Distance education tool selection using novel spherical fuzzy AHP EDAS.

COVID 19 pandemic, which entered our lives suddenly, caused the change of our classical education system and forced higher education institutions to switch to distance education quickly. During this time, the need for a method that can comprehensively and scientifically evaluate the alternatives of videoconferencing tools to be used in distance education has emerged. This paper proposes a novel hybrid multi-criteria group decision-making (MCGDM) model by integrating the analytic hierarchy process (AHP) and the evaluation based on distance from average solution (EDAS) methodologies. The proposed model is developed with spherical fuzzy (SF) sets, which enable decision-makers (DMs) express their membership, non-membership and hesitancy degrees independently, and in a large three-dimensional spherical space. The applicability of the developed spherical fuzzy AHP EDAS is illustrated through a problem of selecting a videoconferencing tool for distance education. For this purpose, three DMs evaluate five popular videoconferencing tools, namely Zoom, Google Meet, Cisco WebEx, Skype and Microsoft with respect to six criteria, which are expanded with 32 related sub-criteria from the literature to more comprehensively handle the problem. The implications, sensitivity and comparative analyses, limitations and future research avenue are also given within the study.

Introduction

Distance education is an education model where students are not physically present in their classrooms, are exempt from learning resources, and are not subject to time limits (Kaplan 2016; Ince et al. 2020). After the COVID 19 outbreak in 2020, the concept of distance education has become compulsory and has been used in all levels of education in many countries (Bansal 2020). And throughout the pandemic, there is an increasing shift to this concept which has required academic staff and students use communication tools such as videoconferencing systems (Correia et al. 2020) due to their abilities to bridge the geographical gap between teacher and students, and replace face-to-face meetings.

However, in the market, there are many available videoconferencing tools having different characteristics, and there are various criteria to be considered for selecting one of them. Higher education institutions that have limited or no experience of distance education experience difficulties in selecting a proper videoconferencing tool for distance education, and a need for a method that can comprehensively and scientifically evaluate available alternatives has emerged.

Although recent studies have already analyzed specific problems around distance education and developed some MCGDM-based models to solve these problems, there is no study presenting a solution to videoconferencing tool selection problem and giving a readily available set of criteria to users for this purpose. Moreover, most of the available studies are developed with crisp numbers without considering the hesitancy of DMs, and since videoconferencing tool selection problems consist many vague criteria, there is also a need for a study that provide an appropriate method to treat the ambiguity and vagueness in the nature of these problems. At this point, fuzzy set theory (Zadeh 1975), by utilizing fuzzy sets, is very suitable to transform linguistic evaluations of DMs to a numerical form and handle vagueness.

Ordinary fuzzy sets have various extensions and the most prominent ones can be summarized as follows: Type-1 intuitionistic fuzzy sets (Atanassov 1986) allow membership and non-membership degrees for the element of a fuzzy set without modeling hesitancy degree separately; however, it can be found since the total of all three degrees is equal to one. Type-2 intuitionistic fuzzy sets (Atanassov 1989) were later developed and provided an increased area of membership and non-membership degrees in which their square sum is less than or equal to one. Neutrosophic sets (Smarandache 1999) are an extended version of intuitionistic fuzzy sets. Unlike intuitionistic fuzzy sets, each element in a neutrosophic set has a degree of truthiness, indeterminacy, and falsity, and these parameters can be considered as membership, non-membership, and hesitancy degrees, respectively. Hesitant fuzzy sets (Torra 2010), on the other hand, were developed to deal with situations where different membership functions are considered as possible. Picture fuzzy sets (Cuong and Kreinovich 2013) are also considered as an extended version of intuitionistic fuzzy sets. Yes, abstain, no and refusal types of opinions of voters may be modeled as positive, neutral, negative and refusal membership degrees, respectively, in a picture fuzzy set methodology. And the q-rung orthopair fuzzy sets (Yager 2016) are more advantageous than intuitionistic fuzzy sets with an increased space for expressing their opinions about membership and non-membership degrees; however, these sets also do not allow to model the hesitancy degree independently.

Spherical fuzzy sets (Kutlu Gündoğdu and Kahraman 2019) were recently developed based on a three-dimensional spherical geometry, and by using the theory of type-2 intuitionistic fuzzy sets and neutrosophic sets. In spherical fuzzy sets, the sphere is considered as a volume rather than a solid, and this allows us to assign membership, non-membership and hesitancy parameters independently and in a larger space. Geometric representations of spherical fuzzy sets, intuitionistic fuzzy sets, neutrosophic sets and type-2 intuitionistic fuzzy sets are given in Fig. 1.

Fig. 1

Geometric representations of spherical fuzzy sets, intuitionistic fuzzy sets, neutrosophic sets and Pythagorean fuzzy sets

In this study, we use spherical fuzzy sets and propose a novel MCGDM based decision support model to aid distance education tool selection problem. Our purpose is to bring together the characteristics of generally accepted mathematical approaches, to ensure objective and logical results, and to apply the created model to a problem encountered and needed in real life. Main advantages of our work according to prior ones can be listed as follows:The main motivation of this study is to develop a decision support model to assist higher education institution managers and other beneficiaries who face distance education tool selection problem. The model is developed in a spherical fuzzy environment by using the AHP (Saaty 1980), which is a relative measurement method that can be implemented to both qualitative and quantitative criteria, and the EDAS (Keshavarz Ghorabaee et al. 2015), which is one of the new MCGDM methods based on the distance of alternatives from the average scores of criteria. In this study, spherical fuzzy sets are used to handle the vagueness of DMs and AHP methodology is integrated to the model to systematically handle the wide and layered criteria set of the problem. For final ranking, EDAS is used since it has an easy methodology, it can handle both objective and subjective criteria, it is explicit and rational, its computation processes are straightforward, and its applicability has been tested in a variety of applications. The contributions of our study to the literature are as follows:The rest of the paper is organized as follows: Section 2 summarizes the literature on the MCGDM studies around distance education, spherical fuzzy AHP applications, and hybrid MCGDMs developed with EDAS. Section 3 gives the methodology. Preliminaries of spherical fuzzy sets are given in this section prior to development of the model. Then, the flowchart and the steps of the proposed model are given in detail. Section 4 gives the application. First, criteria and alternatives are presented by scanning the literature, then the structure of the problem is presented and the proposed model is shown step by step through a numerical example. Section 5 discusses the results by providing sensitivity and comparative analyses, and implications. Section 6 finalizes the paper and gives the limitations and recommendations for future studies.

No previous research handles the problem of selecting videoconferencing tools for distance education.

The uncertainty in the nature of distance education-related problems may significantly affect the results. However, most of the existing studies in the literature either do not address the uncertainty and use crisp numbers or just use type-1 or type-2 fuzzy sets.

Type-1 and type-2 fuzzy sets do not handle the non-membership degrees. However, spherical fuzzy sets provide solution to this parameter.

Spherical fuzzy sets provide users more space to model membership and non-membership than type-1 intuionistic fuzzy sets.

Spherical fuzzy sets enable users model the hesitancy degree of a decision-maker in an independent space. However, type-1 intuitionistic fuzzy sets, type-2 intuitionistic fuzzy sets, q-rung orthopair fuzzy sets and hesitant fuzzy sets do not provide such space for this purpose, and hesitancy degree is directly dependent on other parameters.

Spherical fuzzy sets do not force the membership degree of an element to take values between zero and one like hesitant fuzzy sets.

The parameters of neutrosophic sets are linear in a three-dimensional space; however, in some cases, the linear distance may not be appropriate and in such cases nonlinear distance are more suitable, and spherical fuzzy sets provide membership, non-membership and hesitancy parameters in a nonlinear space.

The EDAS method has not benefited from spherical fuzzy sets for addressing the vagueness. Our framework uses the advantages of both mathematical approaches.

Little has been done for developing sensitivity and comparative analyses in distance education-related MCGDM studies.

This study is unique in terms of both the subject handled and the model developed. Within the framework of distance education, the issue of videoconferencing tool selection problem is discussed and a novel spherical fuzzy MCGDM model is provided to users to solve this problem.

A wide set of criteria that can be used in evaluating the videoconferencing tools is presented by scanning the literature.

EDAS method is developed in a spherical fuzzy environment and strengthened with AHP technique to cope with extended criteria set. The proposed model has the flexibility to be applied not only within the scope of distance education but also in other areas.

Spherical fuzzy sets enable DMs to define membership, non-membership and hesitancy degrees independently and in a larger preference domain than most of other fuzzy extensions.

Sensitivity analysis for both sub-criterion weights and DM weights are performed to validate the result of the proposed methodology.

A comparative study is presented and the effectiveness of the proposed model is compared with five different hybrid MCGDM models.

Spearman’s rank correlation coefficients are calculated to analyze the results obtained from comparative and sensitivity analyses.

Literature review

In this section, the related work is presented for better providing insights into the concepts of this research and more clearly address the novelty of our work. In the literature, distance education, distance learning and e-learning terms can be used interchangeably which may cause ambiguity. In this paper, distance education term is used as an “umbrella term” to describe educational activities in which students are physically separated from their instructors (Yılmaz 2019). In this section, the literature study is discussed under three headings. In Sect. 2.1, MCGDM studies within the scope of distance education are summarized. In Sect. 2.2, the AHP model, which we can call the auxiliary technique of the proposed model, is discussed in general, and then spherical fuzzy AHP studies are summarized. Finally, Sect. 2.3 summarizes previously developed hybrid EDAS studies in the literature.

Literature on MCGDM applications to distance education problems

MCGDM methodology has been widely used in various sectors and they are also applicable to decision problems within a higher education setting, and distance education is one of them. Chao and Chen (2009) evaluated distance web learning system by using fuzzy preference relations with AHP technique. The authors used triangular fuzzy sets in their study. Kurilovas and Vinogradova (2016) used AHP to evaluate distance learning course quality in a triangular fuzzy environment. Garg (2017) used triangular fuzzy sets and proposed an AHP COPRAS WEDBA-based e-learning web site ranking model. Turker et al. (2019) developed an AHP TOPSIS model in a triangular fuzzy environment evaluated a learning management system.Naveed et al. (2020) evaluated critical success factors in implementing E-learning system by using a triangular fuzzy AHP while Jaukovic Jocic et al. (2020) used interval valued triangular fuzzy numbers and applied ARAS method for e-learning course selection.Kabak et al. (2017) used ANP and TOPSIS for distance education website evaluation by using triangular fuzzy sets. Ghannadpour et al. (2018) applied triangular fuzzy ANP and evaluated e-learning center. Khan et al. (2019) used PIV model and handled e-learning website selection problem by using crisp numbers, and Muhammad and Cavus (2017) presented a study by applying DEMATEL to learning management system criteria identification problem by using triangular fuzzy numbers.

Literature on spherical fuzzy AHP

AHP is an effective MCGDM methodology that simplifies issues with complex and many criteria in a hierarchical structure by using pairwise comparison matrices. This methodology generally consists of the following steps: Composing the AHP structure to arrange the objectives, criteria and alternatives in a hierarchical structure, establishing of a pairwise comparison decision matrix to determine the relative weight of criteria, calculation of criteria weighs and testing the consistency (Yu et al. 2011). In AHP technique, consistency ratio is the most important measurement of the results for checking the consistency of the pairwise comparison responses. The author recommends a consistency ratio of 0.1 or less for an accurately structured problem. Although AHP relies on the judgements of DMs for deriving priority scales, the judgements of DMs may not be consistent, and in this case, the concern of AHP is to measure the inconsistency level and improve the judgments of DMs. Priority vector in AHP is a numerical ranking of alternatives and indicates a preference order between them, and the principal eigenvector is the priority vector of a consistent matrix and if inconsistency is allowed in a positive reciprocal pairwise comparison matrix, principal eigenvector is needed to represent the priorities associated with that matrix, on the condition that the inconsistency is less than or equal to a desired value.

In the literature, spherical fuzzy sets are increasingly used with AHP methodology in various MCGDM applications. Industrial robot selection (Gündoğdu and Kahraman 2019), renewable energy location selection (Gündoğdu and Kahraman 2020), food waste management alternative selection (Buyuk and Temur 2020), supplier selection criteria prioritization (Unal and Temur 2020), dynamic pricing model for mobile advertisements, career management activity prioritization (Yildiz et al. 2020) and global supplier selection (Sharaf 2021) are the recent studies developed with spherical sets.

Moreover, spherical fuzzy sets are used also used in hybrid MCGDMs integrated with AHP methodology. Otay and Atik (2020) handled oil station location problem based on AHP WASPAS model. Jaller and Otay (2020) proposed AHP TOPSIS for sustainable vehicle technology evaluation problem, Ayyildiz and Gumus (2020) presented AHP WASPAS for petrol station location selection problem, Mathew et al. (2020) handled manufacturing system selection and developed AHP TOPSIS for this purpose, Otay et al. (2020) integrated AHP to WASPAS and illustrated it through a manufacturer selection problem, and Oztaysi et al. (2020) developed AHP VIKOR for location-based advertisement selection.

Literature on hybrid fuzzy EDAS

EDAS (Keshavarz Ghorabaee et al. 2015) is a new and efficient MCGDM methodology that determines the desirability of alternatives based on total distance of alternatives from their corresponding averages for each criterion. Unlike TOPSIS and VIKOR which use the idea of proximity to ideal solutions, EDAS method takes the average point as a reference and uses two measures, namely positive distance from average and negative distance from average to rank the alternatives. EDAS method has been integrated with different fuzzy extensions in various applications recently. Turskis et al. (2017) used crisp numbers and integrated AHP to EDAS for ranking cultural heritage structures. Some other authors used triangular fuzzy numbers to develop AHP EDAS for various sectors such as organization strategy development (Pehlivan et al. 2018), third party logistics provider selection (Ecer 2018) and supplier evaluation (Stević et al. 2019). On the other hand, Çakır (2018); Juodagalvien et al. (2017) and Kargı (2019) developed crisp-based SWARA EDAS, and applied it to fitness center, house plan shape and steel processing risk level evaluation problems, respectively. Liang et al. (2018) used picture fuzzy sets and integrated ELECTRE to EDAS for cleaner production evaluation problem, and Ghorabaee et al. (2018) developed SWARA CRITIC EDAS in a triangular fuzzy environment and evaluated construction equipment. Asante et al. (2020) handled renewable energy barrier evaluation problem and developed EDAS MULTIMOORA with crisp numbers. Vesković et al. (2020) used trapezoidal fuzzy numbers for business balance solution selection and proposed an EDAS PIPRECIA integration, and Kundakcı (2019) presented a study for steam boiler alternative evaluation by using triangular fuzzy numbers based on EDAS-MACBETH.

Methodology

In this section, the methodology of the proposed spherical fuzzy AHP EDAS is presented as follows: Section 3.1 gives the preliminaries of spherical fuzzy sets prior to model development, and Sect. 3.2 presents the steps of the model in detail.

Preliminaries of spherical fuzzy sets

Definition, basic operators, aggregation operator, normalized Euclidean distance (Kutlu Gündoğdu and Kahraman 2019) and defuzzification operator (Gündoğdu and Kahraman 2020) developed for spherical fuzzy sets are given below:where

andwhere (u), (u), and (u) are the degrees of membership, non-membership, and hesitancy of u to , respectively. Equation 2 becomes as below for on the surface of the sphere,Basic operations Addition, multiplication, multiplication by a scalar, and power of spherical fuzzy sets are given as follows:

Addition.Multiplication.Multiplication by a scalar. ()Power of . ()Aggregation operator The aggregation operation is a function that is used in those situations when we need to find a single value representing the set of various numbers. An aggregation operator converts a number of input data into a single value. In this study, geometric mean operator is used for aggregating the matrices. Geometric mean of n spherical fuzzy numbers is given in Eq. 8.

Spherical Weighted Geometric Mean (SWGM).where

Other operators The transformation of a crisp number to its appropriate fuzzy form is called fuzzification. The inverse of this operation, that is transformation of a fuzzy number to its crisp form, is called defuzzification. The normalized Euclidean distance, on the other hand, is a distance measure used for calculating the differences between two elements, two sets, etc. In fuzzy set theory, it can be useful, for example, for the calculation of distances between fuzzy sets. In spherical fuzzy, use of Euclidean distance becomes meaningful since whole space in the sphere is considered, and Euclidean distance gives the minimum distance between two points in a three-dimensional space. Defuzzification operator and normalized Euclidean distance for spherical fuzzy sets are given in Eqs. 9 and 10, respectively.

Defuzzification operator. Normalized Euclidean distance.

Proposed model: Spherical fuzzy AHP-EDAS

The developed model consists of two stages: In the first stage, criteria weights are determined by using AHP methodology and in the second stage, alternatives are ranked with EDAS. AHP is used as an auxiliary technique to deal with multi-layered and extended criteria set; since it allows structuring the problem and comparing pairs of elements in the hierarchy. The first stage of the proposed model is as follows: Main and sub-criteria are evaluated by DMs in a pairwise manner by utilizing readily given linguistic terms; pair-wise comparison matrices for each main and sub-criteria are constructed from the linguistic evaluations; spherical fuzzy criterion weight matrices are obtained and aggregated by utilizing spherical weighted geometric mean operator; global sub-criterion weights are calculated to be used in the next stage of the model.

In the second stage, EDAS methodology is used for ranking the alternatives since this methodology can be used for objective and subjective criteria; its application is easy; it directly ranks the alternatives by comparing them with the average solution, and it has already been tested in other fuzzy environments. The second stage of the model is as follows: Alternatives are evaluated by DMs with respect to sub-criteria and by utilizing readily given linguistic terms, spherical fuzzy alternative evaluation matrices are constructed from linguistic evaluations of DMs and then aggregated with spherical weighted geometric mean operator; prior weights of sub-criteria are included to the model by using spherical fuzzy multiplication by a scalar operator and final decision matrix is obtained in a spherical fuzzy form; average solution is determined and distances of alternatives to the average solution is calculated by using normalized Euclidean distance; appraisal scores of alternatives are obtained from normalized positive and negative distances from the average solution. The flowchart (Fig. 2) and the steps of the proposed model in detail are given below:

Fig. 2

Flowchart of the proposed model: Spherical fuzzy AHP-EDAS

Define an MCGDM problem with a set of alternatives and the most related main and sub-criteria set. Use a unified criteria type by converting cost criteria to benefit criteria. Let be the selected alternatives as: ; be the criteria set as: and let be the weight vector of all criteria as: where and

AHP stage: Obtaining main and sub-criteria weights Step 1. Evaluate the main and sub-criteria. Let each DM fills a pairwise comparison matrix for each main and sub-criteria set by utilizing linguistic terms presented in Table 1.

Table 1

Linguistic terms and corresponding spherical fuzzy numbers

Priority in pairwise comparisons	μ,ν,π	Score index (SI)
Absolutely more importance	(0.9,0.1,0.0)	9
Very high importance	(0.8,0.2,0.1)	7
High importance	(0.7,0.3,0.2)	5
Slightly more importance	(0.6,0.4,0.3)	3
Equally importance	(0.5,0.4,0.4)	1
Slightly low importance	(0.4,0.6,0.3)	1/3
Low importance	(0.3,0.7,0.2)	1/5
Very low importance	(0.2,0.8,0.1)	1/7
Absolutely low importance	(0.1,0.9,0.0)	1/9

The score indices (Gündoğdu and Kahraman 2020) in Table 1 can be calculated by utilizing Eqs. 11 and 12.where and are the membership, non- membership and hesitancy degrees, respectively.

Step 2. Transform evaluations to spherical fuzzy numbers Transform linguistic pairwise comparison matrices to their corresponding spherical fuzzy numbers and obtain spherical fuzzy pairwise comparison matrices (SFPCMs). Step 3. Check the consistency.Calculate the consistency ratio (CR) of each SFPCM by utilizing readily given corresponding score indices in Table 1, and apply the classical consistency check. If the CR of any SFPCM is less than 0.1, continue to Step 4, otherwise go to Step 1 and let the DM re-evaluate that main or sub-criteria. The random consistency index (RI) values used in the calculation of CR are given in Table 2.

Table 2

Random consistency index (RI)

n	1	2	3	4	5	6	7	8	9	10
RI	0.00	0.00	0.58	0.90	1.12	1.24	1.32	1.41	1.45	1.49

Step 4. Obtain spherical fuzzy criterion weight matrices. Aggregate SFPCMs to obtain spherical fuzzy criterion weight matrices (SFCWMs) by utilizing Eq. 8. Step 5. Aggregate spherical fuzzy criterion weight matrices. Aggregate SFCWMs from all DMs to obtain one aggregated spherical fuzzy criterion weight matrix (ASFCWM) for each main and sub-criteria set by utilizing Eq. 8. Step 6. Convert criterion weight matrix to its crisp form. Obtain score indices by defuzzifying ASFCWMs. Utilize defuzzification operator that is given in Eq. 9. Step 7. Normalize the score indices. Normalize score indices by utilizing Eq. 13 and obtain main criteria weights and sub-criteria local weights.Step 8. Obtain global sub-criteria weights. Obtain global sub-criteria weights by multiplying local weight of each sub-criterion by its corresponding weight of main criterion. Note that the evaluated global weights are the prior weights for EDAS stage. AHP stage of the model ends here.

EDAS stage: Ranking of alternatives

Step 1. Evaluate the alternatives. Let each DM fills the performance evaluation matrix by utilizing linguistic terms presented in Table 1. In this step, the performance of each alternative is evaluated with respect to each sub-criterion.

Step 2. Transform evaluations to spherical fuzzy numbers. Transform linguistic performance evaluation matrix to its spherical fuzzy form and obtain spherical fuzzy alternative evaluation matrix (SFAEM). The structure of the SFAEM is given in Eq. 14.where the evaluation of alternative with respect to criterion

Step 3. Aggregate spherical evaluation matrices. Aggregate SFAEMS from all DMs and obtain one aggregated spherical fuzzy alternative evaluation matrix (ASFAEM) by utilizing Eq. 8.

Step 4a. Obtain spherical fuzzy decision matrix. Construct the spherical fuzzy decision matrix by multiplying ASFAEM by sub-criterion global weights evaluated in AHP stage of the model. Use spherical fuzzy multiplication by a scalar operator that is given in Eq. 6. The structure of the spherical fuzzy decision matrix is given in Eq. 15.Step 4b. Calculate the crisp form of Step 4a for comparison purpose. Obtain crisp form of Step 4a by utilizing defuzzification operator that is given in Eq. 8.

Step 5a. Calculate spherical fuzzy average solution. Determine the spherical fuzzy average solution (SFAS) by utilizing SWGM operator that is given in Eq. 8. The structure of the SFAS is given in Eq. 16.where , and are the membership, non-membership and hesitancy degrees of the average solution.

Step 5b. Calculate the crisp form of Step 5a for comparison purpose. Obtain crisp form of SFAS by utilizing defuzzification operator that is given in Eq. 9.

Step 6. Determine the positions of the alternatives according to the average solution. Determine the positively positioned alternatives and negatively positioned alternatives based on the crisp values obtained in Step 4b and Step 5b.

For each sub-criterion, the alternatives with a higher score value than the SFAS are assigned as positively positioned alternatives; and the alternatives with a lower score value than the SFAS are assigned as negatively positioned alternatives. Note that the crisp values in Step 4b and Step 5b are evaluated only for obtaining positively positioned alternatives and negatively positioned alternatives.

Step 7. Calculate positive and negative distances to average solution. Calculate the total distance between positively positioned alternatives and the SFAS by utilizing normalized Euclidean distance given in Eq. 10 and call it positive distance to average solution (PDAS) for that alternative.

Calculate the total distance between negatively positioned alternatives and the SFAS by utilizing normalized Euclidean distance given in Eq. 10 and call it negative distance to average solution (NDAS) for that alternative.

Step 8. Normalize Step 7. Calculate normalized PDAS and normalized NDAS values by utilizing Eqs. 17 and 18.Step 9. Obtain appraisal scores. Obtain appraisal score for each alternative by utilizing Eq. 19.Step 10. Rank the alternatives. Rank all alternatives according to descending values of appraisal scores. The alternative with the highest appraisal score is the best choice among the candidate alternatives.

Application

In this section, proposed model is applied to a video conferencing selection problem. A higher education institution that has switched to distance education has to choose a video conferencing tool in this context. For this purpose, three DMs from the institution evaluate five selected popular videoconferencing tools with respect to six main, and 32 extended sub-criteria from the literature.

The organization of this section is as follows: Sect. 4.1 presents the main and sub-criteria from the literature; Sect. 4.2 introduces the alternatives and Sect. 4.3 presents the numerical solution in a step by step manner

Main and sub-criteria

According to the literature review, the evaluation criteria for distance education tool selection problem are divided into six main criteria as technical performance; usability; functionality; privacy; security and cost. These six main criteria are extended from the literature to more comprehensively handle the problem, and 32 sub-criteria are obtained as follows:

Technical performance. Technical support (Knapczyk et al. 2005), quality of video (Malinovski et al. 2014), quality of recording (Daman and Mishra 2011), video–audio synchronization (Thenmozhi and Kannan 2018).

Usability. Ease of use of videoconferencing (Mujačić et al. 2014), maximum number of participants (Hughes 2020), user interface (Thenmozhi and Kannan 2018), ease of discussion with others (Correia et al. 2020), ease of sharing data (Mujačić et al. 2014), ease of sharing desktop/whiteboard (Elmoula et al. 2017), ease of starting a meeting, ease of joining a meeting and navigability (Thenmozhi and Kannan 2018).

Functionality. Off-line functionality (Spathis and Dey 2020), integration with other systems and applications (Hurst 2020), potential of customization (Hacker et al. 2020), maximum number of participants per meeting, maximum duration of one group meeting, video calling features (Correia et al. 2020), text messaging features (Elmoula et al. 2017), and audio calling features (Correia et al. 2020).

Security. Confidentiality, integrity and availability (Strohl et al. 2020).

Privacy. Fake avatars (Strohl et al. 2020), cyberbullying (Lorenz 2020), face recognition attacks (Strohl et al. 2020), malware attacks (Venkatraman and Alazab 2018) and information leakage (Kagan et al. 2020).

Cost. Cost per month (Sidpra et al. 2020) and additional costs (Gray et al. 2020).

Alternatives

Videoconferencing tools are widely used for distance education (Martin 2005), and Zoom, Microsoft Teams, Google Meet, Skype and Cisco WebEx are the most preferred video conferencing tools around the world (Hughes 2020). Especially during COVID 19, people in schools have started to use videoconferencing tools such as Google Meet, Microsoft Teams and Zoom very widely. The popular videoconferencing tools used for education are Zoom, Google Meet, Microsoft Teams, Cisco WebEx, and Skype (Amin and Sundari 2020). In this paper, 5 commonly used video conferencing tools namely Zoom; Google Meet; Microsoft Teams; Cisco WebEx and Skype are selected for the MCGDM problem and a brief summary of these alternatives are given below: Zoom. Zoom is a software platform which provides video, audio and messaging services for distance education around the world (Islam et al. 2020) and since it enables face-to-face meetings, it is frequently used by educators. (Stefanile 2020). Google Meet. Google Meet is a web-based videoconferencing tool provided by Google (Sathish et al. 2020). It is used by students for educational purposes during COVID 19 pandemic and can be preferred by universities due to security reasons (Purwanto and Tannady 2020). Microsoft Teams.Microsoft Teams is a communication tool which provides video meetings and chatting (Sathish et al. 2020), and it is a good teaching space providing an elegant solution for distance education purposes after physical closure of higher education institutions during COVID 19 pandemic (Pal and Vanijja 2020). Cisco WebEx.Cisco WebEx is an online tool which makes collaboration of people easier through video, images and sound from anywhere, and is used perfectly within distance education (Tis’ah and Taher 2020). Cisco WebEx can be used for distance education which has become very important during COVID 19 (Harie et al. 2020). Skype.Skype is a telecommunication tool and provides its users to make video and audio calls, send instant messages and exchange files (Kristóf 2020), and it is commonly preferred as a distance education tool (Ashrapova et al. 2020), and used by students to interact (Husein and Purnawarman 2019).

Note that the names of the alternatives are not given in the numerical application of the model due to privacy concerns.

Numerical solution

The structure of the MCGDM problem is given in Fig. 3, and the step-by-step solution is presented below:

Fig. 3

Structure of the MCGDM problem

AHP stage: calculation of main and sub-criteria weights Step 1.Three DMs separately fill pairwise comparison matrices for each main and sub-criteria set by utilizing the linguistic terms.

Step 2. Linguistic pairwise comparison matrices are transformed to SFPCMs. Note that there are 21 SFPCMs.

Step 3.CR is calculated for each SFPCM and it is found that the all CRs are less than 0.1.

Step 4. SFPCMs are aggregated and SFCWMs are obtained. Linguistic pairwise comparison matrices and corresponding SFCWMs for main and sub-criteria are given in Tables 3, 4, 5, 6, 7, 8 and 9 respectively.

Table 3

Lingustic pairwise comparison matrices and corresponding SFCWMs for main criteria

		T	U	F	S	P	C	CR	SFCWM
DM1	T	EI	SMI	SMI	SLI	SLI	SLI	0.039	(0.37,0.62,0.28)
	U	SLI	EI	EI	VLI	LI	SLI		(0.48,0.52,0.32)
	F	SLI	EI	EI	LI	SLI	LI		(0.37,0.61,0.28)
	S	SMI	VHI	HI	EI	EI	SMI		(0.61,0.36,0.31)
	P	SMI	HI	HI	EI	EI	SMI		(0.59,0.37,0.32)
	C	SMI	SMI	HI	SLI	SLI	EI		(0.52,0.47,0.31)
DM2	T	EI	SMI	EI	SLI	SLI	LI	0.089	(0.44,0.54,0.32)
	U	SLI	EI	EI	ALI	LI	SLI		(0.33,0.67,0.26)
	F	EI	EI	EI	LI	LI	LI		(0.39,0.59,0.30)
	S	SMI	AMI	HI	EI	EI	SMI		(0.62,0.35,0.31)
	P	SMI	HI	HI	EI	EI	EI		(0.58,0.37,0.34)
	C	HI	SMI	HI	SLI	EI	EI		(0.56,0.42,0.32)
DM3	T	EI	EI	SMI	LI	SEI	EI	0.089	(0.47,0.48,0.35)
	U	EI	EI	EI	VLI	LI	SLI		(0.38,0.60,0.30)
	F	SLI	EI	EI	LI	SLI	LI		(0.37,0.61,0.28)
	S	HI	VHI	HI	EI	EI	SMI		(0.62,0.34,0.30)
	P	EI	HI	HI	EI	EI	SMI		(0.58,0.37,0.34)
	C	EI	SMI	HI	SLI	SLI	EI		(0.51,0.47,0.33)

Table 4

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of technical performance main criterion

		T1	T2	T3	T4	T5	CR	SFCWM
DM1	T1	EI	VLI	LI	SLI	SLI	0.063	(0.34,0.65,0.26)
	T2	VHI	EI	SMI	HI	HI		(0.65,0.33,0.27)
	...	...	...	...	...	...		...
	T5	SMI	LI	SLI	SLI	EI		(0.43,0.56,0.30)
DM2	T1	EI	LI	LI	SLI	SLI	0.084	(0.37,0.62,0.30)
	T2	HI	EI	SMI	HI	SMI		(0.62,0.36,0.28)
	...	...	...	...	...	...		...
	T5	SMI	SLI	SLI	EI	EI		(0.47,0.50,0.35)
DM3	T1	EI	LI	VLI	SLI	SLI	0.092	(0.34,0.65,0.29)
	T2	HI	EI	SMI	HI	VHI		(0.65,0.33,0.24)
	...	...	...	...	...	...		...
	T5	SMI	VLI	LI	EI	EI		(0.39,0.60,0.32)

Table 5

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of usability main criterion

		U1	U2	...	U9	CR	SFCWM
DM1	U1	EI	SMI	...	VHI	0.010	(0.65,0.34,0.26)
	U2	LI	EI	...	SMI		(0.50,0.47,0.33)
	...	...	...	...	...		...
	U9	VLI	LI	...	EI		(0.34,0.65,0.26)
DM2	U1	EI	EI	...	VHI	0.016	(0.63,0.35,0.29)
	U2	SLI	EI	...	SMI		(0.51,0.45,0.34)
	...	...	...	...	...		...
	U9	VLI	LI	...	EI		(0.33,0.66,0.24)
DM3	U1	EI	SMI	...	VHI	0.027	(0.65,0.34,0.26)
	U2	LI	EI	...	SMI		(0.53,0.42,0.35)
	...	...	...	...	...		...
	U9	VLI	LI	...	EI		(0.34,0.65,0.26)

Table 6

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of functionality main criterion

		F1	F2	...	F8	CR	SFCWM
DM1	F1	EI	LI	...	VLI	0.025	(0.27,0.73,0.20)
	F2	HI	EI	...	SLI		(0.49,0.48,0.33)
	...	...	...	...	...		...
	F8	VHI	SMI	...	EI		(0.59,0.41,0.29)
DM2	F1	EI	LI	...	VLI	0.080	(0.25,0.76,0.19)
	F2	HI	EI	...	SLI		(0.49,0.48,0.33)
	...	...	...	...	...		...
	F8	VHI	SMI	...	EI		(0.56,0.41,0.32)
DM3	F1	EI	ALI	...	VLI	0.055	(0.24,0.76,0.18)
	F2	AMI	EI	...	SLI		(0.51,0.46,0.32)
	...	...	...	...	...		...
	F8	VHI	SMI	...	EI		(0.61,0.39,0.27)

Table 7

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of privacy main criterion

		P1	P2	P3	P4	P5	CR	SFCWM
DM1	P1	EI	SLI	SMI	LI	EI	0.028	(0.45,0.53,0.32)
	P2	SMI	EI	HI	SLI	SMI		(0.55,0.44,0.31)
	...	...	...	...	...	...		...
	P5	EI	SLI	SMI	LI	EI		(0.45,0.53,0.32)
DM2	P1	EI	LI	SMI	LI	EI	0.055	(0.42,0.56,0.30)
	P2	HI	EI	HI	SLI	SMI		(0.55,0.44,0.31)
	...	...	...	...	...	...		...
	P5	EI	SLI	SMI	SLI	EI		(0.47,0.50,0.34)
DM3	P1	EI	SLI	SMI	LI	EI	0.088	(0.45,0.53,0.32)
	P2	SMI	EI	HI	SLI	SMI		(0.55,0.44, 0.31)
	...	...	...	E...	...	...		...
	P5	EI	SLI	SMI	LI	EI		(0.45,0.53,0.32)

Table 8

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of security main criterion

		P1	P2	P3	CR	SFCWM
DM1	S1	EI	SMI	HI	0.033	(0.59,0.37,0.32)
	S2	SLI	EI	SMI		(0.49,0.48,0.34)
	S3	LI	SLI	EI		(0.39,0.59,0.30)
DM2	S1	EI	SMI	VHII	0.031	(0.62,0.35,0.30)
	S2	SLI	EI	SMI		(0.49,0.48,0.34)
	S3	VLI	SLI	EI		(0.34,0.65,0.27)
DM3	S1	EI	SMI	AMI	0.074	(0.65,0.34,0.30)
	S2	SLI	EI	SMI		(0.49,0.48,0.34)
	S3	ALI	SLI	EI		(0.27,0.73,0.23)

Table 9

Linguistic pairwise comparison matrices and corresponding SFCWMs for sub-criteria of cost main criterion

		C1	C2		CR		SFCWM
DM1	C1	EI	SLI		–		(0.45,0.52,0.35)
	C2	SMI	EI				(0.55,0.40,0.35)
DM2	C1	EI	LI		–		(0.39,0.59,0.30)
	C2	HI	EI				(0.59,0.35,0.32)
DM3	C1	EI	SLI		–		(0.45,0.52,0.35)
	C2	SMI	EI				(0.55,0.40,0.35)

Step 5.SFCWMs from all DMs are aggregated and ASFCWMs are obtained. Step 6.Score indices are calculated by defuzzifying the ASFCWMs. Step 7.Main criteria weights and sub-criteria local weights are obtained by normalizing score indices. Step 8.Global sub-criteria weights are obtained by multiplying local weight of each sub-criterion by its corresponding weight of main criterion. ASFCWMs, score indices, local weights, and global weights of main and sub-criteria are given in Table 10. Note that global weights are the prior weights to be used in EDAS stage.

Table 10

ASFCWMs, score indices, local weights and global weights of main and sub-criteria

Main/Sub criterion	ASFCWM			Score index			Local weight				Global weight
T	(0.46,0.51,0.33)			12.207			-				0.156
T1	(0.35,0.64,0.27)			9.247			0.141				0.022
T2	(0.64,0.34,0.28)			17.757			0.271				0.042
...	...			...			...				...
T5	(0.43,0.56,0.31)			11.345			0.173				0.027
U	(0.36,0.63,0.28)			9.282			-				0.119
U1	(0.64,0.35,0.27)			17.867			0.				0.018
U2	(0.51,0.45,0.34)			13.638			0.156				0.014
...	...			...			...				...
U9	(0.34,0.65,0.25)			8.874			0.077				0.009
F	(0.38,0.60,0.29)			9.894			-				0.126
F1	(0.25,0.70,0.19)			6.400			0.062				0.008
F2	(0.50,0.47,0.33)			13.176			0.127				0.016
...	...			...			...				...
F8	(0.59,0.40,0.30)			16.116			0.155				0.020
S	(0.62,0.35,0.31)			16.931			-				0.216
S1	(0.62,0.35,0.31)			17.022			0.440				0.095
S2	(0.49,0.48,0.34)			13.085			0.338				0.073
S3	(0.33,0.66,0.26)			8.589			0.222				0.048
P	(0.58,0.37,0.33)			15.764			-				0.201
P1	(0.44,0.54,0.31)			11.757			0.181				0.036
P2	(0.56,0.43,0.31)			15.119			0.233				0.047
...	...			...			...				...
P5	(0.46,0.52,0.33)			12.037			0.185				0.037
C	(0.53,0.46,0.32)			14.205			-				0.181
C1	(0.43,0.54,0.33)			11.111			0.425				0.077
C2	(0.56,0.39,0.35)			15.056			0.575				0.104

AHP stage of the model ends here. The details of EDAS stage is as follows:

EDAS stage: Ranking of alternatives Step 1. Performance evaluation matrices are filled by three DMs by utilizing the linguistic terms. Step 2.Linguistic performance evaluation matrices are transformed to their corresponding spherical fuzzy forms and three SFAEMs are obtained. Step 3.SFAEMs are aggregated to obtain one ASFAEM as given in Table 11.

Table 11

Aggregated spherical fuzzy alternative evaluation matrix (ASFAEM)

	T1	T2	...	C2
A1	(0.87,0.14,0.06)	(0.68,0.29,0.26)	...	(0.80,0.20,0.10)
A2	(0.73,0.29,0.20)	(0.71,0.27,0.26)	...	(0.70,0.30,0.20)
A3	(0.70,0.31,0.22)	(0.83,0.17,0.08)	...	(0.70,0.30,0.20)
A4	(0.70,0.30,0.20)	(0.83,0.19,0.12)	...	(0.61,0.42,0.23)
A5	(0.66,0.34,0.24)	(0.80,0.20,0.10)	...	(0.77,0.24,0.14)

Step 4a.Spherical fuzzy decision matrix is obtained as presented in Table 12.

Table 12

Spherical fuzzy decision matrix

	T1	T2	...	C2
A1	(0.17,0.96,0.02)	(0.16,0.95,0.08)	...	(0.32,0.85,0.05)
A2	(0.13,0.97,0.04)	(0.17,0.95,0.08)	...	(0.26,0.88,0.09)
A3	(0.12,0.97,0.05)	(0.22,0.93,0.04)	...	(0.26,0.88,0.09)
A4	(0.12,0.97,0.04)	(0.22,0.93,0.03)	...	(0.22,0.91,0.09)
A5	(0.11,0.98,0.05)	(0.21,0.93,0.03)	...	(0.30,0.86,0.07)

Step 4b.Spherical fuzzy decision matrix is defuzzified as presented in Table 13.

Table 13

Defuzzified spherical fuzzy decision matrix

		T1		T2			...				C2
A1		2.165		2.046			...				8.511
A2		2.548		2.622			...				6.477
A3		2.845		4.777			...				6.477
A4		2.840		4.702			...				4.816
A5		3.072		4.138			...				7.744

Step 6.Positively positioned alternatives and negatively positioned alternatives are determined as presented in Tables 15 and 16, respectively.

Table 15

Positively positioned alternatives

	T1	T2	...	C2
A1	(0.17,0.96,0.02)	(0.16,0.95,0.08)	...	(0.32,0.85,0.05)
A2	(0.13,0.97,0.04)	(0.17,0.95,0.08)	...	(0.26,0.88,0.09)
A3	(0.12,0.97,0.05)	(0.22,0.93,0.04)	...	(0.26,0.88,0.09)
A4	–	–	...	–
A5	–	–	...	–

Table 16

Negatively positioned alternatives

	T1	T2	...	C2
A1	–	–	...	–
A2	–	–	...	–
A3	–	–	...	–
A4	(0.12,0.97,0.04)	(0.22,0.93,0.03)	...	(0.22,0.91,0.09)
A5	(0.11,0.98,0.05)	(0.21,0.93,0.03)	...	(0.30,0.86,0.07)

Steps 7 and 8. Normalized PDAS and normalized NDAS are obtained as presented in Table 17.

Table 17

Normalized PDAS and normalized NDAS of alternatives

	PDAS	NDAS		Normalized PDAS	Normalized NDAS
A1	0.019	0.012		1.000	0.315
A2	0.015	0.009		0.784	0.461
A3	0.012	0.015		0.638	0.162
A4	0.015	0.018		0.797	0.000
A5	0.018	0.017		0.938	0.020

Steps 9 and 10. Alternatives are ranked according to descending appraisal scores as presented in Table 18.

Table 18

Appraisal scores of alternatives

Alternative								Appraisal score
A1								0.657
A2								0.622
A3								0.400
A4								0.398
A5								0.479

The ranking of alternatives is obtained as follows:

Alternative A1 is selected since it has the highest appraisal score. Further discussions for the results are given in the next section.

Discussions

In this section, the results of our proposed spherical fuzzy AHP EDAS are analyzed. Section 5.1 gives sensitivity analysis for DM and criterion weights, Sect. 5.2 analyzes the consistency of our model with other fuzzy MCGDM models, and Sect. 5.3 discusses the implications.

Sensitivity analysis

Sensitivity and comparative studies are conducted to validate our model and statistical comparison of the ranks obtained from these studies are analyzed by calculating the Spearman’s rank correlation coefficient. Sensitivity analysis for DM weights For conducting a sensitivity analysis for DM weights, ten different DM weight scenario are obtained by systematically changing their weights as given in Fig. 4. The results show that A1 is the best alternative in eight of the ten scenarios, and A2 is the second. Spearman’s rank correlation coefficients are calculated to analyze these rankings as presented in Table 19, and it is seen that the proposed model gives highly correlated results for different DM weight distribution scenarios.

Fig. 4

Sensitivity analysis results for DM weights

Table 19

Statistical comparison of the ranks obtained from DM weight sensitivity analysis

	1	2	3	4	5	6	7	8	9	10	Av.
1	1.00	1.000	0.964	0.964	1.000	0.964	0.988	0.988	0.855	0.891	0.961
2	–	1.000	0.964	0.964	1.000	0.964	0.988	0.988	0.855	0.891	0.957
3	–	–	1.000	1.000	0.964	0.964	0.988	0.952	0.903	0.806	0.947
4	–	–	–	1.000	0.964	0.964	0.988	0.952	0.903	0.806	0.940
5	–	–	–	–	1.000	0.964	0.988	0.988	0.855	0.891	0.948
6	–	–	–	–	–	1.000	0.952	0.988	0.915	0.867	0.944
7	–	–	–	–	–	–	1.000	0.964	0.867	0.842	0.918
8	–	–	–	–	–	–	–	1.000	0.879	0.903	0.927
9	–	–	–	–	–	–	–	–	1.000	0.806	0.903
10	–	–	–	–	–	–	–	–	–	1.000	1.000

Sensitivity analysis for criterion weights. Sensitivity analysis for criterion weights is conducted as follows: All sub-criterion weights are placed in a total of 32 excel cells and 32 sub-criteria are divided into eight packages with each package containing four cells. Sub-criteria weights are systematically shifted in excel cells and the algorithm is executed for eight different scenarios as shown in Fig. 5. Spearman’s rank correlation coefficients are calculated to analyze these rankings as presented in Table 20. The analysis shows that the proposed model gives quite correlated results; however, it can be said that DM evaluation has an effect on the final ranking of alternatives, which requires selecting a qualified group of DMs for the MCGDM process.

Fig. 5

Sensitivity analysis results for criterion weights

Table 20

Statistical comparison of the ranks obtained from criteria weight sensitivity analysis

		1	2	3	4	5	6	7	8		Av.
1		1.000	0.667	0.786	0.714	0.595	0.667	0.619	0.619		0.708
2		–	1.000	0.643	0.976	0.976	0.690	0.952	0.810		0.864
3		–	–	1.000	0.619	0.619	0.690	0.595	0.690		0.702
4		–	–	–	1.000	0.952	0.738	0.905	0.857		0.890
5		–	–	–	–	1.000	0.786	0.976	0.881		0.911
6		–	–	–	–	–	1.000	0.762	0.952		0.905
7		–	–	–	–	–	–	1.000	0.810		0.905
8		–	–	–	–	–	–	–	1.000		1.000

Comparative analysis

Comparative analysis is conducted by handling our problem with other five MCGDM models developed in spherical fuzzy (SF) and intuitionistic fuzzy (IF) environments. The selected models are IF AHP EDAS; SF AHP TOPSIS; IF AHP TOPSIS; SF AHP CODAS; and IF AHP CODAS. The appraisal scores obtained from six different MCGDM models are presented in Table 21. Spearman’s rank correlation coefficients are calculated to analyze these rankings as presented in Table 22. Analysis of the results shows that the proposed model is in a complete correlation with other five MCGDM models.

Table 21

Appraisal scores of alternatives for six different MCGDM model

	SF AHP EDAS	IF AHP EDAS	SF AHP TOPSIS	IF AHP TOPSIS	SF AHP CODAS	IF AHP CODAS
A_1	0.657	0.897	0.582	0.767	0.022	0.046
A_2	0.622	0.630	0.545	0.571	0.008	0.008
A_3	0.400	0.302	0.436	0.377	-0.026	-0.038
A_4	0.398	0.125	0.461	0.278	-0.010	-0.058
A_5	0.479	0.935	0.521	0.758	0.006	0.043

Table 22

Statistical comparison of the ranks obtained from comparative analysis

	SF AHP EDAS	IF AHP EDAS	SF AHP TOPSIS	IF AHP TOPSIS	SF AHP CODAS	IF AHP CODAS	Av.
SF AHP EDAS	1.000	0.829	0.943	0.943	0.943	0.943	0.934
IF AHP EDAS	–	1.000	0.771	0.943	0.771	0.943	0.886
SF AHP TOPSIS	–	–	1.000	0.886	1.000	0.886	0.943
IF AHP TOPSIS	–	–	–	1.000	0.886	1.000	0.962
SF AHP CODAS	–	–	–	–	1.000	0.886	0.943
IF AHP CODAS	–	–	–	–	–	1.000	1.000

Steps 5a and 5b. SFAS and defuzzified SFAS are presented in Table 14.

Table 14

Spherical fuzzy average solution (SFAS) and defuzzified SFAS

	T1	T2	...	C2
SFAS	(0.13,0.97,0.04)	(0.19,0.94,0.062)	...	(0.27,0.88,0.08)
Defuzzified SFAS	2.480	3.690	...	6.745

Implications

This study has significant implications which can be summarized as follows:

A wide set of criteria is presented by scanning the literature which provides a practical guidance and a better understanding of videoconferencing tool selection problem for distance education and other fields.

A nine-scale linguistic table is presented to evaluate the alternatives. This table can be converted to a spherical fuzzy form which enables capturing the hesitancy and vagueness in the nature of the problem. Spherical fuzzy sets, with their three-dimensional characters, are known to provide a large domain for this purpose.

Hybrid nature of the model provides the advantage of AHP, which divides the problem into sub-particles in a pairwise manner, and makes it possible to deal with problems having many main and sub-criteria.

EDAS part of the model directly determines the positions of the alternatives from average solution, and this makes the methodology more straightforward than most of other MCGDM models.

The presented model can be considered as a kind of decision support tool that has been mathematically proven to be reliable and valid, and flexible enough to be used not only by higher education institutions but also in other sectors.

Conclusion, limitations and future avenue

The aim of this study is to develop a mathematical decision support model for the selection of videoconferencing systems that can be used in distance education. The study is unique in terms of both the subject and the developed model. Within the scope of the study, the AHP technique is integrated with the EDAS methodology. What makes our model truly unique here is that the developed model is presented in a spherical fuzzy environment, which allows us to include the hesitancy and vagueness in the nature of the problem. None of the studies previously presented around distance education topic had the advantages of spherical fuzzy sets to handle the problem. The proposed hybrid model is presented in two stages as AHP and EDAS. Essentially, the AHP part of the model acts as a kind of auxiliary method since the reason using AHP in this model is to calculate the weights of the wide and layered set of criteria. The flagship of our model is the EDAS method where the ranking of the alternatives is carried out. The details of the proposed spherical fuzzy AHP EDAS is given in a step-by-step manner with a flowchart for the whole model. Sensitivity and comparative analyses are also given the study, and Spearman’s rank correlation analysis is conducted to the results obtained from these analyses. The presented model can be evaluated as a comprehensive decision support tool to assist solving problems with multi-layered criteria while modeling the vagueness in the problem. Limitations. The followings can be considered as the limitations of this study:Future research avenue. For future studies, the followings are recommended:

In general, AHP does a good job in problems with multi-layered criteria set; however, the problem of consistency can arise when the number of elements in any matrix exceeds nine.

In this study, 32 sub-criteria are presented under six main criteria, and usability main criterion has nine sub-criteria which means the corresponding matrix has nine elements.

Although the consistency analysis is successfully performed in this study, there may be a problem when the matrix size increases.

Spherical fuzzy division operator has not been developed yet, and this may sometimes cause difficulties in model development. This operator can be developed by other scholars.

The proposed model can be applied to other sectors.

AHP stage of the model can be finalized in spherical environment, and the weights of criteria can be integrated to EDAS stage by utilizing spherical fuzzy multiplication operator that is given Eq. 5.

Machine learning algorithms (i.e., logistic regression) can also be integrated to the model to decrease the expert dependency and make use of available data in the institution.

A technique different from the AHP can be developed which can cope with a wide and layered set of criteria and at the same time not create a consistency problem for matrices having more than nine elements.

Rank reversal phenomenon can be analyzed in future studies.