PERFORMANCE EVALUATION OF SERVICE QUALITY IN HIGHER EDUCATION INSTITUTIONS USING MODIFIED SERVQUAL APPROACH WITH GREY ANALYTIC HIERARCHY PROCESS ( G-AHP ) AND MULTILEVEL GREY EVALUATION

In today’s climate of fierce competition between countries, paying attention to the needs and demands of customers whether in manufacturing or service sector, is considered as a vital competitive edge. Managers in service sector that are under pressure of environmental factors, have focused all their services on customers’ satisfaction and this has led to the continuous improvement in the performance of service organizations. Meanwhile, customers’ expectations should be properly understood and measured. Many efforts have been made to date in order to measure the quality of services using the SERVQUAL model. In this study, we try to investigate the concepts and factors affecting the quality of services according to modified SERVQUAL model and then utilize the proposed model of Grey Analytic Hierarchy Process (G-AHP) and Multilevel Grey Evaluation in order to evaluate the quality of services in the framework of Grey Systems Theory (GST). In order to propose our method, we will conduct a case study of the performance of service quality in higher education institutions of Isfahan-Iran.


Introduction
Every day, we receive services in different sectors such as education, insurance, banking, finance, hotels, transportation, restaurants, healthcare, etc.Some of these are introduced to us as services, while some others as products and finally some as a combination of both.Delivering a product to customers can be accomplished in an either tangible or non-tangible approach (Kotler, 2000).However, the service sector has a significant share of employment, which increases on a daily basis and this has led the quality to be of special importance in services sector.Higher growth rates and intense competition for the quality of provided services in both developed and developing countries, has made its measurement and evaluation a major challenge for every organization.Managers of service organizations today, try to develop the idea and culture of customer orientation in their respective organizations and provide necessary requirements to achieve organization performance improvement while creating a competitive edge through focusing on customers' needs and satisfying their demands, properly.Products and services have many similarities and the quality of services plays essential role to create a unique name.Thus, measuring and improving the quality of services is a necessity in today's life.Higher education institutions, as one of the service organizations, should try to identify their customers' (i.e.students) needs and expectations and to provide them with high quality services to satisfy them and to keep their loyalty to gain a competitive advantage.Providing a high quality service is a necessity for service organizations and educational institutions, especially the universities.
Students as the recipients of university services are considered as the primary sources to identify the educational behaviors of teachers and staff in their own universities.In today's competitive environment, service organizations' managers have found that in order to improve the performance of their organization, it is necessary to evaluate customer satisfaction of the quality of services provided.Therefore, this study evaluates Isfahan University of Technology and Isfahan University in terms of the above-mentioned subjects using a modified SERVQUAL model.We use these factors to measure and assess the performance of quality of services for the institutions mentioned.Since the services consist of non-tangible and non-homogenous factors, measurement of quality in the services sector is much more difficult compared with the manufacturing sector.Because the evaluation are made by considering the linguistic variables and by an evaluator and we also do not have comprehensive and adequate information at our disposal so we need to introduce the foundations of Grey Systems Theory (GST) to measure the uncertainty of the concepts associated with human mind.GST is one of the methods to study the uncertainty, insufficiency and incompleteness of information.We also require an effective instrument to detect and to prioritize the quality of systematic services, an approach that can develop consensus decision-making.Therefore, we will use the theories proposed by Saaty in the 1970s (Saaty, 1980).The Analytic Hierarchy Process (AHP) has been proposed based on the analysis by the human brain for complex problems.It has a widespread use in decision-making.Ranking according to the values obtained by parameters that can be calculated in order to estimate the priority using paired comparisons is an example of this instrument's capabilities (Liu & Hai, 2005).

Service quality in higher education institutions
Paying attention to service quality in higher education began in the 1980s and this interest continued until the early 1990s.This increased attention was due to the necessity of higher education institutions to adapt with financial conditions and customers' pressure to improve service quality (Mostafa, 2006).Since in a competitive market, satisfaction of service is the differentiation factor (Ham & Hayduk, 2003); therefore, students' satisfaction is considered as a decisive factor for the evaluation of higher education institutions.Quality of service is a multidimensional structure obtained from the difference between the existing and the desirable situation from a customer's point of view.Shank et al. (1995) evaluated the service quality in higher education institutions from the professional (teachers) and customer (students) services point of views (Ham, 2003).One of the broad definitions of service quality is paying attention to satisfying the needs or expectations of a customer (Singh & Khanduja, 2010).Quality is a series of activities, processes, actions and interactions offered to customers in order to solve their problems.It is a multidimensional concept.Service quality is an abstract structure, which is very difficult to define and to measure.There is no value in a product or service unless it would be consumed by a customer (Buyukozkan, et al., 2011).A product or service is considered high quality when it complies with demands and needs of customers.

Literature review
Many studies have already been conducted to measure service quality using SERVQUAL.Since the integrated models bring better results; in some of previous studies, SERVQUAL has been integrated into other models.Table 1 reviews the previous studies together with their objectives and results in educational fields and other integrated models for service quality assessment, which are associated with the current study.Selecting the most important dimensions of service quality using the SERVQUAL model.Assurance was the most important dimension while tangibles comprised the least important dimension of service quality.Service quality measurement in the Turkish higher education system with SERVQUAL method (Yilmaz et al., 2007) Education 2007 Evaluating the service quality at two different universities and selecting the most important factor in service quality.Students gave the highest importance to both empathy and responsiveness dimensions.Service quality in higher education : The role of student expectations (Voss et al., 2007) Education 2007 The role of students' expectations and teachers' teaching quality and identifying the most important factors affecting students' satisfaction.Service quality measurement on education service marketing and relationship between perceived service quality and students' satisfaction (Okumuş & Duygun, 2008) Education 2008 Evaluation of service quality in universities.There is a significant difference between perceptions and expectations of students.Students' perception and satisfaction are positively related.
Adaptation and application of the SERVQUAL scale in higher education (Oliveira & Ferrera, 2009) Education 2009 Evaluation of service quality in universities and determining the most important dimensions of improving the quality of service.Prioritizing of the five dimensions of SERVQUAL model in order of their importance: accountability, empathy, reliability, assurance and tangibles.Evaluation of the importance of service quality factor in PMR based on Grey Relation Theory (Yongqing & Jiatao, 2009) PMR 2009 Service quality assessment and selecting the most important factors affecting PMR in order to improve service quality using Grey degree.
Fuzzy application in service quality analysis : An empirical study (Lin, 2010) Commerce 2010 Measuring service quality in four different stores and determining the most important factors to rank commercial stores using fuzzy sets and modified SERVQUAL.Evaluation of E-commerce service quality using the AHP (Yu, 2010)

E-Commerce 2010
Assessing the service quality in e-commerce and determining the most important factors affecting service quality using Analytic Hierarchy Process.
Strategic analysis of healthcare service quality using AHP methodology (Buyukozkan et al., 2011) Healthcare 2011 Measuring and evaluation of service quality in 5 hospital units and their prioritization based on fuzzy AHP model of service quality.Hospital staff should pay more attention to each other.Professionalism and reliability dimensions led to the satisfaction in hospital.Using a modified grey relation method for improving airline service quality (Liou et al., 2011) Airlines 2011 Evaluation of service quality and ranking of 4 airlines in Taiwan using Grey Relation Theory.
Service quality in a Research university: A post _Graduate perspective (Rajab et al., 2011) Education 2011 Evaluating the satisfaction of English language learners at UTM university after graduation and performance quality in students at the end of learning period.Creating the necessary strategies in order to improve the quality before and after the graduation of English language learners.Influence of service quality , university image, and student satisfaction Toward WOM intention : A case study on UPHS university (Jiewanto et al., 2012) Education 2012 Study of the relationship between service quality, satisfaction and increased creditability of the university.History of behavioral intentions such as service quality and customer satisfaction induces an appropriate image of the university through time.This in turn leads to the promotion of the university and higher service quality.Assessment of quality of education in a nongovernment university via SERVQUAL model (Abari et al., 2011) Education 2011 Service quality measurement in Khourasgan Azad University using the SERVQUAL model.A significant difference between expectations and perceptions in all five dimensions of Parasuraman model of service quality.Highest average score belonged to teachers' perception of their knowledge while lowest average score belonged to students' perception of their readiness for their future job.A combined fuzzy AHP and fuzzy TOPSIS based on strategic analysis of electronic service quality in healthcare industry (Buyukozkan & Cifci, 2012) Healthcare 2012 Evaluating the quality of hospital websites using the SERVQUAL model, fuzzy AHP and fuzzy TOPSIS in order to find the most important dimensions and sub-criteria for higher customer satisfaction, and improving the service quality through internet services.Results showed that hospitals should focus more on the allocation of service accuracy (as a sub-criterion) and, reputation and response (as the main criterion).

Grey Systems Theory
In 1982, professor Deng published his first work on the concepts of Grey theory (Deng, 1989).Grey Systems Theory is a very effective technique for solving problems in uncertain conditions with discrete data and incomplete information.A system is called a Grey system if part of it includes known data and another part of it includes unknown data.Fuzzy mathematics usually deals with cases where experts express the uncertainty through the membership function.In cases where the number of experts and their level of experience are low, data are insufficient or there are a few instances available and it is not possible to extract the membership function, we can use the Grey Systems Theory (GST).The advantage of Grey System Theory over Fuzzy Theory is that GST includes fuzzy conditions or in other words, GST works well in fuzzy conditions.A Grey set is defined as a set of uncertain data described by Grey numbers, Grey relations, Grey matrices, etc. Grey number of an interval is a set of numbers that their exact amounts are unknown.If Z is a reference set then X Grey sets of Z reference set with two M x (Z) symbols as upper and lower limits of a Grey set, are defined by Eq. ( 1).
If X(Z)= (Z), then X Grey set becomes a fuzzy set that indicates GST inclusion over the fuzzy condition and its flexibility when dealing with fuzzy problems.

4.1.Grey assessment and ranking
In order to assess M independent options considering N criteria (dimensions) for ranking in a Grey environment, we should act as the following steps (Lin et al., 2008;Chen et al., 2011).
First step: Preference of option πi over the criterion πi through Eq. ( 2).
where  is the value of assessment given by the k th decision-maker for the i th option in terms of the j th criterion that could be shown by  = x , ̅ as a Grey number.
Second step: Creating a Grey decision matrix, where  are linguistic variables, which have been defined based on Grey numbers.
1-If the criteria are positive (the more the better) 2-If the criteria are negative (the lower the better) Fourth step: Determining the ideal positive option or the best answer possible as an option in order to be compared with other options.Assume that there M options defined as u = {u 1 ,u 2 ,…,u m }.Then the best criteria would be u = {u , u , …., u } that can be calculated using the Eq. ( 3).
Fifth step: Using Grey possibility degree to compare each option with u max as the desirable option according to Eq. ( 4) and Eq. ( 5). P{x L*=L(x) + L(y) Considering the relationship of  , .y, four different cases may occur: 1-If = , − then x = y.In that case: P{x ≤ y} = 0.5 2-If > then x < y.In that case: P{x ≤ y} = 1 3-If < then x > y.In that case: P{x ≤ y} = If there is interference and P{ ≤ y} > 0.5 then x < y.
If there is interference and P{ ≤ y} < 0.5 then  > y.
Therefore, it is possible to make the following comparison between the available options u={ u1,u2,…um} and the ideal positive option u max : P{ui≤ } = ∑ {x * ≤ Sixth step: Ranking of options The lower the value of p(ui< ), the better the rank of option i.Conversely, the closer this value to 1, the lesser the importance of the respective option.

Calculation of the relative Grey score
In order to calculate the relative Grey score for options in this study, Grey numbers were used on a scale of 7 according to Table 4.
Step 1: It can be calculated from the following equation for option i and criterion j.
where U is the value of assessment given by the k th decision-maker for the i th option in terms of the j th criterion that could be shown by U = , ̅ as a Grey number.
Step 2: Creating a Grey decision matrix, where U are linguistic variables, which have been defined based on Grey numbers.
Step 3: Normalization of decision matrix that can be calculated based on the type of criteria that are either in form of profit or cost.

D=
If the variables are in form of profit (the more the better): If the variables are in form of cost (the less the better): Step 4: Determining the reference or the ideal option based on the type of problem in order to do the assessment.
Step 5: Calculation of the relative Grey coefficient The relative Grey coefficient between Li and reference options considering the i th criterion, which is shown with £ Oi(j) , is calculated from the following equation: where D OI(J) is the Minkowski distance between the reference options considering the J th criterion.Technical coefficient between the reference options is generally considered according to Wang and ρ is usually 0.5.
Step 6: Calculation of the relative Grey score The relative Grey score between Li and reference options is calculated from the following equation: γ =∑ £ ( )

Grey-AHP
We recommend using the G-AHP model that is comprised of the Grey system and AHP based on AHP model proposed by Saaty, for this study (Saaty, 1980).This model is proposed for service quality assessment in higher education institutions and finding the best institution in terms of service quality performance.The main steps to use G-AHP are as follows: 1.Goal setting: At this stage, the goal is to assess the service quality in 3 higher education institutions and to find the best institution in terms of service quality performance .
2. Determining the Service quality assessment criteria: At this stage, modified SERVQUAL dimensions and important factors extracted from the SERVQUAL model will be selected as the main and sub-criteria, respectively.
3. Introducing options (alternatives): Higher education institutions under assessment are specified as options or choices.
4. Building the hierarchy of decisions: After determining the selection criteria and options, the hierarchy structure is built based on them.The overall objective will be placed on top of this structure and the criteria on lower levels.The available options or choices will then eventually be placed on 3 levels to make decisions.This situation as a general standard framework, regardless of the type of problem, is as described in Fig. 1.
5. Creating the matrix of paired comparisons: This stage includes the paired comparisons and creating the matrix of paired comparisons in each row of the hierarchy in order to answer the realization of objective or to meet its requirements.Each element of this matrix is a Grey number.
6. Normalization of the paired comparisons matrix: x * = 2x ∑ x + ∑ x (9) 7. Calculating the relative weights of criteria and options: The relative weights of factors in each level are calculated using normalized paired comparisons matrix according to Eq. ( 10).The calculated weight is a Grey number.
Wi= ∑ x * , x * (10) 8. Calculating the consistency rate (CR): After creating the paired comparisons matrix and calculating the relative weights of factors, the consistency of the paired comparisons matrix should be investigated.If the consistency rate of the matrix is lower than 0.1, then matrix D (decision-maker judgment about the preference of factors Level 1: Goal Level 2: Criteria Level 3: Alternative under comparison) is acceptable, otherwise the contents of matrix D are too inconsistent to give reliable results.In such cases, it is necessary to repeat the paired comparisons by decision-maker until the consistency rate (CR) reaches to the lower than 0.1.CR can be calculated using Eq. ( 11) to (15).RI in Eq. ( 15) is the mean of consistency rate for the random variable.Table 3 shows the value of RI for each value of n criteria.9. Calculating the weights of each option (alternative): In order to do this, the vector of relative weights of options should be multiplied by the vector of relative weights of criteria.The calculated numbers in this case are also Grey numbers.
10. Ranking of the options: At this stage, ranking is done based on the final weight of each option.Since final weights are Grey numbers, in order to rank them the vector of positive ideal weight will first be defined according to Eq. ( 16).
= , We then use the Grey possibility degree.If the Grey weight of the i th option is [ , ] and = , is the positive ideal option, the Grey possibility degree p ( ≤ ) for each option is calculated according to Eq. ( 4) and the option having the lowest calculated value, will be selected as the best option.

Research methodology
The standard modified SERVQUAL questionnaire with a Grey rating of seven that consists of 41 questions in five dimensions has been used in this study.The validity of this questionnaire was approved by the professors and experts.After assessing service quality and measuring expectations and perceptions, 16 factors out of 41 were selected as the most important ones based on the opinions of students and provided to 8 experts as paired comparisons in an AHP questionnaire format.This study was carried out at three superior higher education institutions of Isfahan (University of Medical Sciences, University of Technology and University of Isfahan).In order to increase the level of accuracy and making the students' judgments closer to reality in this study, linguistic variables were utilized in SERVQUAL questionnaire and G-AHP questionnaire for paired comparisons using the Grey numbers in Table 3 and Table 4, respectively.

Case study
After identifying the best factors affecting service quality according to students' opinions, hierarchy structure was defined as Fig. 2 in order to identify the best higher education institution in Isfahan based on service quality.The paired comparisons matrices were then created by experts in order to give weight to each factor in its respective level (Table 5).

Options
Third level: Sub-dimensions (Sub-criteria) Second level: Dimensions (Criteria) First level: Objective.Relative weight of each option (alternative) is calculated by multiplying the matrix of weight vector for each sub-dimension by weight vectors of university assessment in terms of sub-dimensions.Three superior higher education institution of Isfahan were compared with each other in this study using G-AHP for their service quality performance.Considering all the calculations, the performance ranking of these institutions is as follows:

University of Technology
University of Medical Sciences > University of Technology > University of Isfahan University of Medical Sciences had the best performance among the other universities in this study.This does not mean that the above-mentioned university provides glamorous services.Other universities should in fact improve their service quality based on these criteria in order to provide services to their students compared to the superior university.

Grey score
After following the steps listed, Grey score of dimensions was calculated as follows: Tangibles dimension = 0.5598, Providing a systematic service dimension = 0.60836 Social responsibility dimension = 0.7429 Human factors dimensions = 0.67051 Above results showed that students gave more importance to social responsibility dimension and less importance to tangibles dimension.

10-Grey assessment method
Considering the steps mentioned for Grey assessment: Step1.Grey decisions matrix Step2.we use the normalized vector of dimensions weight: [0.3815, 0.41469, 0.46687, 0.5064, 0.457] The normalized weight matrix is as follows: Results obtained from both Grey assessment and G-AHP were the same, which indicates that both methods confirm each other.In fact, both methods gave the following ranking: 1 st rank: University of Medical Sciences 2 nd rank: University of Technology 3 rd rank: University of Isfahan

Conclusion
This study was carried out in order to develop a model to understand the service quality and assess the performance of some superior universities using the modified SERVQUAL approach.Thus, the objective was first to calculate the gap score for sub-dimensions of five main dimensions and then identifying the most important of them in order to be provided to experts.This model was used to measure the performance of higher education institutions compared to each other.Results showed that universities should focus more on social responsibility and human factors so their services lead to more students' satisfaction.

Table 1
A review of studies, their objectives and results

Table 3
Scale for SERVQUAL linguistic variables section

Table 4
Linguistic variables of the paired comparisons matrix in AHP questionnaire . 2. Hierarchy of service quality assessment based on SERVQUAL model

Table 5
Matrix of dimensions assessment in terms of objective

Table 6
Matrix of sub-dimensions assessment in terms of tangibles

Table 7
Calculation of relative weight for each alternative

Table 8
Calculation of relative weight for each alternative-Continued