Abstract
This study examined whether the tourist experience aligns with the peak–end rule, offering a perspective in the field of travel. This study analyzed real-time data on tourists’ emotional valence, arousal, and satisfaction—key elements of their momentary experiences—and their impact on subsequent retrospective evaluations. The results indicated that these three aspects of tourist experiences during travel significantly impacted retrospective evaluations. In contrast, neither the duration of the experience nor the beginning moments had significant effects. Noteworthy points in the tourist experience, such as the peaks, troughs, and ends of valence and satisfaction, significantly affected loyalty post-trip, thereby supporting the peak–end rule. However, arousal experiences did not have this effect. This study also found that extreme negative experiences in the tourism context may require more consideration than positive peak experiences.
Highlights
Momentary experience sampling method (ESM) data simultaneously capture tourists’ valence, satisfaction, and arousal, revealing their temporal impacts on post-trip loyalty.
The study found that tourists’ valence and satisfaction align with the peak–end rule; arousal does not.
The beginning and duration of travel experiences have no significant impact on loyalty.
Extreme negative moments weigh more on loyalty than positive peak experiences.
Introduction
He told of listening raptly to a long symphony on a disc that was scratched near the end, producing a shocking sound, and he reported that the bad ending ruined the whole experience. (Kahneman, 2011, p. 381)
This incident demonstrates how a negative ending can spoil the overall experience. The peak–end rule theory proposes that in any given experiential process, not all moments have the same significance, with peak and end experiences playing a far more critical role, whereas the average of other momentary experiences or duration of the experience tend to have negligible effects (e.g., Kahneman, 2000; Kahneman et al., 1993; Redelmeier & Kahneman, 1996; Scheibehenne & Coppin, 2020; Schreiber & Kahneman, 2000). This idea is backed by several studies, including the cold pressor test (Kahneman et al., 1993) and colonoscopy experiment (Redelmeier & Kahneman, 1996).
During the travel process, tourists may have a series of different experiences. For example, McCullough et al. (2024) mentioned that interactions with various individuals, such as bar staff, ticketing agents, check-in staff, tour guides, and even cleaning staff, are each distinct but can potentially influence the overall post-trip evaluation of the entire travel experience, further emphasizing that accurately understanding customer evaluations of service experiences requires comprehending how customers assess the series of interactions throughout the service process. Numerous studies have assessed the peak–end rule across a variety of sectors, contexts, and stimuli. However, research in the field of tourism remains scarce. Noteworthy studies include Park et al. (2018), which explored whether adjusting the final leg of the itinerary for a Seoul tour group from a traditional tour bus to a boat ride on the Han River back to the airport could enhance overall satisfaction. Other studies include those focusing on student holidays (Kemp et al., 2008), travelers to Macau (Chark et al., 2022), the experience of tourists at diverse attractions (Kim & Kim, 2019), students visiting a museum (Mitas et al., 2022), and visitors to a theme park (Li, 2021). These findings suggest that understanding peak and end moments can inform the design of better tourism experiences through focusing on the emotional peaks and ends of a trip, which positively impact tourists’ retrospective evaluations. For tourism practitioners, this insight contributes to the structuring of itineraries that enhance visitor experiences at key moments, ultimately influencing visitors’ future behavioral intentions.
Nevertheless, the findings of these studies are not uniform. Kemp et al. (2008) revealed that peak and trough experiences did not significantly impact retrospective evaluations. Chark et al. (2022) suggested that an individual’s overall retrospective evaluation is influenced by the totality of experiences during travel, rather than just peak and end moments. S. Li (2021) discovered that visitors’ satisfaction levels are better aligned with the average emotion intensity than peak moment. While Kim and Kim (2019) focused on tourists’ assessments of individual attractions, Park et al. (2018) did not investigate the impact of peak experiences. Furthermore, Mitas et al. (2022) did not explore the moment-to-moment positive and negative emotional values. Considering the divergent results in tourism research, systematically examining the extent to which tourist experience aligns with the peak–end rule is imperative.
A closer examination of key moments, such as peaks and ends, reveals a lack of consensus on their definition in tourism research. Some studies (Kemp et al., 2008; Kim & Kim, 2019; Li, 2021; Mitas et al., 2022) use emotional valence or arousal to define these crucial instances. In contrast, other studies (Just et al., 2015; Park et al., 2018) rely on levels of satisfaction. These disparate approaches might contribute to the variations found in research outcomes. Thus, the question of whether the peaks or ends of emotions or those of satisfaction more significantly affect tourists’ retrospective evaluations remains. Scheibehenne and Coppin (2020) propose that, in situations with both positive and negative taste experiences, extreme negative experiences more significantly impact retrospective evaluations. However, whether the same holds true for tourist experiences has not been empirically investigated.
In addition to discussions on key moments, the duration of experiences (Kahneman, 2000; Kahneman et al., 1993; Redelmeier & Kahneman, 1996; Scheibehenne & Coppin, 2020) and the beginning moments (Bhargave & Montgomery, 2013; Kemp et al., 2008; Li, 2021) have also been addressed in related research, although such studies are limited in the tourism sector. Whether the duration and beginning moments of a tourist experience influence retrospective evaluations remains to be further explored.
Furthermore, many tourism studies conduct cross-sectional recall surveys post-trip, viewing emotions as a static construct (S. Li et al., 2015). This methodology is also predominant in studies examining the peak–end rule in the field of tourism (Chark et al., 2022; Kemp et al., 2008; Kim & Kim, 2019; Park et al., 2018). However, individuals’ memories of events are typically composed of several notable incidents (Fredrickson & Kahneman, 1993), making it impossible to recall every moment of a holiday (Kemp et al., 2008). Consequently, memory decay and interference may hinder recall surveys (Larsen, 2007). Furthermore, a tourist’s emotions and satisfaction levels during the journey may oscillate, encompassing both positive and negative experiences (Quinlan Cutler et al., 2018). Highly charged single experiences—whether extremely positive or negative—may affect the entire recollection process (Redelmeier & Kahneman, 1996). Thus, this study used real-time recording methods similar to those used in traditional peak–end rule studies to examine the alignment of tourists’ experiences with the pertinent theories.
An accumulating body of evidence underscores the key role of the peak–end rule in shaping individual experiences, with key moments significantly impacting retrospective evaluations. However, it remains unclear whether tourism experiences align with the peak–end rule, or which experiences should be used to define key moments, and there is a lack of research exploring these questions from a momentary level. To address this, this study used the experience sampling method (ESM), which is standard for investigating momentary experiences in the field of psychology (Hektner et al., 2007; Liu et al., 2016; Shoval & Ahas, 2016; Quinlan Cutler et al., 2018). This approach tracks individuals’ momentary experiences in order to investigate the impact of these experiences on retrospective evaluations and provide insights from a standpoint of enhanced ecological validity.
This study aims to systematically examine whether tourists’ experiences align with the peak–end rule, focusing on how key emotional moments—such as peak, trough, and end experiences—along with the duration and beginning of a trip, influence retrospective evaluations. By using the ESM, this study can simultaneously capture tourists’ real-time emotional and satisfaction data, thereby offering a more detailed understanding of which momentary experiences do or do not shape overall trip evaluations, contributing new insights to tourism management and future research.
Literature Review
Peak–End Rule
Kahneman et al. (1993) used two versions of the cold pressor test and found that the participants demonstrated a greater inclination to re-engage with the version that entailed a longer overall discomfort period but was less painful toward the conclusion. This ground-breaking research paved the way for subsequent investigations into the peak–end rule. In 1996, Redelmeier and Kahneman explored this rule in colonoscopy and revealed that the overall post-procedure sensation was chiefly tied to the most discomforting and final moments, rather than the duration of the procedure. Thus, an individual’s comprehensive evaluation of an experience appears to predominantly hinge on the peak and final moments of the event (Kahneman et al., 1993). The number of peak experiences encountered throughout the process does not have a significant impact (Schreiber & Kahneman, 2000). Furthermore, the peak–end rule appears to be more fitting for experiences with well-defined beginnings and endings, and the length of the experience does not markedly influence the retrospective overall assessment (Kahneman, 2000; Scheibehenne & Coppin, 2020).
The peak–end rule has been validated in various contexts, such as watching videos (Fredrickson & Kahneman, 1993) and eating pizza (Just et al., 2015). Scheibehenne and Coppin (2020) explored mixed-valence events; the participants continuously rated a sequence of tastes, each with positive or negative emotional valences. The results revealed that both positive and negative experiences adhered to the peak–end rule. Moreover, in situations including both positive and negative experiences, extreme negative experiences had a stronger influence.
Reitsamer et al. (2020) demonstrated that experiential elements play a critical role in shaping memory dynamics, influencing how key moments are retained and evaluated after consumption. Although research on the peak–end rule in tourism is scarce, the number of studies is gradually increasing. For instance, Kemp et al. (2008) examined 49 students planning to take a holiday, who were required to send daily text messages and provide an emotional assessment of their experiences over the preceding 24 hours, as well as an overall evaluation after the vacation. Significant correlations were found between the average happiness during the holiday, peak–end (average of the two moments), trough-end (average of the two moments), and overall degree of recollected happiness. Surprisingly, the happiness experienced at the end of the holiday was a more potent predictor of the total recollected happiness than the peak or trough. Notably, the duration of the holiday did not impact the retrospective evaluation. However, this study focused on the participants’ autobiographical remembrances of their experiences over the past 24 hours each day during the holiday, rather than instantaneous documentation during each present moment. Such daily reflections could be affected by the peak–end experiences of the preceding 24 hours.
Chark et al. (2022) invited tourists at ferry terminals and Macau International Airport, who were ending their travels, to recall as many episodes as they could from their trip and rate their satisfaction for each activity, as well as provide an overall assessment of their entire trip. The findings identified the most positive, most negative, and last episode from the participants’ recollections. The findings demonstrated that summary evaluations were better predicted using an arithmetic average of all episodic evaluations rather than the peak–end rule. The findings indicated that overall evaluations were impacted by most of the events experienced, rather than only peak–end events, which may be due to the lengthier and more complex nature of travel than that previously investigated by psychologists. As the participants were asked to recollect their travel experiences and events post-journey and not in real-time, memory decay may have occurred (Larsen, 2007). In addition, the events recalled later may have been extreme peak–end events.
Park et al. (2018) manipulated the experiences at the tail end of the itinerary of tour groups heading for Seoul and revealed that tourists who were randomly assigned to take a boat ride back to the airport via the Han River reported elevated levels of overall satisfaction compared with the bus-riding groups. However, this study used an experimental manipulation to alter the end-of-trip experience before conducting a recall survey. The methodology did not include real-time documentation of the tourists’ momentary experiences during the journey or examine peak effects.
Kim and Kim (2019) examined heritage tourism and found that tourists’ responses supported the peak–end rule. However, this study investigated post-visit evaluations of individual heritage sites and overall post-tour evaluations, rather than capturing the tourists’ ongoing experiences throughout the journey in real time. Ellis et al. (2023) conducted a study involving adolescents participating in a summer camp, measuring their experiences across eight recreational activities. Following the completion of all activities, participants provided retrospective evaluations of each activity. The findings revealed that the global average of the recreational experiences had a stronger impact on the post-activity evaluations than the peak-end average. However, it is worth mentioning that the study relied on retrospective evaluations after the activities were completed, rather than recording real-time experiences during the activities.
Few studies have tracked tourist experiences in real time to examine the peak–end rule. For instance, Mitas et al. (2022) monitored students’ emotional arousal through skin conductance readings during museum visits and outdoor guided tours, followed by a questionnaire survey to investigate the students’ inclination to recommend the experience. Emotional arousal during the tour significantly impacted recommendation inclination. However, the study did not examine momentary satisfaction or positive and negative emotional values during the visit. In contrast, S. Li (2021) used real-time skin conductance measurements and recall self-reports during park visits, and found that the emotional valence at the end moment, coupled with the average valence, significantly affected post-visit evaluations. Nevertheless, emotional arousal at key moments did not significantly influence post-visit evaluations. Although this study employed skin conductance readings for real-time measurement of emotional arousal, the emotional valence measurements were based on participants’ post-visit memories rather than real-time documentation of emotional valence.
Tueanrat et al. (2021) reviewed the customer journey literature, emphasizing the need for precise frameworks to capture the dynamic and multifaceted stages of customer experiences, especially in complex contexts like tourism. Consequently, the peak–end rule has garnered increasing attention in tourism research as a framework to understand the influence of critical moments on retrospective evaluations. Peak, trough, and end experiences are significantly more influential, while the other momentary experiences or the duration of the experience have negligible impacts (e.g., Kahneman et al., 1993; Kemp et al., 2008; Redelmeier & Kahneman, 1996; Scheibehenne & Coppin, 2020; Schreiber & Kahneman, 2000). However, the definitions of the peaks and troughs during a tourist’s journey lack uniformity, and techniques to record tourists’ momentary experiences may be inadequate. Thus, the present study examined these aspects.
Main Research Variables and Hypotheses
Existing research presents diverse definitions of peaks and troughs in the tourist experience. Some scholars have examined these moments through the lens of emotions (examples include Kemp et al., 2008; Kim & Kim, 2019; Mitas et al., 2022), whereas others have used satisfaction metrics (such as Chark et al., 2022; Just et al., 2015; Park et al., 2018). To examine the influence of these key moments—as well as the beginning and duration of experiences—on retrospective evaluations, this study further discusses the relationships between variables and proposes research hypotheses.
Retrospective evaluation
Loyalty is a key outcome variable in tourist behavior studies (Han et al., 2010). This study follows Mitas et al. (2022) to use loyalty as an index for retrospective evaluations. Loyalty is intrinsically connected with a tourist’s inclination to return to and endorse a destination (Lin & Liu, 2018), embodying the repeated pattern of tourists frequenting the same travel destination (Tsaur & Sun, 2009). Siebert et al. (2020) further conceptualize loyalty as part of a cyclical process, where predictable and smooth experiences form loyalty loops that strengthen customers’ commitment over time. This perspective aligns with the present study’s focus on loyalty as a retrospective evaluation variable. The present study defines loyalty as a tourist’s tendency to make a return visit to a tourist attraction and willingness to recommend it to others.
Recent studies have suggested the potential impact of emotions and satisfaction on tourist loyalty. For instance, So et al. (2024) demonstrated that emotional experiences during travel significantly contribute to sustaining tourists’ subjective well-being after the trip, underscoring the value of emotional connections in fostering positive post-travel outcomes. Kim and So (2022) conducted a comprehensive review of 2 decades of customer experience research in hospitality and tourism, emphasizing the critical influence of emotional and cognitive dimensions on customer evaluations and behaviors, and identifying loyalty as a key outcome of customer experience. Al-Msallam (2020) found that emotions significantly influence tourists’ loyalty. Similarly, Hultman et al. (2015) suggested that satisfied tourists are likely to share positive opinions about the destination with their social circle after returning home. In a more recent study, Bagheri et al. (2024) confirmed that tourist satisfaction positively affects tourist loyalty. These findings indicate the feasibility of using loyalty as a variable for retrospective evaluations.
Emotions
Tourism literature often represents emotions through two distinct approaches: basic and dimensional (Li, 2021). The basic approach identifies a limited number of key human emotions. For instance, the Consumption Emotion Set (CET) proposed 16 emotions relevant to the consumption context, such as anger, worry, joy, excitement, and surprise (Richins, 1997). Conversely, the dimensional approach divides emotions into two dimensions: valence, which refers to the positive or negative emotional level, and arousal, which indicates the activation level (S. Li et al., 2015; Russell, 1980). However, Russell (1980) argued that these basic emotional categories are not standalone entities but are interconnected, filling a spatial model defined by the axes of valence and arousal. Compelling evidence to support the discrete classification of emotions into categories is lacking (Barrett & Wager, 2006).
The dimensional approach may be fitting for tourism studies that do not focus on a specific emotion (such as Hosany et al., 2021; Li, 2021). Therefore, this study investigated the continuous shifts in emotional valence (positive or negative) and arousal (activation level) in the dimensional framework, thereby illustrating the emotions elicited by the tourist experience. Based on the peak–end rule, this study proposes the following hypotheses:
H1: Tourists’ valence experiences align with the peak–end effect.
1.1. The peak valence has a stronger impact on loyalty than other momentary valence.
1.2. The trough valence has a stronger impact on loyalty than other momentary valence.
1.3. The end valence has a stronger impact on loyalty than other momentary valence.
H2: Tourists’ arousal experiences align with the peak–end effect.
2.1. The peak arousal has a stronger impact on loyalty than other momentary arousal.
2.2. The trough arousal has a stronger impact on loyalty than other momentary arousal.
2.3. The end arousal has a stronger impact on loyalty than other momentary arousal.
Satisfaction
Satisfaction is a consumer’s comparative assessment of the benefits obtained after consumption versus prior expectations; outcome that does not meet expectations may result in dissatisfaction (Oliver, 1980). The concept of satisfaction is widely employed in tourism and marketing research (Lu et al., 2015). Satisfaction can be defined as the degree to which an individual experiences positive or negative feelings (Rust & Oliver, 1994). Most existing research emphasizes the expression facet of satisfaction, implying that satisfaction is an overall evaluation of the purchasing and consumption experience of a product or service (Anderson et al., 1994). Thus, this study defined satisfaction as the degree to which tourists evaluate their current experiences of travel as positive or negative. In line with the peak–end rule, this study proposes the following hypotheses.
H3: Tourists’ satisfaction experiences align with the peak–end effect.
3.1. The peak satisfaction has a stronger impact on loyalty than other momentary satisfaction.
3.2. The trough satisfaction has a stronger impact on loyalty than other momentary satisfaction.
3.3. The end satisfaction has a stronger impact on loyalty than other momentary satisfaction.
Duration and beginning of experiences
Moreover, the duration of experiences is also often an issue discussed in studies evaluating the peak–end rule Kahneman et al. (1993) and Redelmeier and Kahneman (1996) demonstrated that the length of an experience does not significantly influence retrospective evaluations. Similarly, Kahneman (2000) and Scheibehenne and Coppin (2020) found no significant relationship between the length of a movie and the overall impression held after watching it. However, studies mirroring these findings in the tourism sector are lacking. Among the few that discuss the issue of duration is Kemp et al. (2008), who revealed that the length of a holiday taken by students did not impact their memories of overall happiness from the trip. Furthermore, although the peak–end rule does not typically assign a significant role in shaping retrospective evaluations to the beginning of an experience, several researchers argue that these experiences influence later recollections (Bhargave & Montgomery, 2013; Kemp et al., 2008; Li, 2021). Consequently, whether the duration of a touristic experience and its beginning moments affect retrospective evaluations requires additional exploration. This study discusses the peak-end rule from the experiences of valence, arousal, and satisfaction, and thus proposes the research hypotheses:
H4: Tourists’ experiences at the beginning moment do not have a significant impact on loyalty.
4.1. Begin valence does not have a significant impact on loyalty.
4.2. Begin arousal does not have a significant impact on loyalty.
4.3. Begin satisfaction does not have a significant impact on loyalty.
H5: The duration of the tourist experience does not have a significant impact on loyalty.
5.1. The duration of the valence does not have a significant impact on loyalty.
5.2. The duration of the arousal does not have a significant impact on loyalty.
5.3. The duration of the satisfaction does not have a significant impact on loyalty.
Exploring Momentary Experiences With ESM
Literature on the exploration of tourists’ momentary experiences has a noticeable gap (Mannell & Iso-Ahola, 1987). Scholars have proposed that investigating momentary experiential elements in tourism would provide a more valid measure than recall methods owing to the decline in memory accuracy over time (Larsen, 2007). The experience sampling method (ESM) is a frequent strategy for analyzing momentary experiences, as its real-time approach allows the acquisition of subjective experiential data from individuals in distinct time–space contexts (Hektner et al., 2007; Quinlan Cutler et al., 2018). The ESM was developed in the 1970s and primarily used in psychological research (Csikszentmihalyi & Larson, 1987). It collects real-time data in situ, focusing on subjective momentary experiences (Quinlan Cutler et al., 2018). In recent years, the ESM has been validated across various fields and employs interval-, signal-, event-, and place-contingent sampling methods (Birenboim, 2016; Hektner et al., 2007; Quinlan Cutler et al., 2018; Tang & Jhang, 2020). The ESM minimizes recall bias, unlike retrospective evaluations (Stone et al., 2007). However, challenges include fatigue, respondent burden, and attrition, which can be mitigated through proper research design and remuneration (Barrett & Barrett, 2001; Scollon et al., 2009). Moreover, Gabriel et al. (2019) mentioned that sample size recommendations for ESM studies remain limited, and, if possible, it would be helpful to provide sample sizes from related studies in the field for comparison.
The ESM is an intensive data collection method that necessitates repeated responses from participants during specific time periods (Tang & Jhang, 2020). With the recent rapid evolution of mobile communication devices, some studies have used smartphones to conduct ESM surveys (Shoval & Ahas, 2016). For instance, Quinlan Cutler et al. (2018) investigated the momentary experiences and emotions of 21 college students during a trip to Peru, subsequently comparing them with the students’ later memories. Tang and Jhang (2020) used the LINE messaging app to inquire about the 44 participants’ music-listening behavior during their day-to-day activities. Liu et al. (2016) used smartphones to document 51 people’s assessments of the food and wine at a national event. Shoval et al. (2018) used custom-made mobile software to investigate 68 tourists in Jerusalem, recording how their experiences and emotions changed spatially throughout their trip. It appears that ESM studies in tourism have adopted a range of sample sizes. Considering that Shoval et al. (2018) successfully recorded the emotional changes of 68 tourists, this study will aim for a similar sample size.
This highlights the importance of using the ESM to investigate tourists’ momentary experiences. Thus, the present study used the ESM to examine tourists’ experiences at key moments in the journey. As scholars have suggested that initial experiences may influence individuals’ subsequent memories (Bhargave & Montgomery, 2013; Kemp et al., 2008; Li, 2021), initial moments were included in this study’s exploration.
Research Methods
This study used the ESM to examine tourists’ momentary satisfaction, positive or negative emotion (valence), and the activation level of emotion (arousal), while also logging the time of the response (subsequently transformed into the duration from the start of the journey to the response time). After the trip, tourists were invited to provide a retrospective evaluation of their loyalty to the specific trip. Here, the beginning moment was defined as the first response to the ESM survey, with the end moment signifying the final response at the termination of the ESM survey. Peak and trough valence/arousal/satisfaction referred to the experiences at the moments of the highest and lowest levels of valence/arousal/satisfaction, respectively, encountered by tourists during their journey. A retrospective evaluation denoted the assessment of loyalty post-trip. As noted by Godovykh and Tasci (2020), understanding customer experiences requires addressing multiple dimensions to capture the full complexity of emotional responses. This study specifically focuses on three critical dimensions: valence (positive or negative emotions), arousal (activation level of emotions), and satisfaction. By incorporating these dimensions into the Experience Sampling Method (ESM), this study attempts to provide a granular understanding of how key emotional and satisfaction moments influence tourists’ overall evaluations of their journeys.
Measurement Questionnaire
The survey comprised two components: the ESM questionnaire and a retrospective evaluation conducted post-trip. In the ESM questionnaire, participants described their momentary experience, encompassing areas such as emotional states (valence and arousal) and satisfaction levels. After the trip, participants were asked to retrospectively evaluate their loyalty toward the trip.
This study used a dimensional approach to interpret tourists’ emotions in two dimensions: valence and arousal. This strategy helped articulate the emotions incited by the tourist experience. Drawing on previous studies (Hosany et al., 2021; Li, 2021; S. Li et al., 2015; Russell, 1980), the valence dimension was evaluated using the following question: “How negative or positive does this experience make you feel?” Responses were rated on a 7-point Likert scale, ranging from negative to positive. The arousal dimension was assessed by inquiring, “How inactive or active does this experience make you feel?” Responses were rated on a 7-point Likert scale (1 = inactive; 7 = active).
The majority of studies on satisfaction focus on the performance dimension, suggesting that satisfaction is an overall evaluation of a product or consumption experience (Anderson et al., 1994). In line with this, and following Kim and Kim (2019) and S. Li (2021), this study posed the following question: “How satisfied are you with this experience?” Responses were rated on a 7-point Likert scale (1 = dissatisfied; 7 = satisfied).
Loyalty denotes a tourist’s inclination to revisit and endorse a tourist site after their visit (Lin & Liu, 2018; Tsaur & Sun, 2009). This study adopted the loyalty scale based on Chi and Qu (2008). The scale includes four statements: “I would like to visit this site again,” “I would like to visit other sites here in the future,” “I would recommend this site to my friends,” and “I would recommend other sites here to my friends.” Responses were rated on a 7-point Likert scale (1 = disagree; 7 = agree).
Moreover, after the ESM survey, the participants’ demographic information, including gender, age, marital status, and education, was gathered.
Study Area
The study area was Houtong in New Taipei City, Taiwan, a locale that provides visitors with a unique blend of positive and negative experiences. Houtong, once the highest coal-producing area in Taiwan with thousands of miners and their families, declined in 1990 due to the cessation of mining, resulting in severe population loss, leaving behind only the elderly and children (Z. L. Zhou, 2009). In 2007, a cat photographer captured images of the local cats and published a book, attracting a large number of tourists, which turned the cats into an important tourism resource and prompted the government and local businesses to organize various events in response (Lin & Huang, 2020). In recent years, the stories of the old miners have been presented in the Mining Museum Park through old photos, manuscripts, audio recordings, videos, mining tools, and buildings, along with simulations of the harsh working conditions in the tunnels (Z. L. Zhou, 2009). Therefore, visitors to Houtong can not only enjoy interacting with the cats in the village but also experience the emotional hardships of the old miners by visiting the mining museum. This study focuses on exploring tourists’ positive and negative emotions, as well as their retrospective evaluations, in the context of Houtong.
Study Design
This study collected real-time self-reported data throughout the tourists’ journeys, using smartphones. This methodology facilitated an understanding of how key moments during the trip influenced retrospective evaluations after the tour. A preliminary study was conducted in December 2021 before the main research commenced. This pilot study was conducted in the research area with five participants: two male and three female tourists, all invited on-site. It assisted researchers in becoming familiar with the recruitment, explanation, and survey process and confirming the functionality and practicality of the online questionnaire and research protocol. All five participants completed the survey, with travel times ranging from 2 to 3.5 hours.
The main research was conducted on weekends in January 2022. Based on Shoval et al. (2018), who successfully recorded the emotional experiences of 68 tourists, this study aimed for a similar sample size. Considering potential attrition, a total of 80 participants were recruited. Prospective participants were recruited on-site at the research locale through posters. These posters detailed the nature of the study, with participants expected to complete short online surveys on their smartphones three times per hour, totaling more than 2 hours during their visit. Upon completion, the participants received NT 500 (approximately US$16) as reward.
Researchers comprehensively explained the research process, ensuring that the participants understood their right to voluntarily withdraw from the survey at any point. For participants without the LINE messaging app installed on their smartphones, help was provided to set it up and connect them to the dedicated LINE account. This allowed the participants to test recording their experiences and become comfortable with the research procedure. During the journey, a signal-contingent sampling technique was utilized. Based on Birenboim’s (2016) successful implementation of the ESM to record tourists’ experiences, this study also had research assistants send messages three times per hour. These messages contained a link to an online survey. Only responses received prior to the subsequent message were counted as valid. Each participant’s journey spanned a minimum of 2 hours. After their visit, participants were asked to rate their loyalty to the trip.
Previous research has indicated that ESM surveys could interfere with a tourist’s activities, consequently impacting the response rate. Various elements might contribute to this interaction, including the intensity of signals, questionnaire length (Scollon et al., 2009), appreciative communication before the survey, remuneration (Barrett & Barrett, 2001), participants’ backgrounds, conditions when responding, and ease of completing the questionnaire (Tang & Jhang, 2020). As this study sent three messages every hour, which may be considered intense, we minimized the time required for each response to prevent overloading the participants and reducing response rates (Hektner et al., 2007). Moreover, we fostered participant engagement by monitoring response status, delivering timely reminders, and providing suitable rewards. Additionally, Gabriel et al. (2019) noted that because ESM studies emphasize experiences in natural environments, participants usually do not undergo the same type of developmental process, although in some cases similar effects, like the Hawthorne effect, may arise. Given the profound impact of momentary states on evaluations (Schwarz, 2012), this study followed Gabriel and colleagues’ (2019) suggestion by including momentary experiences as control variables in the statistical analysis, which might help mitigate this effect. Upon completion of data collection, this study used SPSS (Version 24) to conduct descriptive analysis on participants and variables, and employed mixed model analysis to examine the impact of momentary experiences during the trip on retrospective loyalty.
Study Results
Participant Characteristics
Among the 80 participants, 15 dropped out before completion, resulting in a total of 65 tourists participated in this study. The number of responses per participant during the ESM process fluctuated between six and 18, amassing 739 unique experiences. Including the 65 post-trip responses, this study gathered 804 experiential data points. Out of the 65 participants, 44.6% were men and 55.4% were women. The majority (66.2%) were 20–29 years old, followed by those aged 30–39 years (15.4%) and 40–49 years (13.8%). Only 4.6% of the participants were 50–59 years old. Furthermore, 27.7% of the participants were married, whereas 72.3% were single. Most participants (72.3%) had a university or college degree, 15.4% had a high or higher vocational school diploma or less, and 12.3% had a Master’s degree or higher.
Descriptive Analysis of Variables
This study drew upon the theory of the peak–end rule and used the ESM to examine the momentary experiences of tourists throughout their journey. The study focused on the impact of experiences at the beginning, peak, trough, and end of an individual’s tour on their subsequent retrospective evaluations. To facilitate this, the study generated 12 binary-coded variables: BeginVal, PeakVal, TroughVal, EndVal, BeginArou, PeakArou, TroughArou, EndArou, BeginSati, PeakSati, TroughSati, and EndSati. The first recorded experience of each participant was assigned a value of 1 under the variables BeginVal, BeginArou, and BeginSati, and all subsequent experiences for the same participant were assigned a value of 0 under these variables. Similarly, the last experience of each participant was designated as 1 under the variables EndVal, EndArou, and EndSati. Moreover, the experiences that represented the highest and lowest points of valence, arousal, and satisfaction for each individual were identified. The highest points were assigned a value of 1 under PeakVal, PeakArou, and PeakSati, and the lowest points were marked as 1 under TroughVal, TroughArou, and TroughSati. We then computed the mean (M) and standard deviation (SD) for each key moment (i.e., moments with any of the 12 variables equaling 1; Table 1). The loyalty score recorded in the post-trip was 5.64 (SD = 1.10).
Descriptive Analysis of Variables
Mixed Model Analysis
The peak–end rule posits that key moments assume a more significant role in retrospective evaluations than other individual experiences (Kahneman et al., 1993). This study used the ESM and examined a series of momentary experiences throughout the tourists’ journeys. To investigate the impact of valence, arousal, and satisfaction at each key moment, rather than other experiences during the tour, three models were used. The data collected using the ESM contained a wealth of information, including repeated measurements of multiple variables for each participant. As the measurements were not independent of each other, the application of a linear regression model was inappropriate. Consequently, this study used a mixed model analysis to estimate within-individual variations.
The Akaike Information Criterion (AIC) estimates the relative quality of a statistical model by approximating the distance between the fitted and true models, and is used for model selection by choosing the one with the lowest AIC value (Taddy, 2019). The valence model showed that the compound symmetry correlation structure had the lowest AIC value at 2283.447. Additionally, the results indicated that the valence of each momentary experience during the tourist’s trip had a significant impact on the post-trip retrospective loyalty score (0.44). Conversely, the duration of the journey did not significantly influence the post-trip retrospective evaluation. Moreover, we examined the impact of experiences at various valence key moments on post-trip loyalty scores after controlling for momentary valence experience and duration. The beginning moment did not have a significant impact on the retrospective evaluation. In contrast, experiences at the peak, trough, and end of valence significantly affected post-trip evaluations (Table 2). The trough of valence (compared with non-trough moments) had the most significant and negative impact on post-trip loyalty (-0.51). This was followed by the impacts of the peak of valence and end moment on loyalty (0.38 and 0.27, respectively). Therefore, this study tends to support research Hypotheses 1.1, 1.2, 1.3, 4.1, and 5.1.
Valence Mixed Model Regression Analysis (Dependent Variable: Loyalty)
Note. 1Compared with non-begin moments. 2Compared with non-peak moments. 3Compared with non-trough moments. 4Compared with non-end moments.
The compound symmetry correlation structure, based on the arousal mixed model analysis, exhibited an AIC value of 2455.104. Momentary emotional arousal during experiences had significant, although relatively small, effects on post-trip loyalty (0.06). However, the duration of this experience did not have a significant effect on post-trip evaluations. In addition, we measured the influence of key moments on post-trip loyalty after controlling for the momentary arousal experience and duration. The findings indicated that the beginning moment, peaks and troughs of arousal, and the end, did not significantly impact loyalty (Table 3). Therefore, Hypothesis 4.2 and 5.2 were supported, while Hypotheses 2.1, 2.2. and 2.3 were not supported.
Arousal Mixed Model Regression Analysis (Dependent Variable: Loyalty)
Note. 1Compared with non-begin moments. 2Compared with non-peak moments. 3Compared with non-trough moments. 4Compared with non-end moments.
The satisfaction model analysis revealed that the compound symmetry correlation structure held an AIC value of 2276.512. The momentary satisfaction derived from each experience during a tourist’s journey significantly affected post-trip loyalty (0.45). However, the duration of the experience did not significantly affect post-trip loyalty. Furthermore, we examined the impact of experiences during key moments on post-trip loyalty after controlling for momentary satisfaction and duration. The experience at the beginning moment did not significantly impact loyalty. However, the experiences during the satisfaction peak, trough, and end significantly influenced loyalty (Table 4). The trough of satisfaction (compared with the non-trough moment) had the strongest—albeit negative—impact on loyalty (-0.64), followed by the peak of satisfaction and the end moment (0.41 and 0.29, respectively). As a result, research Hypotheses 3.1, 3.2, 3.3, 4.3, and 5.3 are supported.
Satisfaction Mixed Model Regression Analysis (dependent Variable: Loyalty)
Note. 1Compared with non-begin moments. 2Compared with non-peak moments. 3Compared with non-trough moments. 4Compared with non-end moments.
Discussion and Implications
Discussion
Since Kahneman and colleagues’ (1993) pioneering study, research has validated the peak–end rule in shaping individual experiences. However, the findings were not consistent in tourism research, necessitating further examination. To evaluate the relative influence of key moments on non-key moment experiences, this study employed three mixed models to individually analyze the peak–end effects of emotional valence, arousal, and satisfaction, and tested five research hypotheses.
Hypotheses 1 and 3 were both supported. Nevertheless, key moments such as emotional valence or satisfaction, experiences at the peak (compared with non-peak), trough (compared with non-trough), and end (compared with non-end) moments significantly impact the retrospective evaluations conducted post-trip. In other words, most positive/negative emotions, highest/lowest satisfaction, and end moments have the strongest influence on tourists’ comprehensive retrospective evaluations after the journey. The reasons for this remain unclear. Some psychology studies have suggested that this could be related to memory-related processes (Geng et al., 2013). For instance, Fredrickson and Kahneman (1993) considered an individual’s portrayal of memories of events as consisting of a few stills rather than a continuous record. Similarly, Kemp et al. (2008) found that people cannot recall their level of happiness at every moment during their holiday. The impact of key moment experiences, such as peaks and troughs, on retrospective evaluations could be due to extreme values being better remembered (Ludvig et al., 2014). Likewise, the impact of the end moment’s experience might be associated with recency effects, which tend to amplify the impact of the information (Hogarth & Einhorn, 1992).
The only unsupported hypothesis in this study is Hypothesis 2. When key moments were defined by arousal, the experiences at the peak, trough, and end moments did not significantly influence retrospective evaluations. This suggests that arousal might not align well with the peak–end rule. This conclusion mirrors Li’s (2021) findings regarding theme park visitors, in which real-time measurements of skin conductance (representing arousal levels) obtained during the participants’ experiences, when analyzed against self-reported data post-recollection, revealed that emotional arousal at various key moments did not significantly influence post-experience evaluations. The related findings offer potential explanations. Lenton et al. (2016) suggest that activities like running and meditation can intensify individuals’ momentary experiences. Interestingly, running tends to elevate arousal (Wegner & Giuliano, 1980), whereas meditation can decrease it (Holmes, 1984), indicating that both heightened (higher) and calmer (lower) emotional states may enrich one’s experience. This may explain why, in the arousal model, the key moments did not significantly influence retrospective evaluations.
Research Hypothesis 4 is supported in this study, although several scholars have suggested that experiences at beginning moments may influence individuals’ subsequent memories (Bhargave & Montgomery, 2013; Kemp et al., 2008; Li, 2021), which contradicts the results of the present study. None of the three models demonstrated that experiences at the beginning moments had a significant impact on retrospective evaluations. The findings of this study were in line with the peak–end rule proposed by Kahneman et al. (1993), indicating that the beginning moment did not play a significant role in retrospective evaluations.
Research Hypothesis 5 is also supported. Among all three models, the duration of experiences did not have a significant impact on loyalty. These findings align with those of Kahneman et al. (1993) and Redelmeier and Kahneman (1996). Moreover, Kemp et al. (2008) found that the length of the holiday did not affect retrospective evaluations of the journey. However, their study examined participants’ recollections of their journey experiences, whereas this study captured participants’ experiences in real-time.
Although previous tourism studies have largely concentrated on positive emotional experiences (Ma et al., 2013), the journey can also encompass a range of negative experiences (Q. B. Zhou et al., 2018). The present study involved real-time tracking of tourists’ experiences and revealed that the instances of the most negative satisfaction or emotion in both valence and satisfaction models had the strongest negative influence on loyalty. This aligns with Scheibehenne and Coppin’s (2020) findings that extremely negative experiences tend to have greater influence than extremely positive ones. Some researchers propose that negative information is more likely to conform to the peak–end rule than positive information. Negative experiences tend to sustain a higher emotional intensity and are more memorable (Vaish et al., 2008). Kensinger (2009) argued that, in terms of neural transmission processes, negative peak (or trough) experiences carry a disproportionately higher weight in memory. From an evolutionary standpoint, negative peak experiences could serve as alerts to potential harm and signal to individuals their capacity to handle similar experiences in the future (Fredrickson, 2000). Thus, the negative trough experiences during a tour may warrant greater attention than the positive peak ones.
Theoretical Implications
This research presents several new findings in contrast to existing work. First, it provides new theoretical contributions by applying the ESM in the tourism context, which enables more accurate momentary measurement of tourists’ emotional and satisfaction levels. Unlike traditional retrospective methods (Chark et al., 2022; Ellis et al., 2023; Kemp et al., 2008; Kim & Kim, 2019; Park et al., 2018), ESM allows for the capture of moment-to-moment emotional shifts, offering a clearer view of how peak–end rule operates during the actual experience, rather than relying on potentially biased recollections. Furthermore, Becker and Jaakkola (2020) provide a theoretical foundation for understanding customer experience as a response to both managerial stimuli and consumption processes. This framework is particularly relevant in dynamic contexts like tourism, where momentary experiences are influenced by complex interactions. Their premises reinforce the importance of using real-time measurement methods, such as ESM, to capture the nuanced dynamics of emotional and satisfaction shifts during the customer journey.
Keiningham et al. (2020) propose that customer experience not only influences immediate satisfaction but also drives strategic innovations, particularly when organizations align customer values with their business objectives. Building on this perspective, this study’s findings show how key moments play a pivotal role in enhancing tourist loyalty, offering actionable insights for customer-centric innovation in the tourism industry. Additionally, given that past research findings are not uniform, this study defines key moments from three perspectives. The findings of this study reveal that emotional valence and satisfaction, rather than arousal, are more likely to align with the peak–end rule in tourism contexts. The role of extreme negative experiences appears to have a much stronger influence on loyalty than other moments. In past arousal-based studies, this observation contrasts with the findings of Mitas et al. (2022), but aligns with the results of S. Li (2021). Given that both higher and lower emotional states can enhance one’s experience (Holmes, 1984; Lenton et al., 2016; Wegner & Giuliano, 1980), this theoretically suggests that arousal may be less suitable for discussing the peak-end effect in tourism research.
While previous research, such as that by Kahneman et al. (1993) and Kemp et al. (2008), has validated the peak–end rule in various contexts, this study contributes to the theory by demonstrating that the most negative trough experiences—particularly in satisfaction and valence models—have the strongest negative impact on loyalty, in the context of tourism. This finding aligns with the notion that negative information tends to weigh more heavily on memory and decision-making, as proposed by Kensinger (2009) and Vaish et al. (2008).
While previous studies have primarily focused on validating the peak–end rule in controlled environments (e.g., Kahneman et al., 1993; Redelmeier & Kahneman, 1996; Scheibehenne & Coppin, 2020), this research uses real-time data to track tourists’ emotional and satisfaction levels during their journeys. By using ESM, this study uncovers the heightened influence of tourists’ extreme negative moments, which traditional retrospective evaluations (e.g., Kemp et al., 2008; Kim & Kim, 2019; Park et al., 2018) may have failed to fully capture. Rahman et al. (2022) propose a comprehensive framework for measuring customer experience, emphasizing the integration of emotional and cognitive dimensions across complex, multifaceted contexts. Their insights reinforce the importance of real-time methods, like ESM, to better understand the nuanced dynamics of tourist experiences and their impact on loyalty. The momentary data provided by ESM demonstrates that tourists’ loyalty is disproportionately affected by negative experiences, offering a more nuanced understanding of the peak–end rule than previously established. Additionally, while the peak–end rule has been extensively examined, this study highlights the disproportionate impact of negative trough experiences over positive peaks, offering a fresh perspective on how tourists’ negative episodes can dominate overall evaluations. These findings suggest that future research on the peak–end rule should consider the weight of negative experiences in shaping tourists’ long-term behavioral intentions, contributing new dimensions to the theoretical understanding of the rule.
Although memories are not always true, they undeniably have an impact on personal decision-making processes (Kahneman, 2011). In the context of tourism, the question arises: Do all experiences during a journey leave an imprint on memory, or is memory primarily influenced by a handful of key moments? Current discourse in the tourism field is inconclusive. This study bridges this gap, suggesting that defining key moments in terms of emotional valence or levels of satisfaction might be more consistent with the peak–end rule, offering a potentially apt metric for subsequent research. Furthermore, the findings of this study highlight the need to focus on the impact of extreme negative experiences on tourists’ long-term behavioral intentions. This study sheds light on tourist experiences and the peak–end rule and furthers understanding of momentary emotional valence, arousal, and satisfaction levels among tourists.
Managerial Implications
Only specific experiences during travel influence tourists’ future behavioral intentions (Hosseini et al., 2023). The results of this study shed light not only on how key experiences influence retrospective evaluations but also on potential focal points for future tourist destination management. Mixed model analyses revealed that emotional valence and satisfaction during travel aligned with the peak–end rule. Specifically, experience duration and beginning moments did not affect retrospective evaluations, whereas the peaks and troughs of emotional valence and satisfaction and the end experiential moments played a central role in shaping the overall retrospective assessment. These findings have practical implications. Elevating tourists’ emotional valence and satisfaction could help amplify long-term positive experiences (Wirtz et al., 2003) and enhance the overall tourist experience (Dixon et al., 2017), thereby affecting future loyalty.
While not all travel experiences are unforgettable, the focus in the competitive tourism market has traditionally been on providing tourists with positive and memorable experiences (Hosseini et al., 2023). However, the findings of this study indicate that greater consideration might be warranted toward addressing negative experiences in tourism management and operations. In the emotional valence model, the most negatively charged emotional trough decreased post-trip loyalty. This effect was significantly stronger than that of momentary, peak, and end valence. Similarly, in the satisfaction model, tourists’ most dissatisfying trough moments decreased loyalty, and this impact was significantly stronger than that of peak momentary, and end satisfaction. This indicates that, while momentary emotional valence and satisfaction significantly impact loyalty, a single emotional valence or dissatisfaction trough has stronger negative effects on future tourist intentions. This can be compared with a grating noise that hinders the experience of savoring music (Kahneman, 2011). This study reveals that negative emotional troughs may have a relatively greater impact on tourists’ loyalty compared to positive peaks, suggesting a need to reconsider the traditional focus on enhancing positive experiences in tourism management.
Thus, a negative episode during a vacation could stain the entirety of travel recollections. Previous studies have emphasized positive emotional experiences (Ma et al., 2013), but this study highlights the need to address extreme negative episodes, such as accidents and unsanitary conditions (Kemp et al., 2008; Q. B. Zhou et al., 2018). This research provides new insights into managing negative emotions and dissatisfaction during the tourist experience. Building on this, this study underscores the significance of real-time monitoring of tourist experiences, highlighting the need to address negative episodes as they occur, a practical consideration that has not been fully explored in the existing peak–end rule tourism literature. By proactively addressing negative emotional troughs, this study demonstrates the potential of real-time management measures to not only mitigate immediate dissatisfaction but also prevent long-term harm to tourist loyalty. This approach suggests a potential shift from reactive to preventive strategies, providing a framework that could contribute to destination management practices.
Specifically, the findings suggest that momentary monitoring and response systems could be implemented to address extreme negative emotion or dissatisfaction troughs before they significantly impact overall loyalty. For instance, integrating customer feedback platforms that notify managers of negative incidents in real-time could help mitigate the negative effects of such troughs. Additionally, the importance of emotional troughs identified in this study highlights the need for staff to be trained to recognize and subsequently provide an opportunity for customers to experience a peak pleasure event or, alternatively, offer a positive event at the end to de-escalate negative emotions and minimize their impact on customers’ long-term behavioral intentions through service recovery strategies. The most dissatisfying incident or the moment when tourists experience the most negative emotion could significantly influence future behavioral intentions. From a managerial perspective, these aspects could be worth prioritizing.
Furthermore, when tourists encounter extremely negative incidents during their journey, providing uplifting experiences at the trip’s end could positively influence their retrospective evaluations. For instance, Just et al. (2015) found that offering gastronomy customers a slice of pizza significantly influenced their overall evaluation, with peak and final pizza ratings having a stronger impact on dining satisfaction. Similarly, Park et al. (2018) demonstrated that tourists ending their Seoul city tour with a boat ride reported higher satisfaction levels compared to those returning by bus. These findings suggest that integrating satisfying end-of-trip activities could enhance overall tourist satisfaction, as also supported by Redelmeier and Kahneman’s (1996) insights on the importance of favorable endings.
This study was based on the peak–end rule and used smartphones to examine the influences of momentary experiences, such as positive and negative emotions, emotional arousal levels, and satisfaction levels, on retrospective evaluations. This could provide insights for destination management entities and travel agencies. Managers should emphasize key moments of the tourist journey, such as peaks, troughs, and ends, and allocate management and marketing resources accordingly. This is particularly true for situations that create extreme negative emotions or intense dissatisfaction.
Limitations and Future Research
The limitations of this study present several opportunities for further investigation. First, the ESM could affect ongoing tourist activities. To mitigate these issues in future studies, it might be beneficial to decrease the frequency of surveys. Second, the travel duration for participants in this study was restricted to 1 day. Future research could extend this timeframe to span multiple days, enabling further exploration using the ESM approach. The process of recruiting participants on-site faced challenges, as the ESM survey process is relatively complex, resulting in lower participation willingness among older tourists. Therefore, the actual participants in this study were skewed toward the 20–29 age group, which may not fully represent the actual distribution of tourists. Future research could simplify the research design to increase the proportion of samples from other age groups. Additionally, researchers could consider collaborating with institutions specializing in senior tourism programs, such as travel agencies, and even accompanying tour groups to assist older participants in recording their experiences using ESM. C. Li et al. (2023) highlighted that revisit intentions naturally decline over time due to fading sensory impressions, suggesting that the timing of post-trip surveys could influence the reliability of retrospective evaluations. In this study, post-trip evaluations were conducted immediately after the journey to minimize memory decay, but future research could explore how varying the timing of surveys impacts data consistency. This study was conducted in a single location, which may limit the generalizability of the findings. However, ESM focuses on capturing momentary data throughout the survey duration, allowing researchers to track temporal causal relationships (Tang & Jhang, 2020). The use of ESM mitigates this concern regarding the research location to some extent. For example, Lenton et al. (2016) and Tang and Jhang (2020) utilized ESM to have participants record their experiences in their everyday lives, with each participant living in their own environment. Nevertheless, Mitas et al. (2022) and S. Li (2021) used ESM to record experiences at a single location. Since this research centers on the dynamics of tourists’ momentary experiences and their causal relationship with post-trip loyalty, the specific location may play a less critical role in influencing the results. This study examined the peak–end effects of variables such as valence, arousal, and satisfaction, and their impact on loyalty. Future studies could integrate additional independent variables to provide deeper insights into their influence on tourists’ prospective behavioral intentions. The peak-end rule may also inspire research in other areas, including leisure, dining, and hospitality. Future studies could investigate whether optimizing the peak experience in leisure activities, such as a memorable campfire session during a group camping trip; enhancing the quality of peak moments and final impressions in dining, such as the main course and dessert; and alleviating trough experiences in hospitality through real-time monitoring of hotel stays to identify issues such as noise disturbances, maintenance problems, or the impact of specific instances of service failures during the stay, can influence loyalty across these fields.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors would like to thank the Ministry of Science and Technology of Taiwan, ROC for the financial support (MOST 110-2410-H-424-006 -SSS).
