Novelty in tourism experiences: the influence of physical staging and human interaction on behavioural intentions

ABSTRACT

Tourists’ tendency to choose new and different experiences and destinations is well known in tourism research and practice. By drawing on social exchange theory and service-dominant logic, the present study investigates how physical staging and human interaction influence behavioural intentions in experiences with varying levels of novelty. This relationship is tested using survey data collected from visitors to a theme park with structural equation modelling. The results confirm the relationship between the variables and the moderating effect of novelty; that is, the level of novelty in the experience influences the effect of both physical staging and human interaction on behavioural intentions. This study contributes to the tourism literature by comparing different experiences with varying levels of novelty and linking them to behavioural intentions. In addition, it differentiates between physical staging and human interaction in the experience. The findings have significant implications for the tourism industry, particularly for tourism companies.

KEYWORDS:

• Novelty
• experiences
• physical staging
• human interaction
• behavioural intentions

1. Introduction

Novelty, a core tendency in tourism, is a change from everyday life when experiencing something new and different, which also implies the search for variety (Lee & Crompton, 1992; Mitas & Bastiaansen, 2018). Prior studies reveal that tourists tend to choose new and different experiences not due to lack of satisfaction, but because they want to experience something that contrasts with their prior experiences (Crompton, 1979; Bigné et al., 2009; Niininen et al., 2004). This is a challenge for the tourism industry, as revisiting tourists is beneficial both economically and in practice (Darnell & Johnson, 2001; Tjørve et al., 2018). As such, further work on theories of consumer behaviour that include revisit behaviour based on positive experiences could assist in better understanding of tourism and other experiential consumption situations (Dolnicar et al., 2015; McKercher et al., 2012).

Several scholars have attempted to determine the possible precursors of behavioural intentions in the tourism field, but call for more research in this area (Choo & Petrick, 2014; Dedeoglu et al., 2018; Vittersø et al., 2017; Wu et al., 2018). Researchers have raised concerns regarding the often taken for granted link between satisfaction and behavioural intentions in tourism experiences (Dolnicar et al., 2015). Researchers studying behavioural and attitudinal loyalty claim that the concept require different treatment in the tourism context due to its unique features (McKercher et al., 2012) and that the precursors may be different in tourism compared to other consumptions situations (Prebensen et al., 2018). Recent studies point to how physical staging and human interaction can positively influence behavioural intentions in experiential cases (Chang et al., 2014; Choo & Petrick, 2014; Dedeoglu et al., 2018; Grissemann & Stokburger-Sauer, 2012; Wu et al., 2018). From the perspective of social exchange theory (SET) and service-dominant logic (SDL), physical staging and human interaction are key activities in the process of co-creating experiences performed by the tourist and the tourism companies (Grönroos & Voima, 2013; Vargo & Lusch, 2008). Research show that tourists taking part in the co-creating process can develop behavioural intentions to the experience (Sugathan & Ranjan, 2019; Zhang et al., 2019). Conceptually, physical staging is the facilitation provided by tourism firms and human interaction denotes the social encounters within the tourism experience (Campos et al., 2017; Prebensen & Foss, 2011). Following the lead of these authors, the present study examines the influence of physical staging and human interaction on behavioural intentions in experiences with various levels of novelty.

Despite the overall tendency of search for novelty in tourism, research shows that visitors pay for and consume experiences within a destination for different reasons (Prebensen & Rosengren, 2016). For instance, an amusement park includes different activities or experiences. Visitors may consume a beach experience at the park to please the children and to relax, which assumes less need for novelty. An activity such as swimming with dolphins is something that typically elicits learning and can trigger feelings such as mastery, assuming higher need for novelty. This may reflect that consumers may perceive tourism experiences with various levels of novelty, meaning the degree of contrast between present experience and previous experiences ranges from new and unique to familiar and common (Bello & Etzel, 1985; Lee & Crompton, 1992). The perceived levels of novelty in this research determine the differences between experiences in terms of either high, medium, or low levels of novelty. This notion is in line with previous research that categorizes experiences according to the emotion they provide (Prebensen & Rosengren, 2016; Roy, 2018; Vittersø et al., 2017).

The explained gap in knowledge linking physical staging, human interaction, and behavioural intentions in experiences with various levels of novelty motivated this research. This study hopes to theoretically contribute by combining these aspects using the lens of SET and SDL. Other researchers emphasize that tourism research could benefit from building on perspectives from SDL (Li & Petrick, 2008; Shaw et al., 2011) and that SDL could benefit from building on SET (Vargo & Lusch, 2017). Earlier studies on behavioural intentions tend to focus on one separate unit of analysis, whereas this study analyses and compares several experiences, which provides a needed methodological contribution to the literature (McKercher et al., 2012). Additionally, this study contributes empirically by answering the call for evidence-based research on both physical staging and human interaction as co-creation activities in the experience and their link to behavioural intentions (Cossío-Silva et al., 2016; Dedeoglu et al., 2018).

2. Literature review

2.1. Social exchange theory and service-dominant logic

Homans (1958) introduced SET, which traces back to sociology, psychology and anthropology, and is often associated with research on organizational behaviour. SET can help identify interactions in which individuals and groups exchange resources (Ap, 1992). Lawler et al. (2000) described social exchange as ‘a joint activity in which two or more actors attempt to produce a flow of benefits better than they can achieve alone or in other relationships’ (pp. 616–617). Social interaction can create positive emotions and continued social interaction among the same people can reduce uncertainty and emotional processes, which can in turn lead to relationships (Lawler et al., 2000).

Other researchers have applied SET to consumption contexts, such as service delivery (Yi & Gong, 2009), service recovery (Choi et al., 2014), customer relationship management (Gefen & Ridings, 2002) and sustainable consumption (Wang et al., 2019). The perspectives of social exchange also gained attention in tourism research, often in the context of residents’ perceptions of and attitudes to tourism (see Hadinejad et al., 2019 for a review). In line with the present study, Choo and Petrick (2014) applied SET to an empirical study emphasizing how social interaction in agro-tourism positively affects behavioural intentions through satisfaction. Additionally, researchers have linked SET to co-creation and SDL (Grissemann & Stokburger-Sauer, 2012).

SDL emphasizes that companies facilitate the customer value creation process, and that value is co-created through mutual interaction and participation of both the company and the customer (Grönroos & Voima, 2013; Vargo & Lusch, 2004, 2008). A co-created tourism experience is ‘the sum of the psychological events a tourist goes through when contributing actively through physical and/or mental participation in activities and interacting with other subjects in the experience environment’ (Campos et al., 2016, p. 23). Prior studies reveal that tourists participating in the co-creation of their tourism experience perceive the experience to be more valuable and are more satisfied (Prebensen & Xie, 2017). In addition to this, recent literature link co-created experiences to behavioural intentions (Sugathan & Ranjan, 2019; Zhang et al., 2019). Sugathan & Ranjan, 2019 suggests that revisit intention is higher in experiences with high co-creation (where tourists take part in creating their own experience) compared to experiences with low co-creation (where tourists do not take part in creating their own experience). Whereas Zhang et al. (2019) emphasized that festival visitors experiencing a satisfactory co-creation festival experience, are more likely to develop behavioural intentions toward the festival. Indicating that tourism companies can increase tourists’ behavioural intentions by inviting the tourists to co-create their experience.

Based on SET and SDL, the current study delineated co-creation activities as physical staging and human interaction. Physical staging includes the facilitation elements performed by the tourism companies (Campos et al., 2016), comprising the physical surroundings, accessibility, maps, and information (Ritchie et al., 2008). Tourists can use these substantial facilities to participate in the co-creation process of their experience. Meanwhile, human interaction refers to the social encounters within the tourism experience, which includes the tourist’s interactions with both staff members and their travel party, in addition to other tourists and locals (Campos et al., 2016; Prebensen & Foss, 2011). Service providers can instruct and support the tourists in the co-creation process though interaction. Physical staging and human interaction are elements that support the tourist, making it easier to engage in the experience (Prebensen & Xie, 2017), which, in turn, likely influences their behavioural intentions.

2.2. Behavioural intentions related to experiences

In the tourism context, behavioural intentions are often described by two factors, revisit intentions and willingness to recommend the experience to others (Bigne et al., 2001). Revisit intentions is the tourist’s intention to have the same experience in the near future (Zeithaml et al., 1996). Whereas the tourists’ willingness to recommend the experience to others often is linked to positive word of mouth (Anderson, 1998).

One possible precursor of behavioural intentions in tourism is physical staging (Chang et al., 2014; Wu et al., 2018). Chang et al. (2014) studied tourists visiting a theme park and found a positive relationship between physical surroundings and behavioural intentions. The study by Wu et al. (2018) showed that perceived experience quality significantly influences both experience satisfaction and theme park image, which has a positive effect on behavioural intentions. Interestingly, they found that within experiential quality the tourists valued the physical environment the most. It is therefore reasonable to believe that physical staging positively influences behavioural intentions in other tourism experiences.

Other researchers approach behavioural intentions using SET to link human interaction to behavioural intentions (Choo & Petrick, 2014; Grissemann & Stokburger-Sauer, 2012). Choo and Petrick (2014) studied a sample of tourists engaged in agro-tourism and found that social interaction effects behavioural intentions through satisfaction. The authors highlight that social interactions between tourists, service providers, and especially companion tourists should be given more attention in these experiences. Grissemann and Stokburger-Sauer (2012) also highlights how company support affects customer loyalty, mediated by customer co-creation activities. In their case, company support was operationalized as the interactions between the company and customer. Therefore, it is reasonable that interactions influence behavioural intentions in other tourism experiences.

Later studies also support a link between both physical staging and human interaction on the evaluations of experiences (Dong & Siu, 2013). Interestingly, Dong and Siu (2013) find that tourists perceived physical staging as more important in their evaluations of a theme park experience compared to human interaction with the staff. Recent research by Dedeoglu et al. (2018) finds that physical as well as non-physical elements of a hotel experience influence both revisit intentions and intentions to recommend the experience to others, through novel and hedonic value. In their study, visitors perceived the non-physical elements, including human interaction, as more important than the physical elements related to physical staging. These studies by Dong and Siu (2013) and Dedeoglu et al. (2018) show that the two factors – physical staging and human interaction – have different influences on behavioural intentions depending on the type of experience (theme park vs. hotel). Hence, the literature points to a possible relationship between physical staging, human interaction, and behavioural intentions (Chang et al., 2014; Choo & Petrick, 2014; Grissemann & Stokburger-Sauer, 2012; Wu et al., 2018). Therefore, the following hypotheses are suggested:

H1: Physical staging is positively related to behavioural intentions in tourism experiences.

H2: Human interaction is positively related to behavioural intentions in tourism experiences.

2.3. Novelty in experiences

Novelty in tourism experiences is ‘the sense that one is experiencing something new, and the sense that one is experiencing something different from usual daily life’ (Mitas & Bastiaansen, 2018, p. 99). Novel tourism experiences are related with unexpectedness and surprise, where the experience departs from the tourist’s expectations (Ma et al., 2013). Recent studies show that novel experiences can trigger strong emotions and enhance attention, which in turn increase memorability of the experience (Skavronskaya et al., 2020). In line with the shift from the experience economy towards the transformation economy (Kirillova et al., 2017; Pine & Gilmore, 2011), novel experiences can be evaluated as peak experiences triggering positive emotions and engagement, which is needed for transformation of one self to happen (Neuhofer et al., 2020). Thus, novelty can be understood as the motivation to travel in the first place (Bello & Etzel, 1985). Novelty can also reflect tourists’ need and desire to learn new skills and obtain knowledge, in addition to feeding their curiosity (Williams & Soutar, 2009). Other researchers have studied novelty in terms of tourist profiles and have found that some tourists seek novelty more than others do (Assaker & Hallak, 2013; Kim & Kim, 2015). A recent literature review on novelty indicate that research on the concept is positioned in personality, behavioural, cognitive and neuropsychology (Skavronskaya et al., 2019). The present study apply perspectives from behavioural psychology, looking at the behavioural reactions to novel tourism experiences, also assuming that experiences is perceived by the touritst to include different levels of novelty.

Tourism includes a bundle of activities, services, and benefits (Medlik & Middleton, 1973) that accommodate different needs, including levels of novelty. Experiences with less novelty are often perceived as familiar, common, and convenient, and have been found to cover a basic need linked to utilitarian experiences (Holbrook & Hirschman, 1982). Low levels of novelty can also be associated with hedonic experiences, often described with feelings of pleasure, satisfaction, and happiness (Vittersø et al., 2017). Experiences with high levels of novelty have been delineated as unique experiences, which increase engagement and interest, can be associated with the feeling of thrill and surprise (Lee & Crompton, 1992), and linked to eudemonic experiences (Vittersø et al., 2017).

Prior studies find a positive relationship between novelty and behavioural intentions in experiences (Chang et al., 2014; Dedeoglu et al., 2018; Jang & Feng, 2007), and that novelty influences behavioural intentions, both in the short and long term (Bigné et al., 2005; Bigné et al., 2009; Mitas & Bastiaansen, 2018). Chang et al. (2014) indicated that higher levels of novelty with the physical surroundings of theme parks positively influences behavioural intentions. Later studies similarly compare experiences with both low and high levels of novelty and variations in the relationship to behavioural intentions that depend on the level of novelty (Kim & Moon, 2009; Roy, 2018; Vittersø et al., 2017). Vittersø et al. (2017) found that tourists who engage in experiences with low levels of novelty (hedonic experiences) are likely to recommend the experience to others, but are not likely to revisit it. On the other hand, tourists who engage in experiences with high levels of novelty (eudemonic experiences) are likely to recommend the experience as well as to revisit it (Vittersø et al., 2017). This result is in line with Roy (2018), who compared two experiences – one with a low level of novelty (bank) and the other with a higher level of novelty (restaurant). The study confirmed that experience quality influences behavioural intentions more for experiences with high levels of novelty than for experiences with low levels of novelty. Kim and Moon (2009) showed how restaurant type moderates’ behavioural intentions according to the feeling of high and low levels of entertainment. This can also be linked to novelty; visitors can perceive a restaurant with high level of entertainment as more novel than one with low level of entertainment. Their results illustrated that the effect of the feeling of pleasure on revisit intention is likely stronger for the novel restaurant experience compared to the less novel restaurant experience. This finding supports the assumption of Dedeoglu et al. (2018) that tourists sensing novelty in the experience will want to obtain the sense again, and might thus show behavioural intentions.

By building on the works of Kim and Moon (2009), Roy (2018), and Vittersø et al. (2017) the present study argue that experiences reflect different emotions, and different levels of novelty lead to differences in behavioural intentions. In addition, experiences with varying levels of novelty may require different degrees of physical staging and human interaction. Based on the above, the following hypotheses are proposed:

H3: Novelty moderates the relationship between physical staging and behavioural intentions in tourism experiences.

H4: Novelty moderates the relationship between human interactions and behavioural intentions in tourism experiences.

A summary of the proposed hypotheses is presented in Figure 1 illustrating the conceptual model of the paper.

Figure 1. Conceptual model.

3. Methodology

3.1. Setting and study population

The context of this study was a marine wildlife park, Zoomarine, in the Algarve region of Portugal. Theme parks are a particular type of tourist attraction (Dybedal, 1998) and there is evidence of their growing popularity among travel consumers (Milman, 2001; Milman et al., 2020). Although the park’s core theme is marine wildlife, visitors can engage in different types of experiences with various levels of novelty, directly or indirectly connected to marine and other types of wildlife. The population included in the study consists of national and international tourists aged over 18 years who visited the park from September to November 2017.

3.2. Instrument and measurement of constructs

The questionnaire to collect data for this study consisted of three sections. The first inquired about the particular experience the tourist engaged in related to novelty, physical staging, and behavioural intentions. The second focused on interactions in general during tourism experiences, while the collected personal information about the respondents was addressed in the last section of the questionnaire. As Table 1 shows, Novelty was measured using five items building on work by Kim et al. (2012) and Williams and Soutar (2009). Physical staging was measured through five constructs based on the concept of company facilitation adopted from Ritchie et al. (2008). A set of five items was used to measure Human interaction following the works of Campos et al. (2016) and Prebensen and Foss (2011). The construct Behavioural intentions was measured with four items covering both the intentions to revisit and intentions to recommend the experience to others, adopted from Zeithaml et al. (1996).

Table 1. Working definitions of the constructs.

Construct	Definition	Reference	Operationalisation
Novelty	The sense that one is experiencing something new and different compared to everyday life.	Mitas and Bastiaansen (2018), Kim et al. (2012), Williams and Soutar (2009)	This activity makes me feel like an adventurer This activity satisfies my curiosity This activity provides authentic experiences This activity is unique This activity is a once-in-a-lifetime experience
Physical staging	The facilitation tourism firms provide to tourists in the experience environment.	Campos et al. (2016), Ritchie et al. (2008)	The information Zoomarine provided is appropriate for this activity For this activity, the maps are informative The physical conditions for this activity are good The activities at Zoomarine are accessible The way staff members communicate with visitors is very important
Human interaction	The tourists’ encounters with other parties of the tourism experience environment.	Campos et al. (2016), Prebensen and Foss (2011)	I like to walk alone Interactions with fellow travellers are important to me Interactions with staff members are important to me I like to interact with random people (other tourists) When travelling, I like to interact with the local people
Behavioural intentions	The tourists’ intentions to revisit the experience and intentions to recommend the experience to others.	Zeitham et al. (1996)	I will repeat this activity as soon as possible I will repeat this activity within a year I will tell my family and friends about this activity I will recommend this activity to anyone who would like to hear about it

The questionnaire was written in English first, then translated into Portuguese before being back translated to English (Harkness, 2003). English was used in addition to Portuguese as the region hosts traditional international travellers familiar and comfortable with English. All scaled questions used the anchors 1 = ‘strongly disagree’ to 7 = ‘strongly agree’ in order to increase response accuracy, as recommended by some literature (Johns, 2010). Table 4 provides the final list of items to measure each construct.

3.3. Experiences in the park and novelty levels

This study compares three experiences in the park involving different levels of novelty, namely: Swimming with dolphins (high novelty), Animal show watching (medium novelty) and The beach (low novelty). Two criteria were used to define these levels of novelty, one quantitative and another one qualitative. The quantitative criterion relied on the computation, for each experience, of the average score of the six items of the construct Novelty listed in Table 1. The results from this computation are in Table 2. As can be observed, Swimming with Dolphins reports the highest mean novelty level among the three experiences (5.86), and thus it was considered as a ‘high novelty’ experience. Then, the following highest mean novelty level was found regarding Animal Show watching (4.89) and, accordingly, it was classified as a ‘medium novelty’ experience. The lowest mean novelty level was achieved for the experience The Beach (4.18), explaining its classification as a ‘low novelty’ experience.

Table 2. Level of novelty in Zoomarine experiences.

Tourism experience	Novelty mean	Novelty level
Swimming with dolphins	5.86	High
Animal show watching	4.89	Medium
The beach	4.18	Low

This quantitative approach for ascribing the novelty levels was complemented with a more qualitative approach, taking into consideration information provided by the park management and prior research conducted at the park (Campos et al., 2017). Indeed, the Swimming with Dolphins experience is an immersive activity consisting of a one-and-a-half-hour direct interaction between the visitors and the dolphins in an enclave of the park. This experience is not included in the park’s admission price; in fact, as documented in the company’s commercial communication to customers, an extra premium price is charged for participation in this activity, often described by participants as an ‘once in a life time experience’, which is both new and unique (Campos et al., 2017). So, this reflection was supportive of our earlier quantitative evaluation of this experience, based on the results in Table 2, and classification as a ‘high novelty’ experience. Animal shows (dolphin, sea lions, and birds of prey) are scheduled sessions where animals perform before an audience, without any direct contact (touch) with the animals. Dolphin shows are attractions found in dolphinaria worldwide and are familiar attractions to mass tourism and recreation (Markwell, 2015). As tourism products, their degree of novelty is lower when compared to swimming with dolphins, so, and also in line with our former quantitative evaluation of this experience and its classification as a ‘medium novelty’ experience. As the Animal shows, The Beach experience is included in the admission price to the park. The beach experience refers to an area in the park that replicates a real beach, where visitors can linger without time restrictions. Beach tourism counts as one of the oldest types of tourism (Picken, 2017). In the Algarve region where the park is located, beach tourism is dominant for the past decades (Turismo de Portugal, 2014), and assumedly presents the lowest degree of novelty, in this way corroborating the previous classification as a ‘low novelty’ experience.

3.4. Data collection and sample

Considering the most conservative estimate for a single proportion (0.5), an unknown target population size, a confidence level of 95%, and a margin of error of 4%, a minimum sample size of 583 was considered in our study (Cochran, 1963). The self-administered questionnaire was delivered to tourists immediately after the experience as they left the premises (e.g. when swimming with dolphins) or the area (e.g. when at the beach). These respondents were approached on assigned dates and times from September to November 2017 taking into consideration the park’s schedules for activities. All visitors of 18 years of age or older had the opportunity to participate in the study. A total of 696 questionnaires were distributed, from those were 592 (85%) completed and considered valid for the analysis. Since no incentive was used to stimulate participation, the high response rate is likely due to the reduced dimension of the questionnaire and the interviewer’s persistence in achieving the minimum defined sample size.

3.5. Data analysis

Data analysis started with the Harman (1967) single-factor test for potential common method bias (CMB). Since measurement of constructs Physical staging, Human interaction and Behavioural intentions was performed by the same seven-point Likert scale, respondents might be tempted to provide similar responses to all items. This could be an important source of CMB that biases the subsequent results (Podsakoff et al., 2012). To detect a potential CMB problem, an exploratory factor analysis (EFA) was conducted considering all items loaded into one common factor (Roni, 2014) using the SPSS 25.0 software package. According to the Harman test, an EFA with a total explained variance for a single factor above 50% indicates CMB. To complement the Harman test, the full collinearity test was applied because it is the most adequate method for assessing CMB in PLS-SEM (Kock, 2015). This test suggests CMB when the variance inflation factors (VIFs) for the exogenous latent variables surpass 3.3.

Subsequently, structural equation modelling (SEM) with the SmartPLS 3.0 (Ringle et al., 2015) software package was run to test the research hypotheses (Chin, 1998; Davcik, 2014). The partial least squares SEM (PLS-SEM) method is becoming increasingly popular in tourism research ( do Valle & Assaker, 2016; Latan, 2018). PLS-SEM is useful when the lack of normality is a concern (Hair et al., 2019; Sánchez-Franco & Roldan, 2005). Indeed, all items used to measure Physical staging, Human interaction, and Behavioural intentions did not follow a normal distribution (tests of normality: p = 0.00). Moreover, the Mardia’s test for multivariate normality rejected this distribution (p = 0.00). PLS-SEM is also useful in exploratory studies on established theories (Hair et al., 2019; Sánchez-Franco & Roldan, 2005), as in our study.

Therefore, using PLS-SEM, the estimated model was evaluated for measurement model fit and structural model fit (Hair et al., 2016, 2019). The measurement model fit was first assessed regarding individual and composite reliability, and then convergent and discriminant validity. To assess individual items reliability, the factor loadings were calculated (Hair et al., 2019). According to Carmines and Zeller (1979), loadings should be higher than 0.708, meaning more shared variance between each item and the construct than error variance. The composite (CR) reliability coefficient and the Cronbach alpha were analysed in order to attest for construct reliability. According to Kline (2005), there is internal consistency for a block of indicators in one construct if these indexes surpass 0.7.

Convergent validity, i.e. the amount of variance shared between each construct and its indicators, was first checked by examining the loadings’ statistical significance, given by the bootstrapping procedure for 5000 sub-samples, as suggested by Hair et al. (2011, 2016, 2019); and then, by calculating the Average Extracted Variance (AVE) for each construct, which should score higher than 0.5. Discriminant analysis indicates to what extent constructs in the model are effectively representing different concepts. In our study, two frequently used criteria were used and checked: the criterion of Fornell and Larcker (1981), according to which the square root of each AVE should exceed the correlations between the construct and the other constructs; and the Heterotrait-Monotrait Ratio (HTMT) criterion, proposed by Henseler et al. (2015), that compares the (mean) items correlations across groups with the (mean) correlations for the items measuring the same construct. According to Henseler et al. (2015), each HTMT should not exceed the threshold of 0.9 (but they are better when lower than 0.85), and must be significantly different from 1.

After measuring the model, data analysis focused on the structural model. Hypotheses 1 and 2 were evaluated by observing the path estimates and their significance. The exploratory and predictive capability of the model was analysed by observing the R² for the endogenous latent variable and also its Q² value, given by the blindfolding routine of SmartPLS 3.0. According to Hair et al. (2011), R² of 0.75, 0.50 and 0.25 can be classified as high, moderate and low. In turn, positive Q² values mean predictive accuracy for the model for that construct. Despite lack of a consensual overall measure of fit in PLS, some measures have been proposed that are implemented in the SmartPLS 3.0 software (Hair et al., 2016), as the Standardized Root Mean Square Residual (SRMR) value (values lower than 0.08 are recommended), and a Normed Fit Index (NFI) (values higher than 0.9 are recommended). These measures were observed in our study. Finally, multi-group analyses were employed to test the moderating effect of novelty (hypotheses 3 and 4). As previously explained in section 3.3, the three levels of novelty were defined based on the mean of novelty mean detected for each experience, as classified in Table 2: the Swimming with Dolphins (high novelty; mean = 5.86), Animal Show watching (medium novelty; mean = 4.89) and The Beach (low novelty; mean = 4.18) experiences had 121, 117, and 71 participants, respectively.

4. Results

4.1. Respondent demographics

Table 3 summarizes the respondents’ characteristics, including gender, age group, country of origin, level of education, marital status, and current occupation. The data show that females slightly outnumber males, and the most representative age groups of this sample are between 21–30 years and 31–40 years old, with 25.9% and 27.1% respectively. Individuals from Portugal and the UK represent almost half of the sample, and 66% of the respondents had higher education degrees. Most of the respondents are married or living as such and 70.4% of the total sample are employed.

Table 3. Sample profile.

Variables	Distribution of answers
Gender	Male: 47.2%; female: 52.8%
Age	<21: 10.4%; 21–30: 25.9%; 31–40: 27.1%; 41-50: 20.8%; 51–60: 11.2%; >60: 4.6%; minimum: 16; maximum: 71; mean: 36.7; standard deviation: 12.56; median: 34
Country of residence	Portugal: 27.5%; the UK: 21.3%; Germany: 12.7%; Spain: 10.1%; Netherlands: 9.6%; Other: 9.5%; France 9.3%
Education	Basic: 2.7%; Secondary: 31.0%; University degree: 66.3%
Marital status	Single: 31.5%; married/living as a couple: 65.2%; divorced: 3.3%
Occupation	Employed: 70.4%; partially employed: 13.8%; unemployed: 15.8%

4.2. Descriptions of indicators and the test for common method bias

Table 4 presents the items included in the model to measure each latent construct. Some preliminary conclusions can be drawn by observing the averages for each item. Concerning Physical staging, all items scored means above 5, indicating that respondents generally agree that Zoomarine provides satisfactory conditions for visitor activities to occur, and the highest means were found for the item measuring the park’s physical conditions. Regarding Human interaction, all item means scored slightly lower; however, contacts with the staff were found to be important (mean = 4.75). Finally, for this sample, the most important manifestation of Behavioural Intentions is to tell friends and relatives about the experiential activity (mean = 5.39).

Table 4. Measurement model assessment.

Dimensions and items	Mean	Loading	CR	AVE	Cronbach’s α
Physical staging			.907	.621	.877
FS1. The information Zoomarine provides is appropriate for the activity	5.42	.778***
FS2. Maps are informative	5.58	.827***
FS3. The physical conditions for this activity are good	5.66	.785***
FS4. Activities are accessible	5.54	.873***
FS5. The way staff members communicate with visitors is very important	5.33	.756***
Human interaction			.887	.724	.809
HI1. Interactions with staff members are important to me	4.75	.823***
H12. I like to interact with unknown people	4.36	.941***
HI3. While traveling, I like to interact with local people	4.37	.886***
Behavioural intentions			.884	.657	.825
RI1. I will repeat this activity as soon as possible	4.49	.812***
RI2. I will repeat this activity within a year	4.13	.699***
RI3. I will tell my family and friends about this activity	5.39	.838***
RI4. I will recommend this activity to all people who are interested in knowing	5.20	.883***

Note: *** bootstrapping p< 0.001.

The EFA of the items measuring Physical staging, Human interaction and Behavioural intentions indicates that the common factor solution has a total explained variance of 36.2%, which is below the 50% threshold proposed by Harman (1967). This result indicates that CMB is likely not an issue with data. The full collinearity test confirms this result, as the VIFs for the exogenous latent variables of the model, Physical staging and Human interactions, are below 3.3 (VIFs for both variables equal 1.071).

4.3. Measurement model assessment

Table 4 reports the results of the assessment of the measurement model, including the final set of items to measure each construct. Two items were removed from the initial set of five items to measure Human interaction because their factor loadings were almost zero, therefore, significantly lower than the threshold of 0.708 (‘I like to walk alone’: loading = −0.065; ‘Interactions with fellow travellers are important to me’: loading = 0.528). Keeping these items in the model would compromise the individual reliability of the construct and its convergent validity (Hair et al., 2011, 2019). In most cases, the standardized loadings surpass .708, confirming the individual reliability of the chosen items to measure the three latent constructs. The exception is for the item RI2 – ‘I will repeat this activity within a year’ (loading = 0.699). However, this item was retained in the analysis given the content validity of the construct, and because its loading is very close to the threshold value. Moreover, deleting them would not improve the model (Bagozzi & Yi, 1988; Hair et al., 2016). Regarding the construct reliability, the composite reliability indexes (CR) and the Cronbach’s alpha values exceed the desired threshold value of 0.7 (Carmines & Zeller, 1979; Kline, 2005). As Table 4 shows, CRs ranged from 0.884 to 0.907 and Cronbach’s alpha values ranged from 0.809 to 0.877.

After assessing reliability, convergent and discriminant validity were verified. Values for the average variance extracted (AVEs) were equal or more than .5, establishing convergent validity (Hair et al., 2011, 2016, 2019). As Table 4 shows, AVESs ranged from 0.621 to 0.724. Significant factor loadings for all items in the corresponding latent variables are another signal of convergent validity. To assess significance, PLS-SEM was used to derive the bootstrapping t-values. The bootstrapping ts are above 1.96 (assuming a 5% significance level) and 2.585 (assuming a 1% significance level), suggesting the significance of the indicators in the corresponding constructs.

To verify discriminant validity, the Fornell and Larcker (1981) criterion was first applied, according to which the square root of each AVE (values in the diagonal) should exceed the correlations between the construct and the other constructs (values out of the diagonal). This criterion was verified in our model (Table 5). To further assess the discriminant validity, the Heterotrait-Monotrait Ratio (HTMT) criterion was assessed (Table 5). As proposed by Henseler et al. (2015), the HTMT values must be lower than 0.85 and significantly different from 1. Data also fulfilled these requirements.

Table 5. Correlations among the latent variables.

Constructs	1	2	3
1. Physical staging	0.788^a
2. Human interaction	0.257 | 0.301^b	0.851^a
3. Behavioural intentions	0.458 | 0.527^b	0.368 | 0.448^b	0.811^a

^a Diagonal values correspond to the squared root value of AVE for each latent variable to assess the Fornell-Larcker criterion.

^b HTMT values.

4.4. Assessing the structural model and hypothesis testing

Table 6 presents the results of the test of Hypotheses 1 and 2. The results support Hypothesis 1, that Physical staging is positively related to Behavioural intentions, as the corresponding path estimate is positive and statistically significant (β = 0.248; bootstrap-t = 6.507; p = 0.000). The same conclusion applies to Hypothesis 2 that Human interactions is positively related to Behavioural intentions (β = 0.495; bootstrap-t = 14.534; p = 0.000). The R² for the endogenous construct, Behavioural intentions, is 0.277, meaning that the joint effect of Physical staging and Human interaction explain 27.7% of the variability of this variable. The Q² for this construct is positive (Q²= 0.178), meaning that the model as enough predictive capacity (Hair et al., 2019). Moreover, the estimated model reported a SRMR value of 0.079 (values lower than .08 are recommended) and an NFI of 0.757. Desirably, NFI should exceed 0.9. However, it tends to penalize models with low complexity, as the proposed model. So, a NFI of 0.757 is likely acceptable as the proposed model only involves three latent variables.

Table 6. Path estimates for the structural model.

Hypothesis	Path coefficients	Bootstrap-t	Support
H1. Physical staging → Behavioural intentions	0.248***	6.507***	Yes
H2. Human interaction → Behavioural intentions	0.495***	14.534***	Yes

Note: ***p < 0.001.

4.5. The moderating effect of novelty

To test the moderating effect of novelty on the relationships established between Physical staging and Behavioural intentions (H3) and between Human interaction and Behavioural intentions (H4), multi-group analyses were conducted for the three experiences in the park: Swimming with dolphins (high novelty), the Animal show (medium novelty), and The beach (low novelty). As explained, the level of novelty is based on the novelty mean measured for each experience (Table 2). Despite the relatively small size in each group (121, 117, and 71 participants, respectively), results have enough statistical power. Indeed, for a model with two exogenous latent variables, the Cohen (1992) table show that a minimum of 47 observations is necessary to achieve a statistical power of 80% for detecting R² values of at least 0.25 (with a 5% probability of error), which approximates the found for the proposed model (0.277).

The most important results of these analyses are shown in Table 7. The path estimates for the relationships in H1 (between Physical staging and Behavioural intentions) and H2 (between Human interaction and Behavioural intentions) are higher for tourists who engaged the highest novelty experience, Swimming with dolphins (β = 0.431; β = 0.393, respectively). The path coefficients for this group are statistically significant (p < 0.01). A similar result was obtained for the path coefficients for the group of tourists who attended the Animal show (p < .01); however, the strength of the relations in H1 and H2 given by the path estimates (β = 0.357; β = 0.272, respectively) is lower in this group when compared to the Swimming with dolphins group. Finally, for participants who chose The beach experience, the path estimates for H1 and H2 were not statistically significant (p > 0.10), meaning that the responses for this group do not support these hypotheses. Thus, the degree of novelty in the experience affects (moderates) the relationships in H1 and H2. Thus, novelty was found to moderate the relationship between Physical staging and Behavioural intentions as well as between Human interaction and Behavioural intentions. In short, the results support H3 and H4.

Table 7. Path estimates: Multi-group analysis and the moderating effect of novelty.

Path relationships	Path estimates
	Swimming with dolphins (High novelty)	Animal show (Medium novelty)	The beach (Low novelty)
H1: Physical staging → Behavioural intentions	.431***	.357***	.297
H2: Human interaction → Behavioural intentions	.393***	.272***	−.149

*p < .10, **p < .05, ***p < .01.

5. Discussion and conclusions

The overarching aim of this study was to investigate how physical staging and human interaction influence behavioural intentions in experiences with varying levels of novelty in the tourism context. The results suggested that physical staging and human interaction influence touristś intentions to recommend and revisit the experience positively. In addition, the current study illustrate that the level of novelty moderates the relationship between physical staging, human interaction and behavioural intentions.

Specifically, the present study is in line with earlier research (Dedeoglu et al., 2018; Dong & Siu, 2013) suggesting that physical staging and human interaction in the experience influence tourists’ behavioural intentions positively. Then again, when comparing the experiences the results found that physical staging is more related to behavioural intentions compared to human interaction, in line with Dong and Siu’s (2013) research on theme parks. One reason for this could be that experiences in theme parks depend heavily on the functional physical surroundings as well as helpful information such as maps, for the tourists to engage and co-create their experience. However, Dedeoglu et al. (2018) find a stronger effect of human interactions compared to physical staging, which contradicts our results. One reason for these conflicting results could be that the current study included interactions with a wider group of different intervening actors with various contributions to the meaning of the experience, specifically interactions with tourists and staff members, but also but also interactions happening between tourists and locals, as well as unknown people (Prebensen & Foss, 2011).

Comparing the three groups according to the perceived levels of novelty, the present study finds differences in the performance of the study’s variables. The results suggested that the effect of physical staging and human interaction on behavioural intentions varies according to the level of novelty in the experience. Indeed, for experiences perceived as having high novelty (Swimming with dolphins), the effect was more substantial compared to both the medium (Animal show watching) and low novelty (The beach) experiences. Thus, physical staging and human interaction is more important in highly novel experiences than it is for lower novelty experiences. This may be because novel experiences mean facing uncertainty and new challenges, which may be easier to overcome depending on the surroundings, the behaviours of others, and interaction. That is, when partaking in a novel experience requiring particular physical involvement or cognitive performance, the tourist’s positive perception of the physical staging and human interaction can determine the success of the experience, thereby influencing the tourist’s judgement about future intentions. Dedeoglu et al. (2018) present a similar argument; that is, customers perceiving novelty regarding a product they use tend to choose the same product again, wanting to experience the same feeling in the future. Vittersø et al. (2017) also support this result, finding that tourists participating in experiences with eudemonic characteristics (high levels of novelty) were more likely to exhibit behavioural intentions and recommend the experience to others.

5.1. Theoretical implications

Other researchers have segmented novel experiences according to trip-related and event related novelty (Skavronskaya et al., 2020), this paper investigates event related novelty and takes the notion one step further by distinguishing between three levels of trip-related novelty and comparing these. This paper also answers the call for research on novelty within activity sectors (Kim & Kim, 2015), in our case comparing three different tourists’ activities. The inclusion of the three levels of novelty within the three different experiences demonstrates that experiences are different, and that the context does matter. The study finds that the level of novelty in experiences influences the perceived importance of physical staging and human interaction on behavioural intentions.

Further, the paper contributes to theory by linking SET and SDL, which is a combination of views that have been called for (Vargo & Lusch, 2017). Additionally, the paper provides evidence-based research on physical staging and human interaction as co-creation activities in the experience and their link to behavioural intentions (Cossío-Silva et al., 2016; Dedeoglu et al., 2018).

Results of the current study, in line with others (Dolnicar et al., 2015; McKercher et al., 2012) found that precursors for behavioural intentions in tourism is complex. Including novelty into this analysis offers a new perspective on perceiving behavioural intentions in tourism experiences.

5.2. Practical implications

Based on our results, practitioners should tailor their physical surroundings and make room for human interaction to encourage tourists to both revisit and recommend the experience. Which is highly relevant for tourism practitioners trying to recover after the COVID-19 pandemic eager to find new ways in order to attract tourists. Tourism companies should seek to understand tourists’ perceptions of novelty in the experiences and adjust the physical and non-physical elements of the experience accordingly. This require companies to have the ability to evaluate the experience they provide, through either gathering feedback from their tourists or engaging other external establishments for this matter. Successful investment here would provide the tourism companies with the knowledge they need in order to tailor their experiences when it comes to physical staging and human interaction.

When an experience is perceived as novel, it can also be seen as unique and often unfamiliar. Thus, tourists might depend more on the physical surroundings and human interaction, which provide support and can enable them to take part in the co-creation process of the experience. Interactions with staff members, other tourists, and local people can make tourists feel more supported, which is important in unfamiliar experiences. Practitioners should make room for conversations between the tourists` before the activity starts, but also during and after the experience for pleasure seeking reasons. It is also important that the staff spend time explaining and clarifying how the activity should be executed for tourists` trying the experience for the first time. This can then have a positive influence on the tourist’s willingness to come back but also recommend the experience to others.

When an experience is perceived as less novel, it can be seen as familiar and known. In these experiences’ tourists are likely to have the knowledge they need in order to participate in the experience, and may therefore not be dependent on the social support of others. Tourist companies should here focus on building the appropriate physical surroundings that enable the tourists to take part in the experience. This can be done through evaluating the premises of the experience at a regular basis, making sure it is both accessible and up to date. In these experience settings tourists tend to spend longer time compared to other service settings, and therefore have the time to evaluate the aesthetics of the premises. Attention should therefore be given not only to the practical design of the experience but also its attractiveness.

5.3. Limitations and future research

One limitation is that the study collected the sample from only one destination in the tourism context (theme park). Therefore, results are difficult to generalize to other destinations and tourism contexts. Any future research should then focus on replicating the study in other destinations and tourism settings by comparing tourism experiences. While our study examines the topic using three different experiences with varying levels of novelty, examining more heterogeneous contexts will continue developing the stream of research on the tourism experience within the SDL and SET theoretical frameworks.

The present study uses SDL and SET as the theoretical framework to investigate behavioural intentions in novel experiences. Further research could apply other theories such as optimal stimulation theory or institutional theory, but also theory from cognitive psychology such as cognitive appraisal theory to further investigate the relationship between novelty and behavioural intentions in tourism.

Moreover, our research model examines only two antecedent factors of behavioural intentions: physical staging and human interaction. Behavioural intention can be explained by other factors rather than only physical staging and human interaction, and the purposed model could be improved by adding other variables. Future work should investigate other antecedents such as mastering or destination image, or moderating variables such as tourist personality and demographics. While our study uses the level of novelty in an experience as a moderating variable, the tourist’s novelty-seeking profile could also affect the results.

Last, the current study would like to emphasize that behavioural intentions and an actual revisit are not the same; intentionality itself does not lead to actual behaviour, so we should not draw conclusions from one to the other. One possible line of research following from this study could examine the relationship between behavioural intention and actual revisits when tourists engage in highly novel experiences.

Acknowledgements

This paper is financed by National Funds provided by FCT- Foundation for Science and Technology through project UIDB/04020/2020.

Disclosure statement

No potential conflict of interest was reported by the author(s).