Proposing a Framework for Airline Service Quality Evaluation Using Type-2 Fuzzy TOPSIS and Non-parametric Analysis

This paper focuses on evaluating airline service quality from the perspective of passengers view. Since now a lot of researches has performed in airline service quality evaluation in the world, a little research has been conducted in Iran, yet. In this research, we proposed a framework for measuring airline service quality in Iran. After reviewing airline service quality criteria, we selected SSQAI model because of its comprehensiveness in covering airline service quality dimensions. We redesigned SSQAI questionnaire items to adopt it with Iranian airlines requirements and environmental circumstances in the Iran's economic and cultural context. This study includes fuzzy decision-making theory, considering the possible fuzzy subjective judgment of the evaluators during airline service quality evaluation. Fuzzy TOPSIS have been applied for ranking airlines service quality performances. Three major Iranian airlines which have the most passenger transfer volumes in domestic and foreign flights, were chosen for evaluation in this research. Results demonstrated Mahan airline has got the best service quality performance rank in gaining passengers' satisfaction with delivery of high quality services to its passengers, among the three major Iranian airlines. IranAir and Aseman airlines placed in the second and third rank, respectively, according to passenger's evaluation. Statistical analysis have been used in analyzing passenger responses. Due to abnormality of data, Non-parametric tests were applied. To demonstrate airline ranks in every criterion separately, Friedman test was performed. Variance analysis and Tukey test were applied to study the influence of increasing in age and educational level of passengers' on degree of their satisfaction from airline's service quality. Results showed that age has not significant relation with passenger satisfaction of airlines, however increasing in educational level demonstrated a negative impact on passengers' satisfaction from airline's service quality.


Introduction
Since increasing in air travel rates, competition between Iranian airlines has grown in recent years.Although a lot of researches has been conducted in airline service quality evaluation in different countries, there is still a little research concerning airline service quality in Iran.
Nowadays, delivering high-quality services has become a marketing requirement for airline companies which want to survive in this competitive environment (Ostrowski, O'Brien, & Gordon, 1993).In this competitive environment, delivering service quality in a desirable manner, is essential for the airline's competitiveness and sustained growth because of passenger's high expectations and rapid development of transport technology that has made the world into the global village.In order to better serve passenger needs, airlines have to pay attention to passenger's expectations from their services.Airlines need to discover new ways to increase focusing on essential service items and reduce the time and energy spent on less important service items (Liou, Hsu, Yeh & Lin, 2011).So they can better manage their budget and have the chance of reducing their prices.In this case, they can cope competitive challenges and avoid losing their passenger with maintaining their perceptions of service quality at a moderate level.Trying to deliver high-quality service to airline passengers, results in retaining existing customers and also, enticing other airlines customers and leads to differentiating airline image from competitors.According to sultan and Simpson Jr (2000) customized services, guarantees, and continuous customer feedback are important factors of a successful service quality strategy as a comprehensive measurement of airlines performance.Chang and Yeh (2002) argue that since service quality is heterogenic, intangible and inseparable, its measurement quality is difficult.In most industries such as airline industry, only customers can investigate the service value and truly evaluate service quality because they are service consumers.In the airline industry, for improving airline service quality performance, airline managers need a framework enabling them to evaluate the quality of services they offer passengers and help them improve quality in required areas.Since the evaluation is produced from the different view of evaluator's linguistic variables, evaluation process must be conducted in an uncertain, fuzzy environment, to gain more accurate data.A fuzzy multi-criteria model is necessary to deal with "qualitative" (unquantifiable or linguistic) or incomplete information (Opricovic & Tzeng, 2003).
Fuzzy MADM techniques are powerful decision-making tools that help managers to involve all aspects of the problem in the decision process.Solving problems and making decision in Fuzzy environment leads to more precise and accurate results in ranking and selecting alternatives.Statistical analysis of passengers' responses empowers airline managers in better understanding of passengers service quality needs and would help them in making effective improvement plans for increasing airlines service quality performance.In this paper, combining Fuzzy MADM and statistical analysis with improving SSQAI scale and redesigning its questionnaire, helped in proposing a stable framework for evaluating airline service quality in Iran.

Service quality in airline industry
Quality is one of the primary drivers of business and is used to differentiate organization's service offering."Quality" lies at the heart of the organization's strategy to gain competitive advantage (Ghobadian, Speller, & Jones, 1994).Offering high-quality services will increase customer satisfaction, leading to consumer retention and encouraging recommendations (Nadiri & Hussain, 2005).Parasuraman, Zeithaml, and Berry (1985) defined the concept of service quality as a comparison between customers' expectations and actual service performance.Parasuraman, Zeithaml, and Berry (1988) argued that, regardless of the type of service, consumers evaluate service quality using similar criteria, which can be grouped into five dimensions.They proposed their five dimensions' model with 22 items measurement scale (called SERVQUAL).The five Dimensions of SERVQUAL are reliability, responsiveness, assurance, empathy and, tangibles which were developed based on Parasuraman et al.'s (1985) study in which they proposed ten dimensions of service quality.Although SERVQUAL has been widely applied to various industries, including airline industry (Nel, Pitt, & Berthon, 1997;Park, Gilbert, & Wong, 2003;Robertson, & Wu, 2004), this scale has been highly criticized in the literature (Bitner, 1990;Bolton & Drew, 1991;Park, Robertson, & Wu, 2006).Cronin and Taylor (1994) consider that SERVQUAL has a naturally flawed concept because of its ill-judged adoption of the ‹ 8 › disconfirmation model.Gilbert and Wong (2003) and Liou et al, (2011) state that however SERVQUAL has been widely used to measure service quality in a variety of industries, no two providers of a service are exactly alike.Park et al. (2006) note that the five dimensions with twenty-two items of SERVQUAL scale can't easily be applied to the airline industry because this scale has not mentioned some of the important criteria in airline service quality such as in-flight meals, seating comfort, seat space and leg room.Cronin and Taylor (1992) developed a performance-based model of service quality called SERVPERF that measures service quality only based on customers' perceptions of the service performance.This model is a variation of SERVQUAL since uses the same criteria of SERVQUAL model.SERVPERF is an applicable and useful tool for measuring service quality.However, Cunningham, Young, and Lee (2004) mentioned that since SERVPERF uses the same dimensions and items of SERVQUAL, it has failed to measure industry-specific dimensions of service quality in the airline industry.As Ghobadian et al. (1994) stated, service quality is a multi-dimensional phenomenon and utility value of its determinants are situation-dependent.Chang and Yeh (2002) assert that attributes of service quality are context dependent and should be selected based on the service environment investigated.Due to this fact, many researchers have adopted different criteria for evaluating airline service, e.g.Elliott and Roach (1993) evaluated timelines, comfort of the seat, luggage transportation, quality of food and beverage, check-in process and inboard service factors in measuring airline service quality.Ostrowski et al. (1993) defined customer-loyalty, timeliness, Food and beverage quality, and comfort of the seat as the service quality and factors.Liou et al. (2011) found twenty-eight criteria classified under dimensions of Booking service, Ticketing service, Check-in, Baggage handling, boarding process, Cabin service, Baggage claim, Responsiveness to realize passengers' satisfaction of airlines service quality.Truitt and Haynes (1994) offered seven criteria for evaluating airline service quality that are customer complaints handling, check-in process, the convenience of transit, process of luggage, timeliness, clearness of seat, and Food and beverage quality.Laming and Mason (2014) expressed that US Department of Transportation Rates airlines quality with on-time performance, customer complaints denied boarding, mishandled baggage.
Recently, evaluating service quality base on the hierarchical concept is taken into consideration by researchers.Brady and Cronin (2001) suppose that customers form their service quality perceptions on the basis of an evaluation of performance at multiple levels and ultimately combine these evaluations to arrive at an overall service quality perception.Dabholkar, Thorpe & Rentz's (1996), Brady and Cronin (2001) and Wu, Lin and Hsu (2011) Suggest that service quality should be based on a hierarchical concept.In hierarchical concept, Customers make their judgments of service quality based on a series of factors that are specific to the evaluated service.Base on this approach, customers form their evaluation of primary dimensions on assessment of the corresponding subdimensions.Wu and Cheng (2013) introduced SSQAI model with eleven criteria in four dimensions specialized for evaluating airline service quality.The SSQAI model is a performance-based measurement scale on the basis of hierarchical structures in measuring service quality.SSQAI (see Fig. 1) is developed based on Dabholkar et al. (1996), Brady and Cronin's (2001) and Caro and Garcia's (2007) models.Park et al. (2006) indicate that many airlines can't find a proper scale to evaluate service quality to assess and improve their service performance.However, many studies have used conventional statistical techniques to test hypotheses and generate airline service quality criteria such as Pakdil and Aydin (2007) and Gursoy, Chen and Kim (2005).In recent years the researchers have tended to apply Fuzzy Multiple Criteria Decision-Making (FMCDM) techniques to strength the comprehensiveness and reasonableness of the decision-making process (Tsaur et al., 2002).The researchers have implemented MCDM methods to measure airlines integrated service quality level and to find weak areas and make suggestions for improvement (Chang & Yeh, 2002;Liou & Tzeng, 2007;Tsaur et al., 2002;Liou et al., 2011).Tsaur, Chang, and Yeh (2002) used SERVQUAL dimensions to derivate service quality attributes and performed AHP and TOPSIS in ranking the airlines.They stated that courtesy, safety, and comfort are the most important attributes.Chang and Yeh (2002) performed fuzzy multi-criteria analysis for ranking four Taiwan's domestic airlines based on the concepts of the degree of optimality and the ideal solution.Fifteen service quality attributes classified in eight dimensions were ranked according to passengers' ISSN 2520-2979 Journal of Sustainable Development of Transport and Logistics,2(2), 2017 ‹ 9 › responses.Liou and Tzeng (2007) applied Fuzzy integral, AHP and Grey Relation Analysis to rank service quality performance of six international airlines.In this paper, the SSQAI model is improved and a framework applicable to measure airline service quality in Iran is designed.

Methodology
After reviewing airline service quality criteria, SSQAI scale was adopted in this study, since it represents a valid and reliable tool for assessing service quality in the airline industry (see Fig. 1). the criteria on the SSQAI model and their symbols used in this study are shown in Table .1.After collecting customer opinions, and using criteria weights determined by experts, ranking of these airlines was calculated using trapezoidal fuzzy TOPSIS.Fuzzy TOPSIS calculation was constructed in excel 2016.Using statistical analysis for analyzing customer reviews, firstly, normalization test was taken to determine using parametric or non-parametric tests.Shapiro-Wilk and Kolmogorov-Smirnov (K-S) normality tests showed collected data are not normal, so non-parametric tests were applied for ISSN 2520-2979 Journal of Sustainable Development of Transport and Logistics,2(2), 2017 ‹ 10 › analyzing passengers' responses.Friedman test was performed to demonstrate airline ranks in every criterion separately.Airlines ranked in all criteria due to customer opinions.Variance analysis and post-hoc Tukey test were applied to study the influence of increasing of age and educational level on degree of passengers' satisfaction with airlines service quality performance.
Three airlines chosen for this research and their symbols are Mahan (A1), IranAir (A2) and Aseman (A3) airlines.These airlines were nominated since they are the three oldest Iranian airlines with a powerful background.Moreover, most flight rates and passenger transportation volume among all airlines in Iran belongs to these airlines.

Experts
Our experts' Community involved 45 respondents from Tehran and Mashhad.Our experts consist of 12 airline manager, 16 Aviation specialist, 17 Frequent fliers of chosen airline's passengers.Tehran is the capital of Iran and most central airline offices are in Tehran, except Iran Air that its central office is located in Mashhad.So, our experts are from both cities.Questionnaire of this research was designed according to experts' opinions.

Passengers
A sample size of 385 respondents was considered in this study to reduce the influence of the statistical assumptions associated with ANOVA.The questionnaire was distributed to passengers in thirteen airline agencies of Mashhad in about four weeks.Mashhad consists of twenty-six regions.Two agencies were selected from each region and the questionnaires were distributed to passengers of this agencies.The questionnaires were distributed doubled because half of the questionnaires were not properly filled and subsequently were dropped.Only candidates who had flown with all of these three chosen airlines in the last recent year at least one time, were selected for participating in answering questionnaires, so data collection was really time-consuming.

Questionnaire design
First, all criteria in evaluating airline service quality were gathered.By consulting Iranian airline experts, it was founded that four dimensions and eleventh sub-criteria of SSQAI model are prober for utilizing in Iran.We tried to redesign and specialize SSQAI instrument questionnaire items to fit with Iran's economic and cultural circumstances and Iranian airlines situations, as well.With the help of airline industry experts, SSQAI items were utilized in a way to be simple and clear, not encountered with the problems such as vacuity of questions of prior models like SERVQUAL.It's believed some of the criteria extracted from literature could be involved in the subset of SSQAI criteria items.So, these criteria were added to our framework questionnaire.Also, some items were changed or dropped due to ensure universality of this model and specializing and localizing this model for using in Iran's airline industry, by taking average scores of experts' opinions in the screening questionnaire.
Each expert had to give scores from 0 to 5 to every item.The average test was applied to scalp questionnaire items and improve stability of the instrument.Items with scores more than 3 were selected to be on final instrument to help with increasing endurance.The final version of our instrument has a total of 64 items representing eleventh criteria of SSQAI airline service quality model (See Table .2).In this paper, the questionnaire was distributed to gather passengers' ratings of three chosen airlines, Mahan, Iran Air and Aseman.Using fuzzy TOPSIS the three Iranian major airlines were ranked based on the passenger satisfaction with these airlines service quality performance.Descriptive statistics of the respondents is shown in Table .3.

Fuzzy Set and Type-2 Fuzzy Number
Fuzzy set theory aids in measuring the ambiguity of concepts that are associated with human being's subjective judgment.Lingual expressions, for example, satisfied, fair, dissatisfied, are regarded as the natural representation of the preference or judgment.The fuzzy linguistic variable reflects different aspects of human language.Its value represents the range from natural to artificial language.When the values or meanings of a linguistic factor are being reflected, the resulting variable must also reflect appropriate modes of change for that linguistic factor (Chen & Chen, 2010).Zadeh (1975) proposed using values ranging from 0 to 1 for showing the membership of the objects in a fuzzy set.The membership degree of the fuzzy set can be described with triangular, trapezoidal, Gaussian, sigmoidal functions or can be formed with different functions.Trapezoidal fuzzy numbers are useful in promoting representation and information processing in a fuzzy environment and their computational simplicity.Trapezoidal fuzzy numbers can be expressed as (n1, n2, n3, n4).A trapezoidal fuzzy number is shown in Fig. 2.

Fuzzy TOPSIS
The technique for order preference by similarity to an ideal solution (TOPSIS) was developed by Hwang and Yoon (1981).Based on the concept, any chosen alternative should have the shortest distance from the ideal solution and the farthest distance from the negative-ideal solution (Opricovic & Tzeng, 2003).Trapezoidal fuzzy numbers are useful in promoting representation and information processing in a fuzzy environment and their computational simplicity.In this study, trapezoidal fuzzy numbers are adopted in the fuzzy TOPSIS calculation.A developed method of Fuzzy TOPSIS offered by Chen (2000) is used in this paper.Fuzzy TOPSIS analysis is conducted as follows:

Define linguistic scale
Linguistic variables used in Fuzzy TOPSIS are shown in Table .4. This scale had been formerly applied in fuzzy TOPSIS analysis by Ertuğrul and Güneş (2007).

Establish the initial decision matrix
If Ai=A1; A2; ….; Am are possible alternatives among which decision makers have to choose, Cj= C1; C2; ….; Cn are criteria with which alternative performance are measured.Xij is the rating of alternative Ai.If we have K passengers participating to compare alternatives (in this paper, the three airlines), the aggregated fuzzy ratings of K passengers can be calculated as: Journal of Sustainable Development of Transport and Logistics,2(2), 2017 ‹ 14 › Wj is the weight of criterion Cj .The aggregated weights can be obtained directly from expert opinions, with the same technique as aggregated fuzzy ratings of passengers, here P defines the number of experts.In this paper, the aggregated weights are generated based on experts' responses.The initial fuzzy decision matrix is constructed in Table 5.  (4,6.46,7,9.13) (2.88,5.49,6,8.38)(2,4.47,5,7.5)0.069135

Calculate the normalized decision matrix
To avoid the complicated normalization formula used in classical TOPSIS, the linear scale transformation can be used to transform the various criteria scales into a comparable scale.The normalized value rij is calculated as:

Calculate the weighted normalized decision matrix
Weights of criteria produced formerly in Fuzzy ANP with experts opinions, are used here.The weighted normalized value is Vij and is calculated as: , .

C11
v is the best value of criteria 'j' respect to alternative 'i', and j v  is the worst value of criteria j respect to alternatives i.

 
12 , ,..., A * shows the positive ideal solution and A -shows the negative ideal solution as demonstrated in Table 8.

Calculate the separation measures
Different from Chen's (2000) approach, Ertugrul, and Gunes (2007) suggest using Euclidean distance for calculating the distance between two fuzzy numbers.The distance between two trapezoidal fuzzy numbers (a1, b1, c1, d1) and (a2, b2, c2, d2) can be calculated by using Euclidean distance as: The distance of each alternative from the fuzzy positive ideal solution (FPIS) and fuzzy negative ideal solution (FNIS) is calculated as: Distance of each alternative from FPIS and FNIS is shown in Table 9.
The closeness coefficient represents the distances to the fuzzy positive ideal solution ( A * ) and fuzzy negative ideal solution ( A − ) closeness coefficient of each alternative (see Table 10) is calculated as:

Rank the preference order
According to the closeness coefficient, the ranking order of three alternatives is A1  Obviously, the best selection is candidate A1.

Kolmogorov-Smirnov and Shapiro-Wilk test
In statistical analysis, we first have to check normality of data.If data were normal, parametric tests are used in data analyzing, else non-parametric tests should be used.so,Kolmogorov-Smirnov and Shapiro-Wilk tests are used for checking normality of data as shown in Table 11.As shown in Results, data are not normal.

Table 2 : Evaluation criteria and Questionnaire items Criteria Items Conduct
1. Cabin crew are kind and polite to me. 2. The employee of (reservation, sales, ticket issuing, identification, and handling) behave respectfully and politely with me. 3. The airline employees' attitude demonstrates their willingness to help me.4. I can depend on the airline employees being friendly.5.The employees' attitude shows me that they understand my needs.6.The employees' behavior allows me to trust their services.7. The pilot's speech during flight is clear and soothing.8.The employees carefully pay attention to passengers depending on the type of traveler (women, men, children, adolescents, persons with disabilities, first class or ...).9.The employees understand my specific needs.10.The employees pay attention to every single traveler.11.The employees always provide me with their best services.12.The employees try their best to provide services to me.When I have a problem, the airline employees show a sincere interest in solving it.20.The employees consume enough time to solve my problem.21.The employees understand the importance of resolving my complaints.22.The employees are able to handle my complaints directly and immediately.The on-site queening at the airport is understanding and predictable.32.I feel comfortable with the volume of noise in the cabin.33.The airlines facility is well designed.34.The layout of airlines serves my need.35.Ticket and travel services offices and counters are pretty and equipped.36.The Quality of meals and drinks on the plane is favorable.37.The Way meals are served on the plane, is perfect.The cabin crew describe how to use safety equipment, such as (oxygen masks, vests, boat, etc.) very well and precisely.39.There are noticeable sprinkler systems in the cabin.Valence 40.I believe that the airline tries to give me what I want.41.I would say that I feel good about what I receive from airlines.42.I would evaluate the outcome of airlines services favorably.43.I will recommend Traveling with this airline flights to my friends and acquaintances.The airline understands that waiting time is important for me.46.Airline employees provide services quickly and in the shortest time.47.I rarely have to wait long to receive the airline services i need.48.There is a rare delay before or during aircraft flight and the flight schedules are accurately according to the announced program.53.Airlines offers adequate and proper flight information to passengers.54.I can Easily access to my required information accurately and up to date in 24 hours a day.55.Website Instructions explaining how to get airline services are legible and understandable.56.The Airline website provides suitable information of various services the company offers.Convenience 57.Airline offers services (before or during the flight) based on schedule formerly announced.58.The Airline web services are desirable and efficient.59.The reservation and ticketing systems are convenient.60.The airline provides me with enough flights and convenient flight schedules 61.Passenger transportation services from the output gate to the aircraft is efficient and desirable.62. Compensation procedure in case of flight delays or cancelation or air accidents, is proper and convenience.63.The passenger load displacement process is convenient and efficient.64.Electronic payment services through airline website are easy and convenient.
Information 49.The airline keep me well-informed about services i need.50.The airline tells me the accurate time on which it provides service.51.The airline understands the information the passengers need.52.Airlines website has interactive features (for example, online answering to questions).‹ 12 ›

Table 5 : Fuzzy design matrix
The normalized fuzzy decision matrix is shown in Table.6.

Calculate the relative closeness (similarity) to the ideal solution A
closeness coefficient CCi is defined to determine the order of all possible alternatives.