Weight of criteria in hotel selection: An empirical illustration based on TripAdvisor criteria

The aim of this paper is to show how to calculate the weight of criteria that influences the decision making process of potential hotel guests. By taking into consideration the growing importance of electronic word-of-mouth and the popularity of hotel reviews websites, this research note exposes the framework to calculate the weight of the 6 TripAdvisor criteria by using the Multi-Criteria Decision Analysis (MCDA) – Analytic Hierarchy Process (AHP) method. 250 tourists in Paris (France) have been interrogated in order to apply the AHP method and to illustrate how to determine the weight of 6 criteria: location, bed quality, comfort & equipment, service, value for money and, cleanliness. By doing so, this research note facilitates the understanding of AHP method and its application to the hotel sector. The empirical study proposes weight for each criterion by taking into consideration tourists’ point of view.


Introduction
Due to the competitive business environment, it is indispensable for hotel managers to understand how potential customers choose their hotel and which criteria have been privileged in their decision-making processes; moreover by considering that Internet has dramatically changed the customer purchasing behaviour (Buhalis & Law, 2008).According to Nielsen (2013) report, 70% of customers trust online reviews.So the reviews, widely known as word-of-mouth (WOM), are considered as one of the most influential factors in customer decision-making process.
WOM is defined as a post-purchase oral communication between people where communicator gives non-commercial information on a brand, a service, a product or a company based on his personal experience (Dickinger & Basu, 1994;Harrison-Walker, 2001).With the development of information

Weight of criteria in hotel selection
Criteria play a vital role in different stages of decision-making process (Yavas & Babakus, 2005).Multiple researches have been done on multi-criteria decision-making process in hotel selection (Chou et al., 2008).Most of them are focused on "Customer's Quality Perception" (Oh, 1999;Albayrak & Caber, 2015) from a post-experience point of view.Accordingly, they try to understand the contribution of hotel attributes on overall customer satisfaction (Tsaur & Tzeng, 1996;Chu & Choi, 2000;Albayrak & Caber, 2015).Studies show that some attributes are highly important to satisfy the customers while others play insignificant roles.On the contrary, very few researches have been focused on the weight of attributes that play a vital role in customer decisionmaking process.TripAdvisor proposes 6 attributes which are subject to customers' evaluation during their stay in the respective hotel: location, bed quality, comfort & equipment, service, value for money and, cleanliness.While a potential customer searches for a hotel on TripAdvisor, he or she goes through the eWOM based on his or her own criteria, which are subjective (Figure 1).According to Crouch (2011), the decision making process is based on an array of information connecting multiple criteria.Serra-Cantallops and Salvi (2014) mentioned the necessity of calculating the weight of these criteria.Therefore, the research gap on the

Table 1. Performance Matrix
Decision-making criteria (C) The decision-maker needs to evaluate and decide on n alternatives represented by the vector A: (Ai; i=1, 2, 3, …n), which could be analysed by m criteria represented by the vector C: (Cj; j=1, 2, 3, …, m).The weight of criteria varies from person to person and it could be represented by the vector W: (Wj; j = 1, 2, 3, ..., m).The assessment of each of the alternatives i on each of the criteria j gives the results Rij as presented in Table 1.

Analytic Hierarchy Process (AHP) and the weight of criteria
AHP method has been developed by Saaty (1977Saaty ( , 1980) ) and it is very useful when the decision maker is unable to construct a utility function (Ishizaka & Nemery, 2013).In this regard, pair-wise comparison of criteria is a solution.The decision maker gives a relative verbal appreciation between two criteria rather than a numerical judgement.(2) = m

If two criteria
Since the observed matrix A might not be consistent, the estimation of w could be calculated as follow (where  max represents the maximum Eigen-Value of matrix A): (3) Aw =  max w The value of the  max could be estimated by normalizing the element of in each column.In order to measure the inconsistency (A is consistent if  max =m and we always have the   max m), consistency ratio (C.R.) could be calculated by using the following steps: In order to calculate the weight of each criterion for potential customers by AHP method, we conducted a pair-wise comparison between these 6 criteria by questioning 250 tourists in strategic tourist spots and some 4 star hotels in Paris (cf.survey copy in Appendix 1).The interviewees confirmed that they always consulted eWOM on TripAdvisor before making any purchase.
In Table 4, a pair-wise comparison matrix has been done which is an example for one particular tourist interrogated in Paris (31 years old, leisure traveller).In order to rapidly evaluate the weight of criteria for each questionnaire, at first, we completed the upper triangular matrix (in bold letters) by using the value from the Table 2.Then, lower triangular matrix was completed by the reciprocal values of the upper triangular matrix.
The next step consists of normalisation of the matrix.Therefore, we got the weight of each criterion from the average of each line.C.R. = 0.1 which means that the result is acceptable.Thus, 250 questionnaires have been calculated individually and the average result shows that the cleanliness has the highest value, which affirms Ryan & Huimin's (2007) study.However, Albayrak & Caber (2015) found that  and Saleh (1992) found that they were subject to duration of stay.
Among the 250 tourists, 69% stayed less than 3 nights and therefore, result shows that the weight of sleep quality and, comfort and equipment increases by the duration of stay.
Interviewees, who earn more than 25000 US dollar annually (36% of respondents), look for better location, comforts and additional services.
The weight of criteria not only fills the previous research gap (Serra-Cantallops & Salvi, 2014) but also gives an insight on eWOM of Trip Advisor for hoteliers in order to adopt the "bestpractice".

Conclusion, limits and perspectives
This research note shows how MCDA-AHP method can be used for calculating the weight of criteria.It is the first research note that measures the weight of criteria of eWOM of TripAdvisor from a potential customers' perspective.AHP method could be used to evaluate others decision-making criteria in order to anticipate the need and desire of international tourists.One of the limitations of this research note is that the sample is based on only 250 respondents.Further studies could be focused on more detailed information such as business vs.leisure traveller, tourists' country of origin, socio-demographic situations, etc.

Figure 1 .
Figure 1.Decision-Making Process on Hotel Selection (developed from Crouch (2011: 32) and based on TripAdvisor criteria).
C h and C k are compared, e hk indicates the evaluation of the criterion C h relatively to the criterion C k .The pair-wise comparison matrix A is composed of all the relative evaluations by considering that e hk = 1/ e kh.By taking into consideration the Eigen-Value formulation, (1) Aw = mw where w = (w 1 , w 2 …, w m ) and mw could be obtained as follow:

( 4 )
Consistency Index (C.I.) = (max -m)/(m -1) Therefore, (5) C.R. = C.I. / R.I. C.R. should be less or equal to 0.1 in order to get the result acceptable.The C.I. of a randomly generated pair-wise comparison matrix (Saaty, 1980) and m = order of matrix.An empirical illustration based on Parisian tourists AHP method is applied.The data was collected from 250 tourists in Paris during 2015 summer, aged between 30 to 65 years, travelling for leisure purpose and used to stay in 3 to 5 stars hotels while travelling and, needless to say, choosing their hotels on TripAdvisor.By taking into consideration TripAdvisor guests' satisfaction attributes, we consider the 6 criteria: c 1 = location of the hotel, c 2 = bed/sleep quality, c 3 = comfort & equipment, c 4 = service/personnel, c 5 = value for money and, c 6 = cleanliness.

Table 2 .
Insight of Saaty Scale weight of criteria in hotel selection based on eWOM has been identified but still remained untouched.

Table 4 .
Pair-wise comparison matrix (example of a tourist)

Table 5 .
Normalisation of matrix (example of a tourist)

Table 6 .
Weight of criteria according to duration of stay