Experiential Marketing of an Underground Tourist Attraction

: The theory of the ‘experience economy’ contends that consumers no longer achieve satisfaction from consuming products but from the experiences they receive while doing so. Producers should therefore actively manage the four experience ‘realms’ of their product o ﬀ erings—entertainment, education, aesthetics and escapism—to provide optimal experiences for their customers. In the case of tourist attractions, however, there is insu ﬃ cient direct empirical evidence to substantiate this recommendation. This study therefore sets out to test the notion of the experience economy in the context of a tourist attraction—in this case, an underground visitor experience in Wales, UK—using partial least squares structural equation modelling. Alternative models are estimated based on three di ﬀ erent mediating variables—arousal, memory and satisfaction—with revisit intention as the dependent variable. The analysis ﬁnds that none of the four experience realms are signiﬁcant predictors of revisit intention in all three of the models, even though all three mediating variables are signiﬁcant predictors of revisit intention. The results therefore suggest that optimal customer experiences do not necessarily need to be built equally upon all four experiences realms. Rather, a customised approach is required to optimise the customer experience for speciﬁc products consumed in particular contexts.


Introduction
Pine and Gilmore's [1,2] theory of the experience economy is now more than two decades old. Its core proposition is that consumers no longer seek simply to purchase products and services but to receive memorable, sense-arousing, satisfying experience encounters as they consume them. The concept was widely expected to revolutionise marketing practice insofar as existing approaches, which tended to focus on functional attributes and quality, would become increasingly outdated [3]. The conceptual model maps the elements of the consumption experience on two axes: passive versus active and absorptive versus immersive. This allows four 'realms' of the consumer experience to be defined (see Figure 1), these being the four 'E's of education, entertainment, escapism and esthetics (note that the US English spelling of aesthetics is conventionally used in discussion of the experience economy). Pine and Gilmore's [2] management proposition is that consumption typically involves all four realms of experience and that when they are all employed to their optimum, a 'sweet spot' is to be found, as represented by the central circle in Figure 1. At this point, the organisation achieves optimum market outcomes, which might be measured in terms of, for example, willingness to recommend the product or repurchase intention. A number of mediating variables could also come into play: memory and sensory arousal being among some of those suggested by later scholars. For the most part, however, the expected shift to experience-led marketing has not taken place. This includes the marketing of tourism products, especially tourism attractions, where experienceled marketing has not hitherto been widespread [4]. This is arguably due to scholars not making their case strongly enough. Indeed, much of the work on experience marketing remains ad hoc, particularly with regard to empirical studies [3]. Moreover, a common criticism of the experience economy concept is that there have been few attempts to develop and test reliable and valid scales for measuring customer experiences across the four experience realms [5]. Other critics have queried whether there is sufficient evidence to suggest that the four experience realms represent distinctive concepts in the minds of customers [3,6]. The few studies that have been undertaken in the tourism context have typically found the predictive abilities of the model to be limited. Mehmetoglu and Engen [7], for example, found only two of the experience realms, escapism and esthetics, to have a significant impact on visitor satisfaction in the case of a music festival, while a different pairing, education and esthetics, were significant in the case of visitors to a museum. Meanwhile the fourth realm, entertainment, had no significant effect on the outcome variable in either case. Mehmetoglu and Engen [7] thus concluded firstly that different experience realms will have greater or lesser importance in different contexts, and secondly that it is not necessary for all four experience realms to play a part in determining behavioural outcomes. This conclusion was also reached by Hosany and Witham [3], who found that while all four experience realms were significant determinants of memory, esthetics had by far the largest magnitude of effect. Such findings challenge the central proposition of Pine and Gilmore's [1,2] model of there being a target 'sweet spot' at which all four experience realms work together to optimise market outcomes. This paper sets out to test Pine and Gilmore's [1,2] experience economy framework, using the measurement scales developed by Oh et al. [6], in the context of an underground tourist attraction located in Wales called 'King Arthur's Labyrinth' (hereafter KAL). Situated in the former workings of the Braich Goch Slate Mines at Corris, Mid-Wales, KAL describes itself as an "underground storytelling adventure and time travel at its best" [8] (n.p.). The storytelling is based on various myths and legends, including those of King Arthur. The latter are told mainly through the stories that have been passed down through the literary heritage of Wales rather than the more widely known versions told in England and France. Earl [9] argues that there is limited evidence that King Arthur was indeed Welsh or even visited Wales. However, the KAL experience achieves a degree of authenticity through careful theming both above and below ground. Visitors to the attraction are given a conducted tour of a system of disused caverns and tunnels of a former slate mine, first by boat and then on foot, during which time stories from the ancient myths and legends of Wales are presented to them in a series of sound-and-light tableaux. Given the nature of the attraction offer, it might be expected that all four experience realms would play a critical role: education, through the telling of stories, some of which will be familiar but others not; escapism, by means of the journey underground; entertainment, through the use of audio-visual tableaux and the drama of the underground journey; and esthetics, generated by the underground ambience and medieval theming of the attraction. For the most part, however, the expected shift to experience-led marketing has not taken place. This includes the marketing of tourism products, especially tourism attractions, where experience-led marketing has not hitherto been widespread [4]. This is arguably due to scholars not making their case strongly enough. Indeed, much of the work on experience marketing remains ad hoc, particularly with regard to empirical studies [3]. Moreover, a common criticism of the experience economy concept is that there have been few attempts to develop and test reliable and valid scales for measuring customer experiences across the four experience realms [5]. Other critics have queried whether there is sufficient evidence to suggest that the four experience realms represent distinctive concepts in the minds of customers [3,6]. The few studies that have been undertaken in the tourism context have typically found the predictive abilities of the model to be limited. Mehmetoglu and Engen [7], for example, found only two of the experience realms, escapism and esthetics, to have a significant impact on visitor satisfaction in the case of a music festival, while a different pairing, education and esthetics, were significant in the case of visitors to a museum. Meanwhile the fourth realm, entertainment, had no significant effect on the outcome variable in either case. Mehmetoglu and Engen [7] thus concluded firstly that different experience realms will have greater or lesser importance in different contexts, and secondly that it is not necessary for all four experience realms to play a part in determining behavioural outcomes. This conclusion was also reached by Hosany and Witham [3], who found that while all four experience realms were significant determinants of memory, esthetics had by far the largest magnitude of effect. Such findings challenge the central proposition of Pine and Gilmore's [1,2] model of there being a target 'sweet spot' at which all four experience realms work together to optimise market outcomes. This paper sets out to test Pine and Gilmore's [1,2] experience economy framework, using the measurement scales developed by Oh et al. [6], in the context of an underground tourist attraction located in Wales called 'King Arthur's Labyrinth' (hereafter KAL). Situated in the former workings of the Braich Goch Slate Mines at Corris, Mid-Wales, KAL describes itself as an "underground storytelling adventure and time travel at its best" [8] (n.p.). The storytelling is based on various myths and legends, including those of King Arthur. The latter are told mainly through the stories that have been passed down through the literary heritage of Wales rather than the more widely known versions told in England and France. Earl [9] argues that there is limited evidence that King Arthur was indeed Welsh or even visited Wales. However, the KAL experience achieves a degree of authenticity through careful theming both above and below ground. Visitors to the attraction are given a conducted tour of a system of disused caverns and tunnels of a former slate mine, first by boat and then on foot, during which time stories from the ancient myths and legends of Wales are presented to them in a series of sound-and-light tableaux. Given the nature of the attraction offer, it might be expected that all four experience realms would play a critical role: education, through the telling of stories, some of which will be familiar but others not; escapism, by means of the journey underground; entertainment, through the use of audio-visual tableaux and the drama of the underground journey; and esthetics, generated by the underground ambience and medieval theming of the attraction.
The present study intends to contribute to the literature in two main ways. Firstly, it will apply the 4Es model in a novel setting-an underground tourist attraction-thereby evaluating the applicability of the experience economy framework in this particular context. Secondly, the paper will provide valuable information for tourist attraction marketers on how to reposition and differentiate their product offerings to compete more successfully in the experience-led marketplace. In order to achieve these aims, the study attempts to estimate the effect of the four experience economy realms via three mediating variables-arousal, memory and satisfaction-on the dependent variable of revisit intentions.
The paper is organised as follows. The following section discusses the theoretical background, focusing particularly on the concept of the experience economy. The four realms of experience are then each reviewed in turn. This is followed by a discussion of the three mediating variables employed in this paper: arousal, memory and satisfaction. An overview of the empirical methods used is then given. Sections presenting the results and findings then follow. Next is a section setting out the conclusions and recommendations. The final section presents the limitations of the study and makes some suggestions for future research.

Conceptual Background
Researchers have long recognised the potential for the concept of the experience economy to be applied to tourism. In their early work, Pine and Gilmore [1,2] repeatedly used the example of Disneyland to illustrate the principles of the experience economy. In their subsequent work, the authors linked the experience economy even more explicitly to tourism [10,11]. The notion of the experience economy is considered particularly relevant to tourism because the products and services offered tend to be intangible in nature and strongly related customer's experiences [12]. As such, almost everything that happens to a tourist can be considered to be either an experience in itself or part of the travel experience as a whole [6].
Defining the term 'experience' has proven controversial and it is not the intention of this paper to rehearse the arguments at length [13,14]. Broadly speaking, an experience can be thought of as an event or occurrence (or set of such) that leaves a lasting impression on a person [3]. It is typically considered to denote an overall or 'holistic' evaluation of consuming a product or service [15], although it may be divided into different contributory factors, some of which may be sub-conscious [16]. The customer's mental and/or emotional impressions of the experiences they have of consuming a product or encountering a service are held in their memory and used to direct their future behaviour [17]. This future behaviour may be actual or intentional, and it is normally directed toward the product or service (e.g., re-purchasing it or recommending it to others). Experiences may also affect the customer's future behaviour through the mediation of sensory arousal, e.g., by providing the emotion of calmness or, indeed, thrill. Another route is through the medium of customer satisfaction, by which consumers who feel satisfied with their purchase may be more inclined to behave positively toward the product in the future, for example by being loyal to the brand, or at least intending to be so. The notion is thus that the emotions generated by consumption experiences influence consumers' future behaviours.
The majority of tourism research using the 4Es model has, however, been conceptual in nature to date [10,32]. Indeed, it is only relatively recently that the concept of the four realms of experience has been investigated empirically within tourism literature. Oh et al. [6], for example, attempted to develop and test an appropriate set of scales for measuring each of the experience realms. Their findings supported the dimensional structure of four realms of experience; they also confirmed the reliability and validity of the scale set. As such, the authors argued that the experience economy concept could successfully be applied to the tourism context and that the measurement framework they developed could be used to study consumer experiences across a range of tourism destination encounters and products. There have been few attempts to deploy and test the scale to date. Of the small number of studies published, however, most were conducted in the tourism destination context, including rural tourism [33], cruise tourism [3,22], resort hotels in Malaysia [34], a community festival [17], senior tourists [28], temple stays [31], science festivals [29], a zoo [25], and a music festival and a museum in Norway [7]. A rare example of an empirical study outside of the tourism context is by Jeong et al. [35], who focused on online shopping in general, rather than online tourism purchases in particular. Another is a study of 'grocerants' in the food service sector of South Korea [36]. The results of these few studies are sufficient, however, to demonstrate the viability of using the four realms of customer experience in the tourism context.

Explanatory Variables: The Four Experience Economy Realms
The four realms of experience economy will next be introduced in the context of tourism and, more particularly, tourist attractions.

Education
Education is frequently identified as a motivator for consuming tourism products and services, including tourist attractions [37]. Prentice [38], for example, argues that people are intrinsically motivated to travel as a means of increasing their self-education and enhancing their personal growth. Visits to tourist attractions can be viewed as opportunities for such consumption acts to take place. This suggests that the educational dimension of 4Es is appropriate for examining the tourist attraction experience.
As noted above, KAL offers its consumers a storytelling experience, using tableaux, sound, lighting and other special effects. The stories are based on various ancient myths and legends of King Arthur that are found in the ancient Welsh texts The Mabinogion and the Book of Taliesin. These will typically be unfamiliar to tourists, so learning is likely to be a major component of their visit.

Esthetics
The esthetic experience is embodied by service-marketing concepts such as 'servicescape' [39] and 'atmospherics' [40]. Research devoted to service design and management of servicescapes (both human-made and natural) has demonstrated how esthetics affect customer experiences [39,41]. Bitner [39] identifies, for example, ambient conditions, spatial layout and functionality, signs, artefacts, symbols and social interactions as all being potentially important. Esthetics can equally well relate to any of the five major senses of sight, hearing, taste, touch and smell, as well as the overall 'feel' of an experience.
Esthetic experiences are therefore likely to be an important part of KAL visitors' overall experience, as well as to influence post-consumption evaluations such as their satisfaction with the visit [6]. In this respect, the KAL experience is particularly interesting as it is based mainly underground in a disused slate mine. As such, the environment provides darkness (the natural darkness is used to help create drama and spectacle) but also conditions that cool and moist. These conditions can be controlled most of the time. While the product offering explicitly de-emphasises the real world-in order to help in the suspension of disbelief and make visitors' journey through the attraction more immersive and engaging-visitors will inevitably witness human-made caverns and signs of industry such as rubble heaps.

Escapism
This realm refers to consumers' desire to get away from the reality and routine of daily life by actively immersing themselves in tourism experiences [1,2]. According to Oh et al. [6], escapism is one of the most frequently recognised motivating factors in the tourism literature. Contemporary society is witnessing the commodification of culture and becoming ever more disconnected from nature [42] (p. 2). Consequently, individuals look for opportunities to experience authenticity in various forms, for instance by travelling and discovering new destinations [2,11,42]. Escaping from daily life is important for the health of individuals and for the wellbeing of society more generally [43]. It might help people to find meaning in life [44] and to look at their life and society from a different perspective [45]. According to Gilmore and Pine [11], even if businesses are not providing an authentic product offer, they must still package it in such a way that customers perceive it to be authentic.
The experience offered by KAL is, by its very nature, based on escapism. Visitors are conducted by a costumed guide through a labyrinth of dimly lit underground tunnels, stopping in various places where stories based on the myths and legends of King Arthur are presented to them. The pretence that they are escaping the real world and travelling back in time is reinforced through a video that is shown to all visitors before they enter the labyrinth, an audio track that is played during their boat journey underground, and a point in the journey at which visitors pass 'through' a waterfall into the mythical realm beyond.

Entertainment
Entertainment involves passive participation and absorption with help of customers' senses and aims to generate pleasure and enjoyment of the environment in which the experience consumption takes place [1,2]. Entertainment is considered to be an essential element of tourism's offerings [3] and one of the motivations for visiting tourist attractions [37]. Oh et al. [6] point out that entertainment has rarely been measured in tourism research. Sheng and Chen [18] identified cultural entertainment as one of five types of experience expectation that influence people's experience at the tourist attraction. Kim and Ritchie [46], meanwhile, suggested that destination marketers should emphasise the entertainment elements and pleasing environments in the promotion of tourism destinations.
The entertainment dimension reflects the appreciation for the spectacle played in front of the viewer. This is built into the KAL experience through a series of sound-and-light shows that are presented to visitors as they move through the labyrinth. It is through these shows that the myths and legends of King Arthur are told. The experience attempts to engage multiple senses and includes a number of surprises along the way. As such, the KAL experience is very much designed to entertain the visitors.

Mediating Variables
Previous studies using the experience economy concept have included a number of possible mediating variables that are situated between the four experience economy realms and the dependent variable. Frequent among these are arousal, memory and satisfaction, which are now discussed in turn.

Arousal
Consumption experiences often elicit emotional reactions from visitors. Indeed, tourism experiences comprise different kinds of events and service encounters that can trigger both positive and negative emotions [27,47,48]. According to Loureiro [33] (p. 3), "emotions are mental states that arise from the evaluation of events or from a consumer's own thoughts". Arousal can be considered to be one of the positive emotions that can result from a visitor experience [6]. Arousal is an emotional state in response to environmental stimuli [49] and is often measured on the continuum from calmness to excitement [6,50]. It refers to the degree of stimulation, excitement or alertness felt by a person [26,35].
A number of studies have attempted to establish a link between the four realms of consumer experience and arousal in the tourism context. These include Oh et al. [6], who found significant relationships between both the educational and esthetic realms of the consumer experience and arousal, measured in terms of how interesting, enjoyable, exciting and stimulating customers found their visit. Loureiro [33], meanwhile, found that all four experience realms were linked to pleasant arousal and this, along with vivid memories, to feelings of place attachment and revisit intentions.
In context of this study, a major part of the environmental stimulus is the journey into and around the underground labyrinth. Visitors are led to expect this journey to be exciting and stimulating. This is supported by the company's promotional brochure, which sells the tourist attraction as 'an underground adventure' guided by 'a mysterious hooded boatman'. Following this reasoning, the positive experiences gained in KAL should lead to higher levels of arousal, and consumers with higher levels of arousal should be more likely to intend to visit the attraction again. The hypotheses to be tested in this study are therefore as follows: Hypothesis 1 (H1). Visitors' educational experience positively influences their arousal.

Memory
In the context of the consumption experience, memory refers to the conscious recollection of past experiences [51]. Researchers have found that memory of past tourism experiences is of great significance for businesses, as it can be the most important source of information used to determine whether they will decide whether to revisit a particular destination in the future [52][53][54]. This is thought to be because the past experiences are perceived to be credible and valuable sources of information that are useful in shaping their future decisions [55]. It is therefore important to measure what kind of experience contributes towards better memory formation [46,48].
The psychology literature provides a strong support for a causal relationship between novelty and memory [56]. It is argued that non-routine events, e.g., vacations, are better remembered than everyday events [57]. Similarly, in their empirical research of memorable tourism experiences, Chandral and Valenzuela [19] found that perceived novelty of experience leads to enhanced visitors' memory. Moreover, according to Kim et al. [48], individuals have a tendency to remember positive experiences more easily than negative ones. The researchers also contend that sensorial and emotional experiences tend to be remembered better by individuals than emotionally neutral events [6,50].
Importantly, some authors [19,50] indicate that experiences are more memorable if they induce feelings of pleasure and arousal. Pine and Gilmore [2] claim that the four realms of experience can all result in strong memories and have other positive post-consumption outcomes. Several of the empirical studies noted above support a significant positive relationship between the four realms of experience and vivid visitor memories [3,17,29,33,34]. Oh et al. [6] also found that well-staged experiences lead to enriched visitor memories of the consumption event. The hypotheses relating to memory are thus: Hypothesis 6 (H6). Visitors' educational experience positively influences their memories.

Consumer Satisfaction
Many studies emphasise the role of customer experiences in determining customer satisfaction [21]. Simply put, if the customer has a good experience of the product, they are more likely to be satisfied with it. Consumer satisfaction is sometimes defined in relative terms. For example, it may be defined in terms of whether (or how far) consumers' expectations and/or psychological needs are met [58]. Spreng and Mackoy [59] point out, however, that meeting or even exceeding predicted expectations will not necessarily satisfy consumers. This is because expectations are not usually the only antecedents of consumers' satisfaction. Other researchers have therefore defined consumer satisfaction in absolute terms. This means that it is defined in terms of customers' post-consumption overall evaluation of the totality of the consumption experience [26,59]. Vanhamme [60] notes, meanwhile, that satisfaction has a dual character in that it embraces both cognitive and affective evaluations: in other words, customer satisfaction may be based on how much customers think they are satisfied with the product as well as how much they feel satisfied with it.
Of the few studies that seek to establish significant linkages between the four realms of consumer experience and satisfaction, fewer still are in the tourism content. Those that do exist do, however, suggest that there may be a significant and relative strong relationship. Oh et al. [6], for example, developed a model in which satisfaction was measured using two axes (very dissatisfied-very satisfied; terrible-delighted), finding positive associations with the esthetic experience in particular. Mehmetoglu and Engen [7], meanwhile, found that up to half of the variation in visitor satisfaction at their two study sites was accounted for by consumer experiences, although it was not the case that all four experience realms were simultaneously in operation.
The present study employs the construct of consumer satisfaction as a mediating variable between customer (i.e., visitor) experiences and re-purchase (i.e., revisit) intentions. The measure used explicitly covers both cognitive and affective evaluations of a visit to the attraction.
Accordingly, the hypotheses relating to visitor satisfaction are as follows: Hypothesis 11 (H11). Visitors' educational experience positively influences their satisfaction.

Dependent Variable: Revisit Intentions
It is recognised that there is a paucity of reliable empirical studies on the future behaviour of tourists [12]. A major hurdle to undertaking good research in this area is that measuring customers' actual experiences is best done at the point of consumption, or very shortly afterwards, particularly for fast-moving consumer goods and point-in-time service encounters such as those typically involved in travel and tourism. Measuring customers' actual behaviour, however, is best done some time after the consumption of the product or the service encounter: for example, visitors to a tourist attraction may choose to revisit it, but perhaps not for many months or even years. Similarly, visitors may recommend the attraction to friends and family but only when they discover that they are travelling to that destination too, which maybe some months or even years hence. Some studies have addressed this difficulty simply by not attempting to measure consequent consumer behaviour but selecting variables such as memory and satisfaction as their outcome variables instead. Mehmetoglu and Engen [7], for example, choose satisfaction as their dependent variable, while Oh et al. [6] use both satisfaction and overall quality perceptions in this way. It might be pointed out, however, that this is not much of a solution, as memories and perhaps even satisfaction can take time to mature.
Some studies have nevertheless attempted to measure the outcome of customer experiences in terms of customers' intentions to remain loyal or to make recommendations to others [24,27,61]. While none of these studies measure experience using the 4Es framework, they demonstrate that behavioural intentions can successfully be used as a proxy for actual behaviour. Indeed, several studies that have adopted the 4Es framework have taken this approach. Hosany and Witham [3], for example, found a significant relationship between cruise tourists' experiences and their intention to recommend through satisfaction as a mediating variable. Manthiou et al. [17] chose loyalty as their behavioural variable, measured by their intention to revisit the community festival in question. In the context of rural tourism, meanwhile, Loureiro [33] found significant positive relationships between tourist experiences and revisit intentions, through the mediation of both arousal and memory. For Hwang and Lyu [62], involvement in golf was found to be a significant moderating variable between golf tournament tourists' experiences and their revisit intentions, which was itself significantly moderated by their well-being perceptions. In the present study, therefore, intention to revisit the attraction in the future will be used as the dependent variable.

Methods
A quantitative methodology was employed to test the 15 hypotheses established for the study, which are summarised in Figure 2. An in situ, self-administered questionnaire was used to collect the data. The sample was selected following a convenience sampling procedure.
Tour. Hosp. 2020, 1 8 golf tournament tourists' experiences and their revisit intentions, which was itself significantly moderated by their well-being perceptions. In the present study, therefore, intention to revisit the attraction in the future will be used as the dependent variable.

Methods
A quantitative methodology was employed to test the 15 hypotheses established for the study, which are summarised in Figure 2. An in situ, self-administered questionnaire was used to collect the data. The sample was selected following a convenience sampling procedure. A non-probability sampling technique [63] was deemed to be the most suitable for this research for two main reasons. First, it is a common sampling method in marketing, tourism and hospitality research, particularly when the research is not intended to provide precise quantitative predictions but to draw conclusions about the relationships between variables, as is the case in this study. Second, convenience sampling is recommended when randomisation is not possible due to the population under investigation not being sufficiently well defined, as was the case with KAL. Following the recommendations of Manfredo et al. [64] and Podsakoff et al. [65], visitors were approached to be surveyed as immediately as possible after finishing their underground tour in KAL (i.e., within less A non-probability sampling technique [63] was deemed to be the most suitable for this research for two main reasons. First, it is a common sampling method in marketing, tourism and hospitality research, particularly when the research is not intended to provide precise quantitative predictions but to draw conclusions about the relationships between variables, as is the case in this study. Second, convenience sampling is recommended when randomisation is not possible due to the population under investigation not being sufficiently well defined, as was the case with KAL. Following the recommendations of Manfredo et al. [64] and Podsakoff et al. [65], visitors were approached to be surveyed as immediately as possible after finishing their underground tour in KAL (i.e., within less than 20 min of them emerging from the labyrinth). This enabled the research to capture more effectively the respondents' experience and their holistic evaluations; it also helped to minimise the bias associated with retrospective recall [65].
The resulting dataset was analysed using structural equation modelling, based on the partial least squares (PLS) method. A PLS model is estimated to establish the causal relationships and is a technique that is frequently used in marketing research [66]. PLS was developed to address limitations inherent in simpler regression methods, such as ordinary least squares (OLS), and as such offers greater flexibility for researchers to test a theory [67]. OLS is a multiple regression method that allows relationships between variables to be tested [66,68,69]. PLS estimation involves two stages: first the development of an outer model (with the emphasis on model validity and the reliability of variables) and then the estimation of an inner model in which variance is explained, effect sizes are determined, etc. [67]. Three PLS models were estimated for this present study: one for each of the mediating variables (arousal, memory and satisfaction).
The scales for measuring each of the eight variables used in this analysis were taken from existing studies and modified for this particular research context. To measure the four dimensions of visitors' experience, scale items from Oh et al. [6] were adopted. Oh et al.'s [6] suggestions for scale adjustment were used, along with a careful analysis of the approaches taken by the other authors who have attempted to adapt these scale items to alternative tourism settings [3,33,34]. Arousal was measured using a four-item set of statements based on Hosany and Witham [3] and Loureiro [33]. These attempted to measure how interesting, enjoyable, stimulating and exciting the visit was considered to be on the seven-point Likert scale. The scales used to measure memory consisted of three items, as used in Ali et al. [34], Hosany and Witham [3], Loureiro [33] and Oh et al. [6] and inspired by Pine and Gilmore [2]. Satisfaction was operationalised using three statements adopted from Hosany and Gilbert [47]. Intention to revisit was assessed using two items based on Loureiro [33]. The items were randomly mixed to avoid common method bias, and space used to remove the dependent variables from the independents, creating separation [70]. The questionnaire was piloted using 15 semi-formal interviews conducted with people who had just completed the survey. Some minor adjustments were made to the wording, but the scales appeared to be robust and useful. The main survey was implemented in the summer of 2015.
The sample yielded 214 usable questionnaires, 56.1% of which were completed by females. Almost three-quarters (73.4%) of the sample were at the attraction with their family, while 83% were first-time visitors to the attraction. The sample was predominantly British, 82% indicating that they were from the UK. Interestingly, 78% of the sample said that they were on holiday, as opposed to travelling from home. See Appendix A for more information. The sample was collected over a number of days. As such, the variables included in the study were tested for mean invariance using ANOVA, but no significant variations were identified. The sample size was over 100 and had more than 10 observations for each variable, so the sample was of an appropriate size for analysis using PLS [66,71]. Indeed, the sample size was substantially greater than many studies that have used [66].

Results
For reliability analysis, each of the multiple-item constructs was analysed using Cronbach's alpha.
For each of the variables tested, the results showed that the alpha was above the minimum of 0.80 as recommended by Hair et al. [66]. In each case, the average variance extracted was also above 0.50, which also indicates convergent validity [72]. At this stage, common method bias was also post-hoc tested using Harman's single-factor test. Using confirmatory factor analysis, it was established that common method bias was not present [70]. Discriminant validity was analysed using a series of chi-square tests, analysing two variables at a time [73]. In each case, it was shown that discriminant validity was present. The variable descriptive information is show in Table 1. Three separate PLS models were estimated: one each for memory, arousal and satisfaction (see Table 2; Table 3). For each model, the item loadings were above 0.70, and significant (p < 0.001) suggesting that all of the items included were loading correctly and confirming convergent validity [69]. In each case, a composite reliability of over 0.70 was achieved, confirming the reliability of the variables [66]. A comparison with standard OLS regression and structural equation modelling (SEM) was also used to ensure path robustness, and in each case the results from the PLS were shown to be robust [67,74]. Table 2. Beta weights and t-statistics of independent variables on mediating variables. The adjusted R-square for the model was 0.721, which means that the model and specification fit well [67]. Only two of the independent variables had a significant effect on the mediating variable arousal: esthetics and entertainment, thus supporting H2 and H4. Entertainment (b = 0.513) had the larger effect, while esthetics (b = 0.235) had around half the effect size. Education and escapism were not significant predictors of arousal: hence neither H1 nor H3 could be confirmed. Meanwhile, arousal (b = 0.717) had a significant positive effect on revisit intentions, confirming H5. The total effects of entertainment (b = 0.368) and esthetics (b = 0.169) were also significant and positive.

Coefficient t-Stat Coefficient t-Stat Coefficient t-Stat
The second model used memory as the mediating variable. The model had an adjusted R-squared of 0.670, which is considered to be a substantive effect [67]. Escapism and entertainment both had significant and positive effects on memory, confirming H8 and H9 respectively. The beta coefficient for entertainment (b = 0.401) suggested that it had nearly twice the effect on memory of escapism (b = 0.201). No significant relationship was found between education and memory on the one hand or esthetics and memory on the other: hence neither H6 nor H7 could be supported. Memory had a positive and significant effect on revisit intention (b = 0.699), confirming H10. As for the total effects, entertainment and escapism had significant effects on revisit intentions. Again, entertainment had the larger effect (b = 0.283) than escapism (b = 0.141).
The final PLS model used satisfaction as the mediating variable. The adjusted R-square was 0.528, which is a moderate result [67]. For this model, only esthetics had a significant effect on satisfaction (b = 0.338). Hence H12 was confirmed but H11, H12 or H14 could not be. Satisfaction, in turn, had a significant effect on revisit intentions (b = 0.649), confirming H15. Meanwhile the esthetic experience had a significant and positive total effect on revisit intentions (b = 0.219). The results of the PLS estimates are contained in Tables 2 and 3.

Discussion
The estimated models indicate a number of important findings, as illustrated by Table 4. Eight of the 15 hypotheses were supported, the other seven not finding support in the data. Firstly, education did not have a significant influence on the mediating variables of memory, arousal or satisfaction in any of the three models estimated; nor was its influence significant in terms of revisit intentions in any of the three models. This suggests that visitors tend to value the stories they encounter as they move through the labyrinth not for their educational content but for other features, such as their spectacle, entertainment value and possibly as a vehicle for escapism. In practical terms, this is perhaps surprising insofar as the stories told in KAL have a clear educational dimension, particularly in terms of their heritage and moral content. It is also surprising because Pine and Gilmore [1,2] suggest that successful consumer product tend to provide optimal experiences across all four experience realms. Previous studies that have applied the experience economy in the tourism context have tended conclude that all four experience realms tend to be significantly involved [3,17,28,32]. The lack of a significant linkage between the educational experience and revisit intentions is, therefore, an unexpected finding of the study. It is not, however, without precedent: Mehmetoglu and Engen [7] found neither education or entertainment to play a significant role in determining visitor satisfaction at a music festival, and escapism and entertainment to play no significant role in determining the satisfaction of visitors to a museum. Hwang and Lyu [62], meanwhile, perhaps surprisingly found esthetics not to be a significant variable in determining revisit intentions to a golf event.
Secondly, it is apparent from Table 4 that esthetic experiences had a significant impact on the mediating variable in all three models: arousal, memory and satisfaction, respectively. Furthermore, in the models mediated by arousal and satisfaction, esthetic experiences had a significant total impact on revisit intentions, although this was not the case with respect to the model mediated by memory. Importantly, however, esthetics did not have the strongest total effect in either of the first two models, suggesting that while esthetics is an important component part of the visitor experience it is not the most influential one. This is in contradiction to most other studies, where esthetics was frequently found to have the strongest effect [3,6,7,17,33]. The finding is not, however, without precedent: Hwang and Lyu [62], for example, did not find esthetics to be a significant determinant of golf tourists' perceptions of well-being.
The third experience realm, escapism, had the lowest coefficients when arousal and satisfaction were the mediating variables and only the second highest coefficient when memory was the mediating variable (entertainment had almost twice the effect). Neither of the first two cases found escapism to significantly affect either the mediating variable concerned or revisit intentions through the mediating variable. Escapism did have a significant effect, however, both with regard to memory and on revisit intentions through the mediation of memory. As such, it might be argued that escapism is a secondary factor in terms of making memorable experiences. This is because it has a small but significant effect on memory, and the effect of memory in turn has a significant effect on revisit intentions. This effect is the weakest in size of any of the three models. This finding finds support in Hosany and Witham [3], who found escapism to be the least influential of the four experience realms, it being only significant in predicting overall perceived quality.
Fourthly, entertainment had a strong and significant effect in two of the models. This variable had the greatest effect both on memory and on revisit intentions through memory, both times around twice the magnitude of its nearest rival, escapism. It also had a substantial and significant effect on arousal and revisit intentions through arousal: around twice that of its nearest rival, which was in this case esthetics. This is in keeping with the findings of, for example, Manthiou et al. [17], whose study concluded that entertainment was a significant factor in determining both memory and visitor loyalty to a community festival. Ali et al. [34] also found entertainment to have a significant effect on memory and customer loyalty in the case of Malaysian resort hotels. Other studies have, however, produced contradictory findings [6,7,33]. Hosany and Witham [3], meanwhile, found esthetics to be most influential in affecting outcome variables, but in this case followed by entertainment. This further supports the conclusions of both Mehmetoglu and Engen [7] and Hosany and Witham [3] that, while all four experience realms may make a contribution to experience outcomes in any given context, the presence and size of such effects is likely to vary from context to context. These conclusions are apparently independent, as neither paper cites the other.

Conclusions
The findings of this paper lend support to the notion that while it is possible for any or all of the four experience realms to be instrumental in shaping customer experiences-along with the consequences of these, including revisit intentions-will depend much on context. In the case of KAL, the experience realms having the greatest impact on revisit intention were firstly entertainment through the mediation of arousal, secondly entertainment mediated by memory, and thirdly esthetics through the mediation of visitor satisfaction. Put another way, revisit intentions were influenced by arousal, which is created mostly through the entertainment provided by the visit; memory, which is also promoted most strongly through the entertainment element; and visitor satisfaction, which is influenced most strongly by the esthetic dimension of the visit. The entertainment realm would thus seem to be the most important in delivering the experience at KAL, followed by the esthetic experience.
These findings do not fully accord with prior expectations. As noted above, in view of the 'underground adventure' basis of the experience, escapism and esthetics might be expected to be the most prominent experience realms. It was also anticipated that all four experience realms would play an important part in determining the outcome variable, as education and entertainment are also integral elements of the KAL experience. The findings of this study thus confirm the inadvisability of simply transferring the findings of experience studies from one context to another, as previously noted by Mehmetoglu and Engen [7] and Hosany and Witham [3].
This study does suggest, however, that the measurement scales used in this study, based on those of Oh et al. [6], are suitable for the purposes of measuring the four experience realms empirically. The scales achieved discriminant validity and were able to explain more than half of the variability in the outcome variable in all three models. Other studies have also noted that these measurement scales perform well, so this study supports their continued and wider use. This study confirms the findings of previous studies that the measurement scales are usable with the PLS methods. This provides further assurance to future researchers that the 4Es approach can be robust and insightful.
Entertainment was found to be a key determinant of the outcome variable, revisit intention, when it is mediated by arousal and memory. This implies that those managing and marketing the attraction should pay special attention to increasing its entertainment value by making this element of the visit more arousing and more memorable. The former might be achieved by augmenting the existing sound-and-light shows, which are the main focus of the attention, with features that either engage more of the senses or engage some of the senses more fully. The present tableaux rely mainly on fixed dioramas, a soundtrack triggered when groups of visitors approach, and lighting that assists in the storytelling by spotlighting different features of the scene. By today's standards, this is relatively simple technology and it might be argued that updating this would make the entertainment element of underground journey more arousing. Introducing image projection technology might be an option, for example. This is not a realistic recommendation for KAL, however, given that the entertainment takes place deep underground, where temperature and atmospheric conditions are not kind to high-technology equipment. Such equipment would also prove difficult to install in a system of underground caverns and tunnels.
The alternative might therefore be to introduce more surprises into the visitor experience. There are already several features scheduled into the underground journey to provide an element of surprise but it might be possible to introduce more. These should be arousing but not frightening, as the visit is aimed at people of all ages, particularly families. It is also important to ensure that the surprises do not detract from the overall experience or be so frequent that they confuse the visitors. Any marketing materials must ensure that while such surprises might be alluded to, they are not given away. Clearly, a cautious strategy is required.
Another potential means of adding entertainment is to make more of the 'mysterious' costumed guide, whose job is mainly to conduct groups of visitors around the labyrinth (so they do not actually become lost) and to ensure their safety. An additional role might be given to the costumed guide to take on the function of storyteller. This might prove difficult to introduce to the schedule, however, as the attraction is strictly timed: visitor groups must enter and leave within specific time windows for logistical reasons, which means there is not a great deal of unused time available in which the costumed guide might be able to provide such entertainment.
Regarding the effect of entertainment on revisit intentions through memory, perhaps there is more scope for those who manage and market KAL to make improvements. Some tourist attractions, for example Disneyland, have built in opportunities for the taking of photographs [75]. These opportunities are engineered, perhaps by providing visitors with themed costumes to wear, backdrops to be photographed against, props to be held or costumed characters be photographed with. This practice is based on the theory that staging 'memorable moments' (sometimes known as 'Kodak moments') that will help visitors to have strong and positive memories of their visit-supported of course by the physical evidence of a photograph-which will in turn enhance behavioural outcomes such as willingness to recommend or revisit intentions. Adopting this practice might be difficult within KAL, however, as visitors are requested not to take flash photographs underground, the reason being that this will illuminate parts of the subterranean environment that are normally shrouded in darkness and not intended to be seen. The solution in the case of KAL might be to provide these photo opportunities above ground, either before the trip has begun or just after its completion.
While entertainment had the greatest effect on revisit intentions through both arousal and memory, the biggest effect on intention to revisit through satisfaction was esthetics. This is another unexpected finding, as prior expectations would be for esthetics to affect revisit intentions through the mediation of arousal or memory. The finding is not, however, entirely without keeping with previous studies. Mehmetoglu and Engen [7] found esthetics to be an important determinant of satisfaction at both their study sites. They reasoned that this was because esthetics tended to feature heavily in the marketing materials in both cases, thus leading visitors to anticipate a particular esthetic in terms of atmosphere, ambience and facilities. The recommendation for KAL would therefore to make sure that their marketing efforts focus particularly on establishing the esthetic experience prior to the visit. For, example, a short video might be used on the attraction's website to introduce visitors to the 'feel' they can expect. Another recommendation would be to ensure that the theme of myths and legends begins as soon as the visitor has arrived and continues right up until the point they leave. This might be done with flags, banners or props to 'dress up' the site. Putting above-ground staff in costume might also assist in this.
Within the underground setting of the labyrinth, it may also be possible to use subtle lighting effects to pick out attractive natural features such as rock formations. Care should be taken in this, however, as it is important to ensure that this does not enable visitors to see parts of the underground environment that would prevent them from 'suspending their disbelief'. If they are reminded that the labyrinth is a real place, with a real history as a slate mine, this may undermine the escapism element of the experience.

Limitations and Suggestions for Future Research
There are a number of limitations associated with this study. First, it has been noted that the dependent variable used in this study was revisit intentions. Other dependent variables could have been used, such as loyalty or recommendation behaviours such as willingness to give positive world of mouth. Other studies have also used dependent variables that are used here as mediating variables, such as memories and satisfaction, while others again have used mediating variables that are not used here, such as pleasure and service quality. It is possible that using different combinations of variables could have revealed patterns of effect that are not noted in this study.
Second, this study does not explicitly incorporate the antecedents of customer experiences into the models it estimates. As noted in the literature review section, other studies have found variables such as visitor expectations, novelty, meaningfulness, surprise, self-image congruence and customer-to-customer interaction to be important in this respect.
Third, the PLS method is a powerful research tool that has advantages over other estimation techniques [76]. There are nevertheless some criticisms of the technique [77]. These issues relate to the measurement used (formative, reflective, single/multiple item), sampling issues, reporting issues and a lack of theoretical rigour [77]. The present study has followed rigorous guidelines and theory to try to ensure such issues do not arise. However, such criticisms of PLS cannot be completely dismissed. It is also recommended that this be further tested using different samples to confirm and further extend these findings.
With regard to further research, the findings of this study support the proposition of Hosany and Witham [3] and Ali et al. [34] that the scale established by Oh et al. [6] is fit for the purpose of measuring customer experiences, provided that minor adjustments are made to the wording of items to ensure that they are contextually appropriate. Future research can, therefore, confidently adopt (and cautiously adapt) Oh et al.'s [6] measurement scales. Again, this would add to the robustness of the findings presented here.
At the same time, the study also confirms the conclusions reached independently in the studies by Mehmetoglu and Engen [7] and Hosany and Witham [3] that effective customer experiences do not necessarily need to be built upon all four realms of experience. Further studies are needed to further investigate this proposition, particularly in different contexts. Particular emphasis needs to be placed on investigating whether the absence of significant effects from one or more of the experience realms actually detracts from the overall experience outcome or is simply not missed when it is absent. If the former is the case, then it would be in the experience provider's interest to make good the absence, but not if the latter is the case. This is the first study to apply the experience economy concept to an underground tourist attraction. While the study did not confirm prior expectations that escapism and esthetics would dominate in determining the experience outcome, esthetics were found to have a significant effect in determining revisit intentions alongside entertainment. These are both passive experience realms, so it would be interesting to determine if this is the case with other underground tourist attractions. For example, it would be especially valuable to determine whether a similar pattern is found with respect to more 'authentic' underground tourist attractions, such as natural caverns and disused mines that are interpreted as industrial heritage.

Acknowledgments:
The authors gratefully acknowledge the research assistance of Magdalena Bylicka, who undertook some of the early conceptualisation and the on-site data collection. They are also immensely grateful to Ian Rutherford, proprietor of King Artur's Labyrinth, for allowing access to the fieldwork site and for his helpful insights on a draft version of the paper.

Conflicts of Interest:
The authors declare no conflict of interest.  Notes: * May add to more than 100% due to rounding. ** = adds to more than 100% when more than one response is possible.