Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Expert Systems with Applications 38 (2011) 9633–9647

Contents lists available at ScienceDirect

Expert Systems with Applications


journal homepage: www.elsevier.com/locate/eswa

Selection effective management tools on setting European Foundation for Quality


Management (EFQM) model by a quality function deployment (QFD) approach
Same Yousefie a,⇑, Mahmood Mohammadi b, Jalal Haghighat Monfared b
a
Islamic Azad University, Central Branch of Tehran, Faculty of Management, Tehran, Iran
b
Islamic Azad University, Central Branch of Tehran, Department of Industrial Management, Tehran, Iran

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

Keywords: EFQM Excellence model literature indicates that using the management tools that are relevant to the
Management tools organization’s needs has become a strategic issue for companies in today’s competitive environment.
European Foundation for Quality By choosing and applying the best management tools among too many management tools, companies
Management (EFQM) can improve their performances and then increase customer satisfaction and gain market shares. The
Quality function deployment (QFD)
aim of this research is to propose an original approach for the management tools selection based on
Fuzzy analytic hierarchy process (FAHP)
Entropy method
the quality function deployment (QFD) approach, a methodology which has been successfully adopted
Symmetrical triangular fuzzy numbers in new products development. Specifically, the research addresses the issue of how to deploy the house
(STFN) of quality (HOQ) to effectively and efficiently improve management tools selection processes and thus
company satisfaction about its excellence achievement. Fuzzy logic is also adopted to deal with the
vagueness nature of the qualitative linguistic judgments required in the proposed HOQ. The model of this
research has been tested by means of a real case application, which refers to an Iranian company oper-
ating in the automotive industry in this case the mixture of 15 categories of management tools with five
EFQM enabler criteria has been characterized by using of the research model. And also the test of the
hypothesis of this research has been done by using spearman correlation coefficient.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction sense, the logic behind the model is that by improving how
the organization operates – the ‘‘management tools’’ – there will
Since its creation in 1991, the main purpose of the European be an inevitable improvement in the ‘‘results’’. It means that
Quality Award (EQA) has been to recognize the organizational management tools are the main part of EFQM excellence model
excellence in European companies. The EFQM excellence model and the fact is, achieving excellence is depend on using the best
is the framework behind this award and it has clearly become management tools that are adopt with the organizations needs
the most commonly applied model in Europe for total quality man- for excellence and have a high level of performance in using of
agement (TQM) (Westlund, 2001). Although in organizational those management tools.
practice the use of the EFQM excellence model is practically Previous studies have also emphasized the need to using man-
unquestioned, some uncertainties still remain at the academic agement tools in developing the excellence in organization (Leon-
level, mainly related to its implementation and assessment of ard & Aadam, 2002). As shown in Fig. 1, Assume EFQM model as a
accomplishing of its criteria in the organizations. pyramid that the fundamental concept and the criteria of EFQM
The EFQM model constitutes a non-prescriptive framework model are in the first and the second level of this pyramid, and also
that assumes there are different approaches to achieving sustain- management tools as the third level of this pyramid (European
able excellence (Ghobadian & Woo, 1996) that derives in the exis- Foundation for Quality Management, 1999).
tence of multiple interpretations around its implementation. So, identifying and using best management tools according to
However, it is made up of certain notions and ideas about the organization’s needs in setting EFQM model and achieving results
general relationships between its elements that have still not in organizations are so important.
been demonstrated empirically (Bou-Llusar et al., 2005). In this The purpose of the current research is to represent a quality
function deployment (QFD) model with both crisp and fuzzy
approaches for the linkages between the EFQM criteria and man-
agement tools, and use of this model for identifying and prioritiz-
⇑ Corresponding author. Address: P.O. Box: 14515/775, Eskandari St., Tehran, ing management tools that are so effective and match with
Iran. Tel.: +98 9122490306. organizational needs for excellence. This aim can be expressed
E-mail addresses: s.yousefie@iauctb.ac.ir, sam.yousefie@gmail.com (S. Yousefie). through the following question.

0957-4174/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eswa.2011.01.166
9634 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

3. Research methodology

3.1. Conceptual model of the research


Fundamental concept

Criteria The conceptual model of this research is an extension of a qual-


ity function deployment (QFD) model in EFQM excellence model
Management tools and management tools selection domain. The model of this re-
search can considered as Fig. 2. It should notice that, it is a first
Fig. 1. Three level of EFQM model. time that a QFD model has been used in EFQM excellence model
and management tools selection filed.
As shown in Fig. 2, it can consider that organization is as a cus-
tomer. But in this model a customer looking for the excellence, and
How organizations can identify, prioritize and select manage-
the needs or criteria of this customer are EFQM enabler criteria. On
ment tools that are effective on setting EFQM model in
the other hand, the technical attributes for satisfying this customer
organizations?
is the effective management tools on setting EFQM model. So, to
The answers to this question would facilitate the understanding
achieving excellence, organization has some needs or organiza-
of what are the most important management tools for an organiza-
tional needs for excellence (ONE) and for satisfying these criteria,
tion that could the improve results, and thus would give some
organization determine some technical attributes for excellence
ideas about the areas where the organizations should concentrate
(TAE). The voices of organization (VOO) about the excellence
their efforts in the management systems, thus allowing a better
should be hear and answered. So now these criteria and technical
implementation of the EFQM model.
attributes should be prioritizing according to the relationship be-
tween themselves. Finally, by use of these prioritized excellence
2. Research questions and hypothesizes
criteria and technical attributes for excellence, the organization
can more concentrate on its core competences and has a basis
The preceding section highlights how the EFQM excellence
for its excellence programming.
model is based on the assumption that management tools drive
The QFD model of this research has named house of excel-
the results and how both enabler criteria and management tools
lence (HOE), because this model basically looking for the excel-
are themselves interrelated. Nevertheless, the EFQM excellence
lence improvement in organization by prioritizes the effective
model is a non-prescriptive framework that allows organizations
management tools on setting EFQM excellence model in organi-
that adopt different approaches to achieve excellence in results.
zation. As it is known, the house of quality (HOQ) in standard
However, the full benefit of the model is obtained when organiza-
QFD model is for improving the quality of the products accord-
tions develop and use best management tools that they are match
ing to the customer’s criteria about the products, but here, orga-
to their needs and have effect on enablers in a way that allows
nization is as a customer that has some needs for excellence and
them to obtain optimal results. This premise leads to our main re-
these needs should be satisfied by technical attributes for excel-
search question, aimed at propose a QFD method for identifying,
lence (TAE).
prioritizing and selection of effective management tools on setting
EFQM model in organization, and making comparison between
both crisp and fuzzy approaches in using QFD method. This re- 3.2. Descriptions of the HOE model
search question is aimed at finding out how enablers should be
ranked to maximize their influence over ranking the using of According to the above preparations, our proposed HOE model
management tools. In other words, we attempt to analyze what (Fig. 3) can be described as follows 9-step procedure. These
is the appropriate method or model within the management tools descriptions, both qualitative and quantitative, are based on the
domain that leads to a maximum improvement in a given excel- ideas from Chan and Wu (2005).
lence profile. Step 1. Identify organization needs for excellence (WHATs): The
This main question can be breakdown in to the four questions producing company should know what organization need for
and tow hypothesizes consist of following items. excellence is important for the company; otherwise you cannot
Questions:

1. What are the effective criteria on setting EFQM excellence


model in an organization?
House of Excellence
2. What are the ranking of effective criteria on setting EFQM excel- (HOE)

lence model in an organization (the research case)?


3. What are the effective management tools on setting EFQM (3)
excellence model in an organization? Technical attributes for
Voice of Organization Excellence (TAE)
4. What are the ranking of effective management tools on setting (VOO)
EFQM excellence model in an organization (the research case)? (1) (4) (2)
Organization Prioritizing
In according to the fuzzy QFD methods and the literature of this Needs for Relation between organizations of
Excellence needs (WHATs) and technical organization
research, the following Hypothesizes were set: (ONE) attributes for excellence (HOWs) needs for
excellence
1. The ranking of effective criteria on setting EFQM model in an (WHATs)
organization (the research case) are the same in crisp and fuzzy
(5)
approaches.
Prioritizing of technical attributes
2. The ranking of effective management tools on setting EFQM for excellence
excellence model in an organization (the research case) are
the same in crisp and fuzzy approaches. Fig. 2. Conceptual model of the research.
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9635

Step 5
Generate Technical
Attitude for excellence
(HOWs)

Step 3
Step 1 Step 2 Identify Competitors, Step 4
Identify Determine Step 6 Conduct organization Determine
Organization Relative Determine Competitive Analysis Final
Needs for Importance Relationships Between & Importance
excellence Ratings of WHATs and HOWs Set organization Ratings of
(WHATs) WHATs Performance Goals for WHATs
WHATs

Step 7
Determine Initial
Technical Ratings of
HOWs

Step 8
Conduct Technical
Competitive Analysis &
Set Technical Performance
Goals for HOWs

Step 9
Determine Final Technical
Ratings of HOWs

Fig. 3. House of excellence (HOE): a 9-step model.

know how to satisfy your organization and thus how to keep your one of the most useful multi criteria decision making methods
business successful and achieve excellence. Available methods to for rating alternatives.
collect organization needs include focus group, individual inter- Step 3. Identify competitors and conduct organization competi-
views, listening and watching, and using existing information. It tive analysis: Competitors who produce the similar products
is suitable and economical to gather organization needs through should be identified by the company under study. Knowing the
focus group (American Supplier Institute, 1994). Grouping related company’s strengths and constraints in all aspects of excellence
organization needs into a category is helpful in analyzing the and in comparison with its main competitors is essential for a
needs. Affinity Diagram (Cohen, 1995), a method of arranging ran- company if it wishes to improve its competitiveness in the rele-
dom data into natural and logical groups, can be used to determine vant markets. This kind of information can be obtained by asking
organization needs. Cluster analysis can also be used for this pur- the experts to rate the relative performance of the company and
pose (Griffin & Hauser, 1993). Usually organization needs can be its competitors on each WHAT and then to aggregate the expert’
organized as a tree-like structure with an increasing number of ratings. Useful ways of conducting this kind of comparison anal-
items moving from left/top (higher levels) to right/bottom (lower ysis are also via questioner. Denote the company in question by
levels). Suppose that, through appropriate ways, K experts have C1. Suppose that L  1, competitors are identified, denoted as
been selected and M organization needs have been identified based C2, . . ., CL. Then the K experts are requested to provide their per-
on the opinions of these K experts. The M organization needs are ceptions on the relative performance of these L companies’
denoted as W1, . . ., WM. These needs could be classified into some excellence’s criteria of the similar type in terms of the M organi-
meaningful categories according to practical situation. zation needs. Suppose that expert k supplies a rating xmlk on
Step 2. Determine the relative importance ratings of organization company C1’s performance in terms of Wm using scale (2), where
needs for excellence: Organization needs for excellence (WHATs) xmlk is one of the nine crisp numbers or STFNs in scale (1). Then
usually are of different degrees of importance and it is a common the performance rating of company C1 on organization needs for
practice for the company to focus more on the important WHATs. excellence Wm is given as:
The relative importance of the WHATs is usually expressed as a set
of ratings that can be determined by letting the experts reveal their X
K

perceptions on the relative importance of the WHATs and then X ml ¼ ðX ml1 þ X ml2 þ    þ X mlK Þ=K ¼ X mlk =K;
K¼1
averaging their perceptions. The appropriate ways of obtaining ex-
perts’ perceptions are by analytic hierarchy process (AHP). AHP is m ¼ 1; 2; . . . ; M; l ¼ 1; 2; . . . ; L ð1Þ

very poor poor neutral good Very good

ð2Þ
1 2 3 4 5 6 7 8 9
[0,2] [1,3] [2,4] [3,5] [4,6] [5,7] [6,8] [7,9] [8,10]
9636 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

Thus, the companies’ performance ratings on the customer needs Step 6. Determine the relationships between HOWs and WHATs:
can be denoted by an ML matrix, called customer comparison This is an important work in HOE/QFD which is performed care-
matrix: fully and collectively by experts. The relationship between a
HOW and a WHAT is usually determined by analyzing to what
extent the HOW could technically related to and influence the
WHAT. All these relationships form a matrix with the WHATs
as rows and the HOWs as columns. It is suitable to complete this
matrix in a column- or HOW-wise manner since once a HOW is
defined we usually begin establishing to what extents it relates
to the WHATs (American Supplier Institute, 1994). Let the rela-
Based on this X information, experts competitive priority ratings on tionship value between technical attitude Hn and organization’s
the WHATs for the producing company C1 can be obtained, as need Wm be determined as rmn according to scale (4). Then we
e = (e1, e2, . . ., em) where em is company C1’s priority rating on cus- can form the following relationship matrix between the HOWs
tomer need Wm. This set of priority ratings derived by the more and the WHATs:
objective entropy method as introduced in the Appendix.

very weak weak moderate relation strong Very strong

ð4Þ
1 2 3 4 5 6 7 8 9
[0,2] [1,3] [2,4] [3,5] [4,6] [5,7] [6,8] [7,9] [8,10]

According to company C1’s current performance on the WHATs


in relation to its competitors’ performance, performance goals on
the WHATs can be set for the company. These goals should be
set competitively and realistically by the company, which is a
highly strategically activity involving many considerations from
relevant management. Assume that for organization need Wm, a
proper performance goal am has been set according to scale (1).
Step 7. Determine initial technical ratings of HOWs: Initial techni-
Thus the company has a goal performance vector in terms of the
cal ratings of HOWs are decided by two factors, final importance
organization needs, denoted as a = (a1; a2, . . ., aM). In most cases,
ratings of WHATs and the relationships between the HOWs and
each goal performance level should not be lower than current per-
the WHATs. These ratings indicate the basic importance of the
formance level, implying the need or desire for further improve-
HOWs developed in relation to the WHATs. They are usually com-
ment. From this we can also set the company’s improvement
puted using the simple additive weighting (SAW) method. That is,
ratio for Wm as um = am/xml. It is obvious that the higher the
for technical attitude Hn, its initial technical rating is computed as
improvement ratio, the more the company should work on the
the following simple weighted average over its relationships with
WHAT, and thus the more important the WHAT for the company.
the WHATs:
Step 4. Determine the final importance ratings of customer needs:
organization needs for excellence with higher relative importance X
M
tn ¼ f1  r 1n þ f2  r2n þ    þ fM  r Mn ¼ fm  rmn ;
perceived by experts and higher competitive priorities and m¼1
improvement ratios should receive higher attention. Thus, organi-
n ¼ 1; 2; . . . ; N: ð5Þ
zation needs Wm’s final importance rating for the company is
determined jointly by its relative importance gm, competitive pri- Other methods to obtain comprehensive ratings for a set of choices
ority em and improvement ratio um as: in relation to a number of performance criteria, such as the tech-
nique for order preference by similarity to ideal solution (TOPSIS)
fm ¼ um  g m  em ; m ¼ 1; 2; . . . ; M ð3Þ (Yoon & Hwang, 1995) and the operational competitiveness rating
(OCRA) procedure (Parkan & Wu, 2000), can also be used to com-
WHATs with high such final ratings indicate both importance and pute initial technical ratings (Chan & Wu, 1998). We will denote,
potential business benefit to the company. Of course, a (weighted) in any case, the HOWs’ initial technical ratings by a vector,
sum of um, gm and em can also produce a reasonable fm, if preferred. t = (t1, t2, . . ., tN).
In any case, we will denote the final importance ratings for the orga- Step 8: Perform technical competitive analysis: Although some
nization needs as a vector, f = (f1, f2, . . ., fM). technical parameters and know-HOWs of the competitors’ organi-
Step 5. Generate technical measures (HOWs): After organization zation cannot be easily obtained and some may even be kept con-
reveal it needs for the product, the company’s excellence team (ex- fidential, the producing company should make every effort to
perts) should develop a set of HOWs to capture the organization acquire this information and failing to do so may result in an unfa-
needs. HOWs could be generated from current organization’s stan- vorable position for the company in the market place. In case of ex-
dards or selected by ensuring through cause–analysis that the treme difficulty in obtaining the technical attributes for excellence
HOWs are the first-order causes for the WHATs (Hauser & Clausing, of the competitors’ on some HOWs, careful technical assessments
1988). Assume that N technical attitude have been developed, de- should be made to give reliable scores (in a suitable scale such as
noted as H1, H2, . . ., HN. Their measurement units and improving 4) representing the technical performance of the competitors’
directions should also be determined, which is usually easy to do products on the said HOWs.
and important for the company to conduct technical competitive Let the technical parameter or performance score of company
analysis for the HOWs. C1’s excellence needs on technical attitude Hn be determined as
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9637

Table 1
Quantitative descriptions of HOE model.

Step 5
Technical Measures (HOWs):
H1 H2 ... HN

Step 1 Step 2 Step 6 Step 3 Step 4


Organization Relative Relationships Between Organization Final
Needs for Importance WHATs and HOWs (R): Competitive analysis Importance
excellence (X) Ratings (f):
Ratings (g): H1 H2 ... HN
(WHATs): C1 C2 ... CL (e) (a) (u)
W1 g1 W1 r11 r12 ... r1N W1 x11 x12 ... x1L e1 a1 u1 f1
W2 g2 W2 r21 r22 ... r2N W2 x21 x22 ... x2L e2 a2 u2 f2
... ... … ... ... ... ... ... ... ... ... ... ... … … ...
WM gM WM rM1 rM2 ... rMN WM xM1 xM2 ... xML eM aM uM fM
Step 7
Initial Technical Ratings (t):
t1 t2 ... tN

Step 8
Technical Competitive
Analysis (Y):
H1 H2 ... HN
C1 y11 y21 ... yN1
C2 y12 y22 ... yN2
... ... ... ... ...
CL y1L y2L ... yNL
(z) z1 z2 ... zN
(b) b1 b2 ... bN
(v) v1 v2 ... vN
Step 9
Final Technical Ratings (s):
s1 s2 ... sN

ynl. Then we can form the technical comparison matrix of the com- Step 9. Obtain final technical ratings of the HOWs: Those HOWs
panies’ excellence needs on the HOWs: with higher initial technical ratings (tn’s), higher technical compet-
itive priorities (zn’s) and higher improvement ratios (vn’s) indicate
working focuses and market opportunities for the producing com-
pany. Final technical rating is a useful measure to reflect this point
which, with respect to Hn, can be computed for the company’s
excellence by integrating all these factors using a formula similar
to (3) for computing the final importance ratings for the WHATs:
From this Y information technical competitive, priority ratings on Sn ¼ V n  t n  z n ; n ¼ 1; 2; . . . ; N ð7Þ
the HOWs can be obtained for the producing company using the en-
tropy method as described in the Appendix. We will denote these Thus we have a final technical rating vector on the HOWs,
ratings as a vector, z = (z1, z2, . . ., zN), where zn represents the com- s ¼ ðs1 ; s2 ; . . . ; sN Þ. HOWs with higher final technical ratings, imply-
pany’s technical competitive priority with respect to Hn. Based on ing greater importance for the company’s excellence to be success-
the above matrix Y, company C1 could also set performance goals ful in the competitive markets.
on the HOWs. It should be noted that these goals are different from The above quantitative descriptions of our proposed HOE model
design specifications. Essentially they represent levels of perfor- are summarized in Table 1.
mance on the HOWs which the company believes is required for In the next section, we will demonstrate step by step the con-
its level of excellence to be of technical competitiveness in the cepts and operations of the model through an easy-to-understand
relevant markets in comparison with its competitors. The goals automotive Iranian company. To make our HOE model fully oper-
should also be reachable according to the company’s technical able, we will use fuzzy method to handle the vagueness of people’s
resources. Suppose that the company sets a goal performance level linguistic assessments and entropy method to derive competitive
bn on Hn, then we have a technical performance goal vector, b = priority ratings and also group analytic hierarchy process (GAHP)
(b1, b2, . . ., bN). Compared to these goals we can define improvement with both crisp and fuzzy approaches for determining the relative
ratios vn’s for the current performance of company on the HOWs: importance rating of excellence criteria or WHATs. A brief intro-
Vn = bn/yn1 for Hn to be maximized; or when bn P yn1 for Hn to duction to these three methods is given in the Appendix.
meet target;
Vn = yn1/bn for Hn to be minimized; or when bn < yn1 for Hn to 3.3. Population and data
meet target.
Total population of this research was selected from the experts
Or in a uniform manner: of an automotive company as our case in this research. The total
population was of 30 experts, and we obtained 30 valid responses.
V n ¼ maxfyn1 =bn g ¼ minfyn1 =bn g ð6Þ And data were collected from a self-constructed questioner.
9638 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

3.4. Measurement
are helpful for practitioners to follow. Here we present an automo-
Data were collected from the experts and from all part of that tive Iranian company example to illustrate the concepts and com-
company through a questionnaire. Following the QFD methodol- putations in our proposed HOE model in details. An automotive
ogy, a questionnaire covering aspects of each steps of that method- Iranian company, called company C1, wishes to make an improve-
ology. The questionnaire method consists of the elaboration of a ment level of it’s excellence in response to the competition of other
basic survey for QFD methodology in EFQM domain. Thus, all of competitors in the same district. HOE technique can help C1 make
the steps of QFD methodology assessed through several questions, the appropriate decision resulting in better improvement.
which are evaluated on a scale (for instance, from 1 to 9 points). The basic idea is (i) to understand what are organization needs
This approach enables the evaluation of the organization’s commit- for excellence (such as EFQM enabler criteria like leadership, strat-
ment towards each criterion and management tools and relation- egy, etc.) and then to identify the important ones through criteria
ships between management tools and enablers criteria to be analysis, and (ii) to associate the organization’s needs for excel-
carried out and provides a score that quantifies the consistency be- lence with appropriate technical attitudes or solutions (such man-
tween organizational needs and QFD model and the EFQM excel- agement tools like strategic management, production
lence model. Consequently, questionnaires are useful to generate management, etc.) and then to find the important ones through
a quality pattern and to identify discrepancies in the organizational technical analyses. In what follows we will build the HOE model
needs for excellence results. for this example step by step according to the qualitative and
After an exhaustive examination of a QFD methodology, the quantitative descriptions in Section 3.
questionnaire was operationalised through four parts. These four
parts were grouped into separately aggregated scales correspond- Question 1: What are the effective criteria on setting EFQM
ing to each part. Cronbach’s alpha ranged between 0.89 and 0.94, excellence model in an organization?
indicating that the items deal with the same underlying construct.
The managers assessed the items using a five-point Likert scale Step 1: At first the company must determine the experts to re-
ranging from 1 (very low performance, very weak relation or abso- veal their various perceptions about the questions. Here, for illus-
lutely not important) to 9 (very high performance, very strong rela- tration purpose, thirty experts of the company from all part of the
tion, or absolutely important). The validity of questionnaire was company are selected to help conduct the HOQ analysis (i.e.,
checked by some experts that they were justified the validity of K = 30). By a complete literature review about the effective criteria
the questionnaire, so the validity of this research is content on setting EFQM mode, five enabler criteria of EFQM model as
validity. organization’s needs for excellence (ONE) or WHATs are found
(i.e., M = 5). They are: ‘‘leadership’’, ‘‘strategy’’, ‘‘employee’’, ‘‘shar-
ing & resources’’ and ‘‘processes’’.
3.5. Statistical procedure
Question 2: What are the ranking of effective criteria on setting
In this research, we propose an alternative method for estimat-
EFQM excellence model in an organization (the research case)?
ing the relationships between the enablers and the management
tools with a QFD Model, based on the crisp and fuzzy approaches.
Step 2: The five WHATs can hardly be of same importance to the
So for testing the hypothesis of this research, we used spearman
organization. So by use of pair wises comparison between each tow
correlation coefficient, as this methodology implies the adoption
criteria, the thirty experts are asked to reveal their perceptions on
of a global approach in the study of the EFQM Excellence Model.
the relative importance of the five WHATs using the five linguistic
Spearman correlation coefficient analysis is a statistical multivari-
terms in scale, so each expert should answer ten questions in pair
ate technique that summarizes the relations between two sets of
wise form and then by use of group analytic hierarchy process
variables. If our data are not normally distributed or have ordered
(GAHP), the relative importance of each criteria were calculated.
categories, we can choose Spearman, which measure the associa-
Table 2 shows the final pair wises comparison of all the aggrega-
tion between rank orders. Correlation coefficients range in value
tion of experts and the relative importance of each enabler criteria
from 1 (a perfect negative relationship) and +1 (a perfect positive
with crisp numbers. Table 3(a) shows the final fuzzy pair wises and
relationship). A value of 0 indicates no linear relationship. When
Table 3(b) shows the fuzzy relative importance of each criterion
interpreting your results, be careful not to draw any cause-and-
with fuzzy numbers which are the results of using a fuzzy analytic
effect conclusions due to a significant correlation. Although
hierarchy process (FAHP), as shown in Appendix. Suppose that ex-
spearman correlation coefficient can be conducted using standard
pert 1 rates the importance of W1 by comparison to W2 as ‘‘high’’.
statistical software (e.g. SPSS), in this paper we use spss.13.
Using scale (2), these linguistic assessments of the WHATs’ relative
importance can be converted to crisp numbers or symmetrical tri-
4. Results and discussion angular fuzzy members (STFNs) according to practical need. For
example, expert 1 considers W1 as having ‘‘high’’ importance than
In this section the questions of this research will answer and W2, which can be represented by a crisp number 7 or an STFN [6, 8]
also hypothesizes of this research will be test step by step with according to scale (2) so we show this kind of fuzzy numbers as
HOE model. Complete QFD examples to fully illustrate the proce- M = (l,m,u). for above example it can be represented by (6,7,8) as
dure of QFD do not appear frequently in the literature, but they a STFN.

Table 2
Relative importance of each criterion by a GAHP method with crisp numbers.
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9639

Table 3a
Final fuzzy comparisons matrix of criterion.

Table 3b
Fuzzy relative importance of each criterion by a FGAHP method.

Criteria Fuzzy relative importance Defuzzified relative importance Relative importance (g fw )


L M U
W1 0.065588 0.094353 0.124516 0.0947 9.04
W2 0.079553 0.115859 0.166 0.11932 11.39
W3 0.105682 0.156506 0.220246 0.15974 15.25
W4 0.195478 0.311328 0.501182 0.32983 31.48
W5 0.202162 0.321954 0.530163 0.34406 32.84

Applying the entropy method as illustrated in the Appendix, we


can obtain company C1’s competitive priority ratings on the 5 orga-
Table 4
Final excellence competitive analysis matrix X = [xml]57. nization’s needs for excellence based on the above excellence com-
parison matrix X. For example, the ‘‘leadership’’ of W1 on the seven
WHATS C1 C2 C3 C4 C5 C6 C7 em
companies’ excellence performance is composed of the seven com-
W1 6.46 5.46 4.93 5.56 5.66 4.16 6.90 0.200003 panies’ performance ratings on W1: (6.46, 5.46, 4.93, 5.56, 5.66,
W2 5.56 5.40 4.53 5.26 5.36 3.83 6.70 0.199318
4.16, 6.90), which is the first row of matrix X. Then we can compute
W3 5.33 5.46 4.96 5.23 5.73 4.06 6.53 0.200302
W4 5.93 5.20 4.66 5.03 5.20 3.86 6.53 0.199979 the total score of W1: x1 = x11 + x12, . . ., + x16 + x17 = 6.46+ 5.46 +,
W5 5.53 5.06 4.80 5.23 5.46 4.60 6.83 0.200399 . . ., + 4.16 + 6.90 = 39.17, and obtain the ‘‘probability leadership’’
of W1:

The AHP methodology of Satty (1980) provides a consistency ra-


P11 = X11/X1 = 6.46/39.17 = 0.165106
tion to measure any inconsistency whit in the judgments in each
P12 = X12/X1 = 5.46/39.17 = 0.139574
comparison matrix as well as for the entire hierarchy. The ration
P13 = X13/X1 = 4.93/39.17 = 0.125957
can be use to indicate whether or not the largest can be arranged
P14 = X14/X1 = 5.56/39.17 = 0.142128
in an appropriate order of ranking and how consistent are the pair
P15 = X15/X1 = 5.66/39.17 = 0.144681
wise comparison matrixes. If the calculated consistency rate of a fi-
P16 = X16/X1 = 4.16/39.17 = 0.106383
nal comparison matrix is less than 0.1, then the consistency of the
pairwise judgment can be thought as being acceptable. Otherwise P17 = X17/X1 = 6.90/39.17 = 0.176170
the judgments expressed by the experts are considered to be
inconsistent, and the decision makers have to repeat the pairwise The entropy of W1 is then computed using (A.20) as:
comparisons. In this research after calculating the consistency rate
of the entire comparison matrix and also the final comparison ma- X
7
EðW 1 Þ ¼ U Pl1 lnðPl1 Þ
trix that made by a geometric mean method, it was found they are
l¼1
all less than 0.1. Therefore, the consistency of the judgment in all
the comparison matrices is acceptable.
¼ ½0:165106 ln 0:165106 þ    þ 0:17617 ln 0:17617
Step 3: This step is for company C1 to identify competitors and ¼ 1:93442
conduct excellence competitive analysis. In the district’s automo-
We can obtain in the same way the entropy for each of the 5 orga-
tive market, company C1 has six main competitors, called company
nizations’ needs for excellence as:
C2, C3, C4, C5, C6 and C7, each of which makes a similar type of prod-
ucts. In order to understand the automotive market and its relative ðEðW 1 Þ; EðW 2 Þ; . . . ; EðW 5 ÞÞ
position in the market, and to finally find out the priorities for fur- ¼ ð1:93444; 1:92782; 1:93734; 1:93421; 1:9383Þ
ther improvement, company C1 asks all the experts to rate the rel-
ative performance of its own company and the six competitors’ Finally, according to (A.21) we can obtain company C1’s competitive
similar products in terms of the five WHATs using scale (2). For priority ratings on the Wj’s:
example, expert 1 rates the performance of C2’s performances on
e ¼ ðe1 ; e2 ; . . . ; e5 Þ
W3 as ‘‘neutral’’ using scale (2), which corresponds to a crisp num-
ber of 5, i.e., X321 = 5. We will not consider using STFNs to represent ¼ ð0:200003; 0:199318; 0:200302; 0:199979; 0:200399Þ
performance assessments since it is too complex to incorporate where, for example,
STFNs into the following entropy computations. X
The final excellence competitive analysis of the thirty experts’ e1 ¼ EðW 1 Þ= EðW m Þ ¼ 1:93444=ð1:93444 þ    þ 1:9383Þ
assessments are shown in Table 4, where, according to the thirty’
¼ 1:3137=9:05144 ¼ 0:200003
assessments of the relative performance of the seven company’
similar products in terms of the 5 WHATs, a excellence comparison This set of competitive priority ratings are shown in the last column
matrix X = [Xmn]57 can be obtained by averaging the customers’ of Table 4 from which we know that W5 is of the highest competi-
assessments. tive priority for the company, followed by W3; W4 and W2.
9640 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

Table 5 Applied to a triangular fuzzy number FN = (fL, fM, fU), for defuzzifica-
Improvement ratio of WHATs. tion of a STFN, the Facchinetti, Ghiselli Ricci, and Muzioli (1998) ap-
WHATs am U = GOAL/Xm1 proach produces a score identified by the value:
W1 8 1.23711 Score ¼ ðfL þ 2f M þ fU Þ=4 ð8Þ
W2 7 1.25749
W3 7 1.31250 f ¼ ðf1 ;2 ; f3 ; f4 ; f5 Þ
W4 7 1.17978
W5 7 1.26506 ¼ ð0:023432; 0:029906; 0:041994; 0:077817; 0:087224Þ

So the final importance of WHATs is as:

Based on the resources available and the relative performance of W5 > W4 > W3 > W2 > W1
the seven company on the 5 WHATs, company C1 can set improving
Both sets of ratings indicate that W5 is the most important WHAT,
goals on each WHAT to better satisfy the organizations’ needs for
followed by W4 and W3. But the crisp approach show that W1 is
excellence. After various considerations, company C1 decides the fol-
more important than W2 but in fuzzy approach W2 is more impor-
lowing performance goals on the WHATs using scale (2):
tant than W1.
a ¼ ða1 ; a2 ; a3 ; a4 ; a5 Þ ¼ ð8; 7; 7; 7; 7Þ These final importance ratings of the WHATs, expressed as
both crisp numbers and STFNs, are shown in the second column
This set of goals is shown in Table 5. It is noted that all goal perfor-
of Table 6.
mance levels are higher than C1’s current performance levels repre-
sented by the first column of excellence comparison matrix X. If C1’s
Hypothesize 1: The ranking of effective criteria on setting
performance on a WHAT is poorer or much poorer than the perfor-
EFQM model in an organization (the research case) are the same
mance of most of its competitors, then the goal level is set to be
in crisp and fuzzy approaches.
much higher than its current level to be of competitiveness. Other-
wise, if C1’s performance on a WHAT is better than the performance
In order to be comparable, the crisp and fuzzy final importance
of most of its competitors, then the goal level is only set to be
ratings are tested by spearman correlation coefficient. According to
slightly higher than its current level which is enough for C1 to keep
the Table 6, with percentage results of crisp and fuzzy ranking, the
and enhance its established competitiveness. We do not consider
spearman coefficient correlation for these tow type of data is 0.9
setting goals in STFN form either, since these results in some com-
and there is a very strong positive correlation between fuzzy and
putational and explanatory difficulties.
crisp importance ranking. So the first hypothesis of this research
According to company C1’s current and goal performance levels
that maintains: ‘‘The ranking results of effective criteria on setting
on the five WHATs, its improvement ratios with respect to the
EFQM model in an organization (the research case) are the same in
organizations’ needs for excellence can be easily computed accord-
crisp and fuzzy approaches.’’ were supported.
ing to the formula um = am/xm1:
u ¼ ðu1 ; u2 ; u3 ; u4 ; u5 Þ ¼ ð1:2371; 1:2575; 1:3125; 1:1798; 1:2651Þ Question 3: what are the effective management tools on setting
EFQM excellence model in an organization?
Step 4: According to each WHAT’s relative importance rating, com-
petitive priority rating and improvement ratio, company C1 could Step 5: Now it is time to convert organizations’ needs for excel-
now reach the final importance rating of the WHAT using (3). In lence (ONE) into technical attributes specifications. After careful
case that the relative importance ratings are crisp numbers, the fi- considerations and literature review (Darrell, 2007; EFQM, 2000;
nal importance ratings are also given as the following crisp Ignacio, 2005), 15 technical attributes for excellence (HOWs) that
numbers: relate to and can help realize the five WHATs are proposed as:
f ¼ ðf1 ; f2 ; f3 ; f4 ; f5 Þ ¼ ð0:0286; 0:02807; 0:0369; 0:0680; 0:0870Þ
H1 = inventory management
Here, for example, the final importance rating of W1 in crisp form, f1, H2 = total quality management
is computed by (3) as: H3 = human resources management
f1 ¼ u1  g 1  e1 ¼ 1:23711  0:1157  0:200003 ¼ 0:0286 H4 = knowledge management
H5 = technology management
From f we can finally rank the importance of the five WHATs in the H6 = information management
following order: H7 = energy management
W5 > W4 > W3 > W1 > W2 H8 = project management
H9 = financial management
where ‘‘>’’ means ‘‘more important than’’. H10 = change management
If relative importance ratings are STFNs, final importance rat- H11 = customers relationship management
ings are also given as STFNs: H12 = supply chain management
H13 = business process management
f f ¼ ðf1f ; f2f ; f3f ; f4f ; f5f Þ
H14 = strategic management
¼ ð½0:016228; 0:030808; ½0:019939; 0:041606;
H15 = production management
½0:027784; 0:0579; ½0:046119; 0:118244;
½0:051251; 0:134405Þ Question 4: what are the ranking of the effective management
tools on setting EFQM excellence model in an organization (the
Here, for example, the final importance rating of W1 in STFN form,
research case)?
f f1, is computed by (3) and the scalar multiplication rule of STFNs
as:
Step 6: Then the experts begin to establish the relationships be-
f1f ¼ u1  g f1  e1 ¼ 1:2371  ½0:065; 0:124  0:200003 tween the HOWs and the WHATs, or to examine to what extent each
HOW is related to each WHAT. This step is usually done simulta-
¼ ½0:016228; 0:03808 neously with Step 5 since in the process of generating HOWs, each
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9641

Table 6
Normalization and determine the percentages of the five WHATs.

WHATs Crisp (fM) Fuzzy (fMf )

Crisp weights Nor. Per. % Fuzzy weights Defuzzi. Nor. Per. %


W1 0.0286 0.115058 11.50 [0.016,0.030] 0.023432 0.089994 9.00
W2 0.02807 0.112926 11.30 [0.020,0.042] 0.029906 0.114858 11.47
W3 0.0369 0.148449 14.84 [0.028,0.058] 0.041994 0.161284 16.13
W4 0.0680 0.273565 27.36 [0.046,0.118] 0.077817 0.298867 29.90
W5 0.0870 0.350002 35.00 [0.051,0.134] 0.087224 0.334996 33.50

Table 7
Final relationship matrix between WHATs and HOWs with both crisp and STFNs.

Final matrix HOWs


WHATs L Inventory management U L Total quality management U L Human resource management U
W1 4.1 5.1 6.1 5.5 6.5 7.5 6.2 7.2 8.2
W2 3.7 4.7 5.7 4.4 5.4 6.4 5.6 6.6 7.6
W3 4.1 5.1 6.1 4.5 5.5 6.5 4.8 5.8 6.8
W4 5.0 6.0 7.0 3.7 4.7 5.7 4.0 5.0 6.0
W5 4.7 5.7 6.7 4.7 5.7 6.7 4.2 5.2 6.2
L Knowledge management U L Technology management U L Information management U
W1 4.8 5.8 6.8 4.5 5.5 6.5 5.1 6.1 7.1
W2 4.6 5.6 6.6 4.1 5.1 6.1 5.1 6.1 7.1
W3 4.1 5.1 6.1 5.1 6.1 7.1 4.4 5.4 6.4
W4 4.2 5.2 6.2 5.3 6.3 7.3 4.8 5.8 6.8
W5 4.9 5.9 6.9 5.0 6.0 7.0 4.4 5.4 6.4
L Energy management U L Project management U L Financial management U
W1 4.5 5.5 6.5 3.9 4.9 5.9 5.3 6.3 7.3
W2 4.6 5.6 6.6 4.0 5.0 6.0 4.5 5.5 6.5
W3 3.4 4.4 5.4 4.0 5.0 6.0 4.8 5.8 6.8
W4 4.2 5.2 6.2 4.0 5.0 6.0 5.0 6.1 7.1
W5 4.0 5.0 6.0 3.8 4.8 5.8 5.0 6.0 7
L Change management U L Customer relationship management U L Supply chain management U
W1 4.6 5.6 6.6 4.5 5.5 6.5 4.4 5.4 6.4
W2 4.7 5.7 6.7 4.2 5.2 6.2 3.6 4.6 5.6
W3 4.8 5.8 6.8 5.1 6.1 7.1 4.3 5.3 6.3
W4 4.6 5.6 6.6 4.9 5.9 6.9 5.0 6.0 7.0
W5 5.0 6.0 7.0 4.1 5.1 6.1 4.0 5.8 6.8
L Business process management U L Strategic management U L Production management U
W1 5.1 6.1 7.1 4.7 5.7 6.7 4.0 5.0 6.0
W2 4.3 5.3 6.3 3. 4.9 5.9 4.2 5.2 6.2
W3 5.1 6.1 7.1 5.5 6.5 7.5 5.2 6.2 7.2
W4 5.5 6.5 7.5 4.3 5.3 6.3 6.0 7.0 8.0
W5 5.0 6.0 7.0 4.8 5.8 6.8 6.0 7.0 8.0

HOW’s relationships with the WHATs are always examined once the Here, for example, crisp initial technical rating of H1, t1, is computed
HOW is considered. The relationships between the HOWs and the as the weighted average over H1’s crisp relationship values with the
WHATs are determined by technical analysis and empirical judg- five WHATs, r11, r21, . . ., r51, which correspond to the crisp part of the
ment, and usually may not be precise. So it is quite appropriate to relationship matrix that is bolded in Table 7, and the weights are
use STFNs to represent this kind of relationships. For each HOW the crisp final importance ratings of the five WHATs, f1, f2, . . ., f5, i.e.,
with respect to each WHAT, the experts determine the relationship
X
5
first in linguistic term using scale (4) and then convert this relation- t1 ¼ fm  r m1
ship into corresponding crisp number and STFN, for example, the m¼1
expert consider the relationship between H1 and W1 as ‘‘very ¼ 0:0286  5:1 þ 0:02807  4:7 þ 0:0369  5:1
strong’’ that corresponds to a crisp number of 9 and an STFN [8, 10]. þ 0:068  6:0 þ 0:087  5:7 ¼ 1:38
The full matrix of these relationships, both in crisp numbers and
STFNs, are shown in Table 7 where be obtained by averaging the ex- From these crisp initial technical ratings, the technical measures
pert’ assessments about the relationship between WHATs and (HOWs) can be ranked in the following order:
HOWs.
Step 7: According to the WHATs’ final importance ratings and H15 > H13 > H9 > H5 > H10 > H6 > H14 > H12 > H11 > H3
the relationship values between the HOWs and the WHATs, the > H1 > H4 > H 2 > H 7 > H 8 ð9Þ
HOWs’ initial technical ratings can be computed usually through
the simple additive weighting (SAW) formula (5). When crisp If fuzzy numbers of the relationship matrix are used, the fuzzy ini-
numbers are used, the initial technical ratings are given as tial technical ratings are also given as STFNs:

t ¼ ðt 1 ; . . . ; t 15 Þ tf ¼ ð½0:73; 2:52; ½0:72; 2:46; ½0:75; 2:52; ½0:73; 2:51; ½0:8; 2:67;
¼ ð1:38; 1:35; 1:40; 1:38; 1:48; 1:42; 1:27; 1:23; 1:49; 1:44; ½0:76; 2:57; ½0:66; 2:33; ½0:64; 2:28; ½0:8; 2:68; ½0:77; 2:60;
1:39; 1:40; 1:51; 1:41; 1:61Þ ½0:74; 2:52; ½0:74; 2:55; ½0:82; 2:72; ½0:76; 2:56; ½0:88; 2:89Þ
9642 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

Table 8
The crisp and fuzzy initial technical rating of HOWs.

HOWs
L Inventory management U L Total quality management U L Human resource management U
tf 1.38 1.35 1.4
tf 0.73 2.52 0.72 2.46 0.75 2.52
L Knowledge management U L Technology management U L Information management U
tf 1.38 1.48 1.42
tf 0.73 2.5 0.79 2.6 0.76 2.56
L Energy management U L Project management L Financial management U
tf 1.26 1.23 1.49
tf 0.66 2.33 0.64 2.2 0.80 2.68
L Change management U L Customer relationship management U L Supply chain management U
tf 1.43 1.39 1.40
tf 0.77 2.60 0.74 2.52 0.74 2.55
L Business process management U L Strategic management U L Production management U
tf 1.51 1.41 1.61
tf 0.82 2.72 0.76 2.56 0.88 2.89

Table 9
Final technical competitive analysis matrix Y = [ynl]157.

HOWs C1 C2 C3 C4 C5 C6 C7
Strategic management 5.56 5.00 4.43 4.96 5.40 4.50 6.43
Business process management 6.30 5.06 3.86 5.20 5.33 4.66 6.20
Supply chain management 6.53 5.30 5.03 5.33 5.56 4.76 6.66
Customer relationship management 6.13 5.46 5.10 5.20 5.60 4.33 6.70
Change management 5.20 4.73 4.76 4.60 4.66 4.33 6.36
Financial management 6.03 4.43 4.86 5.23 5.33 4.73 6.66
Project management 5.63 5.03 4.80 4.83 5.10 5.00 6.56
Energy management 5.16 4.4 4.70 4.86 5.3 4.63 6.13
Information management 5.13 4.86 4.66 5.13 4.96 4.6 6.9
Technology management 4.80 4.70 4.63 5.00 4.93 4.4 6.46
Knowledge management 5.03 5.10 4.50 5.03 4.93 4.86 6.53
Human resource management 5.76 4.83 4.53 5.33 5.50 4.76 6.60
Total quality management 5.70 5.16 4.23 4.86 5.23 4.56 6.53
Inventory management 5.36 4.83 4.96 4.90 5.16 4.93 6.23
Production management 6.46 5.23 5.03 5.26 5.23 5.13 7.03

Here, for example, the initial technical rating of H1 in STFN form, t f1 , mance on the HOWs, company C1 must try all the means to obtain
is computed as the weighted average over H1’s STFN form relation- this valuable information in order to know its technical strengths
ship values with the five WHATs, rf11 ,rf21 ; . . . ; r f51 , which correspond and weaknesses and hence to improve or enhance its competitive-
to the first column of the STFN form relationship matrix Rf, and ness. Through a lot of efforts company C1 obtains all the technical
the weights are the final importance ratings of the five WHATs in parameters of its own and its competitors in terms of the 15
STFN form, f1f ; f2f . . . f5f , i.e., HOWs. This information forms a technical comparison matrix
Y = [ynl]157 as shown in Table 9.
X
5
t f1 ¼ fmf  r fm1 Applying entropy method to Y in the same manner as in excel-
m¼1 lence competitive analysis (Step 3), technical competitive priority
¼ ½0:016; 0:0308  ½4:1; 6:1 þ    þ ½0:05; 0:134  ½4:7; 6:7 ratings can be obtained for company C1 on the 15 HOWs:

¼ ½0:73; 2:52 z ¼ ðz1 ; z2 ; . . . ; z14 ; z15 Þ


According to the principle in the Appendix, these fuzzy ratings have ¼ ð0:071431; 0:071285; 0:071443; . . . ; 0:07157; 0:071409Þ
the following ranking order for the HOWs’ initial importance:
From these ratings which we know that H9; H15 and H5 are of the
H15 > H13 > H9 > H5 > H10 > H6 > H14 > H12 > H11 > H3 highest competitive priorities. According to the technical perfor-
mance of its own and the other six competitors company in terms
> H1 > H 4 > H 2 > H 7 > H8 ð10Þ
of the 15 HOWs, company C1 could set technical performance goal
It is noticed from (9) and (10) that the crisp and fuzzy ratings exhibit on each of the HOWs for itself.
the same ranking order. Both sets of ratings indicate that H15 is of the To better fulfill the customer needs. These goals should be
highest initial importance, followed by H13, H9 and H5. The crisp and determined both competitively and realistically. Company C1’s rel-
fuzzy initial technical ratings of the 15 HOWs are shown in Table 8. evant experts agree with the following performance goals on the
Step 8: Now turn to technical competitive analysis which is to HOWs for further improvement:
find and establish competitive advantages or to further enhance
b ¼ ðb1 ; b2 ; . . . ; b14 ; b15 Þ ¼ ð7; 8; 8; 8; 7; 8; 7; 7; 7; 6; 7; 7; 7; 7; 8Þ
the existing advantages for company C1, through comparing all
the company’ similar products in terms of their technical perfor- From these goal (bn) and current (yn1) technical performance levels,
mance on the 15 identified HOWs. Although it is always not easy improvement ratios for company C1 to be competitive in terms of
to acquire the technical performance levels of competitors’ perfor- the HOWs can be easily computed using (6) as:
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9643

v ¼ ðv 1 ; v 2 ; . . . ; v 14 ; v 15 Þ caused by the setting of performance goals (bn’s) or improvement


¼ ð1:25749; 1:26984; . . . ; 1:30435; 1:23711Þ ratios (vn’s): (i) H15’s improvement ratio (1.23711) is lower than
H9’s (1.3636), and (ii) H8’s improvement ratio (1.3548) is higher
Step 9: This is the last step of our proposed HOE model. Integrat- than H7’s (1.2426).
ing the initial technical ratings, technical competitive priority rat- If initial technical ratings are STFNs, then the final technical rat-
ings and improvement ratios of the HOWs, final technical ratings ings are also given as STFNs:
can be computed by (7). If initial technical ratings are crisp num-
bers, the final technical ratings are also crisp numbers and given sf ¼ ðsf1 ; . . . ; sf15 Þ
as: ¼ ð½0:066; 0:227; ½0:065; 0:223; . . . ; ½0:071; 0:239; ½0:078; 0:256Þ:
s ¼ ðs1 ; s2 ; . . . ; s14 ; s15 Þ ¼ ð0:12411; 0:12247; . . . ; 0:13315; 0:13205Þ
Here, for example, the final technical rating of H1 in STFN form, sf1,
Here, for example, the final technical rating of H1 in crisp form, s1, is is computed by (7) and the arithmetic of STFNs as:
computed by (7) as:
sf1 ¼ v1  t f1  z1 ¼ 1:25749  ½0:73; 2:52  0:071431
s1 ¼ v 1  t 1  z1 ¼ 1:25749  1:38  0:071431 ¼ 0:12411 ¼ ½0:066; 0:227
From s we can rank the final technical importance of the nine HOWs
These fuzzy ratings produce the following ranking order for the
in the following order:
HOWs’ final importance:
H9 > H15 > H5 > H11 > H6 > H13 > H14 > H10 > H4 > H1
H9 > H15 > H5 > H11 > H6 > H13 > H14 > H10 > H4
> H3 > H2 > H12 > H8 > H7 ð11Þ
> H1 > H3 > H2 > H12 > H8 > H7 : ð12Þ
This final technical importance order differs from the initial techni-
It is noticed from (11) and (12) that the crisp and fuzzy ratings show
cal importance order (9) in two aspects: (i) H15 is of higher initial
an almost identical ranking order for the HOWs’ final technical
technical importance but lower final technical importance than
importance. Both sets of ratings indicate that H9 is the most impor-
H9; and (ii) H8 is of lower initial technical importance but high final
tant HOW, followed by H15 and then by H11 and H6, and that H7 is
technical importance than H7. Since technical competitive priority
the least important HOW, preceded by H38 and H12.
ratings (zn’s) do not vary too much, these two differences are mainly
These crisp and fuzzy final technical ratings of the HOWs are
shown in Table 10. In order to be comparable, they are both scaled
Table 10 to have maximum rating or upper limit of unity, which are also
Crisp and fuzzy final technical ratings of the 15 HOWs. shown in Table 10. From these scaled ratings we can see again that,
HOWs Final technical ratings Scaled final technical ratings although the crisp and fuzzy ratings exhibit an identical trend,
crisp ratings always tend to be close to the upper limits of the cor-
Crisp (sn) Fuzzy (sfn ) Crisp (sn) Fuzzy (sfn )
responding fuzzy ratings. This shows that fuzzy ratings are more
H1 0.124113 [0.066, 0.227] 0.854326 [0.252, 0.868] representative of the possible variations of the HOWs’ technical
H2 0.122475 [0.065, 0.223] 0.843047 [0.249, 0.853] importance, which would make the technical improvement more
H3 0.122697 [0.066, 0.220] 0.844581 [0.252, 0.846]
H4 0.128393 [0.068, 0.233] 0.883786 [0.261, 0.897]
flexible and the design process more feasible.
H5 0.142687 [0.077, 0.257] 0.982174 [0.294, 0.987] The above nine steps complete the HOE process for improving
H6 0.134406 [0.072, 0.243] 0.925178 [0.276, 0.930] the company’s improvement trends on excellence. The corre-
H7 0.112604 [0.059, 0.207] 0.77510 [0.225, 0.794] sponding tables of results, after appropriate arrangement, can
H8 0.119689 [0.062, 0.221] 0.82387 [0.238, 0.847]
form an HOE like Fig. 2 which links organization needs to tech-
H9 0.145276 [0.079, 0.261] 1.00000 [0.301, 1.000]
H10 0.128473 [0.069, 0.232] 0.884335 [0.264, 0.889] nical considerations and exhibits all the relevant elements and
H11 0.138175 [0.074, 0.251] 0.951116 [0.282, 0.959] their relationships. As a result of this HOE model, it is concluded
H12 0.121494 [0.065, 0.221] 0.836298 [0.247, 0.847] that H7 could be deleted from further consideration (in QFD’s
H13 0.133152 [0.072, 0.239] 0.916544 [0.276, 0.915] second phase, parts deployment) to save technical efforts with-
H14 0.132053 [0.071, 0.239] 0.908977 [0.271, 0.914]
H15 0.142942 [0.078, 0.256] 0.983933 [0.299, 0.980]
out decreasing organization satisfaction. If resource or budget
considerations require to further cut down the number of HOWs,

Table 11
Normalization and determine the percentages of the 15 HOWs.

HOWs Crisp (sn) Fuzzy (sfn)


Crisp weights Nor. Per. % Fuzzy weights Defuzzi. Nor. Per. %
H1 0.124113 0.063693 6.36926 [0.066,0.227] 0.135156 0.063825 6.38254
H2 0.122475 0.062852 6.28517 [0.065,0.223] 0.133169 0.062887 6.288714
H3 0.122697 0.062966 6.29661 [0.066,0.220] 0.13305 0.062831 6.283073
H4 0.128393 0.065889 6.58889 [0.068,0.233] 0.139687 0.065965 6.59649
H5 0.142687 0.073224 7.32241 [0.077,0.257] 0.154875 0.073137 7.313718
H6 0.134406 0.068975 6.89749 [0.072,0.243] 0.145948 0.068922 6.892164
H7 0.112604 0.057786 5.77861 [0.059,0.207] 0.122847 0.058013 5.801264
H8 0.119689 0.061422 6.1422 [0.062,0.221] 0.130701 0.061721 6.172143
H9 0.145276 0.074553 7.45531 [0.079,0.261] 0.157568 0.074409 7.440888
H10 0.128473 0.06593 6.59299 [0.069,0.232] 0.139532 0.065892 6.589179
H11 0.138175 0.070909 7.09086 [0.074,0.251] 0.150141 0.070902 7.090178
H12 0.121494 0.062349 6.23485 [0.065,0.221] 0.132208 0.062433 6.243329
H13 0.133152 0.068331 6.83312 [0.072,0.239] 0.144356 0.06817 6.816961
H14 0.132053 0.067767 6.7767 [0.071,0.239] 0.143414 0.067725 6.772494
H15 0.142942 0.073355 7.33552 [0.078,0.256] 0.154942 0.073169 7.316866
9644 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

H8; H12 and H2 form a good deleting order that will not signifi- many management tools, companies can improve their perfor-
cantly influence the fulfillment of the organization needs. And mances and then increase customer satisfaction and gain market
also, according to the Table 10, by use of these crisp and fuzzy shares. But for the organizations, that adopted excellence models
importances, the importance weighting of each management tool such as EFQM, to improve their performances, selection and choos-
can be computed. In this way, first of all, the crisp importance ing these management tools has been a big challenge in today’s dy-
values of management tools should be normalized, and then namic environment. This paper presents a systematic and
the percentages of them should be calculated. But for the fuzzy operational approach to HOE to help resolve this problem. This
importance values, at first, the defuzzification of the fuzzy values study has addressed the applicability of QFD in the organizational
of management tools should be determined, for this step has excellence context. More specifically, an original methodology has
done by the Facchinetti et al. (1998) approach, and then the nor- been proposed and adopted to rank viable EFQM excellence criteria
malization and determine the percentage of them, should be and the management tools a firm can undertake to improve excel-
done (see Table 11). lence performances.
The methodology developed could be rightly considered as a
Hypothesize 2: The ranking results of effective management useful tool for selecting the most efficient and effective manage-
tools on setting EFQM excellence model in an organization ment tools leverages to reach organizational excellence. We pro-
(the research case) are the same in crisp and fuzzy approaches. pose a 9-step HOE model, which is basically a QFD model, to
unify the HOE process and a few 9-point scales to unify the mea-
In order to be comparable, the crisp and fuzzy final importance surements in HOE to avoid arbitrariness and incomparability.
ratings are tested by spearman correlation coefficient. According to We especially address the various ‘‘voices’’ in the HOE process
the Table 11, with percentage results of crisp and fuzzy ranking, and suggest the use of symmetrical triangular fuzzy numbers
the spearman coefficient correlation for these tow type of data is (STFNs) to reflect the vagueness in expert’s linguistic assessments.
0.993 and there is a very strong positive correlation between fuzzy Furthermore, we employ the quantitative entropy method to con-
and crisp importance ranking. So the first hypothesis of this re- duct competitive analysis and derive competitive priority ratings.
search that maintains: ‘‘The ranking results of effective manage- All information required, computations involved and feasible
ment tools on setting EFQM excellence model in an organization methods are clearly indicated to give an applicable framework
(the research case) are the same in crisp and fuzzy approaches.’’, for practitioners to perform HOE analysis without confusions and
were supported. difficulties. To fully illustrate our proposed HOE model, we present
an automotive company example that involves five organizational
needs for excellence (EFQM enabler criteria), 15 technical attri-
5. Conclusions butes for excellence (management tools) and seven competitor
companies.
Using the useful management tools that are relevant to the In a similar manner, the weighted importance of management
organization’s needs for excellence has become so important. By tools allows the firm to identify the key factors of intervention in
choosing and applying the best management tools among too order to improve the perceived excellence. As an example, pro-

Inventory management (6.38)

Total quality management (6.29)


Leadership (9.00)
Human resources management (6.28)

Knowledge management (6.60)

Strategy (11.47) Technology management (7.31)

Information management (6.89)

Energy management (5.80)

Employee (16.13) Project management (6.17)

Financial management (7.44)

Change management (6.59)


Resources (29.90)
Customer relationship management (7.09)

Supply chain management (6.24)

Business process management (6.82)


Processes (33.50)
Strategic management (6.77)

Production management (7.32)

Fig. 4. The basis for programming and organizational resources allocation for excellence.
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9645

cesses emerges in step 4 as the most important factor from experts’ This means that, for example, the membership value or ‘‘possibility’’
point of view, and it should be considered as the key excellence cri- that leadership is assigned a number 9 is l[8,10](9) = 1, the ‘‘possibil-
terion to improve the performance of the organization. In order to ity’’ that leadership is assigned a number 8.5 or 9.5 is l[8,10](8.5) =
assess and rank viable management tools, in the approach pro- 0.5 or l[8,10](9.5) = 0.5. So assigning leadership a number 8.5 or
posed we have introduced entropy method, which considers the 9.5 is acceptable or ‘‘possible’’ to the degree of 50%. The basic arith-
competition of implementation for each ‘‘what’’ and ‘‘how’’. The metic rules for STFNs are as follows:
entropy can be directly adopted as a synthesis parameter to select
the most suitable EFQM enabler criteria and management tools
that have the most competitive importance to implement. Accord- Addition : ½a; b þ ½c; d ¼ ½a þ c; b þ d ðA:3Þ
ing to step 3 and step 8, it is considered that WHATs and HOWs,
both of them have the same competitive importance. Since per- Subtraction : ½a; b  ½c; d ¼ ½a  c; b  d ðA:4Þ
sonal judgments are required when building the HOE, fuzzy logic
has been adopted as a useful tool. Through fuzzy logic linguistic Scalar multiplication : k  ½a; b ¼ ½ka; kb k > 0 ðA:5Þ
judgments an expert gives to weights, relationships and correla-
tions have been appropriately translated into triangular a fuzzy
Multiplication : ½a; b  ½c; d  ½ac; bd; a0c0 ðA:6Þ
number. Moreover, fuzzy logic has allowed to cope well with
Division : ½a; b ½c; d  ½a=c; b=d; a  0; c > 0 ðA:7Þ
uncertainties and incomplete understanding of the relationships
between ‘‘WHATs’’ and between ‘‘HOWs’’ and ‘‘WHATs’’. In addi-
For any two STFNs, E e 1 = [a, b] and, E
e 2 = [c, d], if one interval is not
tion, fuzzy logic becomes fundamental to dealing with several
parameters that seem difficult to express in a quantitative mea- strictly contained by another then their ranking order can be easily
sure. As an example, detailed information about relationships be- and intuitively determined. That is
tween management tools and EFQM excellence criteria are
If d > b and c P a, or d P b and c > b, then e E2 > e
E 1 , where ‘‘>’’
usually not available, while linguistic judgments on them can be means ‘‘is more importance or preferred than’’.
easily obtained.
If a = c, b = d, then e
E2 ¼ e E1
By use of the fuzzy importance percents ranking of EFQM crite-
ria and management tools, from Tables 6 and 11, the basis for pro- But if one interval is strictly contained by another, i.e., if d < b and
gramming and allocating of organization resources for the c > a, or d > b and c < a, then the ranking problem becomes complex
improving of excellence performances, can provided. It is shown and many possibilities may occur. For more details about fuzzy set
in Fig. 4. The methodology proposed does not deal with the practi- theory, STFNs and fuzzy ranking methods, see Zimmermann
cal implementation of management tools. Future work may be (1987).
thus directed to extend a similar QFD approach from a strategic le-
vel to tactical and operational ones. Specially, future work can ex- A.2. Fuzzy AHP
tend sub-set of each management tools in to the other phases of
QFD approach. To apply the process depending on this hierarchy, according to
the method of Chang’s (1996) extent analysis, each criterion is ta-
ken and extent analysis for each criterion, gi; is performed on,
Appendix A
respectively. Therefore, m extent analysis values for each criterion
can be obtained by using following notation (Kahraman, Cebeci,
A.1. Fuzzy methods
& Ruan, 2004):
Fuzzy set theory was developed for solving problems in
which descriptions of objects are subjective, vague and impre- M1gi ; M 2gi ; M 3gi ; . . . ; M m
gi

cise, i.e., no boundaries for the objects can be well defined. Let
X = {x} be a traditional set of objects, called the universe. A fuzzy where gi is the goal set (i = 1, 2, 3, 4, 5, . . ., n) and all the Mjgi
set eE in X is characterized by a membership function leðxÞ that (j = 1, 2, 3, 4, 5, . . ., m) are triangular fuzzy numbers (TFNs). The
E
associates each object in X with a membership value in the steps of Chang’s analysis can be given as in the following.
interval [0, 1], indicating the degree of the object belonging to Step 1: The fuzzy synthetic extent value (Si) with respect to the
e
E. A fuzzy number is a special fuzzy set when the universe X ith criterion is defined as Eq. (A.8)
is the real line R1 : 1 < x < +1. A symmetrical triangular fuzzy " #1
number (STFN), denoted as e E ¼ ½0; 1, is a special fuzzy number X
m Xn X
m
Si ¼ M jgi  Mjgi ðA:8Þ
with the following symmetrical triangular type of membership
j1 i¼1 j¼1
function:
To obtain Eq. (A.9);
leE ðxÞ ¼ 1 j x  ðc þ aÞ=2 j =½ðc  aÞ=2; a  x  c ðA:1Þ
X
m
M jgi ðA:9Þ
STFN is widely used in practice to represent a fuzzy set or concept j¼1
e
E = ‘‘approximately b’’ where b = (a + c)/2. For example, if an EFQM
enabler criterion leadership is rated as having ‘‘very high’’ impor- Perform the ‘‘fuzzy addition operation’’ of m extent analysis values
tance by a decision maker, then traditionally we may assign leader- for a particular matrix given in Eq. (A.10) below, at the end step of
ship a number 9 using crisp scale. To capture the vagueness of the calculation, new (l, m, u) set is obtained and used for the next:
decision maker’s subjective assessment, we can according to the !
same scale assign leadership an STFN [8, 10] which means ‘‘approx- X
m X
m X
m X
m
M jgi ¼ lj mj uj ðA:10Þ
imately 9’’ and is represented by the following membership j¼1 j¼1 j¼1 j¼1
function:
where l is the lower limit value, m is the most promising value and u
l½8;10 ðxÞ ¼ 1 j x  9 j; 8  x  10: ðA:2Þ
is the upper limit value. And to obtain Eq. (A.11);
9646 S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647

" #1
Xn X
m forms much better than any other companies in terms of organiza-
M jgi ðA:11Þ tional needs for excellence Wm, then further improvement may not
i¼1 j¼1 be urgently needed and thus a lower priority could be assigned to
Perform the ‘‘fuzzy addition operation’’ of M jgi (j = 1, 2, 3, 4, 5, . . ., m) Wm. At the other extreme, if C1 performs much worse than many
values give as Eq. (A.12): other companies on Wm, then it may be difficult for C1 to build a
! competitive advantage within a short period of time. In both cases,
X
n X
m X
n X
n X
n
Wm could be assigned a lower priority rating. However, if most com-
Mjgi ¼ lj mj uj ðA:12Þ
panies perform quite similarly on Wm, not too much improvement
i¼1 j¼1 i¼1 i¼1 i¼1
effort from C1 may result in a better performance of its excellence
And then compute the inverse of the vector in Eqs. (A.12) and (A.13) and give C1 a unique competitive advantage. Thus a higher priority
is then obtained such that could be assigned to Wm. In particular, if all companies’ perfor-
" #1 " # mances on Wm are the same, it implies a great excellence opportu-
Xn X
m X
n X
n X
n
nity since any improvement would create a significant competitive
M jgi ¼ 1= ui ; 1= mi ; 1= li ðA:13Þ
i¼1 j¼1 i¼1 i¼1 i¼1
advantage. So the highest priority could be assigned to Wm. This ba-
sis of assigning priorities is interestingly related to the entropy con-
Step 2: The degree of possibility of cept in information theory. Entropy is a measure for the amount of
M2 = (l2, m2, u2) P M1 = (l1, m1, u1) is defined as Eq. (A.14): information (or uncertainty, variations) represented by a discrete
VðM 2  M 1 Þ ¼ sup½minðlM1 ðxÞ; lM2 ðyÞÞ ðA:14Þ probability distribution, p1, p2, . . ., pL:
Yx
X
1
And x and y are the values on the axis of membership function of EðW 1 Þ ¼ UL Pml lnðPml Þ ðA:19Þ
each criterion. This expression can be equivalently written as given l¼1

in Eq. (A.15) below: where UL = 1/ln(L) is a normalization constant to guarantee


8
0 6 E(p1, p2, . . ., pL) 6 1. Larger entropy or E(p1, p2, . . ., pL) value im-
<1
> if m2 P m1 ;
plies smaller variations among the pl’s and hence less information
VðM2 P M 1 Þ ¼ 0 if l2 P u1 ; ðA:15Þ
>
: l1 u2 contained in the distribution. For the mth row of the customer com-
Otherwise
ðm2 u2 Þðm1 u1 Þ parison matrix X corresponding to the organizational needs for
P
To compare M1 and M2; we need both the values of V(M2 P M1) excellence need W m ; xm1 ; xm2 ; . . . ; xml , let xm ¼ xml be the total
and V(M1 P M2): score with respect to Wm. Then according to (A.19), the normalized
Step 3: The degree possibility for a convex fuzzy number to be ratings pml = xml/xm for l = 1, 2, . . ., L can be viewed as the ‘‘probabil-
greater than k convex fuzzy numbers ity distribution’’ of Wm on the L companies with entropy as

M i ði ¼ 1; 2; 3; . . . ; kÞ can be defined by VðM  M 1 ; M2 ; . . . ; Mk Þ X


1
EðW 1 Þ ¼ UL Pml lnðPml Þ ¼ EðW 1 Þ
¼ V½ðM  M1 Þ&ðM  M 2 Þ&    &ðM  Mk Þ ¼ min VðM  M i Þ; l¼1
i ¼ 1; 2; . . . ; k X
1
¼ UL xml lnðxml =xm Þ ðA:20Þ
Assume that Eq. (A.16) is l¼1

l
d ðAi Þ ¼ min VðSi  Sk Þ ðA:16Þ It is clear that the larger the E(Wm) value, the less information con-
tained in Wm or smaller variations among the pml’s (or xml’s). If all
For k = 1, 2, 3, 4, 5, . . ., n; k – i. Then the weight vector is given by
companies’ performance ratings on Wm, xm1, xm2, . . ., xmL, are the
Eq. (A.17):
same, Wm has zero variations and E(Wm) achieves its maximum of
l l l
W l ¼ ðd ðA1 Þ; d ðA2 Þ; . . . ; d ðAn ÞÞT ðA:17Þ 1. So E(Wm) can be used to reflect the relative competitive advan-
tage in terms of the organizational needs for excellence Wm. All
where Ai (i = 1, 2, 3, 4, 5, 6, . . ., n) are n elements. these E(Wm) values, after normalization:
Step 4: Via normalization, the normalized weight vectors are gi- ,
ven in Eq. (A.18): X
M
em ¼ EðW m Þ EðW m Þ; m ¼ 1; 2; . . . ; M ðA:21Þ
W l ¼ ðdðA1 Þ; dðA2 Þ; . . . ; dðAn ÞÞT ðA:18Þ m¼1

where W is non-fuzzy numbers. em can be considered as the excellence competitive priority ratings
for company C1 on the M organizational needs for excellence, with a
A.3. Entropy method for competitive priority ratings larger em indicating higher competitive priority for the correspond-
ing Wm. For more on entropy and its applications (Chan, Kao, Ng, &
In our HOE model, step 3 is to obtain and analyze the following Wu, 1999).
excellence comparison matrix:
References

American Supplier Institute (1994). Quality function deployment (service QFD): 3-day
workshop. Dearborn, MI: ASI Press.
Bou-Liusar, J. C., Escring-Tena, A. B., Roca-Pluig, V., & Beltran-Martin, I. (2005). To
what extent do enablers explain results in the EFQM excellence model? An
empirical study. International Journal of Quality & Reliability Management, 22(4),
337–353.
Chan, L. K., Kao, H. P., Ng, A., & Wu, M. L. (1999). Rating the importance of customer
where xml is the performance of company C1’s on organizational needs in quality function deployment by fuzzy and entropy methods.
needs for excellence (ONE) Wm, perceived by the experts. Based International Journal of Production Research, 37(11), 499–518.
on this X information, the company C1 may set priorities on the M Chan, L. K., & Wu, M. L. (1998). Prioritizing the technical measures in quality
function deployment. Quality Engineering, 10(3), 467–479.
organizational needs for excellence in order to achieve a relative Chan, L.-K., & Wu, M.-L. (2005). A systematic approach to quality function
competitive advantage over other companies. If company C1 per- deployment with a full illustrative example. Omega, 33, 119–139.
S. Yousefie et al. / Expert Systems with Applications 38 (2011) 9633–9647 9647

Chang, D. Y. (1996). Applications of the extent analysis method of fuzzy AHP. Ignacio, J. (2005). Theoretical foundation of EFQM model. The resource based view.
European Journal of Operational Research, 95, 649–655. Total Quality Management, 16(1), 17–22.
Cohen, L. (1995). Quality function deployment: how to make QFD work for you. Kahraman, C., Cebeci, U., & Ruan, D. (2004). Multiattribute comparison of catering
Reading, MA: Addison-Wesley. service companies using fuzzy AHP: The case of Turkey. International Journal of
Darrell, R. (2007). Management tools and techniques. California Management Production Economics, 87, 171–184.
Reviews, 43(2). Leonard, D., & Aadam, R. M. C. (2002). The role of the business excellence model in
European Foundation for Quality Management (1999). EFQM model for business operational and strategic decision making. Management Decision, 40, 17–25.
excellence: company guidelines. Parkan, C., & Wu, M. L. (2000). Comparison of three modern multi-criteria decision
European Foundation for Quality Management (2000). EFQM advice booklets. making tools. International Journal of Systems Science, 31(4), 497–517.
Facchinetti, G., Ghiselli Ricci, R., & Muzioli, S. (1998). Note on ranking fuzzy Satty, T. L. (1980). The analytic hierarchy process. New York, NY: McGraw-Hill.
triangular numbers. International Journal of Intelligent Systems, 13, 613–622. Westlund, A. H. (2001). Measuring environmental impact on society in the EFQM
Ghobadian, A., & Woo, H. S. (1996). Characteristics, benefits and shortcomings of system. Total Quality Management, 12(1), 125–135.
four major quality awards. International Journal of Quality and Reliability Yoon, K., & Hwang, C. L. (1995). Multiple attribute decision making: An introduction.
Management, 13(2), 10–44. Thousand Oaks, CA: Sage.
Griffin, A., & Hauser, J. R. (1993). The voice of the customer. Marketing Science, 12(1), Zimmermann, H. J. (1987). Fuzzy set decision making and expert system. Boston:
1–27. Kluwer.
Hauser, J. R., & Clausing, D. (1988). The house of quality. Harvard Business Review,
66(3), 63–73.

You might also like