Tourism Management 32 (2011) 1381e1388

Contents lists available at ScienceDirect

Tourism Management
journal homepage: www.elsevier.com/locate/tourman

Using a modified grey relation method for improving airline service quality
James J.H. Liou a, *, Chao-Che Hsu b, Wen-Chien Yeh c, Rong-Ho Lin a
a
Department of Industrial Engineering and Management, National Taipei University of Technology, No. 1, Section 3, Chung-Hsiao East Road, Taipei 106, Taiwan
b
Department of Transportation Management, Tamkang University, 151 Ying-Chuan Road, Tamsui, Taipei 251, Taiwan
c
Department of Air Transportation, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County 338, Taiwan

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

Article history: This study applies a modified grey relation method to improve service quality among domestic airlines in
Received 25 May 2010 Taiwan. First, we replace the referential alternative (sequence) with an aspired alternative, which should
Accepted 19 January 2011 better reflect the reality of today’s competitive markets. Second, because the compared alternatives do
not usually have the same criteria/aspects, traditional methods are unsuitable to deal with them. Our
Keywords: model fixes this problem, allowing decision-makers to understand the gaps between alternatives and
Service quality
aspired levels in practice. Third, we develop a new ranking index to measure the airlines’ competitive-
Customers’ needs
ness in terms of service quality. To validate the effectiveness of our model, we conduct a large sample
Airline
Grey relation
survey. We also provide managerial improvements needed by each carrier to achieve the aspired level of
SERVQUAL customer satisfaction.
Multiple-criteria decision-making (MCDM)
2011 Elsevier Ltd. All rights reserved.

1. Introduction in multi-criteria problems. However, in practice, decision-makers

must often simultaneously evaluate the rate of progress achieved
In today’s competitive environment within the airline industry, for one or several alternatives with different criteria. They therefore
delivering high-quality service has become a marketing require- need to identify unimproved gaps for the alternatives before they
ment. However, with the global economic downturn, most airlines can improve them to achieve the minimum gaps. Traditional
are struggling just to survive. They are forced to cut costs and methods are unsuitable for ranking these unimproved gaps in the
services as much as possible. Therefore, it is imperative for airline alternatives, because each alternative has different criteria, and is
managers to determine what their customers do and do not want. compared with a referential alternative rather than an “aspired”
What airlines need are ways to keep the essential service items and alternative. The referential alternative is usually obtained from a set
minimize efforts spent on the less important service items while of existing alternatives but these do not necessarily satisfy
still maintaining passenger perceptions of airline service quality. It customers’ needs. This study thus proposes a modified grey relation
is thus the primary purpose of this paper to evaluate the service method for solving these problems and reducing the gaps.
level of Taiwan’s domestic airlines and to identify some of the larger Furthermore, based on the Technique for Order Preference by
gaps between what airlines provide and what the customer’s needs Similarity to an Ideal Solution (TOPSIS) developed by Chen and
(desires) are, so that airlines can prioritize their improvement Hwang (1992) and compromise solutions (Yu, 1973; Zeleny, 1982),
strategies. we develop a new ranking index that considers both the aspired
Many studies have been done to assess the service quality of and worst alternatives. The new index might be more reliable than
airlines using traditional statistical testing (Chen & Chang, 2005; the original model, which is based on referential alternatives only.
Gilbert & Wong, 2003; Gursoy, Chen, & Kim, 2005; Pakdil & The modified grey relation method assigns a multi-criteria grey
Aydin, 2007), while others have used multiple-criteria decision- grade based on the particular measure of closeness/similarity to the
making (MCDM) methods to accomplish the same goals (Chang & ideal/aspired level solution. The method is introduced as an
Yeh, 2002; Liou & Tzeng, 2007; Tsaur, Chang, & Yeh, 2002). Most applicable technique to implement within MCDM process.
MCDM methods use the same criteria to compare all alternatives With the help of the Civil Aviation Administration of Taiwan
based on synthesized scorings for ranking or selecting alternatives (CAAT) and four major domestic airlines, we formulated an evalu-
ation framework of service quality and conducted a large-scale
investigation acquiring data directly from passengers. After exam-
* Corresponding author. Tel.: þ886 2 27712171x2300; fax: þ886 2 27317168. ining the importance/weight and performance that passengers
E-mail address: jamesjhliou@gmail.com (J.J.H. Liou). place on each service-criterion for satisfying the desired level, we

0261-5177/$ e see front matter
2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tourman.2011.01.013
1382 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388

applied the modified grey relation method to look at the gaps There are a number of studies that focus on airline service
between passengers’ perceived performance and the aspired levels. quality. These include one by Gilbert and Wong (2003), who
Using the new ranking index, we compared the competitiveness of developed a 26-item questionnaire incorporating measures of
four airlines in terms of service quality. Finally, based on the anal- reliability, assurance, facilities, employees, flight patterns, cus-
ysis, we developed strategies to help each carrier to improve. tomization and responsiveness dimensions, comparing the differ-
The remainder of this paper is organized as follows: Section 2 ences in passengers’ expectations to the actual perceived airline
summarizes important research studies on airline service quality; service quality. Chen and Chang (2005) categorized airline service
Section 3 introduces our research methods; Section 4 describes our processes into ground and in-flight services. They then applied
empirical data; Section 5 describes the results of data analysis; and SERVQUAL to investigate the gap in service quality between
in Section 6 we offer some conclusions. customers’ expectations and management’s perceptions of these
expectations in the ground and flight stages. Pakdil and Aydin
2. Service quality in the airline industry (2007) proposed a 35-item questionnaire based on SERVQUAL to
measure service quality in the airline industry. They used factor
Airline service may not be physically complex like high-tech analysis to extract eight elements from the questionnaire:
products, but it still represents a very intricate mix of intangibles employees, tangibles, responsiveness, reliability and assurance,
(Mazanec, 1995). Service quality is a composite of various interac- flight patterns, availability, image and empathy.
tions between a passenger and airline employees, as well as Although most studies have used traditional statistical tech-
anything that is likely to influence passengers’ perceptions, such as niques to test some hypotheses, others have applied MCDM models
an airline’s image (Gursoy et al., 2005). Service quality can be to investigate an airline’s integrated service level and to make
defined as a consumer’s overall impression of the relative efficiency suggestions for improvement. Tsaur et al. (2002) used the MCDM
of the organization and its services (Park, Robertson, & Wu, 2004). model and concluded that the most important attributes of airline
It can also be defined as a chain of services in which the entire service are courtesy, safety and comfort. Based on the fact that
service delivery is divided into a series of processes (Chen & Chang, airline service quality is heterogeneous, intangible and inseparable,
2005). In brief, there is no universal and exact definition of service as mentioned above, Chang and Yeh (2002) used fuzzy multi-
quality; service quality thus means different things to different criteria analysis to describe the ambiguity between criterion
industries. This seems to suggest that the concept of service quality weights and airline service-quality ratings. They proposed a list of
is context-dependent. Its measurements should thus reflect the 15 attributes to measure airline service quality, with flight safety
operational environment being investigated. being the most important. Liou and Tzeng (2007) applied a non-
Understanding exactly what customers expect is the most additive fuzzy integral model to investigate the service quality of
crucial step in defining and delivering high-quality service eight international airlines. Most of the previous MCDM models
(Zeithaml, Parasuraman, & Berry, 1990). Chang and Yeh (2002) have focused on ranking and selecting from a set of alternatives
argue that quality in airline services is difficult to describe and based on the synthesized scorings for each alternative compared
measure due to its heterogeneity, intangibility and inseparability. with the referenced or relatively best alternative. Our modified grey
Only customers can truly define service quality in the airline relation method allows for the solving of MCDM problems with
industry. Gronroos (1984) suggests that customer perceptions of conflicting and non-commensurable criteria and provides a solu-
service quality can be divided into technical quality and functional tion that is the closest to the aspired level and farthest from the
quality. Technical quality focuses on the quality evaluation of the worst level. Our model revises the original SERVQUAL and IPA
core service that the customer receives from the seller. Functional which includes weights in gap analysis and applies performance-
quality involves the evaluation of the service delivery process, only with a setting aspired level to measure service quality. This
which reflects customers’ experiences of service quality. study also replaces the use of a fixed, common number of criteria
Parasuraman, Zeithaml, and Berry (1985) developed a frame- for all alternatives and proposes a method for decision-makers to
work that defines service quality as the degree and direction of rank the unimproved gaps.
discrepancy between customers’ expectations and perceptions.
Their model was further developed and became known as 3. The modified grey relation method
SERVQUAL, which contains 5 dimensions with 22 attributes of
service quality (Parasuraman, Zeithaml, & Berry, 1988). However, Grey theory was developed by Deng (1982) to study the degree
the traditional gap analysis does not consider the weight of crite- of relationship between various criteria in an MCDM problem. Grey
rion and the overall performance value. Furthermore there is little if theory is a very useful mathematical tool for dealing with system
any theoretical or empirical evidence supporting the relevance of analysis with limited information. In comparison with conventional
an expectationseperformance gap as the basis for measuring methods, which require massive amounts of data, grey analysis
service quality (Carman, 1990). Cronin and Taylor (1992) suggested possesses the following advantages (Deng, 1982; Shi, 1990): (1) it
that service quality can be measured using a performance-only uses simple and easy calculations; (2) it requires a only reasonable
index as opposed to the gap-based SERVQUAL scale. Brady, Cronin, amount of sample data; (3) it does not require a typical distribution
and Brand (2002) argue the superiority of a performance-only of samples; (4) the quantified outcomes from the grey relational
approach for the measurement of service quality. Importance- grade do not result in conclusions that are contradictory with the
performance analysis (IPA) (Martilla & James, 1977) is another qualitative analysis; (5) the grey relational grade model is a transfer
widely used analytical technique that yields prescriptions for the functional model, which is very suitable and effective in dealing
management of customer satisfaction. However, Oh (2001) has with discrete data.
suggested that more empirical evidence is needed on the actual
contribution of IPA to strategic decisions in industries. Matzler, 3.1. Four axioms of the grey relational grade
Bailom, Hinterhuber, Renzl, and Pichler (2004) show empirically
that the managerial implications derived from IPA are misleading We assume that the alternatives are denoted as a1, a2,.,ai,.,am
and the method needs to be revised. We therefore develop a model and are assessed by criteria n1, n2,.,ni,.,nm, respectively. We
using a performance-based measure to evaluate service quality and denote wi(j) as the weight/importance of criterion j for alternative
consider importance based on different criterion weights. ai; ai(j) is the performance value for alternative ai at criterion j. Let
 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388 1383

A ¼ fai ji ¼ 0; 1; .; mg be an alternative set for the grey relation, Table 1

where ao ˛A is the referential sequence, and ai ˛ A (i ¼ 1, 2,., m) is The performance table for the modified grey relation model.

the comparative sequence. If the grey relational coefficient in ai(j)

corresponding to ao(j) is g(ao(j), ai(j)), then the grey relational grade
of ai corresponding to ao, g(ao, ai), must satisfy the following four
axioms:

a. Norm interval

0 < gðao ; ai Þ  1; ci

gðao ; ai Þ ¼ 1 iff ao ¼ ai

gðao ; ai Þ ¼ 0 iff ao ; ai ˛f;

where f is the empty set;

   
b. Duality symmetric

 min mina* ðjÞ  ai ðjÞ þ z max maxa* ðjÞ  ai ðjÞ
i j i j
g a* ðjÞ; ai ðjÞ ¼   ;
x; y˛X0gðx; yÞ ¼ gðy; xÞ iff X ¼ fx; yg; ja* ðjÞ  ai ðjÞj þ z max maxa* ðjÞ  ai ðjÞ
i j

(1)
c. Wholeness
where z˛½0; 1 is the distinguished coefficient and represents the
gðao ; ai Þ s gðai ; ao Þ iff X ¼ fxi ji ¼ 0; 1; 2; .; ng; n > 2; significance of max max ja* ðjÞ  ai ðjÞj , whose value is generally set
i j
often at z ¼ 0:5 (Guo, 1985). From the above definition, we can define the
weight grey gap between ai and a* with respect to criterion j as
d. Approachability h
i
gi ðjÞ ¼ wi ðjÞ  1  g a* ðjÞ; ai ðjÞ : (2)
gðao ðjÞ; ai ðjÞÞ decreases with increasingjao ðjÞ; ai ðjÞj:
Also, the grade of grey relation (larger is better) is as follows:
If we let the criterion j be the X-axis and the alternative ai(j) be
the Y-axis, then we can draw m þ 1 lines from the ao to am
 X
ni

sequences. The degree of relation among the various comparative g a* ; ai ¼ wi ðjÞg a* ðjÞ; ai ðjÞ : (3)
and referential sequences can be determined by the degree of j¼1
similarity between the lines. Based on the geometry in conjunction
Similarly, we can obtain the grey relational coefficient in ai
with the four axioms, the requirements, and a quantified model of
corresponding to the tolerable (worst) level a, g(a(j), ai(j)) and
grey relational grade can be identified (Tzeng, Chang, Lin, & Hung,
the grade of grey relation (smaller is better) as
2002).
However, the referential sequence ao is obtained from the    

 min mina ðjÞ  ai ðjÞ þ z max maxa ðjÞ  ai ðjÞ
existing set A and its performance at criterion j is decided by the i j i j
g a ðjÞ;ai ðjÞ ¼   ;
best value of criterion j within existing alternatives ai. In the real ja ðjÞ  ai ðjÞj þ z max maxa ðjÞ  ai ðjÞ
world, however, each alternative is evaluated according to its own- i j

criteria, an ideal point, as in the original grey relation method, and (4)
ðao ðjÞ ¼ max ai ðjÞÞ cannot be set. Therefore, the benefit or cost
i
must be reset according to the expectation of the decision-makers
 X
ni

(passengers) for each criterion of each alternative. We call the best g a ; ai ¼ wi ðjÞg a ðjÞ; ai ðjÞ : (5)
a*(j) the aspired level and the worst a(j) the tolerable level. These j¼1
functions are expressed as follows:
After we have obtained the grade of grey relation for both
a*(j) ¼ aspired_a(j) (or a*(j) ¼ aspired_level),
aspired and tolerable levels, we define the ranking index as in
a(j) ¼ tolerable_a(j), (or a(j) ¼ tolerable_level).
Eq. (6), which was extracted from the ideas of TOPSIS (Chen &
To simplify our questionnaire, the values are set in our case
Hwang, 1992) and compromise solutions (Yu, 1973; Zeleny, 1982).
where 5 and 1 for best/aspired and worst/tolerable levels, respec-
 
tively. According to the new definitions for aspired to and tolerable g a* ; ai
Ri ¼   : (6)
levels, we can rewrite the performance table for the modified grey g a ; ai
relation method as follows (see Table 1):where ni is the number of
criteria in each alternative ai, because each alternative has its own
assessing criteria. 4. The data for the empirical case

Since the completion of the high-speed railroad, domestic

3.2. Quantified model of the grey relational grade airlines in Taiwan have faced strong market challenges. At first,
airlines tried to reduce prices to attract more customers, but they
According to the four axioms and the definitions for aspired to soon realized that this was a no-win situation. Service quality is the
and tolerable levels, the modified grey relational coefficient in ai fundamental element needed to survive in this highly competitive
corresponding to aspired level a*, g(a*(j), ai(j)), can be derived from domestic market. Supported by the CAAT and four major domestic
Eq. (1) as follows (Deng, 1989): airlines, we surveyed passengers to obtain their perceptions of
1384 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388

airline services. The data collection and analysis procedures are Table 2
illustrated below. Passenger profiles.

Attributes/distribution Sample number Frequency (%)

4.1. Questionnaire design Gender
Male 3426 61.7
Although SERVQUAL has been widely used to measure service Female 2127 38.3
quality in a variety of industries, no two providers of service are Age
exactly alike (Gilbert & Wong, 2003). We find that we needed to 20 or younger 291 5.2
adapt SERVQUAL so that, in actuality, it serves only as a framework for 21e30 1322 24.0
31e40 1357 24.4
this study. The questionnaire design took place in several steps. First, 41e50 1294 23.3
the service items from SERVQUAL and a Gallup survey delegated by 51e60 967 17.4
the CAAT in 2000 were taken into consideration. Even though 61 or older 322 5.7
SERVQUAL measures general quality attributes for service industries, Occupation
it does not include specific attributes that reflect the specific opera- Government employee 1386 24.0
tional environment that is being investigated. Therefore, we Private-sector employee 1170 21.1
proposed a 32-item questionnaire that included airline service Student 384 6.9
Private business 555 10.0
quality dimensions consistent with the SERVQUAL model and with
Management 547 9.8
the Gallup survey. We met with four customer-service managers of Others 1511 28.2
domestic airlines and officers of CAAT who had experience in the
Education
Gallup survey of 2000 to refine the questions. Throughout a 4-h Junior high or below 519 9.3
brainstorming session, these experts deleted and added questions Senior high 1447 26.1
from the original 32 items, ending up with 30 service attributes. College 2851 51.3
During this discussion process, the items of safety and reliability were Graduate school 736 13.7
deleted by airline managers and CAAT experts, although they have
been mentioned in other studies (e.g., Gilbert & Wong, 2003; Tsaur
many times the passenger flies with the airline per month, class of the
et al., 2002). This is partly because safety improvement items
seat, the purpose of the flight, and what booking and ticketing
include many aspects (e.g., maintenance, training, culture and oper-
channels they used. Part 2 deals with airline service attributes.
ation) that are difficult for passengers to understand or evaluate.
Respondents were asked to indicate the importance of each attribute
Passengers are very easily influenced by short-term events or media.
using a 5-point Likert scale with anchors of “1 ¼ least important” to
Similarity, reliability is also a long-term consideration that is not easy
“5 ¼ most important.” The performance measure for each attribute
to observe just from passengers’ flying experience. The CAAT has
was rated using satisfaction levels on a 5-point Likert scale where
long-term statistical data about the reliability of each airline. The
“1 ¼ strongly dissatisfied” and “5 ¼ strongly satisfied.” The last part of
purpose of this study is to focus on the passenger’s view point of
the questionnaire collected demographic information, such as sex,
service process items, so these items were eliminated. The refined
age, education and occupation.
questionnaires were pre-tested by 45 passengers. The results from
the pre-test were used by the experts to revise the questionnaire to
include 8 dimensions with 28 service attributes. The content validity 4.3. Results of the survey
of the questionnaire was deemed adequate. The questions were
adapted to reflect industry circumstances in Taiwan and specific We distributed 25,334 questionnaires and received 5598,
passenger-perceived service contexts. It should be noted here that a return rate of 22%. Eliminating those samples that contained
most domestic routes are simultaneously covered by three airlines incomplete answers, we obtained 5553 useful samples. The useful
(A1, A2 and A3). Therefore, any variations in service quality between questionnaires completed were 2917 for A1, 1303 for A2, 1056 for A3,
the three airlines due to the routes should be very minor. The biggest and 277 for A4. The distribution matches the domestic marketing
difference is for A4 which mainly operates between some small
outlying islands using small airplanes. Some of the services items are Table 3
thus not applicable for A4 due to short flight time and limited space in Passenger flight information.
these small aircraft. However, our proposed model can handle situ- Attributes/distribution Sample number Frequency (%)
ations like this where the alternatives have different service items. Number of flights per month
1 or less 3761 67.7
4.2. Primary survey 2e4 1469 26.5
5e7 215 3.9
8 or more 108 1.9
To conduct the primary survey, we randomly picked one weekday
and one weekend to target all passengers of various backgrounds who Booking channel
used the four domestic airlines. To preserve confidentiality, the four Telephone 293 5.3
Internet 1326 23.9
airlines are referred to as A1, A2, A3 and A4. We trained graduate Travel agency 1357 24.4
students to administer the surveys, who were then sent to 16 Airline counter 1295 23.3
domestic airports in Taiwan. Our coworkers distributed a question- Others 1282 23.1
naire and a pen to each passenger at each boarding gate of the 16 Seat class
domestic airports. Others collected their answers at the exit doors of Economy 5418 97.6
the 16 airports after the baggage claim point. In our primary survey, Business 135 2.4
we approached all passengers taking flights from the four airlines in Purpose of travel
question and asked them for three types of data: (1) information on Business 2078 37.4
their flights, (2) their importance and satisfaction level regarding each Visiting friends/relatives 1552 27.9
Tourism 1701 30.6
service attribute, and (3) their personal profiles. Part 1 of the ques-
Other 222 4.1
tionnaire gathered flight information, such as the airline name, how
 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388 1385

Table 4
Importance and satisfaction for service criteria of economy class passengers.

Dimensions/criteria A1 (n ¼ 2842) A2 (n ¼ 1269) A3 (n ¼ 1030) A4 (n ¼ 277)

Imp. Sat. Imp. Sat. Imp. Sat. Imp. Sat.

Booking service
Convenience of booking (1) 3.72 3.82 3.87 3.86 3.93 3.91 3.77 3.76
Promptness of booking (2) 3.62 3.83 3.81 3.88 3.94 3.89 3.77 3.73
Courtesy of booking employee (3) 3.91 3.85 4.00 3.92 4.01 3.91 3.87 3.77

Ticketing service
Convenience of buying ticket (4) 3.92 3.84 3.95 3.89 3.99 3.90 3.88 3.73
Promptness of buying ticket (5) 3.89 3.85 3.94 3.92 4.01 3.92 3.88 3.72
Courtesy of selling employee (6) 3.94 3.87 3.98 3.92 4.02 3.92 3.97 3.80

Check-in
Convenient check-in (7) 3.96 3.88 3.99 3.91 4.05 3.95 3.94 3.75
Efficient check-in (8) 3.95 3.89 3.94 3.93 4.03 3.92 3.92 3.74
Courtesy of check-in employee (9) 3.95 3.88 4.00 3.94 4.00 3.94 3.94 3.76
Check-in information (10) 3.94 3.87 3.97 3.9 3.94 3.89 3.90 3.75

Baggage handling
Convenience of baggage handling (11) 4.01 3.74 4.01 3.74 3.95 3.78 3.96 3.68
Courtesy of baggage handling employee (12) 3.91 3.72 3.96 3.75 3.81 3.75 3.89 3.68

Boarding process
Clarity of announcement (13) 3.91 3.85 3.92 3.88 3.93 3.88 3.95 3.73
Promptness of ID check (14) 3.98 3.85 4.01 3.86 4.00 3.85 4.00 3.72
Courtesy of boarding employee (15) 3.97 3.85 4.01 3.91 4.03 3.86 3.97 3.72

Cabin service
Cabin safety demonstration (16) 4.01 3.84 4.02 3.90 4.05 3.86 3.89 3.69
Variety of newspapers and magazines (17) 3.76 3.75 3.77 3.77 3.81 3.75 # #
Courtesy of flight attendants (18) 4.03 3.87 4.07 3.93 4.07 3.89 # #
Flight attendants’ willingness to help (19) 4.02 3.86 4.01 3.88 4.02 3.88 # #
Clean and comfortable interior (20) 3.99 3.88 4.03 3.91 4.13 3.92 3.75 3.68
In-flight facilities (21) 3.98 3.84 3.98 3.87 4.05 3.87 3.78 3.65
Captain’s announcement (22) 3.94 3.83 3.96 3.86 4.02 3.86 3.69 3.66

Baggage claim
Convenient baggage claim (23) 3.66 3.64 3.66 3.70 3.70 3.65 3.76 3.61
Courtesy of baggage claim employee (24) 3.65 3.62 3.68 3.66 3.68 3.62 3.71 3.57

Responsiveness
Fair waiting-list call (25) 3.55 3.46 3.54 3.41 3.81 3.77 3.72 3.68
Handling of delayed flight (26) 3.53 3.47 3.52 3.40 3.69 3.65 3.63 3.48
Complaint handling (27) 3.77 3.76 3.71 3.64 3.96 3.92 3.77 3.45
Missing baggage handling (28) 3.76 3.75 3.69 3.56 3.89 3.87 3.68 3.65

Note: # Not applicable; Imp. ¼ importance; Sat. ¼ satisfaction.

share, as A4 has a relatively small fleet and mainly operates based on the modified grey relation method, we prioritize the gaps
between small outlying islands. The passenger profiles and flight for improvement of the four airlines. Finally, we analyze the
information are presented in Tables 2 and 3. different aspects of the airlines’ overall performance.
In terms of reliability, internal consistency methods were widely
used. The Cronbach’s alpha was used to evaluate internal consis-
5.1. Importance and satisfaction levels of each criterion
tency. Cronbach’s alpha is the average of all possible split-half
coefficients resulting from different ways of splitting the scale
Passengers’ opinions about the importance and satisfaction
items, and a value of 0.6 or less generally indicates unsatisfactory
levels of the four airlines for each criterion are given in Table 4. The
consistency reliability (Malhotra, Hall, Shaw, & Crip, 1996). In this
criteria for cabin service are considered to be the most important
study, the Cronbach’s alphas were found to be 0.98 for importance-
factor related to service quality (3.94e4.13), with the exception of
related items and 0.99 for satisfaction-related items. According the
the “variety of newspapers and magazines” (3.76e3.81). This
results in Table 3, almost 98% of the respondents travelled in
should not be difficult to understand, because cabin service
economy class. It can be expected that these travellers have
occupies more of a passenger’s travelling time than other aspects of
different importance values and satisfaction than do business class
service. Other services are relatively basic in the domestic airline
travellers. Therefore, we removed the business class data from the
market. Conversely, respondents rated “baggage claim” as the least
sample and calculated the data, focussing only on economy class
important attribute of service. This is reasonable because most
passenger requirements only in the following analysis.
domestic passengers are travelling on business trips and thus do
not have large bags to claim. Among the attributes of cabin service,
5. Results of empirical analysis “courtesy of flight attendants” and “flight attendants’ willingness to
help” are the two attributes that passengers rated most important.
Based on the data collected as described above, we first discuss This suggests that airlines should concentrate on these two factors
the importance and satisfaction levels for each criterion. Then, as the foundation of a good service rating. The behaviour of
1386 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388

Table 5
The weighted grey gap of four airlines for each criterion.

A1 A2 A3 A4

Gap (102) Rank Gap (102) Rank Gap (102) Rank Gap (102) Rank
Booking service
Convenience of booking (1) 0.226 10 0.165 12 0.075 21 0.366 23
Promptness of booking (2) 0.204 14 0.127 16 0.112 17 0.417 18
Courtesy of booking employee (3) 0.185 18 0.059 23 0.077 20 0.358 24

Ticketing service
Convenience of buying ticket (4) 0.203 15 0.114 17 0.095 19 0.430 17
Promptness of buying ticket (5) 0.184 20 0.057 25 0.058 23 0.447 15
Courtesy of selling employee (6) 0.151 23 0.058 24 0.058 23 0.310 25

Check-in
Convenient check-in (7) 0.133 26 0.078 20 0.000 28 0.400 20
Efficient check-in (8) 0.115 28 0.039 27 0.057 25 0.416 19
Courtesy of check-in employee (9) 0.133 27 0.020 28 0.020 27 0.382 22
Check-in information (10) 0.151 23 0.100 18 0.112 17 0.396 21

Baggage handling
Convenience of baggage handling (11) 0.378 6 0.375 7 0.301 7 0.525 7
Courtesy of baggage handling employee (12) 0.400 5 0.354 8 0.336 4 0.516 9

Boarding process
Clarity of announcement (13) 0.185 18 0.131 15 0.130 15 0.437 16
Promptness of ID check (14) 0.189 16 0.170 10 0.186 8 0.461 13
Courtesy of boarding employee (15) 0.188 17 0.078 20 0.169 10 0.457 14

Cabin service
Cabin safety demonstration (16) 0.208 12 0.097 19 0.170 9 0.499 10
Variety of newspapers and magazines (17) 0.339 7 0.307 9 0.336 4 # #
Courtesy of flight attendants (18) 0.155 22 0.040 26 0.116 16 # #
Flight attendants’ willingness to help (19) 0.172 21 0.134 14 0.133 14 # #
Clean and comfortable interior (20) 0.134 25 0.078 20 0.060 22 0.498 11
In-flight facilities (21) 0.206 13 0.151 13 0.152 12 0.550 5
Captain’s announcement (22) 0.222 11 0.169 11 0.169 10 0.521 8

Baggage claim
Convenient baggage claim (23) 0.485 4 0.400 6 0.467 2 0.608 4
Courtesy of baggage claim employee (24) 0.510 3 0.458 5 0.504 1 0.659 3

Responsiveness
Fair waiting-list call (25) 0.687 1 0.734 2 0.306 6 0.497 12
Handling of delayed flight (26) 0.672 2 0.740 1 0.466 3 0.766 2
Complaint handling (27) 0.324 9 0.489 4 0.057 25 0.836 1
Missing baggage handling (28) 0.339 7 0.589 3 0.146 13 0.535 6

employees plays an important role in customers’ expectations performance of other employees. If we consider the negative
about the service quality. This suggests that most air travellers are attributes in greater detail, we find that airlines can enhance levels
still concerned with face-to-face interactions. Therefore, airlines of satisfaction by better handling unusual cases. The results also
should educate employees about the importance of their attitude indicate whose efforts are appreciated by customers and whose
towards service quality. The results also suggest that providing attitudes still need to be improved. Another important observation
a wide selection of newspapers and magazines is a nonessential is that the average rating for baggage handling (3.68e3.78) is
service, and that this is an area where airlines can cut costs without obviously higher than that for baggage claim (3.57e3.70). On
sacrificing passenger perceptions of service quality. domestic flights, baggage handling at check-in is usually done at
The levels of passenger satisfaction with service attributes the counter by airline employees, while the baggage claim of some
ranged between 3.40 and 3.95. Generally speaking, the reservation, airlines is managed by ground-handling companies. This result
ticketing, check-in and boarding processes received higher service indicates that airline managers should request that their
satisfaction levels (3.72e3.95), while baggage claim, complaints outsourcing providers (ground-service companies) handle baggage
and handling of delays showed lower satisfaction levels carefully. Otherwise, passengers will form a negative image of the
(3.40e3.77). The results indicate that the service attributes recog- airline, as most passengers do not differentiate the airline from the
nized by passengers are those that are highly computerized, while ground-handling company.
those that require personal contact received lower satisfaction
ratings. Although most service attributes associated with personal 5.2. Weighted grey gap analysis
contact received relatively low ratings, the courtesy shown by
check-in employees was the only exception, obtaining the highest According to the procedures described in Section 3 and in
levels of satisfaction (3.76e3.94) among all service attributes. This Table 1, we can combine the passengers’ importance (weighting)
interesting result indicates that passengers are happy with the and satisfaction levels (performance) to obtain the grey gaps with
frontline employees, but they do not seem to be satisfied with the respect to the aspired level. We also rank the gaps so that the airline
 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388 1387

can prioritize their strategies for improvement. For example, the passengers’ perception of service quality is dynamic and sensitive
first item in Table 5 shows that A1 airline needs to close the gap to recent major incidents, such as accidents or any negative news,
between the aspired to and satisfaction level of service related to which are not necessarily promptly reflected in the markets.
fair calls from the waiting-list. A1 has the largest market share for Although A1 is temporarily in the leading position in terms of
domestic flights, but passengers are not satisfied with its system of market share, there is still some room to improve services and to
wait-listing passengers. Its wait-list calling is completely depen- maintain its leading status. On the other hand, other airlines should
dent on the check-in staff and is based on passenger signatures. improve their side to catch up with competitors and attract more
Hence, travellers do not instantly know their updated status. A2 passengers.
needs to improve the handling of delayed flights. Due to the
changeable weather and crowded skies of Taiwan, delays seem to 6. Conclusion
have become very common. However, airlines can still provide
better service in this situation. For example, they could either This study uses a modified grey relational model to investigate
arrange to have a better place for passengers to wait or inform them the service quality of four major domestic Taiwan airlines. Rather
in advance so that they can change flights. As for A3, its largest gap than the original referential alternative, we use the aspired alter-
is in the category “courtesy of baggage claim employees.” The native in the analysis, which should be closer to the reality of
baggage handling at check-in is done at the counter by A3’s today’s competitive markets. The proposed method for weighted
employees, while the baggage claim is outsourced by a ground- grey gap analysis can be used to help airline managers realize the
handling company. This gap analysis indicates that A3 should relative performance of each service attribute with respect to
request that their outsourcing ground-service companies handle customers’ needs. We also provide some managerial recommen-
baggage more carefully. For A4, complaint handling has the largest dations for each airline in Section 5. Our results indicate that
service gap. This could be improved by providing some standard passengers deem the cabin service criteria to be the most important
operating procedures and instructing their employees how to factors affecting service quality (with the exception of the “variety
handle various types of abnormal situations. In contrast, service of newspapers and magazines”). The results suggest the impor-
items with small grey gaps indicate where airlines are already tance of training flight attendants to offer heartfelt courteous
doing well or where passengers think the items are less important, service. It also tells airlines how they can cut costs (i.e., not needing
indicating that airlines could reduce their efforts to satisfy to provide such a wide selection of newspapers and magazines)
passenger needs for these items. without sacrificing passenger perceptions of service quality. On the
other hand, respondents considered “baggage claim” as the least
important service attribute, largely because of the large amount of
5.3. Competitive service analysis based on the modified grey
business travellers. However, our weighted grey gap analysis shows
relational model
the largest gap for baggage claim. Therefore, airlines should focus
on improving their baggage claim process or ask their ground-
To examine the airlines’ relative competitive strengths in rela-
service personnel to handle baggage carefully to meet the needs of
tion to the service criteria identified by customers, a modified grey
non-business travellers. Generally speaking, those service attri-
relational analysis can be carried out based on the survey results.
butes that are highly computerized (e.g., reservation and ticketing)
Note that because the planes used by A4 are small and flight times
showed smaller grey gaps, while fair wait-listing calls, complaint
short, some service attributes are not applicable. That is, its criteria
handling, courtesy of baggage claim employees, and handling of
are different from other airlines. In the traditional MCDM methods,
delays, all of which need personal contact, showed larger grey gaps.
each alternative is assumed to have the same criteria for evaluation,
These results suggest that airlines could increase their level of
but our modified grey relational model can handle this situation.
computerization and save on labour costs in some processes, such
Furthermore, our model also considers that the weights of criteria
as waiting-list calls. Although most service attributes associated
for each alternative are different. Using our modified grey relation
with personal contact received relatively large gaps, the courtesy
model as described in Section 3 we obtain the results shown in
shown by check-in employees an exception. This is an advantage
Table 6. The grey grade and ranking index results are based on both
airlines should maintain. Check-in employees are valuable assets to
the aspired to and tolerable levels. It should be noted that in the
an airline. In addition, the ranking index derived from our modified
grey scale the grade for a tolerable level performance is smaller for
grey relational model provides a performance ranking of service
better performance.
quality indicative of the airline’s relative position in the market.
If we define A_B, meaning that A outranks B, then the ranking
This ranking index is more reliable than the other traditional
of service quality based on the Ri value for the surveyed airlines is as
MCDM methods for prioritizing alternatives since it considers both
follows: A3 _A2 _A1 _A4 . It is interesting to note that assessment
negative and positive ideal points (Opricovic & Tzeng, 2007).
of the service quality is not strongly reflected in the market share so
To summarize, this study explored the grey gaps and rankings of
that A1 airline ranks third in service quality but has the largest
four domestic airlines in terms of service criteria by using a modi-
market share. This result indicates that even though service quality
fied grey relational model modified to fit Taiwan’s domestic airline
has a big impact on airline markets, other factors, such as safety and
market, within which each airline has different service criteria. We
reliability, might also play an important role. Furthermore,
replaced the referential sequence used in the original model with
an aspired level, which should be more reasonable in the real
Table 6 world. We modified the original gap analysis using performance-
Results of the modified grey relational analysis. based measure and considering the importance of each criterion.
Grey grade for Grey grade for Ranking index
Moreover, we developed a new ranking index that considers both
aspired level tolerable level the aspired and tolerable levels in order to prioritize the alterna-
A1 0.925 0.932 0.993 (3) tives. The analysis further provides the surveyed airlines with
A2 0.937 0.929 1.008 (2) summaries of their weaknesses and suggestions for improvement.
A3 0.951 0.923 1.031 (1) Our contribution here is not only to point out the critical items on
A4 0.877 0.949 0.924 (4) which the airlines should focus, but also to identify the service
Parentheses ( ) denotes rankings. items where they are already doing well and for which they can
1388 J.J.H. Liou et al. / Tourism Management 32 (2011) 1381e1388

reduce their efforts without affecting the overall service level. The Gronroos, C. (1984). A service quality model and its marketing implications. Euro-
pean Journal of Marketing, 18(1), 36e44.
findings of this study are likely to significantly help airlines learn to
Guo, H. (1985). Identification coefficient of relational grade of grey systems. The
use grey gap analysis to identify their strengths and weaknesses. In Journal of Fuzzy Mathematics, 3(2), 55e58.
future, it would be worthwhile to apply the proposed model in Gursoy, D., Chen, M. H., & Kim, H. Y. (2005). The US airlines relative positioning
other markets to further show its usefulness. based on attributes of service quality. Tourism Management, 26(1), 57e67.
Liou, J. H. J., & Tzeng, G. H. (2007). A non-additive model for evaluating airline
Although the present study makes a significant contribution to service quality. Journal of Air Transport Management, 13(3), 131e138.
the literature, it still has some limitations. To reduce the complexity Malhotra, N. K., Hall, J., Shaw, M., & Crip, M. (1996). Marketing research an applied
of the questionnaire, we set the level of satisfaction of aspired orientation. Sydney: Prentice-Hall.
Martilla, J. A., & James, J. C. (1977). Importance-performance analyses. Journal of
alternative to be the same for each airline. A different value could be Marketing, 41(1), 77e79.
set for each airline which would be more reasonable in the real Matzler, K., Bailom, F., Hinterhuber, H. H., Renzl, B., & Pichler, J. (2004). The asym-
world due to various operating environments. Furthermore, the metric relationship between attribute-level performance and overall customer
satisfaction: a reconsideration of the importance-performance analysis. Indus-
conclusions drawn from our study are based on cross-sectional trial Marketing Management, 33(4), 271e277.
data, and thus we only took a snapshot of this model. A longitudinal Mazanec, J. A. (1995). Positioning analysis with self-organization maps: an
study could be carried out in future from which we would be able to exploratory study on luxury hotels. The Cornell Hotel and Restaurant Adminis-
tration Quarterly, 31(2), 84e91.
investigate our model across different time periods and make Oh, H. (2001). Revisiting importance-performance analysis. Tourism Management,
comparisons, thus providing more insight into the phenomenon of 22(6), 617e627.
service quality. Opricovic, S., & Tzeng, G. H. (2007). Extended VIKOR method in comparison with
outranking methods. European Journal of Operation Research, 178(2), 514e529.
Pakdil, F., & Aydin, O. (2007). Expectations and perceptions in airline service: an
analysis using weighted SERVQUAL scores. Journal of Air Transport Management,
References 13(4), 229e237.
Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1985). A conceptual model of service
Brady, M. K., Cronin, J. J., & Brand, R. R. (2002). Performance-only measurement of quality and its implications for future research. Journal of Marketing, 49(4),
service quality: a replication and extension. Journal of Business Research, 55(1), 41e50.
17e31. Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1988). SERVQUAL: a multiple-item
Carman, J. M. (1990). Consumer perceptions of service quality: an assessment of the scale for measuring customer perceptions of service quality. Journal of Retailing,
SERVQUAL dimensions. Journal of Retailing, 66(1), 33e55. 64(1), 12e40.
Chang, Y. H., & Yeh, C. H. (2002). A survey analysis of service quality for domestic Park, J. W., Robertson, R., & Wu, C. L. (2004). The effect of airline service quality on
airlines. European Journal of Operational Research, 139(1), 166e177. passengers’ behavioral intentions: a Korean case study. Journal of Air Transport
Chen, F. Y., & Chang, Y. H. (2005). Examining airline service quality from a process Management, 10(6), 435e439.
perspective. Journal of Air Transport Management, 11(2), 79e87. Shi, K. (1990). Grey relation theory and its applications. ISUMA90, U.S.A.
Chen, S. J., & Hwang, C. L. (1992). Fuzzy multiple attribute decision making: Methods Tsaur, S. H., Chang, T. Y., & Yeh, C. H. (2002). The evaluation of airline service quality
and applications. Berlin: Springer-Verlag. by fuzzy MCDM. Tourism Management, 23(2), 107e115.
Cronin, J. J., & Taylor, S. A. (1992). Measuring service quality: a reexamination and Tzeng, G. H., Chang, J. R., Lin, J. D., & Hung, C. T. (2002). Non-additive grey relation
extension. Journal of Marketing, 56(3), 55e68. model for the evaluation of flexible pavement condition. International Journal of
Deng, J. (1982). Control problems of grey systems. Systems and Control Letters, 5(2), Fuzzy Systems, 4(2), 715e724.
288e294. Yu, P. L. (1973). A class of solutions for group decision problems. Management
Deng, J. (1989). Introduction to grey system theory. The Journal of Grey Systems, 1(1), Science, 19(8), 936e946.
1e24. Zeithaml, V. A., Parasuraman, A., & Berry, L. L. (1990). Developing quality service:
Gilbert, D., & Wong, K. C. (2003). Passenger expectations and airline services: Balancing customer perceptions and expectations. New York: The Free Press.
a Hong Kong-based study. Tourism Management, 24(5), 519e532. Zeleny, M. (1982). Multiple criteria decision making. New York: McGraw-Hill.