linkage between soft and hard total quality management:
evidence from Malaysian manufacturing industries, Jordan Journal of Mechanical and Industrial
Engineering, 7(1), 57-65.
Bai, C. and Sarkis, J., 2013, A grey-based DEMATEL model for evaluating business process management
critical success factors, International Journal of Production Economics, 146(1), 281-292. doi:10.1016/
[Link].2013.07.011
Baykasoğlu, A. and Gölcük, İ., 2017, Development of an interval type-2 fuzzy sets based hierarchical MADM
model by combining DEMATEL and TOPSIS, Expert Systems With Applications, 70, 37-51. doi:10.1016/
[Link].2016.11.001
Box, G. E. P. and Meyer, R. D., 1986, An analysis for unreplicated fractional factorials, Technometrics, 28(1),
11-18. doi:10.1080/00401706.1986.10488093
Büyüközkan, G. and Güleryüz, S., 2016, An integrated DEMATEL-ANP approach for renewable energy
resources selection in Turkey, International Journal of Production Economics, 182, 435-448. doi:10.1016/
[Link].2016.09.015
Cebi, S., 2013, Determining importance degrees of website design parameters based on interactions and types of
websites, Decision Support Systems, 54(2), 1030-1043. doi:10.1016/[Link].2012.10.036
Chen, J. K., 2018, The influence of behavioural intention on third-party e-commerce payment, South African
Journal of Economic and Management Sciences, 21(1), a2157. doi:10.4102/sajems.v21i1.2157
Falatoonitoosi, E., Leman, Z. and Sorooshian, S., 2012, Casual strategy mapping using integrated BSC and
MCDM-DEMATEL, Journal of American Science, 8(5), 424-428. doi:10.7537/marsjas080512.41
Fotopoulos, C. B. and Psomas, E. L., 2009, The impact of “soft” and “hard” TQM elements on quality
management results, International Journal of Quality & Reliability Management, 26(2), 150-163.
doi:10.1108/02656710910928798
Gabus, A. and Fontela, E., 1973, Perceptions of the world problematique: communication procedure,
communicating with those bearing collective responsibility, DEMATEL Report (No. 1), Battelle Geneva
Research Centre, Geneva, Switzerland.
Garcia, I., 2011, Knowledge management, soft TQM and hard TQM, and organizational performance,
International Forum, 14(1), 70-85.
Habtoor, N., 2016, Influence of human factors on organisational performance: quality improvement practices as
a mediator variable, International Journal of Productivity and Performance Management, 65(4), 460-484.
doi:10.1108/IJPPM-02-2014-0016
Hidayanto, A. N., Abednego, N., Aminah, S. and Sucahyo, Y. G., 2015, Analysis of cloud adoption determinants
by using BOCR analysis and DEMATEL, International Journal of Business Information Systems, 18(2),
221-248. doi:10.1504/IJBIS.2015.067265
Ho, L.-H., Feng, S.-Y., Lee, Y.-C. and Yen, T.-M., 2012, Using modified IPA to evaluate supplier’s performance:
multiple regression analysis and DEMATEL approach, Expert Systems With Applications, 39(8), 7102-
7109. doi:10.1016/[Link].2012.01.034
Horng, J.-S., Liu, C.-H., Chou, S.-F. and Tsai, C.-Y., 2013, Creativity as a critical criterion for future restaurant
space design: developing a novel model with DEMATEL application, International Journal of Hospitality
Management, 33, 96-105. doi:10.1016/[Link].2012.06.007
Hsu, C.-C., 2012, Evaluation criteria for blog design and analysis of causal relationships using factor analysis and
DEMATEL, Expert Systems With Applications, 39(1), 187-193. doi:10.1016/[Link].2011.07.006
Idris, M. A. and Zairi, M., 2006, Sustaining TQM: a synthesis of literature and proposed research framework,
Total Quality Management & Business Excellence, 17(9), 1245-1260. doi:10.1080/14783360600750535
96 Multi-Layer Hierarchical DEMATEL Method: Analysis of Soft Factors in TQM Practice

Imeri, S., Kekäle, T., Takala, J. and Liu, Y., 2014, Understanding the impact of “hard” and “soft” elements
of TQM in South-East European firms, Management and Production Engineering Review, 5(3), 9-13.
doi:10.2478/mper-2014-0022
Jeng, D. J.-F., 2015, Generating a causal model of supply chain collaboration using the fuzzy DEMATEL
technique, Computers & Industrial Engineering, 87, 283-295. doi:10.1016/[Link].2015.05.007
Khanam, S., Siddiqui, J. and Talib, F., 2016, A DEMATEL approach for prioritizing the TQM enablers and
IT resources in the Indian ICT industry, International Journal of Applied Management Sciences and
Engineering, 3(1), 11-29. doi:10.4018/IJAMSE.2016010102
Kumar, A. and Dixit, G., 2018, An analysis of barriers affecting the implementation of e-waste management
practices in India: a novel ISM-DEMATEL approach, Sustainable Production and Consumption, 14, 36-
52. doi:10.1016/[Link].2018.01.002
Lee, Y.-C. and Wu, C.-H., 2014, A duo-theme DEMATEL approach for exploring the driving factors of online
luxury goods sales: e-retailers’ perceptions, Journal of Information and Optimization Sciences, 35(2), 177-
202. doi:10.1080/02522667.2013.876773
Lenth, R. V., 1989, Quick and easy analysis of unreplicated factorials, Technometrics, 31(4), 469-473. doi:10.1
080/00401706.1989.10488595
Lewis, W. G., Pun, K. F. and Lalla, T. R. M., 2006, Exploring soft versus hard factors for TQM implementation
in small and medium-sized enterprises, International Journal of Productivity and Performance
Management, 55(7), 539-554. doi:10.1108/17410400610702142
Li, M.-h., Xia, J.-b. and Chen, C.-q., 2012, Network pedigree model with reachable effect factor based on
DEMATEL & ISM, Journal of Jilin University (Engineering and Technology Edition), 42(3), 782-788.
doi:10.13229/[Link].jdxbgxb2012.03.023
Michnik, J., 2013, Weighted influence non-linear gauge system (WINGS)—an analysis method for the systems
of interrelated components, European Journal of Operational Research, 228(3), 536-544. doi:10.1016/
[Link].2013.02.007
Modgil, S. and Sharma, S., 2017, Impact of hard and soft TQM on supply chain performance: empirical
investigation of pharmaceutical industry, International Journal of Productivity and Quality Management,
20(4), 513-533. doi:10.1504/IJPQM.2017.082834
Powell, T. C., 1995, Total quality management as competitive advantage: a review and empirical study, Strategic
Management Journal, 16(1), 15-37. doi:10.1002/smj.4250160105
Prajogo, D. I. and McDermott, C. M., 2005, The relationship between total quality management practices and
organizational culture, International Journal of Operations & Production Management, 25(11), 1101-
1122. doi:10.1108/01443570510626916
Psomas, E., Vouzas, F. and Kafetzopoulos, D., 2014, Quality management benefits through the “soft” and “hard”
aspect of TQM in food companies, The TQM Journal, 26(5), 431-444. doi:10.1108/TQM-02-2013-0017
Rahman, S.-u. and Bullock, P., 2005, Soft TQM, hard TQM, and organisational performance relationships: an
empirical investigation, Omega, 33(1), 73-83. doi:10.1016/[Link].2004.03.008
Rajesh, R. and Ravi, V., 2015, Modeling enablers of supply chain risk mitigation in electronic supply
chains: a grey-DEMATEL approach, Computers & Industrial Engineering, 87, 126-139. doi:10.1016/
[Link].2015.04.028
Saleh, R. A. and Sweis, R. J., 2017, The relationships between soft/hard total quality management practices and
operational performance in Jordanian manufacturing organisations, International Journal of Management
Concepts and Philosophy, 10(4), 345-377. doi:10.1504/IJMCP.2017.087261
Saleh, R. A., Sweis, R. J., Saleh, F. I. M., Sarea, A. M., Eldin, I. M. S. and Obeid, D. N., 2018, Linking soft
and hard total quality management practices: evidence from Jordan, International Journal of Business
 Journal of Quality Vol. 29, No. 1 (2022) 97

Excellence, 14(1), 49-86. doi:10.1504/IJBEX.2018.088315

Seleem, S. N., Attia, E.-A. and El-Assal, A., 2016, Managing performance improvement initiatives using
DEMATEL method with application case study, Production Planning & Control, 27(7–8), 637-649. doi:10
.1080/09537287.2016.1165301
Shakeri, H. and Khalilzadeh. M., 2020, Analysis of factors affecting project communications with a hybrid
DEMATEL-ISM approach (a case study in Iran), Heliyon, 6(8), e04430. doi:10.1016/[Link].2020.e04430
Song, W. and Cao, J., 2017, A rough DEMATEL-based approach for evaluating interaction between
requirements of product-service system, Computers & Industrial Engineering, 110, 353-363. doi:10.1016/
[Link].2017.06.020
Sun, C.-C., 2015, Identifying critical success factors in EDA industry using DEMATEL method, International
Journal of Computational Intelligence Systems, 8(2), 208-218. doi:10.1080/18756891.2015.1001945
Thiagaragan, T., Zairi, M. and Dale, B., 2001, A proposed model of TQM implementation based on an
empirical study of Malaysian industry, International Journal of Quality & Reliability Management, 18(3),
289-306. doi:10.1108/02656710110383539
Thiagarajan, T. and Zairi, M., 1997, A review of total quality management in practice: understanding the
fundamentals through examples of best practice applications—part I, The TQM Magazine, 9(4), 270-286.
doi:10.1108/09544789710181899
Tsai, S.-B., Zhou, J., Gao, Y., Wang, J., Li, G., Zheng, Y., et al., 2017, Combining FMEA with DEMATEL
models to solve production process problems, PLoS ONE, 12(8), e0183634. doi:10.1371/journal.
pone.0183634
Tzeng, G.-H. and Huang, C.-Y., 2012, Combined DEMATEL technique with hybrid MCDM methods for
creating the aspired intelligent global manufacturing & logistics systems, Annals of Operations Research,
197(1), 159-190. doi:10.1007/s10479-010-0829-4
Wang, T.-C., 2012, The interactive trade decision-making research: an application case of novel hybrid MCDM
model, Economic Modelling, 29(3), 926-935. doi:10.1016/[Link].2012.02.001
Wilkinson, A., 1992, The other side of quality: “soft” issues and the human resource dimension, Total Quality
Management, 3(3), 323-330. doi:10.1080/09544129200000038
Wu, L., Zhang, Q., Shan, L. and Chen, Z., 2013, Identifying critical factors influencing the use of additives by
food enterprises in China, Food Control, 31(2), 425-432. doi:10.1016/[Link].2012.10.028
Wu, W.-W., 2008, Choosing knowledge management strategies by using a combined ANP and DEMATEL
approach, Expert Systems With Applications, 35(3), 828-835. doi:10.1016/[Link].2007.07.025
Wu, W.-W., Lan, L. W. and Lee, Y.-T., 2011, Exploring decisive factors affecting an organization’s SaaS
adoption: a case study, International Journal of Information Management, 31(6), 556-563. doi:10.1016/
[Link].2011.02.007
Zhang, P.-k. and Luo, F., 2017, Influencing factors of runway incursion risk and their interaction mechanism
based on DEMATEL-ISM, Tehnički Vjesnik, 24(6), 1853-1861. doi:10.17559/TV-20170928105858
Zhou, D.-q., Zhang, L. and Li, H.-w., 2006, A study of the system’s hierarchical structure through integration
of DEMATEL and ISM, in Proceedings of the 2006 International Conference on Machine Learning and
Cybernetics, Dalian, China, 1449-1453. doi:10.1109/ICMLC.2006.258757
 品質學報 29卷1期﹐84-98頁 (2022年2月)
Journal of Quality Vol. 29, No. 1 (2022) DOI:10.6220/joq.202202_29(1).0005

多層級DEMATEL法分析TQM實踐的軟性因素
陳日光*
肇慶學院經濟與管理學院

(multi-layer hierarchical
decision making trial and evaluation laboratory)

收件日：109/10；修改日：109/12；接受日：110/02
*
聯絡作者：肇慶學院經濟與管理學院，526061廣東省肇慶市端州區肇慶大道1號。
E-mail: jkchen1146@[Link]

View publication stats