A relativistic value-based approach to interpreting e-rating and e-complaining behaviour in the hospitality sector

This study broadens the understanding around the topics of e-rating and e-complaining by adopting a value-based approach. In the consumer behaviour literature, value is understood as a “relativistic preference experience” (i.e. an evaluation outcome that varies across subjects, objects and contexts). E-ratings and e-complaints can be analysed through this theoretical lens to investigate whether any significant differences exist relating to tourists’ socio-demographic characteristics (variation across subjects: gender and nationality), the characteristics of the hotels rated (across objects: hotel category and size) and the party with which the tourists are travelling and the time of their stay (across contexts). This study measures and interprets these differences in a sample of 727 reviews in which tourists expressed their experience of hotels located in Arzachena-Costa Smeralda (Sardinia, Italy). The findings reveal that significant variations in travellers’ rating behaviour exist only across objects (i.e. hotel category and size) and contexts (i.e. time of the stay). However, when complaining behaviour is considered, significant variations are reported to exist across subjects (i.e. gender and country of origin), objects (hotel category and size) and contexts (i.e. travel party). Contributions to the theory and managerial implications are discussed, and suggestions for further research are made.


Introduction
User-generated content (UGC) has introduced a new venue for easily voice both positive and negative word-of-mouth (WOM) recommendations (e.g. Ekiz et al., 2012). UGCs have been found to direct and influence tourists' choices (Mauri and Minazzi, 2013;Xiang et al., 2015), especially when they are uploaded in Online Travel Agencies . Further, they often alter decisions about accommodation after having obtained further information online (Del Chiappa, 2011b), and can even induce tourists to change the accommodation suggested by a traditional travel agency (Del Chiappa, 2013;. This explains why UGC are considered to exert a relevant influence on hotel bookings (e.g. Ye et al., 2011), with ecomplaints being extremely harmful to companies (Bambauer-Sachse and Mangold, 2011).
All that said, online consumer feedback has become essential to understand customers' satisfaction/dissatisfaction (Berezina et al., 2015;Kim et al., 2016;Li et al., 2013), to recover and improve hotel services (Dellarocas, 2003;Stringam and Gerdes, 2010) and to achieve higher levels of customer satisfaction/commitment , customer retention and loyalty (Stauss and Schoeler, 2004;Zheng et al., 2009). Said in other words, the ubiquity of peer-to-peer platforms has provided new sources of information for studying tourists' behaviour and their satisfaction with service providers (e.g. Chen et al., 2016;Zhou et al., 2014).
Despite the huge attention given to the field of online reviews, there is still a need to further develop academic research devoted to this area, with particular attention to the analysis of e-rating, e-complimenting (Khoo-Lattimore and Ekiz, 2014) and e-complaining behaviour (Berezina et al., 2015). In particular, there is a need do deepen the analysis of the variable that can moderate these behaviours. This study was therefore carried out to contribute to this research area through the lens of a valuebased approach. This approach is deeply rooted in marketing and consumer behaviour literature (Gallarza et al., 2011;Holbrook, 1999) and has been used in tourism literature for consumers' evaluations in many settings, including hotels (Gallarza et al., 2016). According to Marketing Science (2012) value and experience are considered among the durable topics, and currently there is a call for future research aiming to re-interpret these concepts "integrating behavioural theory and marketing frameworks into big data marketing" (Marketing Science Institute, 2016).
To our best knowledge there has been no published paper on e-rating and/or ecomplaining which have followed an experiential perspective, adopting a relativistic value-based approach, and therefore providing a new, albeit traditional, appraisal of the understanding of these electronic behaviours, grounding their nature in the classical consumer behaviour literature.
Analysing data from UGCs concerning e-rating and e-complaining through the lens of the value concept is an attempt to respond to the MSI call. Using a value-based approach where value is "relativistic" (Holbrook, 1999) -that is, it changes according to subjects, objects and contexts -this study aims to better understand whether e-complaining behaviours follow different patterns based on the variability of guests (e.g. gender and nationality), the context of the service consumption (e.g. travel party, time of the stay) and the objectivity of the hotel business (e.g. hotel size, hotel category). To achieve this aim, the study presents and discusses the findings of an empirical analysis of 727 online reviews in which tourists expressed their experience of hotels located in Arzachena-Costa Smeralda, a famous 3S (sun, sea and sand) tourism destination in Sardinia (Italy).
This study deepens the current body of knowledge related to e-rating and ecomplaining behaviour offering a value-based interpretation; meanwhile, its findings support hotel practitioners in their attempt to better understand how to direct their limited resources to the features of the hotel that receive the highest number of complaints and that reflect the characteristics of their market targets and/or of their hotel businesses the most.
The paper is organized as follows. First, we provide a brief explanation of the research trend, born in the 1980s, concerning the experiential aspects of consumption, ending in Holbrook's (1999) value conceptualization. Second, we organize our review on e-WOM, seeking to highlight its experiential nature.
Third, the empirical investigation into the relativistic character of e-WOM is addressed, in which two forms of e-WOM in hotels (i.e. erating as a form of customer satisfaction and ecomplaining behaviour) are tested as subjective, objective and contextual behaviours. The final section includes the conclusions and recommendations for readers and researchers.

Literature review
Consumer value and the experiential approach to consumption phenomena If it was possible to choose one main and durable concept in marketing research this would probably be "consumer value", as it has continuously been used and revisited by marketing and consumer behaviour scholars for the last 40 years (e.g. Arnould, 2014 and the literature review in Gallarza et al., 2017). In addition to the aforementioned MSI research priorities (Marketing Science Institute, 2012, in services marketing literature, value is also considered as one of the 12 priorities for service research (Ostrom et al., 2015, p. 128). It is also noticeable how the contemporary approach to Service-Dominant Logic (SDL) has thrown new light onto this old concept (Helkkula et al., 2012), as value co-creation is at the core of this approach since it started (e.g. Vargo and Lusch, 2008), but also in its very latest reformulation, the word "value" being considered in two of the four axioms of the SDL (Lusch and Vargo, 2016).
Back to the history of the concept, one of the most well-known conceptual approaches to the notion of value corresponds to Holbrook, with early works Holbrook and Hirschman, 1982) and more recent ones (Holbrook, 2006;Holbrook 2007a, b and c) in which he posted, following a phenomenological approach, a conceptual framework on value as an experiential paradigm for studying consumer behaviour (Holbrook, 1999).
This particular approach, after two decades of research, was capitalized in a conceptual framework on value (Holbrook, 1994(Holbrook, , 1999 that contains both a definition and a typology. Holbrook's definition of value is referred to as an axiology (that is, as a judgement of goodness/badness) and thus pursues a philosophical approach that conceives consumer value as "an interactive relativistic preference experience" (Holbrook, 1999, p. 5). Value is understood as interactive because it entails an interaction between a subject (a consumer) and an object (a product or service); it is also preferential because it embodies a preference judgement, mainly emotional and very often related to words such as affect (pleasing vs. displeasing), attitude (like vs. dislike) and/or evaluation (good vs. bad). By experience it is understood that value resides not in the product purchased or in the brand chosen but rather in the consumption experience derived from the purchase. The last characteristic of this definition claims that value is relativistic because it is understood as personal (it varies across people), comparative (with differences among objects) and situational (specific to the context).
This well-known approach to consumer behaviour research has been the object of broad theoretical discussions among academicians (in chronological order Babin et al., 1994;Rust and Oliver, 1994;Oliver, 1999;Smith, 1999;Boksberger and Melsen, 2011;Arnould, 2014;Gallarza et al., 2016;Gallarza et al., 2017), and it has resulted in many empirical studies with value scales derived from Holbrook's conceptualization, for instance in retailing (e.g. Mathwick et al., 2001), travel and tourism (e.g. Babin and Kim, 2001;Gallarza and Gil-Saura, 2006), service restaurants (e.g. Sánchez-Fernández et al., 2009) and hospitality settings (e.g. Gallarza et al., 2016;Gallarza et al., 2015). Our work aims to contribute further to this previous trend of applications and discussions on consumer behaviour through Holbrook's conceptualization of value by taking one step further, relating consumer value to electronic behaviour (erating and e-complaining).
Many works on value have added to value research after Holbrook (1999)'s proposal, complementing and further enhancing the relevance of such a concept for researchers and marketers (e.g. Helkkula et al. 2012;Khalifa, 2004;Payne et al., 2017;Vargo and Lusch, 2016). For the purpose of this study, we have focussed our attention on work by Helkkula et al. (2012, p. 59) who characterises value phenomenologically as experiential, defining it "…as an ongoing, iterative circular process of individual, and collective customer sense making": this view is clearly in line with our proposal of adopting a value-based approach for understating e-behaviours, which are necessary "ongoing", "iterative", and "individual/collective".

Electronic WOM as an experiential behaviour: a value-based approach to erating and e-complaining
In the last two decades, a myriad of empirical works on the use of UGC by consumers have swamped our marketing and consumer behaviour journals. As the field is so prolific, a wide range of verbal expressions used by academicians and marketers can be found, all of them referring to the same phenomenon: customers' e-complaints (e.g. Lee and Hu, 2004;Zheng et al., 2009), word-of-mouse rating (e.g. Stringam and Gerdes, 2010), complaining in cyberspace (e.g. Sparks and Browning, 2010), digitalization and digital of word of mouth (e.g. Crotts et al., 2009;Dellarocas, 2003) or more simply e-complaints (e.g. Au et al., 2014), online complaints (e.g. Ekiz et al., 2012), e-word of mouth (e.g. Zhang et al. 2010) and online users' reviews (e.g. Ye et al., 2009). This wide semantic scope evidences the prolific empirical fieldwork on the use of UGC, where e-complaint has definitely been a preferred field among e-behaviours. We have revised the underpinnings of complaint behaviour, to better understand the peculiarities of e-complaining, and therefore the relevance of a value based approach to explain it.
According to Singh (1988), consumer complaint behaviour fits into two categories, namely behavioural (i.e. action) and non-behavioural (i.e. no action), with different types of action that can be public (i.e. seeking a refund from the company) or private (i.e. word of mouth to friends and relatives). In line with this vein of research, e-WOM, in our value-based approach, can be considered as a public behavioural action a) that has an experiential outcome that occurs during and/or at the end of a dynamic process of evaluation, b) that contains intrinsically embedded affection and emotions and c) that allows consumers to socialize with other peers regarding their consumption experience (i.e. social focus).
Firstly, negative WOM can occur during and/or at the end of a consumption experience when consumers, based on value assessments (e.g. Gallarza et al., 2011), face a service failure  and feel dissatisfied with the service provider (Oliver, 1997). Accordingly, a value-based approach can be adopted to deepen our understanding of erating (as a way of expressing satisfaction) and e-complaining (as the end of a process of evaluation) behaviour.
Secondly, based on the literature review provided by Verhoef et al. (2009, p. 3), many authors have considered that "emotion is key to customer experiences". According to Sparks and Browning (2010, p. 799), "a negative online review is an expression of dissatisfaction that is posted with the possible aims of venting emotions, engaging with others, and sharing information". That said, it is evident that 2.0 travel applications have certainly provided consumers with a new and easy way to air grievances (Ekiz et al., 2012) that often retrace the negative emotions that were elicited by the service failure (Del Chiappa and Dall'Aglio, 2012;Maurer and Schaich, 2011). In other words, social media and UGC are the contemporary way of dealing with emotions, that is with "fantasies, feelings, and fun" as used in Holbrook and Hirschman's (1982) original expression for naming experiential values in consumption. Hence, e-complaining behaviour can be interpreted and explained effectively through the lens of the value-based experiential approach to consumption.
Thirdly, in line with value being both self-and other-oriented (Holbrook, 1999), the social focus of e-complaining behaviour is also clear, as it is a kind of behaviour that, through sharing information and knowledge about the service provider, aims to support social interactions with peers and to help them to make the right choices when selecting a service provider (Dwivedi et al., 2007;Maurer and Schaich, 2011). In this sense it could be argued that a richer view of social interactions is possible through e-WOM in comparison with traditional WOM. Indeed, especially when individuals interact in virtual spaces over a long period of time, they often develop a sense of community and trust the comments posted online (Dwivedi et al., 2007) to a great extent (Del Chiappa, 2011b). e-WOM offers individuals the chance to activate interrelationships between customers before and after experiencing the products and/or service, making this interaction with others a key driver of e-complaining behaviour (Litvin et al., 2008). Expressions such as "online feedback forum" (Tuzovic, 2010) and "Internet complaint forum" ) make this social dimension, and therefore its experiential nature, evident.
In sum, because of its sequential, emotional and social nature, e-WOM (both positive and negative) can be understood as experiential behaviour, allowing us to analyse it through a value-based approach. In addition to this, an extra rationale for adopting this approach is that e-WOM (e-rating and e-complaining) is clearly preferential. As consumer value is "an interactive relativistic preference experience" (Holbrook, 1999, p. 5), complaint behaviour is derived from axiological behaviour, as it assumes a preference judgement. These preference judgements, as true experiential behaviour, in Holbrook's nomenclature, should be "relativistic", varying among persons, products or services and situations; following this conceptual rationale that bridges e-WOM with an experiential value judgement, it is its relativistic nature that our empirical study aims to prove.

Conceptual background and research questions' postulation
This empirical study aims to contrast the relativistic nature (subjective, objective and contextual-based) of e-WOM behaviour (erating and e-complaining) within a hotel experience. This is performed by testing whether differences in e-rating and e-complaining do exist according to subjects (customers' demographic traits, i.e. gender and nationality), objects (the characteristics of the hotel business, i.e. hotel size and hotel category) and contexts (the circumstances under which the consumption experience took place, i.e. the time of the stay and the social interaction occurring given to the travel party).
Methodologically, the empirical objective is therefore to test differences on e-rating and ecomplaining (dependent variables) by subjects, objects and contexts (independent variables). The research questions supporting these tests of differences are presented in Table 1 and explained in sections 3.1, 3.2 and 3.3. The rationale for those assumptions are based on Holbrook's (1999) conceptualization of experiential consumption. However, in the following paragraphs, to reinforce the rationale, additional studies devoted to the analysis of the relativistic nature of e-WOM, tested inside and outside tourism settings, were considered.
The existing literature has suggested that complaining behaviour can be influenced by psychographic characteristics (Morganosky and Buckley, 1987), contextual-based situationspecific attributes (Stauss and Seidel, 2004) and operational/business indicators, such as hotel class, average daily rate and popularity index (Del Chiappa and Dall'Aglio, 2012;Jeong and Jeon, 2008). Accordingly, our methodological approach states that e-WOM can be understood as a relativistic consumption experience; more precisely, the value-based approach followed in our study considers erating and e-complaining as "relativistic" because the behaviour in question is simultaneously subjective, objective and contextual.
We thus seek to assess valuations of service consumption (hospitality) that are meant to be subjective (different based on the sociodemographic characteristics of the guests), varying between different objects (different hotels in terms of star category and size) and evaluated in different contexts (period of the year and travel party).

E-rating and e-complaints vary between subjects (they are personal)
Generally, early empirical research has shown that any consumption behaviour is influenced by demographic characteristics such as gender, age and level of education (e.g. Richins, 1983). Moreover, experiential marketing highlights the intrinsically personal relativity of consumption behaviour (Holbrook, 1999;Holbrook and Hirschman, 1982) and considers that clear differences exist in experiential behaviour among different individuals. In hospitality-based studies, personal differences in e-WOM have been found between men and women (e.g. Del Chiappa and Dall'Aglio, 2012; Maurer and Schaich, 2011), between consumers with different nationality (e.g. Au, Buhalis and Law, 2014;Del Chiappa and Dall'Aglio, 2012;Ekiz and Au, 2011;Ngai et al., 2007) and between leisure and business travellers (e.g. Lewis and McCann, 2004). As far as the objectives of our study are concerned, these previous works have suggested the adoption of an experiential value-based approach to analyse e-rating and e-complaining behaviour and proposed the following research questions:

E-rating and e-complaining vary between objects (they are comparative)
The perception of the outcome of any consumption experience must always be comparative (Holbrook, 1999;Oliver, 1999) and always varies between objects experienced by consumers. That said, it has been proven that consumers cannot take into account all the positive and negative aspects of the experience at once (Nilson, 1992). Similarly, consumers cannot compare simultaneously their perceptions relating to many different objects (Siracaya et al., 1996). Despite this, it has been proven that complaining behaviour is influenced by operational/business indicators, such as the hotel class, average daily rate and popularity index (e.g. Park et al., 2008). Hence, for the purposes of our study, we postulate the following research questions with the aim of investigating whether significant differences exist in e-complaining behaviour across different objects.

E-rating and e-complaining vary with the dimensions of space and time (they are contextual)
Dimensions of time and space (i.e. context) are extremely important in the evaluation process of any products and services (Gabott and Hogg, 1998;Holbrook, 1999;Zeithaml, 1988). According to Gabott and Hogg (1998), the nature of value depends on the products themselves, on the individual who buys them and on the situation or context. As far as complaining behaviour (both offline and online) is concerned, it is rather obvious to assume that it can be influenced significantly by the context of consumption (De Matos et al., 2012;Stauss and Seidel, 2004). In tourism-related settings, the circumstantiality of consumption judgements is even greater (Van Raaij, 1986). Prior studies (e.g. Del Chiappa and Dall'Aglio, 2012; Park et al., 2008) have reported that the travel party influences complaining behaviour. Hotel services are both seasonal and perishable, which raises the marketing stress level for hospitality managers, and their staff, struggling in trying to guarantee a productive capacity (e.g. staff, labour, equipment, etc) as efficiently and as profitably as possible based on the imbalances in tourism demand that usually occur based on the time of stay of their guest (e.g Lovelock, 1984;Rao and Singhapakdi, 1997), the main distinction between low season, middle season and high season being especially considered. That said, we decided to consider in our analysis this temporal-based contextual variable which, to the best of our knowledge, has not been previously investigated in e-rating (ecomplaining behaviour). Hence, the contextual nature of consumer value suggests the following research questions:

Method
In the tourism-related literature, many studies have used UGC as secondary data to assess guests' satisfaction with hotel services (e.g. Zhou et al., 2014) or to investigate e-rating and e-complaining behaviour (e.g. Crotts et al., 2009;Lee and Hu, 2004;Pan et al., 2007;Ye et al., 2009;Zhang et al., 2010). Hence, for the purposes of the current study, 727 online reviews of hotels in Arzachena-Costa Smeralda (Italy) were sourced from an online travel agency (OTA) with booking and rating/review functions (booking.com).
Arzachena, a municipality with 13,561 inhabitants located in the northeast of Sardinia (Italy), was selected as the research site for this study for its relevance in the national and international tourism market. Firstly, the region of Sardinia as a whole, the second largest island in the Mediterranean Sea, can be considered as a tourism destination with an international appeal given its unique history, authenticity, culture, environment and remoteness. Secondly, Arzachena includes the well-established and famous Costa Smeralda, one of the best-known luxury tourism destinations in the world, originally created by the Prince Aga Khan and currently mostly owned in its real estate by the Qatar Holdings Investment. On the whole, Arzachena offers a wide range of three, four and five-star hotel facilities both independent and hotel chains (i.e. Starwood, ITI Hotels), with five-star hotels located mostly in the Costa Smeralda area. According to ISTAT (http://dati.istat.it/), the Italian national statistics office, in 2015 Arzachena had 71 hotel facilities of which ten were five-star hotel (14.08%), 31 were four-star hotels (43.66%), and 23 three-star hotels (32.39%).
For the purposes of this study all the online reviews about the hotels were sourced from Booking.com, thus resulting in 727 comments to be analysed. The full text of all each comment was copied and pasted into a spreadsheet for later manual coding, as in previous works (e.g. Tse and Zhang, 2013), including information related to the socio-demographic (i.e. gender culture/country of origin) and trip-related characteristics (i.e. travel party and period of stay) of the consumers posting the comments and several hotel operational/business indicators (i.e. hotel class and size).
Booking.com only accepts hotel reviews from travellers who have made a reservation for that hotel and paid for it on their website. Therefore, compared with other studies (e.g. Jeong and Jeon, 2008;Maurer and Schaich, 2011;Zheng et al., 2009), we were able to reduce/overcome the risk of building our analysis on online reviews posted by competitors. Booking.com users can assess their overall satisfaction with the hotel services using a ten-point Likert scale applied to a list of six attributes (cleanliness, comfort, location, services, staff and value for money). These six attributes are related to the literature on value dimensions, for instance efficiency and excellence in Holbrook (1999) or value for money and functional value in Sweeney and Soutar (2001), thus providing a rationale for the use of these secondary data when adopting an experiential value-based approach to e-WOM. Additionally, people are allowed to post both positive and negative comments on the Booking.com site.
A content analysis method was adopted for the purpose of this study, more specifically we applied a manual holistic approach (Kim et al., 2016). This choice was made for several reasons. First, we found that some satisfaction (dissatisfaction) indicating reviews contained negative (positive) experiences about the hotel services despite the fact that the customers were providing a positive (negative) overall rating to the hotel. Further, we found that quite often some words depicting certain service features were commented in a way that its nature (compliment vs. complaint) could not have been ascertained without a proper semantic contextualisation with the whole linguistic structure of the comment. As a matter of fact, similarly to what was highlighted in previous studies (Kim et al., 2016), a computerbased analytical program would have never allowed us to discern the differences in the semantic meaning of the frequency by which certain features-related words were cited in the comment. Finally, the use of a computer analytical program would not have allowed us to organise the complaints by the variables (e.g. gender, travel party, hotel size, etc) we were aiming to consider as moderating dimensions of e-complaining behaviour. This explains why we opted for an in-depth manual review method. First, with the aim of establishing a coding table, a comprehensive list of hotel attributes and a typology of satisfiers/dissatisfiers were adopted based on prior studies (Kim et al., 2016). Hence, the list was shared with a group of local hotel managers (with a working experience of at least 10 years); by doing this, the list was partially added/re-worded. This was done to increase the ability of our findings to provide information that would be able to consider at the greatest extent the views/needs of local managers and suggest effective managerial implications. The coding list was useful to establish valid categories for data coding. The data analysis was done following the method and process adopted in previous studies (e.g. Sparks and Browning, 2010;Zhou et al., 2014). Specifically, to enhance the external validity and ensure the reliability of the human coding, the study followed a team approach to analyse the data obtained from the reviews. Initially, one researcher read the reviews and carried out basic open coding. Subsequently, the initial codes were reviewed by the research team and an independent person was brought in to revise the coding and decide whether he/she agreed with the codes. Whenever the research team and the independent reviewer disagreed about the coding adopted, they discussed the point until an agreement was reached, and the final coding was then undertaken.
At the end of the process, it emerged that all the negative comments were triggered by, or targeted at, specific features internal or external to the hotel. In particular, and according with the coding list, we considered the following 12 categories: 1. "rating and promises" (i.e. coherence with the classification rating, expected services not delivered), 2. "location and surroundings" (i.e. position of the hotel, view, external environment, etc.), 3. "access and parking" (i.e. accessibility, parking facilities, etc.), 4. "design, look and feel" (i.e. shape, decoration, atmosphere, general maintenance, etc.), 5. "staff" (i.e. empathy, competence, responsiveness, attitude, etc.), 6. "organization" (i.e. general organization, room assignment, opening hours of guest services, etc.), 7. "room experience" (i.e. comfort, cleanliness, space, equipment, air conditioning, etc.), 8. "bathroom" (shower, towels, courtesy products, etc.), 9. "breakfast" (i.e. quality, quantity, variety, etc.), 10. "food and drink" (i.e. quality/quantity of food and drink in the restaurant and/or bar), 11. "facilities and services" (i.e. common spaces, garden, swimming pool, beach, shuttle services, Internet, etc.) and, lastly, 12. "price and qualityprice ratio". In addition to their general correspondence with value dimensions such as service quality, efficiency, playfulness and aesthetics (Holbrook, 1999), these categories are in accordance with previous works satisfiers/dissatisfiers in online hotel review (Kim et al., 2016) or on service failures in the hospitality industry; in particular, "staff", "room experience" and "breakfast" are similar to the categories of Lewis and McCann (2004, p. 10), and "rating and promises" are coherent with previous works such as Lee and Hu (2004), who found that the most frequent complaints include "service provided is not as agreed".
When manually coding all the reviews, the research team extracted the negative feelings expressed in each of them; it was then possible to calculate the number of negative comments and the corresponding percentage for each of the 12 features above. If a guest complained about more than one aspect for the same feature, only one complaint was counted.
Overall, 430 e-complaints were identified. The data were coded and analysed using SPSS (version 17.0), then a series of ANOVA and chi-square tests were applied for the purpose of testing the research questions of the study.
As far as the socio-demographic profile of the guests uploading comments is concerned, 45% were male, 21% were female and 34% did not declare this information. In terms of their travel party, most uploaders were in a young (37%) or mature (24%) couple, followed by those travelling in a group/group of friends (13%), families with babies/young children (12%) or older children/teens (5%) and people travelling alone (8%). Individuals were international (69%) or national (31%) travellers who spent their holiday on the Emerald Coast in summer time (83%) or in mid-season (16%). Most of the comments were related to stays in four-star hotels (62%), followed by three-star (33%) and five-star (5%) hotels. In terms of hotel size, the comments were mostly related to middle-size hotels (50-150 rooms: 49%) followed by small hotels (< 50 rooms: 46%) and big hotels (> 150 rooms: 5%). S.D.=1.33). However, surprisingly, four-star hotels were found to be assessed higher than five-star hotels. This could be because of unmet promises in higher hotel categories and/or because of the positive rating given by guests to the personal touch and organization offered by smaller, often family-run, hotels.

Is e-rating a relativistic behaviour?
Table 3 also shows that significant differences exist in tourists' assessment based on the hotel size (F (4.122, p<0.05) (RQ2.2.a). Specifically, big hotels (those with more than 150 rooms) seem to be more exposed to lower ratings (M=7.39, S.D. =1.79). It could be argued that this occurs because big hotels lack the "human touch" found in small and mid-size hotels. Surprisingly, the overall assessment is higher in winter than in summer. This could be explained by arguing that in winter people enjoy relaxing in un-crowded hotels.

Is e-complaining a relativistic behaviour?
Out of the total of 727 reviews analysed, 430 included negative elements. Tables 5 to 8 record the most frequently mentioned complaints about hotel services (numbers in bold) made by travellers based on their gender, their cultural background (subjects' effect in our experiential approach), the hotel class/size (objects' effect), the time of year when they  stayed at the hotel and the party with whom they were travelling (context's effect). Firstly, regarding RQ1.2.a and RQ1.2.b, Table 5 shows that men generally complain more than women. In particular, women complain significantly more than men about "rating and promises" (X2=4.048, p<0.05), whereas men complain more about the "bathroom" (X2=5.163, p<0.05).
The following services/aspects, "rating and promises", "location and surroundings", "access and parking", "design, look and feel", "staff", "organization", "breakfast", "facilities and services" and "price/quality ratio", receive the most complaints from international guests, while Italian travellers complain more about the other aspects. In particular, international guests complain about "location and surroundings" and "price/quality ratio" significantly more than Italian travellers (organization: X2=4.090, p<0.05; food and drink: X2=4.552, p<0.05).
Secondly, Table 6 (RQ2.1.b and RQ2.2.b) shows that significant differences in complaining behaviour can be found for both  hotel class and hotel size. For example, threestar hotels receive complaints more frequently than four-and five-star hotels for "location and surroundings" (X2=9.848, p<0.01) and "room experience" (X2=6.439, p<0.05). Large hotels (those with more than 150 rooms) receive significantly more complaints for "design, look and feel" (26.3%) and "room experience" (36.8%) than medium-sized and small hotels (design, look and feel: X2=18.989, p<0.01; room experience: X2=11.569, p<0.01), thus showing the need to refurnish and upgrade hotel services and facilities. Interestingly, medium-sized hotels (50 to 150 rooms) receive the most frequent complaints for "staff" (7.6%) and "organization" (10.6%). This could be explained by their difficulty in reaching the "minimum efficient organizational dimension".
Thirdly, now analysing the effect of context, regarding RQ3.1.b, according to Table 7a, tourists complain differently depending on the party with whom they are travelling. For example, "room experience" is the aspect of most concern for families with older children/teenagers (25.6%), while families with babies/small children and those with older children and teenagers complain more frequently than the other groups about "access and parking" (4.5%). However, there are significant differences only for "staff", with singles making complaints more frequently than the other categories (X 2 =19.253, p<0.05), and price/quality ratio, with young couples making the most complaints (X 2 =11.717, p<0.01).
Finally, to provide an answer to RQ3.2.b, Table  7b shows that significant differences in complaining behaviour cannot be found based on the time of the stay.
However, the data show that hotels receive more complaints in summer than in any other period for "rating and promises" (34.6%), "access and parking" (2.2%), "design, look and feel" (8.9%), "staff" (7.6%), "room experience" (21.5%), "bathroom" (8.6%), "facilities and services" (10.3%) and "price/quality ratio" (6.1%). On the contrary, in mid-season hotels receive more complaints than in any other period for "facilities and services" (10.3%), breakfast (9.5%) and "food and beverage experience" (7.8%). This could be explained by the mid-low service profile that hotels deliver to consumers in low season to be able to cover their costs with the lower rates that they are able to adopt.

Conclusion
This paper has offered a value-based approach to negative e-WOM (i.e. e-complaining behaviour) and interpreted this behaviour as a relativistic experiential-based public action that consumers adopt after having experienced a service failure and dissatisfaction with hotel services.
The idea that negative e-WOM can be understood as an experiential outcome is based on several assumptions, namely that negative e-WOM is the outcome of a consumption process that is highly relativistic in nature and into which emotional and social aspects are intrinsically embedded. In fact, emotions and socialization are clear dimensions of e-WOM along with, on the other hand, the chance of activating interrelationships between customers before and after experiencing the offers.
Through this theoretical lens of experiential relativistic value, we presented an empirical study aiming to investigate whether significant differences exist in e-rating and e-complaining behaviour based on subjective-based characteristics (i.e. gender and nationality), objective-based characteristics (i.e. hotel size and hotel category) and contextual-based characteristics (i.e. time of the stay and travel party).
Based on an experiential and relativistic valuebased approach, our findings showed that significant variations in travellers' e-rating behaviour exist based on subjects (i.e. gender), objects (i.e. hotel category and size) and contexts (i.e. time of the stay). Specifically, the overall assessment was higher for females, four-star hotels, small-size hotels and for guests enjoying their stay in the low season. These findings confirm previous research in which four-star hotels were given a significantly more positive assessment than three-star hotels (Jeong and Jeon, 2008). However, surprisingly, smaller hotels and four-star hotels were found to be assessed higher than bigger accommodation and five-star hotels. This could be because of unmet promises in higher hotel categories and/or because of the positive rating given by guests to the personal touch and organisation offered by smaller, often familyrun, hotels. Further, they add to the current body of knowledge by highlighting that the assessment significantly differs based on the time of the stay, being higher in the low season rather than in mid or high season, as we could have expected. This could be explained by arguing that in winter and mid-season people enjoy relaxing in un-crowded hotels. Furthermore, it could be argued that this occurs because there are fewer guests staying in the hotel during the mid-season, resulting in less "organisation stress" and better customer service. Hospitality services are both seasonal and perishable, which raises the marketing stress level for hospitality managers and staff (Rao and Singhapakdi, 1997).
This study also provided an insight into the most frequently made complaints for the 12 hotel features that we investigated based on the aforementioned variables. Considering ecomplaining behaviour, significant variations were reported to exist across subjects (i.e. gender and country of origin), objects (hotel category and size) and contexts (i.e. travel party); no significant differences were reported to exist based on the period of the stay. Taking gender into account, our findings reported men generally complaining more than women, thus confirming prior research (e.g. Maurer and Schaich, 2011). However, a more detailed attribute-based analysis shows that women complain significantly more than men about "rating and promises". Quite surprisingly, men were reported to complain more than women about the "bathroom". International guests complain significantly more than Italian travellers about the "location and surroundings" and "price-quality ratio", thus further confirming that cultural differences need to be taken into account when analysing e-complaining behaviour (Ekiz and Au, 2011;Ngai et al., 2007). Regarding the travel party, singles were found to complain about "staff" significantly more than the others. Young couples complain significantly more about the "price-quality ratio". In relation to the hotel category, threestar hotels receive complaints more frequently than four and five-star hotels for "location and surroundings" and "room experience". Big hotels receive significantly more complaints for "design, look and feel" and "room experience".
These contrasting results (significant for some dimensions and not for others) show the complexity and multidimensionality of hospitality behaviour and call for a further analysis, similar to the one conducted by Lewis and McCann (2004) in relation to another contextual variable, the reasons for travelling. They found significant differences in hospitality complaint behaviour regarding this variable, with business travellers being more likely than leisure travellers to mention slow restaurant service, access to business facilities, inefficient staff and a poor variety of food. Unfortunately, this study could not consider this variable given that Arzachena, and Sardinia as a whole, mostly depend on leisure tourism (Zara and Cao, 2015), thus rendering it particularly hard to collect data (both traditional and through online review) from leisure and business travellers. Thus, future study could consider the possibility of taking into account this variable by selecting as a research site a tourism destination characterised by a reasonable mix of leisure and business travellers.
Discussion of these results has shown that the relativistic nature of complaining is also reported in other studies (e.g. Au et al., 2014;Maurer and Schaich, 2011;Ngai et al., 2007), but it also adds to literature on value, as our confirmatory results are in accordance with contemporary works on theories and value models such as Khalifa (2014, p. 661) that proposes future research questions on "negative disconfirmation of expectations as a customer" (i.e. reasons for e-complaining) being tested across segments to see "if all value models appeal to all type of customers". Using Khalifa (2014)'s and others models on value (i.e. SDL), for integrating e-behaviours into value-based approaches can build on the MSI priorities that guided our research (i.e. Integrate behavioural theory and marketing frameworks into big data marketing).
These conclusions are relevant both to researchers and to hospitality managers. From a theoretical perspective, to the best of our knowledge, the study is the first aimed at deepening the scientific debate around the topic of e-rating and e-complaints by adopting an experiential and relativistic valued-based approach. Accordingly, our findings confirmed previous research showing that consumers are often dissatisfied with basic services, including the failure to deliver a service or inappropriate behaviour of personnel (e.g. Lee and Hu, 2004). Further, they highlighted that e-complaining behaviour varies across subjects, objects and contexts of consumption. Interestingly, our results confirmed the pioneering studies analysing the influence that culture exerts on the complaining behaviour of tourists (e.g. Au et al., 2014;Ekiz and Au, 2011). Further, they add to the current body of knowledge about ecomplaining behaviour considering the moderator effect of variables (i.e. travel party, hotel size, and time of stay) that to our best knowledge have not been investigated so far.
Regarding marketing practices, these results provide useful information to hotel managers struggling in finding ways to direct their limited resources to the features of the hotel services that are the most frequently mentioned in online reviews and that reflect the characteristics of their market targets and/or of the hotel businesses that they manage, so that their overall rating on 2.0 platforms can increase and the number of e-complaints that they receive can be reduced. According to prior studies, in return this would allow them to increase their sales activity and their market share (e.g. Ye et al., 2011). Further, our findings that hotel managers must recognise the nuances in the way subjective, objective and contextual variables can affect the ecomplaining behaviour of their actual and prospective guests so that they can effectively develop service design strategies where the mix of the different service features (considered both quantitatively and qualitatively) is selected to better reflect the characteristics of their guests (subjective), of their business (objective) and of the context of the experience consumption (contextual). For example, hotel managers need to implement actions to keep under control the price-quality ratio of their offer (lowering the price and/or increasing the standards of their services) when they target international travellers; in fact, these guests were found to be complaining about this aspect more than Italians. Hotel managers working in three-star hotels should seriously take into account the need/opportunity to enhance the room experience (e.g. improving the equipment, internal atmosphere, comfort, cleanliness, air conditioning, etc). Given that the number of complaints that the biggest hotels received regarded "design, look and feel" when compared to smaller ones, our findings seem to suggest that it is worth investing more to recover and improve their shape, decoration, atmosphere and general maintenance. Finally, young couples were reported to complain significantly more than others guests about the price-quality ratio; hence, hotel managers seeking to attract and to satisfy the needs of this market segment should seriously keep this price-quality under control.
Finally, as the Internet is increasingly used to obtain information, it has become important for managers to analyse and manage their firms' and products' online brand reputation. The huge amount of UGC that the Internet makes available allows hotel managers to access consumers' judgement about their hotel services very easily and directly. Given the relevant role that UGC plays in influencing service expectations and in driving hotel choices, it is essential for hospitality managers to monitor online comments regularly and respond to each of them as soon as possible with the aim of protecting their e-brand reputation and positively influencing consumers' attitudes toward the property (e.g. Litvin and Hoffman, 2012).
That said, it could be argued that Web 2.0 technologies have become an important tool for hospitality marketers to manage complaints, although, unfortunately, it would seem that several hospitality companies show an unbalanced exploitation of Web 2.0 (e.g. Sigala, 2011) and often take no notice of ecomplaints, as it has been shown for the case of the Italian (e.g. Del Chiappa, 2011a) and Spanish hotel sectors (e.g. Del Chiappa, 2014).
Although this study helps to fill a gap in the existing knowledge in the literature and offers some implications for practitioners, its scope was restricted. Firstly, our study was based on a single online travel site and was limited to one tourism destination. Therefore, our findings cannot be generalized to other hotels operating in other tourism destinations. It could be argued that further research in different destinations is needed to establish whether these findings can be confirmed and/or whether the study should take a location factor into consideration. Secondly, when analysing CCB, we only considered how often complaints were made for each hotel feature and did not consider at all the extent to which each feature can influence customers' overall satisfaction. Previous research has shown that reviews including comments about negative experiences do not necessarily imply an overall negative assessment (Del Chiappa and Dall'Aglio, 2016;Jiang et al., 2010). It could be argued that a relatively high number of complaints for a relatively non-significant hotel feature would not be a priority for hotel managers (Zainol et al., 2010). Further research is needed, adopting an experiential and relativistic value-based approach, to recognize the most important factors (i.e. subjective, objective and contextual-based factors) that exert the most influence on the overall satisfaction of guests. Thirdly, future study could be run selecting tourism destinations with significant flows of both leisure and business travellers, in order to consider the moderating effect that this contextual factor could exert on e-rating and e-complaining behaviour. Finally, the analysis presented and discussed here was conducted interpersonally (among different subjects). Longitudinal studies could be undertaken with the aim of testing whether significant differences in e-complaining behaviour occur when different objective and contextual-based circumstances of consumption are considered for a given traveller.