Facial-Expression Recognition: an emergent approach to the measurement of tourist satisfaction through emotions

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ “This is an Accepted Manuscript of an article published by Elsevier in Telematics and Informatics on 2020, available at: https://doi.org/10.1016/j.tele.2020.101404.” 2 Facial-Expression Recognition: an emergent approach to the measurement of tourist satisfaction through emotions M.Rosario González-Rodríguez, M.Carmen Díaz-Fernández, Carmen Pacheco Gómez M.Rosario González-Rodríguez, Universidad de Sevilla, [email protected] (corresponding author) M.Carmen Díaz-Fernández, Universidad de Sevilla, cardi[email protected] Carmen Pacheco Gómez, Universidad de Sevilla, [email protected] Abstract Purpose - The employment of facial-expression recognition to analyse emotions constitutes a potential instrument for the measurement of customer satisfaction in the tourism sector. The study aims to assess the functionality of Artificial Intelligence to measure tourists’ emotions and hence their satisfaction with the quality of the service provided on a guided tour when visiting a UNESCO heritage site. Design/methodology/approach - The methodology comprises the following stages. Firstly, the emotions are analysed through data recorded by using a software application on facial-expression recognition on a sample of tourists visiting a heritage site. Secondly, the tourists were asked to rate their overall satisfaction with the guided tour visit. Finally, a structural equation modelling approach is used to validate the strong relation between emotions and satisfaction. Findings - The results achieved confirm that the information obtained from facialexpression recognition demonstrated that it is as valid an instrument as that offered by the self-administered questionnaires for the measurement of customer satisfaction. The findings from the application reveal that a change in the scientific and professional field is emerging in the measurement of customer satisfaction focused on the emotions from a digital approach. 3 Research limitations/implications - This research is mainly based on the use of specific software for facial-expression recognition with its intrinsic measurement of emotions with and in a specific heritage scenario. Other scenarios and software of a more sophisticated nature implemented in the tourism and hospitality industry are necessary for the in-depth comprehension of the significant role played by emotions in the improvement of service quality. Practical implications - The recent application of recording emotions in the Tourism Industry provides practitioners with useful insights for the detection of deficiencies in their services and therefore the means to boost their reputation and destination image. Originality/value- Artificial Intelligence presents a new paradigm in the measurement of satisfaction by substituting self-administered surveys with a method based on the use of innovative software that recognizes faces and detects emotions through facial expressions. The paper contributes to the literature by using an Artificial Intelligence approach to measure satisfaction through emotions in the tourism sector. Keywords: Artificial Intelligence; Facial-expression recognition; Emotions; Satisfaction; PLS-SEM; Emotionalyser. Paper type Research paper 1. Introduction Traditionally, structural surveys have been employed to assess customer satisfaction in order to attain in-depth knowledge on the areas of potential improvement in the provision of a specific service (Coghlan & Pearce, 2010; Jacobs, Fehres & Campell, 2012; Walters, Sparks, & Herington, 2012; Oh and Kim, 2017). However, with the digital revolution, Artificial Intelligence (AI) tools are emerging through which the data collection process is greatly improved (Pittman & Reich, 2016; Lowe-Calverley et al., 2019). These innovative instruments are able to recognize faces, and to detect, via facial expressions, the emotions felt by users at the moment of the consumption of the service (Mauss & Robinson, 2009; Wang & Minor, 2008). Bearing in mind that emotions have a significant impact on the majority of human activities and functions (Faria et al., 2017), facial-expression recognition has already begun to be implemented experimentally in certain hotel chains, such as that of the Bluebay Group (bluebayresorts.com). Thanks to its use, several of the less efficient 4 features of the hotel could be identified, such as the service provided by the advertising leaflets, through the sensors placed on the shelf devoted to these leaflets. Recently, as a pioneering initiative in Spain, the Campanile hotel chain has been using sensors to recognize facial expressions of emotions to evaluate and improve its protocols in customer services (imotionanalytics.com). The method based on the recognition of facial expressions allows hoteliers to give an immediate response, make effective decisions, and act on the problems that arise on a daily basis. This technology appears to be an effective tool for managers to solve such difficulties in their hotels. In addition, through facial-recognition technology, there is a reduction in the risk of bias in social desirability appearing in responses (Morin, 2011; Paulhus, 2002; Poels & Dewitte, 2006; Ravaja, 2004; Shanshi, Scott & Walters, 2014). With results of greater accuracy, practitioners can act more effectively by applying measures that optimize the quality of service delivery, or experience, and hence increase the perceived value and other variables that directly affect customer satisfaction. Emotions play a fundamental role as antecedents of satisfaction (De Rojas & Camarero, 2008; Yuksel, Yuksel & Bilim, 2010; Walters, Sparks & Herington, 2012; González-Rodríguez, et al., 2019) and become the value-base object (measurement object) of facial-feature recognition (Lang, Greenwald, Bradley & Hamm, 1993; Tarnowski et al., 2017; Kirana et al., 2018). In this respect, negative emotions are related to a lack of satisfaction and tourists would perceive a negative experience either in the tourist destination or in the quality of the service provided. Conversely, the greater the accumulation of positive emotions, the higher the level of customer satisfaction. Knowledge regarding the causes of negative emotions, which enables dissatisfaction to be suitably addressed, can be gathered with the proper location of intelligent sensors. Hence, facial-expression recognition sensors can replace questions from selfadministered questionnaires related to service quality, experience quality, and satisfaction in the tourism destination and hospitality sector (Li, Scott & Walters, 2015). The unquestionable importance of feelings as an antecedent factor of satisfaction, together with the emergence of AI devices, form the basis of the present research devoted to the exploration of emotions through facial-expression recognition and their link with customer satisfaction measured by a self-administered questionnaire. The accuracy of how well emotions are related to customer satisfaction provides the validity of the technology used. Thus, the main aims of the present research, directly related to improving the perceived quality of a specific service offered on a UNESCO world 5 heritage site are as follows: analyse customer satisfaction from the measurement of the emotions that are recorded from tourist facial expressions during the guided tour; and demonstrate the importance of emotions as a key antecedent of customer satisfaction, which in turn would help practitioners improve the quality of a service. The research strives to emphasise the use of artificial intelligence (AI) based on facial-expression recognition and emotions, and to reveal that the results provided by AI are sufficiently optimal to truly be considered as an effective instrument for the validation of or the combination with traditional surveys in order to better understand individuals’ emotional responses (Critchley & Harrison, 2013; Kreibig, 2010; Lambie & Marcel, 2002; Mauss, Levenson, McCarter, Wilhelm & Gross, 2005; Mauss & Robinson, 2009). In order to comply with the objectives above, a literature review has been carried out on emotions as an antecedent that influences the satisfaction with the quality of the service as perceived by the customers, and on the advantages and disadvantages of the application of self-report questionnaires to capture individual emotions. A powerful facialrecognition tool, Emotionalyser, has been employed to collect data on emotions. Structural equation modelling, PLS-SEM, has been applied in order to test the effectiveness of the facial-expression recognition used in the measurement of customer satisfaction, which in turn would provide information to practitioners regarding improvements to service quality. 2. Literature review 2.1. Satisfaction and emotions Traditionally, studies on consumer satisfaction have been addressed from a cognitive perspective where satisfaction was generally modelled as the result of a process of comparison between the expectations and the perceived results after a consumption experience (Bigné et al., 1997a; Oliver, 1977, 1981; Oliver & Desarbo, 1988; Parasuraman, Zeithamal & Berry, 1985, 1988, 1994). However, a new trend in addressing satisfaction from a cognitive-affective approach emerged in the late eighties in the academic literature on customer behaviour (Bagozzi, Gopinath & Nyer, 1999; Bigné & Andreu, 2004; Dubé & Menon, 2000; Richins, 1997; Yu & Dean, 2001; González- Rodríguez et al., 2016). These authors claim that the cognitive component of satisfaction is insufficient to understand the consumer's response to a consumption experience since satisfaction is a process of cognitive evaluation that produces affective responses that 6 influence a customer’s decision-making and therefore the subsequent consumer behaviour. The understanding of the influence of emotions on individual responses such as satisfaction and behaviour has appeared as a recurrent theme in many disciplines, including travel and tourism. Emotions originally studied from consumer psychology have been understood as an antecedent driver of responses from consumers exposed to advertisements, service interactions, use of products, visiting experiences, or to certain environmental conditions, to name but a few. Although there is no standard definition for emotion, it is commonly accepted that “Emotions arise in response to events that are important to the individual's goals, motives, or concerns” (Frijda, 1988, p. 351). In an individual, emotions trigger affective, cognitive, physiological, and behavioural responses (Brave & Nass, 2002). In this vein, in behavioural science, it is argued that emotion is a key construct for understanding consumer preference with regard to consumption of products and services (Dai et al., 2015). Moreover, pleasant feelings and satisfying results are identified as emotion-related factors that determine consumption experiences (Stock, 2011). Accordingly, in the empirical research, the domain of emotion is divided into positive and negative emotions (Dai et al., 2015; Tarnowski et al., 2017). In the literature, emotions have been generally measured through the self-report questionnaire method and observation techniques (Coghlan & Pearce, 2010; Isomursu, Tähti, & Kuutti, K., 2007; Lee & Kyle, 2012; Wang & Minor, 2008). While self-report techniques are mainly based on dimensional theories of emotions since they focus on cognitive aspects of a subjective or memorable experience, observation techniques are mostly related to categorical theories since they observe physiological and expressive responses derived from a stimulus. Traditionally, in the tourism and hospitality fields, self-report methods based on previous measurement scales of emotion have been used to explore how travellers’ emotions experienced during a trip are strong predictors of satisfaction (Bigné & Andreu, 2004; Bigné, Andreu & Gnoth, 2005; Faullant, Matzler, & Mooradian, 2011; San Martin & del Bosque, 2008; González-Rodríguez et al., 2019) and behavioural intention (Bigné, Mattila, & Andreu, 2008; Donovan, Rossiter, Marcoolyn, & Nesdale, 1994; Machleit & Eroglu, 2000; Rodríguez Molina, Frías-Jamilena, & Castañeda-García, 2013; White & Scandale, 2005; Yüksel & Yüksel, 2007). A literature review on tourism and consumer behaviour reveals that the empirical approach for the measurement of emotions based on self-report questionnaires relies on either unipolar (San Martin, 2005) or bipolar items 7 (De Rojas & Camarero, 2008; Alegre & Garau, 2010; Žabkar, Brenčič & Dmitrović, 2010) to capture data concerning mainly valence and arousal. In the bipolar approach, emotions are conceived as bipolar states where extreme positions are mutually exclusive (Bagozzi et al., 1999; Oliver, 1997), such as happy/unhappy, entertained/bored. The use of bipolar measurement scales is not suitable when individuals experience the two extreme positions in the same experience. This situation could happen when the product or service provided presents multiple features leading the respondents to experience independent states of happiness and unhappiness, or when the service is provided over a period of time and hence the respondents might have both positive and negative experiences (Andreu, Bigne, Chumpitaz & Swaen, 2006). However, with the unipolar scales, the individual indicates the intensity or frequency with which each emotional state (positive and negative) is experienced by the respondent. The main limitation of the unipolar state is that the answers for positive and negative emotional states may be just the opposite and therefore redundant. Self-report measures have been useful and widely employed in tourism mainly because they are simple and economical methods for the capture of emotions (Healey, Nachman, Subramanian, Shahabdeen & Morris, 2010; Micu & Plummer, 2010; Poels & Dewitte, 2006; Teixeira,Wedel & Pieters, 2012; Domínguez-Quintero et al., 2019). However, selfreport questionnaires present major limitations when measuring travellers’ emotions. First, the self-report methods may be biased towards providing socially acceptable answers (Brønn & Vidaver-Cohen, 2009; Podsakoff & Organ, 1986; Worthington, Ram & Jones, 2006), even when respondents have been assured confidentiality and anonymity. Second, the self-report measures may involve cognitive bias basically due to the respondents’ conscious awareness of emotions (Clore & Ortony, 1988; Frijda et al., 1995; Kolodyazhniy, Kreibig, Gross, Roth & Wilhelm, 2011; Poels & Dewitte, 2006; Winkielman & Berridge, 2004; Wilhelm, Grossman & Mueller, 2012) and the ability to remember or explain the emotions they have experienced throughout the visit (Paulhus & Vazire, 2007; Poels & Dewitte, 2006; Wilhelm & Grossman, 2010). Third, a significant bias when measuring emotions involves the time elapsed between the moment travellers experience the emotion to the time those emotions are reported (Micu & Plummer, 2010; Urry, 2009). Mostly, in the tourism sector, emotions are considered as simple cognitive appraisals by being measured at a specific point in time and therefore represent the overall emotional assessments of the visit or trip (De Rojas & Camarero, 2008; Hosany & Gilbert, 2010). Furthermore, emotions are dynamic (Forlizzi & Ford, 2000) and a user can 8 experience a variety of emotions while the service or product is being provided. Only a few studies measure emotions by asking respondents the same questions at regular intervals throughout the visit (Graham et al., 2008; Tussyadiah & Fesenmaier, 2009), although, for these studies, there is still a time lapse between the moment the emotion is experienced and the moment said emotion is reported (Kim & Fesenmaier, 2015). Fourth, other limitations have also been recognized as being related to the items utilised to identify emotions (Chamberlain & Broderick, 2007): certain measures grounded in emotions either contain terms which are not familiar to the respondents or ignore emotions which are part of people’s daily lives. Furthermore, those measurements related to semantically differential items where the two anchor points are not clear opposites (e.g., bored and relaxed) are particularly confusing for respondents. A growing interest in the collection of information on emotions through psychophysiological indicators has been stressed in the literature on tourism to overcome the limitations of the self-report measures (Bagozzi et al., 1999; Chamberlain and Broderick, 2007; Li, Shanshi, Scott, Noel, Walters & Gabbi, 2014; Li, Walters, Packer & Scott, 2018; Li, Walters, Packer & Scott, 2016; Shanshi, Scott & Walters, 2014). Psychophysiological measurements can be conducted continuously, thereby providing a vast number of measurements for the detection of the dynamism of emotions (Wang & Minor, 2008). Furthermore, since these measurements do not depend on verbal scales, they prevent users from experiencing any confusion with message processing (Ravaja, 2004) and require no cognitive effort or memory. The psychophysiological instruments available for the collection of data on emotions are essentially those that measure physiological reactions, such as sweating, and pupil dilation (Kreibig, 2010; Partala et al., 2000), or that measure expressions mainly based on facial or vocal expressions (Ekman & Friesen, 1978; Kaiser & Wehrle, 1994; Litman & Forbes, 2003). Eye tracking captures gaze direction (reflecting attention) and blink frequency (associated with mental load). Scan paths of the eyes reveals how people look at websites and advertisements: which parts of a web, an advertisement, or picture are actually noticed and how long people look at various items. Modern eye-tracking technology can easily collect this data, which makes it increasingly popular in behavioural studies. Recently, studies in tourism have used eye-tracking methodologies in a laboratory setting to explore visual attention distribution of consumers while navigating on an online interface (Espigares-Jurado et al., 2020), while looking at evocatively beautiful and emotional pictures, tourism advertisements, or hotel marketing images (Li, Huang and 9 Christianson, 2016; Scott, Green and Fairley, 2016; Scott et al., 2019; Wang et al., 2018). However, eye tracking is not without limitations: eye movements can be influenced by excessive blinking or tears and also by biases when participants have previous experience with eye-tracking applications. In this vein, the combination of eye tracking and facialexpression analysis with other physiological data to collect tourist emotional responses can take research to a higher level (Li, Scott and Walters, 2015). Electrodermal activity analysis (EDA) measures the activation of the Autonomic Nervous System (ANS), which controls most of the organs and muscles and is responsible for arousal such as excitement and anticipation. When people are exposed to a stimulus, the ANS is activated leading to a change in skin conductivity. Skin conductance (SC) is a tool to measure EDA and therefore to capture emotion responses. EDA has also been used in tourism research to assess the emotional arousal experienced by participants in laboratory settings whilst observing stumuli (Babakhani, Ritchie & Dolnicar, 2017; Brodien Haparai, Walters & Li, 2018; Hadinejad, Moyle & Scott, 2019). The progress in technology regarding the availability of wearable and mobile physiological sensors offers researchers the opportunity to obtain tourists’ physiological responses, such as electrodermal activity when exposed to a variety of cognitive, physical, and emotional experiences, in real time and throughout the entire experience (Kim & Fesemmaier, 2015; Shoval, Schvimer & Tamir, 2018). The major limitations of EDA are related to the sensitivity of the equipment used and the demand for accurate analysis since skin conductance levels (SCLs) differ across individuals (Li, Scott & Walters, 2015). Facial-expression recognition techniques analyse the shapes and patterns of facial features as result of emotional responses when individuals are exposed to external and internal stimuli. Hence, the impact of any content, product, or service on how humans respond can be assessed by continuously observing the changes elicited in their emotional states as reflected in their facial features. Coding facial expressions during the time the individual is exposed to the stimuli can be employed to complement and enrich selfreports with quantified measures of emotional responses towards products or services that are of a more unconscious nature. The majority of facial-expression recognition devices or applications that gather facial images using video or thermal imaging (Huopio, 1998) involve the following steps (Figure 1): face detection, that is, the position of a face in a video frame, which can be achieved by applying classifier algorithms; detection of facial landmarks or facial features 16 detected face expresses a certain basic emotion (Stöckil et al., 2019), whereby the sum of all the detected emotions for an individual is 100%. Figure 2. Facial-recognition expressions: mobile device application All the images obtained during the visit were edited and fed into the application to obtain the data. Photographs were taken of 250 individuals. However, the final sample consisted of 230 individuals whose images were of a sufficiently high quality to be valid for their analysis by the software. For the final sample, 56% of the participants were Gender: Female Moustache: 0.00 Beard: 0.00 Smile: 0.00 Anger: 0.00% Contempt: 0.00% Disgust: 0.00% Fear: 0.00% Happiness: 95% Neutral: 5% Sadness: 0.00% Surprise: 0.00% 17 female and 44% male. Of the participants, 30% were between the ages of 20 and 35, 48% were 35 to 55, and 22% were over 55 years old, while 42% were international tourists and a large percentage had higher or secondary education (67%). The data obtained was analysed to understand the pattern the sample in terms of emotions. In order to go into further depth in the analysis to confirm the validity of the facial-expression recognition application in its ability to measure satisfaction with the service provided, a structural equation model (SEM) was estimated by using Partial Least Squares (PLS-SEM). From the SEM model, the relevance of the emotions in the explanation of customer satisfaction was tested. To this end, the SmartPLS v3 software was used. 4. Results 4.1. First-Stage Analysis A first analysis has been carried out in order to observe how the sample behaves in terms of emotions. For all respondents, different scores were given to the emotions (ANGER (Anger), CONTEMPT (Disappointment), DISGUST (Aversion), FEAR (Fear), HAPPINESS (Happiness), NEUTRAL (Neutral), SADNESS (Sadness), and SURPRISE (Surprise)). A person can present different emotions at the same time if these are similar emotions, such as Happiness and Surprise. A respondent with a high score in Happiness presents low scores in the negative emotions. Thus, the scores higher than 50% are the threshold for deciding the most predominant emotion in the individual. The descriptive statistics for the emotions are displayed in Table 1. Table 1. Descriptive statistics Surprise Neutral Fear Sadness Disgust Anger Contempt Happiness Average 1.937 10.630 0.230 2.010 0.7836 0.951 0.592 77.272 Std. Deviation 3.920 23.874 1.033 3.389 3.001 3.021 1.695 20.394 Pearson’s coeff. 2.024 2.246 4.492 1.686 3.829 3.176 2.863 0.264 From Table 1, it can be observed that Happiness is the emotion which has been detected as the most frequent emotion on the faces of participants while Fear was detected 18 with the lowest frequency by the Emotionalyser App. Moreover, most of the emotions present high variability based on Pearson’s Coefficient of variation (Fear, Disgust, Anger, Contempt, Neutral, Surprise), whereby the emotion Fear followed by Disgust and Anger are those with the highest variability, in contrast with Happiness which has the lowest variability. The distribution of the emotion scores for each respondent is also observed in Figure 3. While most of their scores for Happiness remained above 80%, for neutral emotions most fell below 20%. With only a few exceptions, as observed in Figure 3, all the negative emotions plus that of Surprise most frequently present scores below 5%. 19 Figure 3. Participants’ Emotion distribution 0 0 50 100 050 100 150 200 250 Happiness 0 10 20 30 40 0100 200 300 Surprise 0 50 100 150 050 100 150 200 250 Neutral 0.0 5.0 10.0 15.0 20.0 0100 200 300 Fear 0 10 20 30 0100 200 300 Sadness 0 10 20 30 050 100 150 200 250 Anger 0 10 20 30 0100 200 300 Disgust 0 5 10 15 20 050 100 150 200 250 Contempt Happiness scores Suprise scores Neutral scores Fear scores Sadness scores Anger scores Disgust scores Contempt scores 20 4.2 PLS-SEM results Partial Least Squares structural equation modelling (PLS-SEM) was employed to test the hypotheses. While PLS-SEM can be described as a composite-based approach that uses linear combinations of the indicator variables as proxies of the conceptual variables with a potential to explain the variance of the dependent variables, Covariance-Based SEM (CB-SEM) defines constructs as common factors that explain covariation between their indicators (Rigdon et al., 2017). Hence, the decision to use PLS-SEM is mainly made by setting the constructs for emotions in the model as composites defined as linear combinations of their indicators or dimensions (Becker, Rai & Rigdon, 2013; Henseler et al., 2014; Rigdon, 2012; Sarsted et al., 2016). The constructs for positive emotions (Happiness and Surprise) and for negative emotions (Disgust, Contempt, Fear, Anger, and Sadness) have been modelled as composites and estimated in Mode B (regression weights) since the existence of correlated items or internal consistency is not assumed for these latent variables. The weights provide evidence of the relative contribution of each indicator to their respective composite. 4.3. Measurement Model In order to evaluate the measurement models in PLS-path modelling, it is necessary to distinguish between Mode A composites and Mode B composites. The Mode B composites (EC) for negative and positive emotion constructs were evaluated at the indicator level by assessing multicollinearity and regression weights (Hair et al., 2017). Unlike composites in Mode A, discriminant validity, composite reliability, and Average Variance Extracted are not applicable. First, to test the potential multicollinearity between items, the statistics of the variance inflation factor (VIF) were calculated for the two constructs. The VIF values obtained for the indicators were below the 3.3 threshold (Petter, Straub and Rai, 2007). Hence, no multicollinearity was observed between the items for the two B composites. Second, the significance of regression weights was addressed via the bootstrapping technique based on n=5,000 subsamples. Table 2 shows the VIF values and the statistical significance of the regression weights. 21 Table 2. Measurement model for Mode B composites: regression weights and VIFs Mode B Composites Weights VIF Bootstrapping 95% Confidence Intervals BC Lower Upper Negative Emotions Contempt 0.227* 1.124 0.101 0.325 Disgust 0.427* 1.176 0.214 0.651 Fear 0.195* 1.280 0.092 0.320 Sadness 0.114* 1.037 0.05 0.321 Anger 0.109* 1.109 0.08 0.221 Positive Emotions Happiness 0.587* 1.563 0.285 0.653 Surprise 0.483* 1.734 0.258 0.514 Notes: BC: Bias-Corrected. 5,000 bootstrap samples; * p<0.05 (two-tailed t-distribution) 4.4. Structural Model In order to assess the structural model, potential multicollinearity between constructs has been checked. The VIF values for the exogenous constructs indicate that multicollinearity is not an issue in our research model since the VIFs remain below 3.3. Table 3 shows the main effects of the negative, positive, and neutral emotion constructs on satisfaction. To test for the significance of the path coefficients, the Bootstrapping procedure has been applied with 5,000 samples. It provides t-values and confidenceinterval bias corrected at a 95% confidence level for the assessment of the statistical significance of the path coefficients (Hayes & Scharkow, 2013; Roldán & Sánchez- Franco, 2012). From Table 3, it is observed that positive emotions have positive and significant effects on satisfaction (β=0.557), whereas a negative relationship between negative emotions and satisfaction has been found (β=-0.322) as expected. Hence, hypotheses H1 and H2 are confirmed. It is also interesting to notice that the contribution of positive emotions is greater than that of negative emotions on satisfaction. 22 A high value for the coefficient of determination (R-square) has been obtained, which reveals the great explanatory power of emotions regarding satisfaction (Hair et al., 2011; Henseler et al., 2009). Thus, 68.89% of the variability of satisfaction is explained by the respondents’ emotions. The model has also been assessed through the analysis of the cross-validated redundancy index (Q2) for the dependent variable of satisfaction. A Q2 value of 0.348 (greater than 0) implies that the model shows satisfactory predictive relevance. The high explanatory power and predictive relevance achieved in the research model might indicate that the algorithm of the Emotionalyser App is a very powerful tool for the explanation of satisfaction through facial-expression recognition. Finally, to evaluate the goodness of fit of the research model, the standardized root mean square residual (SRMR) (Henseler et al., 2014) has been obtained. The SRMR is of 0.067, which lies below the thresholds of 0.10, and the more conservative threshold of 0.08 (Hu & Bentler, 1999). Table 3. Hypotheses testing, path, and confidence interval Bootstrapping 95% confidence interval BCa Path coefficients Lower Upper H1: Positive Emotions Satisfaction 0.557** *(8.891) 0.297 0.672 H1a: Positive Emotions*gender Satisfaction 0.073* (2.256) 0.004 0.187 H1b: Positive Emotions*agegroup1Satisfaction 0.004 (0.978) -0.031 0.978 H1b: Positive Emotions*agegroup2Satisfaction 0.0061 (0.1553) -0.0825 0.0748 H2: Negative Emotions  Satisfaction -0.322***(4.133) -0.523 -0.185 H2a: Negative Emotions*genderSatisfaction 0.087 *(2.246) -0.007 -2.011 H2b: Negative Emotions*agegroup1Satisfaction 0.0032 (0.098) -0.0004 0.147 H2b: Negative Emotions*agegroup2 Satisfaction 0.007(0.831) -0.0023 0.876 Notes: BCa, Bias-Corrected and accelerated. 5,000 bootstrap samples. *p<0.05;**p<0.01, ***p<0.001 (based on t-statistics, one-tailed test). +Gender=1 for female and Gender=0 for male; +agegroup1=1 for age between 20 and 35, and agegroup1=0 for age between 35 and 55 (this being the reference group); +agegroup2=1 for age over 55, and agegroup2=0 for age between 35 and 55. 23 Having considered the possibility of heterogeneity being caused by demographic variables such as age and gender in the relationships analysed by hypotheses H1 and H2, the article employs those variables as moderating factors. The results shown in Table 3 demonstrate that there are no significant differences expressed by different age groups regarding the relation between positive emotions and satisfaction and the relation between negative emotions and satisfaction. Hypotheses H1b and H2b are therefore not supported. However, by considering gender as a moderating variable, the relation between positive emotions and overall satisfaction is shown to be stronger for female participants, whereas the relation between negative emotions and overall satisfaction is stronger for male visitors. Hypotheses H1a and H2a are consequently confirmed. 5. Discussion and Conclusion The aim of the study is to analyse individuals’ immediate emotional responses when an on-site tour-guide service is provided in a heritage place. We assume that the emotional responses are good indicators of the perception of the quality of the service provided. To support this assumption, the paper provides a literature review on research which focuses on the role of emotions in the measurement of tourist satisfaction. In order to test the hypotheses that emotions detected via individuals’ facial expressions influence satisfaction, participants were also invited to comment on their overall satisfaction with the service at the end of the guided visit, and structural equation modelling was used that was estimated by means of PLS. The study found that Happiness, with its scores higher than those of other emotions, was the most frequent emotion response. Happiness and Neutral emotions present more variability in comparison to negative emotions. Positive and neutral emotions exert a positive and significant influence on overall satisfaction, whereas negative emotions have a negative and significant influence on overall satisfaction. This study concludes that emotional responses as coded by the facial-expression recognition device, Emotionalyser, constitute a good indicator for overall satisfaction derived from the service provided on a tourist heritage site. The findings achieved are supported by the academic literature that shows 24 empirical evidence of emotions playing a relevant role in the customer experience in tourism (Lee & Kyle, 2012; Mcintosh & Siggs, 2005; Prayag, Hosany & Odeh, 2013; Yuksel, Yuksel & Bilim, 2010) and therefore in behaviour intention (Bigné et al., 2005; Prayag, Hosany & Odeh, 2013; Tsaur, Chiu & Wang, 2007). The moderating influence of gender on the relation between emotions overall satisfaction is also supported by previous studies in the tourism literature (Wu et al., 2016; Yan & Wu, 2018), since positive emotions are more commonly associated with women and neutral and negative emotions are more frequent in men. 5.1. Theoretical Implications In the last decade, scholars have claimed the need to employ real-time and real-world settings to attain physiological measurements (Hunziker, Buchecker & Hartig, 2007; Walls et al., 2011). However, a few studies applying psychophysiological metrics have been found in the literature and the most frequent methodology to measure emotions in tourism is through surveys via either open-ended questions (Prayag, Hosany & Odeh, 2013; Walters, Sparks & Herington, 2012) or self-report methods (Chamberlein & Broderick, 2007; Jacobs, Feheres & Campell, 2012; Wang & Minor, 2008). However, self-report questionnaires are not exempt from limitations since they are subject to a bias towards social desirability, the scale of measurement employed, or seek responses regarding emotional experiences of which respondents remain unaware. Measuring emotions through facial-expression recognition is still in its infancy in the tourism sector (Kim & Fesenmaier, 2015; Mills, Meyers & Byun, 2010), and has largely been applied in the fields of marketing and publicity (Lee, Broderick & Chamberlein, 2007; Micu & Plummer, 2010) and salient organizations and large-scale governmental programs (Mills, Meyers & Byun, 2010). It is broadly recognized that researchers and managers should understand the major role of emotions in customer experiences in the tourism destination and in the hospitality industries (Li, Ashkanasy, & Ahlstrom, 2014). Our study responds to a claim in the literature on tourism with respect to recording facial-expression recognition spontaneously rather than in a laboratory (Matsumoto & Hwang, 2015). 25 Furthermore, this paper combines both psychophysiological measures and self-report questionnaires as a hybrid approach towards confirming the potentiality of the application of AI in the tourist sector. The combination of these methods provides a potential direction for future research in the tourism industry at a time when artificial intelligence instruments (facial-expression recognition, eye tracking, etc.) are continuously improving in their ability to detect emotions while the subjects remain unaware of the spontaneous measurements being taken. 5.2. Managerial Implications In recent decades, major efforts have been made in information technology that enable a better understanding of people’s emotions and their influence on behaviour (Blaters & Steirnert, 2015; Campellone, & Kring, 2013; Firmin, Luther, Lysaker, Minor & Salyers, 2016; Lerner, Li, Valdesolo, & Kassam, 2015). For over forty years, the facial-expression recognition method has been under development for the collection of accurate information regarding emotional responses to a stimulus (Wilhelm & Grossman, 2010). However, artificial intelligence is starting to be more widely appreciated in the tourism industry since it can improve business by measuring customer experience through the codification of emotional experiences (Kim & Fesenmaier, 2015). Since the user experience is dynamic (Forlizzi & Ford, 2000), the accurate measurement of emotions provides Destination Marketing Organizations (DMOs) and practitioners with valuable insights regarding the perception of the experience by the customers in real time. In general, tourist destinations and the hospitality industry strive to provide tourists with memorable experiences that go beyond mere satisfaction. Emotional responses might therefore be used not only to improve the quality of the services in tourist destinations and in the hospitality industry, but also to design attractions, activities, and hotels, and to allocate resources more efficiently. The facial-coding approach provides the tourism industry with a means to measure the impact of the service regarding retaining customers, and to attract new customers by creating a differential value based on the customers’ emotional experiences, and therefore to improve its ability to provide customers with a 32 Domínguez-Quintero, A. M., González-Rodríguez, M. R., and Roldán, J. L. (2018). The role of authenticity, experience quality, emotions, and satisfaction in a cultural heritage destination. Journal of Heritage Tourism Vol. 14 No. 5, 491-505 Donovan, R. J., Rossiter, J. R., Marcoolyn, G., and Nesdale, A. (1994). Store atmosphere and purchasing behavior. Journal of Retailing, Vol. 70 No. 3, 283-294 Dubé, L., and Menon, K. (2000). Multiple roles of consumption emotions in postpurchase satisfaction with extended service transactions. International Journal of Service Industry Management, Vol. 11 No. 3, 287-304 Ekman, P. (1992). An argument for basic emotions. Cognition and Emotion, Vol. 6, 169- 200 Ekman, P. (1999). Basic emotions. Handbook of cognition and emotion. John Wiley & Sons Ltd Ekman, P., and Friesen, W. V. (1978). Manual for the facial action coding system. Consulting Psychologists Press Espigares-Jurado, F., Muñoz-Leiva, F., Correia, M. B., Sousa, C. M., Ramos, C. M., and Faísca, L. (2020). Visual attention to the main image of a hotel website based on its position, type of navigation and belonging to Millennial generation: An eye tracking study. Journal of Retailing and Consumer Services, Vol. 52, 101906. Faria, A. R., Almeida, A., Martins, C., Gonçalves, R., Martins, J., and Branco, F. (2017). A global perspective on an emotional learning model proposal. Telematics and Informatics, Vol. 34 No. 6, 824-837. Faullant, R., Matzler, K., and Mooradian, T. A. (2011). Personality, basic emotions, and satisfaction: Primary emotions in the mountaineering experience. Tourism Management, Vol. 32 No. 6, 1423-1430 Firmin, R. L., Luther, L., Lysaker, P. H., Minor, K. S., and Salyers, M. P. (2016). Stigma resistance is positively associated with psychiatric and psychosocial outcomes: A meta-analysis. Schizophrenia Research, Vol. 175 No. 1-3, 118-128. Forlizzi, J., and Ford, S. (2000). The building blocks of experience: an early framework for interaction designers. In Proceedings of the 3rd conference on Designing interactive systems: processes, practices, methods, and techniques (pp. 419-423). Frijda, N. H. (1988). The laws of emotion. American Psychologist, Vol. 43 No. 5, 349- 358 33 Frijda, N. H. (1993). Moods, emotion episodes, and emotions. In M. Lewis & JM Haviland (Eds.), Handbook of emotions (pp. 381-403). New York, NY, US: Guilford Press González-Rodríguez, M. R., Martínez-Torres, R., and Toral, S. (2016). Post-visit and previsit tourist destination image through e-WOM sentiment analysis and perceived helpfulness. International Journal of Contemporary Hospitality Management, Vol. 28 No. 11, 2609-2627. González-Rodríguez, M. R., Domínguez-Quintero, A. M., and Paddison, B. (2019). The direct and indirect influence of experience quality on satisfaction: the importance of emotions. Current Issues in Tourism, 1-19, doi.org/10.1080/13683500.2019.1668917 Graham, S. M., Huang, J. Y., Clark, M. S., and Helgeson, V. S. (2008). The positives of negative emotions: Willingness to express negative emotions promotes relationships. Personality and Social Psychology Bulletin, Vol. 34 No. 3, 394-406 Hadinejad, A., Moyle, B. D., Kralj, A., and Scott, N. (2019). Physiological and self-report methods to the measurement of emotion in tourism. Tourism Recreation Research, Vol. 44 No. 4, 466-478. Hadinejad, A., Moyle, B. D., Scott, N., and Kralj, A. (2019). Emotional responses to tourism advertisements: the application of FaceReader™. Tourism Recreation Research, Vol. 44 No. 1, 131-135. Hair Jr, J. F., Sarstedt, M., Ringle, C. M., and Gudergan, S. P. (2017). Advanced issues in partial least squares structural equation modeling. SAGE Publications Hair, J. F., Ringle, C. M., and Sarstedt, M. (2011). PLS-SEM: Indeed a silver bullet. Journal of Marketing theory and Practice, Vol. 19 No. 2, 139-152 Hayes, A. F., and Scharkow, M. (2013). The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: Does method really matter? Psychological Science, Vol. 24 No. 10, 1918-1927 Healey, J., Nachman, L., Subramanian, S., Shahabdeen, J., and Morris, M. (2010). Out of the lab and into the fray: towards modeling emotion in everyday life. In International Conference on Pervasive Computing (pp. 156-173). Springer, Berlin, Heidelberg. Hensel, J. S. (1970). Physiological measures of advertising effectiveness: A theoretical and empirical investigation. Doctoral dissertation, The Ohio State University Henseler, J., Ringle, C. M., and Sinkovics, R. R. (2009). The use of partial least squares path modeling in international marketing. In New challenges to international marketing (pp. 277-319). Emerald Group Publishing Limited 34 Henseler, J., Dijkstra, T. K., Sarstedt, M., Ringle, C. M., Diamantopoulos, A., Straub, D. W., and Calantone, R. J. (2014). Common beliefs and reality about PLS: Comments on Rönkkö and Evermann (2013). Organizational Research Methods, Vol. 17 No. 2, 182-209 Hetland, A., Vittersø, J., Fagermo, K., Øvervoll, M., and Dahl, T. I. (2016). Visual excitement: analyzing the effects of three Norwegian tourism films on emotions and behavioral intentions. Scandinavian Journal of Hospitality and Tourism, Vol. 16 No. 4, 528-547. Hirose, M. (2001). Combined voluntary gaze direction and facial muscle activity as a new pointing technique. In Human-computer Interaction: INTERACT'01: IFIP TC. 13 International Conference on Human-Computer Interaction, 9th-13th July 2001, Tokyo, Japan (p. 100). IOS Press Hosany, S., and Gilbert, D. (2010). Measuring tourists’ emotional experiences toward hedonic holiday destinations. Journal of Travel Research, Vol. 49 No. 4, 513-526 Hu, L. T., and Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural equation modeling: a multidisciplinary journal, Vol. 6 No. 1, 1-55 Hunziker, M., Buchecker, M., and Hartig, T. (2007). Space and place–two aspects of the human-landscape relationship. In A changing world (pp. 47-62). Springer, Dordrecht Huopio, S. (1988). Biometric identification. Eye. Paper presented at Seminar on Network Security: Authorization and Access Control in Open Network Environment, Helsinki University of Technology, Dept. of Computer Science, available at: www.tml.tkk.fi/Opinnot/Tik-110.501/ 1998/papers/12biometric/biometric.htm Isomursu, M., Tähti, M., Väinämö, S., and Kuutti, K. (2007). Experimental evaluation of five methods for collecting emotions in field settings with mobile applications. International Journal of Human-Computer Studies, Vol. 65 No. 4, 404-418 Jacobs, M. H., Fehres, P., and Campbell, M. (2012). Measuring emotions toward wildlife: a review of generic methods and instruments. Human Dimensions of Wildlife, Vol. 17 No. 4, 233-247 Kaiser, S., and Wehrle, T. (1994). Emotion research and AI: some theoretical and technical issues. Geneva Studies in Emotion and Communication, Vol. 8 No. 2, 1-16 Kayser, D. (2017). Using facial expressions of emotion as a means for studying musicinduced emotions. Psychomusicology: Music, Mind, and Brain, Vo. 27 No. 3, 219 35 Kim, J., and Fesenmaier, D. R. (2015). Measuring emotions in real time: Implications for tourism experience design. Journal of Travel Research, Vol. 54 No. 4, 419-429 Kim, M. H., Joo, Y. H., and Park, J. B. (2005). Emotion detection algorithm using frontal face image. 제어 로봇 시스템 학회 국제 학술 대회 논문집, 2373-2378 Kirana, K. C., Wibawanto, S., and Herwanto, H. W. (2018). Emotion Recognition using Fisher Face-based Viola-Jones Algorithm. Proceedings of the Electrical Engineering Computer Science and Informatics, Vol. 5 No. 5, 173-177. Kolodyazhniy, V., Kreibig, S. D., Gross, J. J., Roth, W. T., and Wilhelm, F. H. (2011). An affective computing approach to physiological emotion specificity: Toward subject´s independent and stimulus´ independent classification of film‐induced emotions. Psychophysiology, Vol. 48 No. 7, 908-922. Kreibig, S. D. (2010). Autonomic nervous system activity in emotion: A review. Biological Psychology, Vol. 84 No. 3, 394-421. Lambie, J. A., and Marcel, A. J. (2002). Consciousness and the varieties of emotion experience: A theoretical framework. Psychological Review, Vol. 109 No. 2, 219-259 Lang, P. J., Greenwald, M. K., Bradley, M. M., and Hamm, A. O. (1993). Looking at pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology, Vol. 30 No. 3, 261-273 Lee, J., and Kyle, G. T. (2012). Recollection consistency of festival consumption emotions. Journal of Travel Research, Vol. 51 No. 2, 178-190 Lee, N., Broderick, A. J., and Chamberlain, L. (2007). What is ‘neuromarketing’? A discussion and agenda for future research. International Journal of Psychophysiology, Vol. 63 No. 2, 199-205 Lerner, J. S., Li, Y., Valdesolo, P., and Kassam, K. S. (2015). Emotion and decision making. Annual Review of Psychology, Vol. 66, 799-823 Li, Y., Ashkanasy, N. M., and Ahlstrom, D. (2014). The rationality of emotions: A hybrid process model of decision-making under uncertainty. Asia Pacific Journal of Management, Vol. 31 No. 1, 293-308 Li, S., Scott, N., and Walters, G. (2015). Current and potential methods for measuring emotion in tourism experiences: A review. Current Issues in Tourism, Vol. 18 No. 9, 805-827 Li, S., Walters, G., Packer, J., and Scott, N. (2015). Innovative techniques for measuring emotion in tourists' decision making processes and experiences. CAUTHE 2015: 36 Rising Tides and Sea Changes: Adaptation and Innovation in Tourism and Hospitality, 602. Li, S., Walters, G., Packer, J., and Scott, N. (2016). Revisiting the influence of emotion on tourism advertising effectiveness using psychophysiological methods. Travel and Tourism Research Association: Advancing Tourism Research Globally. 14. https://scholarworks.umass.edu/ttra/2016/Grad_Student_Workshop/14 Li, Q., Huang, Z. J., and Christianson, K. (2016). Visual attention toward tourism photographs with text: An eye-tracking study. Tourism Management, Vol. 54, 243- 258. Li, S., Walters, G., Packer, J., and Scott, N. (2018a). A comparative analysis of self-report and psychophysiological measures of emotion in the context of tourism advertising. Journal of Travel Research, Vol. 57 No. 8, 1078-1092. Li, S., Walters, G., Packer, J., and Scott, N. (2018b). Using skin conductance and facial electromyography to measure emotional responses to tourism advertising. Current Issues in Tourism, Vol. 21 No. 15, 1761-1783. Li, S. (2019). Emotional Appeals in Tourism TV Commercials: A Psycho-Physiological Study. Journal of Hospitality & Tourism Research, Vol. 43 No. 6, 783-806. Litman, D., and Forbes, K. (2003). Recognizing emotions from student speech in tutoring dialogues. In Automatic Speech Recognition and Understanding, 2003. ASRU'03. 2003 IEEE Workshop on (pp. 25-30). IEEE Lowe-Calverley, E., Grieve, R., and Padgett, C., (2019), A Risky Investment? Examining the Outcomes of Emotional Investment in Instagram, Telematics and Informatics, doi: https://doi.org/10.1016/j.tele.2019.101299. Machleit, K. A., and Eroglu, S. A. (2000). Describing and measuring emotional response to shopping experience. Journal of Business Research, Vol. 49 No. 2, 101-111 Magdin, M., Benko, Ľ., and Koprda, Š. (2019). A case study of facial emotion classification using affdex. Sensors, Vol. 19 No. 9, 2140 Matsumoto, D., & Hwang, H. C. (2014). Judgments of subtle facial expressions of emotion. Emotion, Vol. 14 No. 2, 349-357 Mauss, I. B., and Robinson, M. D. (2009). Measures of emotion: A review. Cognition and Emotion, Vol. 23 No. 2, 209-237 Mauss, I. B., Levenson, R. W., McCarter, L., Wilhelm, F. H., and Gross, J. J. (2005). The tie that binds? Coherence among emotion experience, behavior, and physiology. Emotion, Vol. 5 No. 2, 175-190. 37 Mcintosh, A. J., and Siggs, A. (2005). An exploration of the experiential nature of boutique accommodation. Journal of Travel Research, Vol. 44 No. 1, 74-8 Micu, A. C., and Plummer, J. T. (2010). Measurable emotions: How television ads really work: Patterns of reactions to commercials can demonstrate advertising effectiveness. Journal of Advertising Research, Vol. 50 No. 2, 137-153 Mills, J. E., Meyers, M., and Byun, S. (2010). Embracing broadscale applications of biometric technologies in hospitality and tourism: is the business ready? Journal of Hospitality and Tourism Technology, Vol. 1 No. 3, 245-256 Morin, C. (2011). Neuromarketing: the new science of consumer behavior. Society, Vol. 48 No. 2, 131-135 Oh, H., and Kim, K. (2017). Customer satisfaction, service quality, and customer value: years 2000-2015. International Journal of Contemporary Hospitality Management, Vol. 29 No. 1, 2-29. Oliver, R. L. (1977). Effect of expectations and disconfirmation on post-exposure product evaluations. Journal of Applied Psychology Vol. 62 No. 4, 480-486 Oliver, R. L., and DeSarbo, W. S. (1988). Response determinants in satisfaction judgments. Journal of Consumer Research, Vol. 14 No. 4, 495-507 Oliver, R.L. (1981). Measurement and evaluation of satisfaction processes in retail settings. Journal of Retailing, Vol. 57 No. 3, 25-48 Parasuraman, A., Zeithaml, V. A., and Berry, L. L. (1988). Servqual: A multiple-item scale for measuring consumer perception. Journal of Retailing, Vol. 64 No. 1, 12-37 Parasuraman, A., Zeithaml, V. A., and Berry, L. L. (1994). Reassessment of expectations as a comparison standard in measuring service quality: implications for further research. Journal of Marketing, 111-124 Partala, T., Jokiniemi, M., and Surakka, V. (2000). Populi. Eye Tracking, Research & Applications Symposium: 123–129 Paulhus, D. L. (2002). Socially desirable responding: The evolution of a construct. In H. Braun, D. N.Jackson, & D. E. Wiley (Eds.), The role of constructs in psychological and educational measurement (pp. 67–88). Hillsdale, NJ: Erlbaum Paulhus, D. L., and Vazire, S. (2007). The self-report method. Handbook of research methods in personality psychology, Vol. 1, 224-239 Petter, S., Straub, D., and Rai, A. (2007). Specifying formative constructs in information systems research. Management Information Systems Quarterly, 623-656 38 Petrick, J.F., Morais, D.D. and Norman, W.C. (2001). An Examination of the Determinants of Entertainment Vacationers’ Intentions to Revisit. Journal of Travel Research, Vol. 40, 41-48. Pittman, M., and Reich, B. (2016). Social media and loneliness: Why an Instagram picture may be worth more than a thousand Twitter words. Computers in Human Behavior, Vol. 62, 155-167. Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in organizational research: Problems and prospects. Journal of Management, 12(4), 531-544 Poels, K., and Dewitte, S. (2006). How to capture the heart? Reviewing 20 years of emotion measurement in advertising. Journal of Advertising Research, Vol. 46 No. 1, 18-37 Prayag, G., Hosany, S., and Odeh, K. (2013). The role of tourists' emotional experiences and satisfaction in understanding behavioral intentions. Journal of Destination Marketing & Management, Vol. 2 No. 2, 118-127 Ravaja, N. (2004). Contributions of psychophysiology to media research: Review and recommendations. Media Psychology, Vol. 6 No. 2, 193-235 Regulation (EU) 2016/679, 2016 O.J. (L 119) 1 Richins, M. L. (1997). Measuring emotions in the consumption experience. Journal of Consumer Research, Vol. 24 No. 2, 127-146 Rigdon, E. E. (2014). Rethinking partial least squares path modeling: breaking chains and forging ahead. Long Range Planning, Vol. 47 No. 3, 161-167 Rigdon, E. E., Sarstedt, M., and Ringle, C. M. (2017). On comparing results from CBSEM and PLS-SEM: five perspectives and five recommendations. Marketing Zfp, Vol. 39 No. 3, 4-16 Rodríguez Molina, M. Á., Frías-Jamilena, D. M., and Castañeda-García, J. A. (2013). The moderating role of past experience in the formation of a tourist destination's image and in tourists’ behavioural intentions. Current Issues in Tourism, Vol. 16 No. 2, 107-127 Roldán, J. L., and Sánchez-Franco, M. J. (2012). Variance-based structural equation modeling: guidelines for using partial least squares in information systems research. In Research methodologies, innovations and philosophies in software systems engineering and information systems (pp. 193-221). IGI Global San Martín, H. (2005): Estudio de la imagen de destino turístico y el proceso global de satisfacción: Adopción de un enfoque integrador. Tesis doctoral, Universidad de Cantabria 39 San Martín, H., and Del Bosque, I. A. R. (2008). Exploring the cognitive–affective nature of destination image and the role of psychological factors in its formation. Tourism Management, 29(2), 263-277 Sarstedt, M., Hair, J. F., Ringle, C. M., Thiele, K. O., and Gudergan, S. P. (2016). Estimation issues with PLS and CBSEM: Where the bias lies! Journal of Business Research, Vol. 69 No. 10, 3998-4010 Sedenberg, E., & Chuang, J. (2017). Smile for the camera: privacy and policy implications of emotion AI. arXiv preprint arXiv:1709.00396. Shanshi, L. I., Scott, N., and Walters, G. (2014). Potential methods for measuring emotion in Chinese tourism advertising. G20 First East-West Dialogue on Tourism and the Chinese Dream, 66. Shanshi, L. I., Scott, N., andWalters, G. (2014). Potential methods for measuring emotion in Chinese tourism advertising. G20 First East-West Dialogue on Tourism and the Chinese Dream, 66. Shoval, N., Schvimer, Y., and Tamir, M. (2018). Real-time measurement of tourists’ objective and subjective emotions in time and space. Journal of Travel Research, Vol. 57 No. 1, 3-16 Söderlund, M., & Sagfossen, S. (2017). The depicted service employee in marketing communications: An empirical assessment of the impact of facial happiness. Journal of Retailing and Consumer Services, Vol. 38, 186-193 Stock, R. (2011). How does product program innovativeness affect customer satisfaction? A comparison of goods and services. Journal of the Academy of Marketing Science, Vol. 39, 813-827. Stöckli, S., Schuttle-Mecklenbeck M., Borer, S. and Samson, A.C. (2019). A Practical Guide for Automated Facial Emotion Classification in Schulte-Mecklenbeck, M., Kühberger, A., & Johnson, J. G. (Eds.). A handbook of process tracing methods. Routledge (Chapter 13). Tarnowski, M. P., Kołodziej, A. Majkowski, and Rak, R. J. (2017). Emotion Recognition using Facial Expressions. International Conference on Computational Science, pp. 1175–1184. Teixeira, T., Wedel, M., and Pieters, R. (2012). Emotion-induced engagement in internet video advertisements. Journal of Marketing Research, Vol. 49 No. 2, 144-159. 40 Tsaur, S. H., Chiu, Y. T., and Wang, C. H. (2007). The visitors behavioral consequences of experiential marketing: An empirical study on Taipei Zoo. Journal of Travel & Tourism Marketing, Vol. 21 No. 1, 47-64 Tussyadiah, I. P., and Fesenmaier, D. R. (2009). Mediating tourist experiences: Access to places via shared videos. Annals of Tourism Research, Vol. 36 No. 1, 24-40 Urry, H. L. (2009). Using reappraisal to regulate unpleasant emotional episodes: Goals and timing matter. Emotion, Vol. 9 No. 6, 782-797 Walls, A. R., Okumus, F., Wang, Y. R., and Kwun, D. J. W. (2011). An epistemological view of consumer experiences. International Journal of Hospitality Management, Vol. 30 No. 1, 10-21 Walters, G., Sparks, B., and Herington, C. (2012). The impact of consumption vision and emotion on the tourism consumer’s decision behavior. Journal of Hospitality & Tourism Research, Vol. 36 No. 3, 366-38 Wang, Y. J., and Minor, M. S. (2008). Validity, reliability, and applicability of psychophysiological techniques in marketing research. Psychology & Marketing, Vol. 25 No. 2, 197-232 Wang, T. C., Tsai, C. L., and Tang, T. W. (2018). Exploring Advertising Effectiveness of Tourist Hotels’ Marketing Images Containing Nature and Performing Arts: An Eye- Tracking Analysis. Sustainability, Vol. 10 No. 9, 3038. White, C. J., and Scandale, S. (2005). The Role of Emotions in Destination Visitation Intentions: A Cross-Cultural Perspective. Journal of Hospitality & Tourism Management, Vol. 12 No. 2, 168-179 Wilhelm, F. H., and Grossman, P. (2010). Emotions beyond the laboratory: Theoretical fundaments, study design, and analytic strategies for advanced ambulatory assessment. Biological Psychology, Vol. 84 No. 3, 552-569 Wilhelm, F. H., Grossman, P., and Mueller, M. (2012). Bridging the gap between the laboratory and the real world: Integrative ambulatory psychophysiology. Handbook of research methods for studying daily life, 210-234. Worthington, I., Ram, M., and Jones, T. (2006). Exploring corporate social responsibility in the UK Asian small business community. Journal of Business Ethics, Vol. 67 No. 2, 201-217 Wu, X., Chen, J., Jia, T., Ma, W., Zhang, Y., Deng, Z., and Yang, L. (2016). Cognitive bias by gender interaction on N170 response to emotional facial expressions in major and minor depression. Brain topography, Vol. 29 No. 2, 232-242. 41 Yan, Q., Zhou, S., and Wu, S. (2018). The influences of tourists’ emotions on the selection of electronic word of mouth platforms. Tourism Management, Vol. 66 No. 1, 348-363. Yu, Y. T., and Dean, A. (2001). The contribution of emotional satisfaction to consumer loyalty. International Journal of Service Industry Management, Vol. 12 No. 3, 234- 250 Yüksel, A., and Yüksel, F. (2007). Shopping risk perceptions: Effects on tourists’ emotions, satisfaction and expressed loyalty intentions. Tourism Management, Vol. 28 No. 3, 703-713 Yuksel, A., Yuksel, F., and Bilim, Y. (2010). Destination attachment: Effects on customer satisfaction and cognitive, affective and conative loyalty. Tourism Management, Vol. 31 No. 2, 274-284 Žabkar, V., Brenčič, M. M., and Dmitrović, T. (2010). Modelling perceived quality, visitor satisfaction and behavioural intentions at the destination level. Tourism Management, Vol. 31 No. 4, 537-546.