Predicting Private and Public Pro-Environmental Behaviors in Rural Tourism Contexts Using SEM and fsQCA: The Role of Destination Image and Relationship Quality

Abstract

The importance of pro-environmental behavior in tourism has been established, but explaining its sub-dimensions, especially in the private and public dimensions, is under-researched. Existing literature on tourism research mainly uses SEM to analyze tourist pro-environmental behavior, while fsQCA is scarcely implemented. In this study, SEM is applied to reveal the links among destination image, relationship quality, and pro-environmental behavior, while fsQCA is utilized to investigate configurations predicting pro-environmental behavior. Responses of 285 tourists were collected and analyzed to test the proposed hypotheses. The SEM results showed that (1) destination image directly and positively affected relationship quality (including satisfaction and destination trust); (2) relationship quality was found to positively and directly influence private and public pro-environmental behaviors; (3) relationship quality did mediate the influence of destination image on private pro-environmental behavior partially, while it played a full mediating role in the effect of destination image on public pro-environmental behavior. The findings from fsQCA indicated that (1) three sufficient configurations consistently lead to a high level of private pro-environmental behavior: (a) high destination image and satisfaction, (b) high destination image and trust, (c) high relationship quality; (2) there was only one sufficient causal configuration for a high level of public pro-environmental behavior: high relationship quality. The results provide tenable evidence that relationship quality can be a vital factor enhancing the sub-dimensions of pro-environmental behavior. The integration of these two methods helps to open the black box of tourist pro-environmental behavior in rural tourism contexts in a more systematic and holistic way.

1. Introduction

As the foundation of the survival and development of mankind [1], land is becoming increasingly scarce as a strategic resource. People’s lives and work will face enormous challenges without the sustainable use of land resources, which is not only related to the well-being of land stakeholders, but also affects the ecological and economic security of a region, and even a country. In this sense, the importance of the sustainable use of land has been widely recognized in land research [2].

Rural tourism is a type of tourist activity in the natural rural land and is closely related to the tourism adaptability of land [3]. The conservation and utilization of the key attractiveness of rural tourism, i.e., the life community of mountains, rivers, forests, and lakes, etc., is of great significance to the ecological transformation and sustainable development of rural land [4]. As prior scholars note, rural residents’ non-agricultural activities, including developing rural tourism, significantly reduce their dependence on forest land, and contribute to the ecological restoration of rural forest land and positive changes in the rural ecological environment [5,6]. From the perspective of rural tourism evolution, rural tourism has developed rapidly worldwide and been advocated and promoted by a large number of international organizations. It has become an irreplaceable element of global tourism [7]. China is no exception. Over the past four decades, the country’s rural tourism has experienced remarkable achievements in accelerating rural economic growth, optimizing the rural industrial structure, and improving the income of rural residents [8]. Tourism is an indispensable tool for rural revitalization; thus, rural tourism has developed rapidly in China [9] and become the most dynamic consumption segment and a growth point for the country’s tourism sector [10]. The COVID-19 pandemic has inflicted immeasurable damage on the global tourism industry [11], including the Chinese tourism market. The first wave of the pandemic exerted a severe impact on the economy, society, production, and life. The tourism industry has not been spared. With the implementation of China’s “dynamic clearing” policy, different industries have gradually recovered from the pandemic, including the tourism sector [12]. After the pandemic stabilized in China, people’s attention to tourism steadily increased, while the growth of China’s outbound tourism has been stagnating [13]. Rural tourism, as a pivotal segment of the tourism industry [7], has been preferred by tourists since it provides unique ecological resources and lifestyle during the post-pandemic period [14]. Tourists prefer to visit rural destinations rather than urban or outbound destinations due to the pandemic and consequent prevention and control measures. For instance, rural destinations were one of tourists’ favorites during the Labor Day Holiday of 2021 [10]. Thus, the rapid expansion of rural tourist destinations offers a new industrial choice for the rapid and sustainable development of rural land. In addition, developing rural tourism is one of the best options for rural land transformation and upgrade, as well as the reform of rural land use [15,16].

Nevertheless, the rapid rise of rural tourism is a double-edged sword, which brings new challenges to the sustainable use of rural land. The unsustainable pattern of land use is characterized by the destruction of the rural ecological environment caused by the increase in visitation (e.g., overcrowding, pollution, waste disposal, and vegetation deterioration) and unreasonable development of the rural landscape [17,18], such as the case of the Longji Terraces in Zhang et al.’s research [19]. A precondition for the sustainable development of rural tourism is to provide high-quality ecological resources and environments. Pristine ecological environments significantly enhance the appeal and competitiveness of tourist destinations. Pro-environmental behaviors support the belief that clear and clean waters and lush mountains are invaluable assets [20] and help to mitigate the negative effects of tourism activities [21]. The ideal state of rural tourism development is to pursue the sustainability of rural land use and make it a model of the sustainable use of rural land [22]. Consequently, to maintain the sustainability of rural tourism and rural land use, it is essential to focus on reducing the negative effects on rural destinations and improving tourist pro-environmental behaviors.

The sustainable development of destinations is inseparable from maintaining healthy natural ecological environments, while the pro-environmental behavior of users is also pivotal to these efforts. Thus, pro-environmental behavior has been widely researched [23]. Current studies focus on World Natural and Cultural Heritage, nature reserves, cultural tourism, community tourism, urban tourism, island tourism, ecological tourism, national parks, national wetland parks, urban parks, and resort areas [24,25,26,27,28]. The research on rural destinations is more scarce, justifying more in-depth research.

Rural tourism is associated with beautiful natural landscapes, folk traditions, and customs in rural areas, which support positive destination images [29]. Positive destination images are crucial and intangible assets for the destination marketing and management of rural tourism areas. They support attractive and recognizable destination brands by presenting favorable rural land images, which enhance the long-term competitiveness and sustainability of rural tourism in an increasingly competitive marketplace [30]. Previous scholarly efforts have revealed that destination image exerts a significant and positive impact on revisit intention and word of mouth [31], but also promotes pro-environmental behavior. When researching an ecological area in Southern Taiwan, Chiu et al. argued that destination image significantly influenced pro-environmental behavior [32]. The focus of prior research on destination image and pro-environmental behavior was cultural heritage and ecological tourism [31,32]. However, such studies in rural tourism are scant. Previous research also reveals that the image of rural tourism destinations has significant effects on sustainable tourism development, local employment, and the harmonious development of the local ecology, economy, society and culture [29], but the impact of destination image on pro-environmental behavior remains obscure, especially regarding its sub-dimensions (such as private and public pro-environmental behavior). Since there are considerable differences between these two sub-dimensions [33], it is worthwhile to explore how destination image affects them.

Scholars suggest that relationship marketing includes all marketing efforts aimed at building, developing, and maintaining successful relationships [34]. This crystallizes the value of building a continuous bond with customers and other stakeholders [35]. Relationship marketing has been widely employed in investigating consumer behavior and posits that brand image influences consumer loyalty, purchase intentions, and word of mouth [34,36,37]. It is an important paradigm in marketing and its core lies in relationship quality [38]. Along similar lines, relationship quality in tourism affects user behavior, including variables such as loyalty, satisfaction, and trust [30,39] and pro-environmental behavior [8,40,41]. However, few studies have examined the effect of relationship quality on the sub-dimensions of pro-environmental behavior in the rural tourism context.

Liu et al. (2011) define relationship quality as an emotional state derived from interactive experience evaluation, including trust and satisfaction [42]. Within the framework of appraisal theory, emotion is viewed as an individual’s adaptive response to the external environment [43]. Cognitive emotional appraisal represents a person’s emotional responses, and the cognitive evaluation of environments determines emotional responses [44]. By applying this theory, it has been verified that the positive effects of overall cognitive evaluations of destinations on positive emotions are significant [45]. Thus, destination image is a critical source of relationship quality, and this has been confirmed by the extant research [46]. By underpinning the appraisal theory of emotion and the relationship marketing paradigm, the present study investigated the links among destination image, relationship quality, and pro-environmental behavior in rural tourism through SEM to reveal how destination image affects the sub-dimensions of pro-environmental behavior.

SEM is employed in quantitative analysis to examine the linear relationships between variables, but it cannot reveal the configuration effects between conditional variables [47]. Traditional relevant theoretical frameworks only emphasize the simple and symmetric relationships between certain antecedents and outcomes, not complex multi-factorial and concurrent causality [48]. However, systematic clarification of the causal logic is necessary to explain complicated social phenomena that reflect the complex aggregation relationships among several concurrent conditions and outcomes. Moreover, the causal mechanism of an individual condition and its outcomes change under various conditions [49]. The QCA method has been applied to examine whether an individual condition or condition configuration meets the requirement and sufficiency for producing the outcome. The causality deduced from the aggregation relationships is concurrent and asymmetric [50]. As a new research paradigm, the QCA method is employed to analyze the causality of concurrent conditions and has been widely applied in management studies [51]. There may be multiple implementation paths with equivalent results for specific strategies of pro-environmental behavior; however, traditional quantitative studies do not explain the interdependent complex causality of multiple antecedents [49]. Recently, some scholars have adopted the combination of SEM and fsQCA in tourism management and presented positive results [52,53]. Accordingly, the current study introduced both SEM and fsQCA to examine factors influencing private and public pro-environmental behavior. This is expected to offer new insights for the sustainability of rural tourism.

Based on the aforementioned knowledge gap, the present study attempted to fill the void with three specific objectives: (1) to employ the appraisal theory of emotion and relationship marketing to explain the links among destination image, relationship quality, and pro-environmental behavior; (2) to uncover the mediating roles of relationship quality (including satisfaction and destination trust); and (3) to explore the causal configurations that result in private and public pro-environmental behaviors through the application of fsQCA. Theoretically, this study may extend the pro-environmental behavior literature by revealing the effects of destination image and relationship quality on sub-dimensions of tourist pro-environmental behavior within the framework of “destination image–relationship quality–pro-environmental behavior”, as well as their differences in the effects on pro-environmental behavior. From the aspect of methodology, the combination of SEM and fsQCA allows for a more comprehensive approach to understanding the mechanism forming the tourist pro-environmental behavior in rural tourism contexts.

2. Literature Review and Hypothesis Development

2.1. Theoretical Background

2.1.1. Relationship Marketing Paradigm

Relationship marketing was first conceptualized by Berry (1983) as a tool to identify, establish, and enhance customer relationships for satisfying the needs of companies and relevant stakeholders [54]. It has a focus on long-term trustful and mutually beneficial relationships with valuable consumers. Previous studies found that the more marketing resources allocated to maintaining these customers, the more effective marketing became [55], including increasing brand loyalty and retaining existing customers [54]. Practitioners thus gradually realized the importance of maintaining the relationships with customers and the significance of relationship marketing [56].

The essence of relationship marketing is to measure relationship quality through company–customer relationships [54]. Relationship quality is an emotional condition drawn from interactive experience assessment, encompassing trust and satisfaction [42]. Fostering customer satisfaction is viewed as a crucial constituent of developing quality relationships [57,58]. Satisfaction is a driver of employee organizational citizenship behavior [59,60]. In the tourism context, satisfaction refers to visitors’ affective reactions to experienced behaviors during the visit [61]. Satisfaction has a profound impact on individuals’ behavioral-making processes such as loyalty [39], product consumption [62,63], and pro-environmental behavior [40,64,65].

Trust is a willingness to rely on one’s own trusted exchange partners [66]. Consumer trust includes believing that the trusted genuinely cares about the principal and the trusted is capable of fulfilling obligations in the relationship [67,68]. Thus, trust is an integral element of building relationships [34,67].

Increasing the trust in destinations is considered an important means of building assets, as it fosters the bond between customers and destinations and improves the quality of relationships [69]. Trust is the belief that the destination is reliable and will deliver on its promises, while trust is measured by the satisfaction with the services provided by the destination [58]. Satisfaction with service quality induces trust, and trust predicts positive word of mouth and revisit intentions [70] and loyalty [30].

Local environment conservation is inseparable from tourist behaviors in developing rural tourism destinations. Consequently, the current research places the relationship marketing paradigm into the context of rural tourism destination development to investigate the links between tourists in rural destinations and their pro-environmental behaviors. Considering that trust and satisfaction are two essential components of relationship quality, this study explored how satisfaction and trust affected pro-environmental behavior by introducing the relationship marketing paradigm. The employment of this paradigm represents a conducive attempt at understanding tourist pro-environmental behavior in rural tourism contexts.

2.1.2. Appraisal Theory of Emotion

The appraisal theory of emotion posits that emotion is people’s adaptive reaction to external environmental factors [43]. Therefore, the cognitive evaluation of emotion means people’s emotional reactions, i.e., the cognitive evaluation of the environment determines people’s emotional reactions [44]. Due to factors such as beliefs, attitudes, or personality, different individuals may trigger dissimilar cognitive evaluations under the same environmental stimuli, resulting in varied emotions [71]. This theory provides a basis for explaining the reasons why different people have divergent emotional responses to the same environment, and also explains emotional processing in marketing. As such, it is considered one of the most prevalent theories attempting to examine the antecedents and consequences of consumer sentiment [72,73]. Empirical research in various tourism settings has shown that overall cognitive evaluations of destinations have a significant beneficial impact on positive emotions, and the appraisal theory of emotions is effective in explaining behavior and has become an important conceptual framework for explaining emotional experiences [45,74].

The findings of empirical studies in multiple tourism settings have indicated that the appraisal theory of emotions is applicable in explaining the links between cognition and emotions. As prior studies have confirmed that destination image influences relationship quality [75,76], it is reasonable to examine how destination image affects relationship quality within the framework of the appraisal theory of emotions, providing an important theoretical perspective in the field of pro-environmental behavior research in rural tourism.

2.2. Hypothesis Development

2.2.1. Link between Destination Image and Relationship Quality

Destination image has long been a popular subject in tourism research [77,78]. A large number of studies have examined the role of destination image in individuals’ behavior, in which multi-faceted influences have been found, such as subjective mental states, consumption behaviors, and choices of destinations [79,80,81]. These findings all demonstrate the importance of image for destination marketing. In today’s dynamic and competitive tourism markets, creating and maintaining a positive destination image significantly influences how a destination creates marketing tactics [82]. Image is also a powerful management tool for remaining competitive in the tourism market [83].

Crompton (1979) proposed that image encompasses beliefs, impressions, thoughts, and perceptions that people have toward objects, behaviors, and events [84]. Some researchers [77,85,86] note that destination image is a three-dimension construct including cognitive image, affective image, and conative image, while some argue that there is a fourth component, i.e., the overall image [87]. The tourism literature also identifies four features of destination image, namely being complex, multiple, relativistic, and dynamic [88]. A meta-analysis of 66 studies indicated that destination is a multi-dimensional concept [89]. Though the focus of destination image research varies, researchers generally agree with the concept of overall image [84]. Moreover, in terms of destination image measurement, early researchers proposed that destination image can be measured as an overall construct [82], which has been supported by other researchers [90]. Consequently, the use of overall image has been found to be viable. The meta-analytic work of Zhang et al. pointed out that overall image serves as a good substitute for other dimensions (such as affective, cognitive, or conative dimensions) to measure destination image. It has strong explanatory power for destination image perceptions [91]. A recent meta-analysis reviewing 63 studies on overall image also characterized it as a synthetic and formative construct [83]. Based on the above discussion, the researchers followed Josiassen, Assaf, Woo, and Kock (2016) in viewing overall image as an interchangeable construct of destination image [92], and defined it as tourists’ overall perceptions and evaluations of the tourist destination. Findings of prior research indicate that the research on destination image mostly uses subjective attitudes, such as positive emotions, satisfaction, as well as behavioral intentions and choices. On the one hand, destination image affects positive emotions toward destinations [45] and is significantly and positively related with satisfaction [93]. On the other hand, destination image directly or indirectly drives behavioral intentions and future destination choices [94].

Satisfaction is an expression of the emotional or affective responses to a tourism product or service [95]. It articulates comfort, pleasant feelings, and acceptance of consuming the product or service [96]. The findings of some studies reveal a direct effect of destination image on satisfaction [93]. Wang and Hsu (2010) argue that “overall tourism destination image has an indirect impact on behavioral intentions through satisfaction” [97]. When destination image is more positive, the level of satisfaction will be higher, and a destination will attract more tourists. In contrast, the lack of a favorable destination image influences satisfaction in the opposite direction [98]. Lam concluded that satisfaction is influenced by destination image when studying online reviews on social media platforms [99]. On the other side, the influence of destination image on trust is not to be ignored. People’s confidence and belief in tourism products or service providers are generally defined as destination trust [100]. Song et al. (2019) demonstrated that positive brand images prompt consumers to show brand loyalty and trust [101]. Similarly, a positive destination image stimulates and enhances destination trust [75]. In studies of Spain–Portugal border areas [102] and a World Cultural Heritage Site [76], researchers found that destination image acts as a direct antecedent of trust.

The positive effects of destination image on relationship quality, especially in its two sub-dimensions (satisfaction and trust), have been verified over the years. For example, when studying international tourists visiting the Angkor temple complex in Cambodia, Chen (2013) found that destination image significantly and positively affected satisfaction and trust [75].

Based on the preceding discussion, the existing literature has emphasized the positive impact of destination image on relationship quality variables (satisfaction and trust). Since destination image can be recognized as a cognition, and relationship quality is an emotional state, the current investigation, based on the appraisal theory of emotions, sought to answer the question of how destination image affects relationship quality in rural tourism. Thus, it was hypothesized that:

Hypothesis 1 (H1).

Destination image directly and positively affects satisfaction.

Hypothesis 2 (H2).

Destination image directly and positively affects destination trust.

2.2.2. Link between Tourist Satisfaction and Destination Trust

Satisfaction and trust are two critical relationship quality variables. The association between them has aroused great interest in academia. In consumer behavior, if consumers express satisfaction with a brand, they will have more trust in the brand [103]. Overall satisfaction has positive effects on trust [104], which has been confirmed by findings in studies on e-services [105].

Research in tourism has also demonstrated the positive influence of satisfaction on trust. Osman and Sentosa (2013) found that satisfaction influences trust in a positive way, providing new evidence regarding the positive association between customer satisfaction and trust [106]. Various case studies have confirmed this conclusion. For example, foreign visitors’ satisfaction with the World Heritage Site at Angkor Wat was found to be related to trust [75]. Similarly, when tourists at a seaside resort showed more satisfaction with the destination, their trust also increased [107]. Similar results were found with rural tourist destinations [69].

Therefore, the third hypothesis was:

Hypothesis 3 (H3).

Satisfactiondirectlyandpositivelyaffectsdestination trust.

2.2.3. Link between Relationship Quality and Tourist Pro-Environmental Behavior

Pro-environmental behavior is closely related with the sustainable development of destinations [22]. The factors influencing pro-environmental behavior are a popular topic in tourism research [108,109,110]. Existing studies have a focus on the effects of emotion on behavior, rather than on cognitive elements [41,111]. As an emotional state generated from interactive experiences, the importance of relationship quality is recognized by studies in the shipping business, retailing, and catering. These academic efforts confirm that relationship quality affects consumer behavior [62,63]. Tourism studies demonstrate similar results. For instance, in the hospitality industry, relationship quality significantly and positively influences customer loyalty [39].

Similarly, relationship quality significantly affects pro-environmental behavior, which has been highlighted by studies from the perspective of tourists [40,41,111]. In addition, satisfaction, as an integral part of relationship quality, is instrumental to pro-environmental behavior. When tourists feel satisfied with the interactive experiences at a destination, they are more inclined to engage in pro-environmental behavior. Prior studies reveal that satisfaction enhances environmentally responsible behavior [37,112]. In other contexts, such as environment-friendly behavior involving plastic bag use and island tourism, satisfaction has been found to promote environmentally responsible behavior [41,64,65]. Overall, the results of these empirical tests confirm the role of satisfaction in predicting pro-environmental behavior in tourism [30,70].

In the same vein, the role of trust in pro-environmental behavior cannot be neglected. Trust has demonstrated its major impact on behavior in various settings. For example, research has revealed that employee well-being is directly affected by organizational trust and identification, while employee well-being improves environmentally friendly behavior [113]. A study of Muslim tourists traveling abroad concluded that satisfaction and trust are key indicators for tourist behavioral intentions [95].

Based on these previous academic efforts, satisfaction and destination trust can be viewed as the driving factors of pro-environmental behavior. However, pro-environmental behavior is usually examined as a single-dimensional construct, and research on the sub-dimensions of this behavior is at an underdeveloped stage for rural tourism destinations. Recently, researchers have begun to view pro-environmental behavior as a multi-faceted concept [114]. Generally, it is divided into private pro-environmental behavior and public pro-environmental behavior [115]. From the private behavior perspective, some research conceptualizes pro-environmental behavior as types of behavior that involve material conservation or energy saving [116]. Some define it as behavior that “harms the environment as little as possible, or even benefits the environment” [117]. These definitions all agree that the aim of behavior is to conserve or reduce damage to the environment [118]. Based on the extant literature and the rural tourism setting, private pro-environmental behavior in this research is defined as discretionary behavior that impacts environmental quality directly by lifestyle changes [115], such as conserving electricity and water, no littering, and protecting wildlife [26,119]. For pro-environmental behavior in the public dimension, Stern (2000) defines this as behavior that exerts an influence on the environment indirectly, such as being involved in pro-environment campaigns, contributing to environmental organizations, and supporting environmental regulations [114]. Considering the context of rural tourism destinations, public pro-environmental behavior is behavior that positively influences the environment indirectly through working as a volunteer to help the destination environment, donating money to support destination environmental protection, joining in the destination cleanup efforts to protect the environment, and writing letters, online messages, or emails in support of destination conservation.

The level of effort for pro-environmental behavior varies with the behavioral types. Compared with private pro-environmental behavior, it takes more energy and effort to perform public pro-environmental behavior. Various dimensions of relationship quality potentially may have dissimilar effects on the sub-dimensions of pro-environmental behavior. However, extant studies mainly examine the impact of relationship quality on tourist pro-environmental behavior as a single construct [58]. The current study categorizes tourist pro-environmental behavior into private and public pro-environmental behavior for a better understanding of how relationship quality affects the two sub-dimensions of pro-environmental behavior differently in the rural tourism context. To this end, this research proposed the following hypotheses:

Hypothesis 4 (H4).

Satisfaction directly and positively affects private pro-environmental behavior.

Hypothesis 5 (H5).

Destination trust directly and positively affects private pro-environmental behavior.

Hypothesis 6 (H6).

Satisfaction directly and positively affects public pro-environmental behavior.

Hypothesis 7 (H7).

Destination trust directly and positively affects public pro-environmental behavior.

2.2.4. Link between Destination Image and Tourist Pro-Environmental Behavior

For the link between destination image and pro-environmental behavior, prior studies verify that destination image is a driver of pro-environmental behavior in different settings, such as ecological areas, green hotels, and restaurants [32,120]. However, there is a lack of such research for rural tourism. Moreover, previous studies treated pro-environmental behavior as a single variable, which is not sufficient for explaining the specific effects of destination image on the sub-dimensions of pro-environmental behavior [32]. According to the aforementioned, the following hypotheses were put forward to examine the effects of destination image on the sub-dimensions of pro-environmental behavior in rural tourism:

Hypothesis 8 (H8).

Destination image directly and positively affects private pro-environmental behavior.

Hypothesis 9 (H9).

Destination image directly and positively affects public pro-environmental behavior.

2.3. Conceptual Model

Based on the literature review and hypothetical propositions, this study presents the conceptual model as shown in Figure 1.

3. Method

3.1. Measurement

Scale items were rigorously validated before being modified and employed to evaluate the constructs for the present research. To measure destination image, four items (e.g., I have a good impression of this rural destination) were adapted from Nguyen and Leblanc (2001) [121]. To measure satisfaction, three items (e.g., Overall, I am satisfied with my visit to this rural destination) were adapted from He (2011) [122]. To measure destination trust, four items (e.g., I trust this rural destination) were adapted from Wu et al. (2018) and Han et al. (2018) [123,124]. To measure private pro-environmental behavior, four items (e.g., I conserve water at this rural destination) were adapted from Tonge et al. (2015) [27]. To measure public pro-environmental behavior, four items (e.g., I work as a volunteer to help the environment of this rural destination) were adapted from Lee et al. (2013) [125].

Table 1. Detailed measurements.

Construct	Item	Item Label	Source
Destination image	I have a good impression of this rural destination.	DI1	Nguyen and Leblanc (2001) [121]
	In my opinion, this rural destination has a good image in the minds of tourists.	DI2
	I believe that this rural destination has a better image than its competitors.	DI3
Satisfaction	Overall, I am satisfied with my visit to this rural destination.	TS1	He (2011) [122]
	Compared to my needs, I am satisfied with my visit to this rural destination.	TS2
	Compared to my expectations, I am satisfied with my visit to this rural destination.	TS3
Destination trust	This rural destination takes care of my needs as a tourist.	DT1	Wu et al. (2018); Han et al. (2018) [123,124]
	I trust this rural destination.	DT2
	I have confidence in this rural destination.	DT3
	This rural destination is reliable.	DT4
Private pro-environmental behavior	I conserved electricity at this rural destination (e.g., I switched off lights and electronic equipment if I was not using them.)	PRPEB1	Tonge et al. (2015) [27]
	I conserved water at this rural destination (e.g., I turned off the tap if I am not using it).	PRPEB2
	I did not litter at this rural destination.	PRPEB3
	I took care of animals and plants at this rural destination.	PRPEB4
Public pro-environmental behavior	I work as a volunteer to help the environment of this rural destination.	PUPEB1	Lee et al. (2013) [125]
	I donated money to support the environment protection of this rural tourist destination.	PUPEB2
	I joined in this rural destination’s cleanup efforts to protect the environment.	PUPEB3
	I wrote letters, online messages or emails in support of the conservation of this rural destination.	PUPEB4

presents the measurements in detail. In this work, 5-point Likert scales anchored from 5 (“strongly agree”) to 1 (“strongly disagree”) were employed.

3.2. Statistical Analysis Method

This research applied a multi-method approach for statistical analysis. As a variable-oriented method, SEM analyzes the linear relationship between variables with a focus on the net effect; contrarily, fsQCA assumes that the relationship between variables is asymmetric, providing a better understanding of the non-linear effect [126]. The fsQCA is a supplement to SEM by offering new insights into the configurations of various antecedents for an outcome [51]. Extant studies mainly used SEM to analyze correlations between variables. To this end, this study employed SEM to analyze the linearity between destination image, relationship quality (including satisfaction and trust), and pro-environmental behavior, while fsQCA was used to examine the combinational factors predicting pro-environmental behavior.

3.3. Pretest of the Measurements

All the items of the measurements were translated into Chinese for the field survey and translated back to English later. In order to review and validate the content, six experts (including three destination practitioners and three tourism scholars) were invited to conduct a pretest before the invitation of 50 qualified Chinese tourists for the sample survey. As per Zheng et al. (2022)’s suggestion, the receivable reliability and validity of the pretest study was identified, respectively, through the computation of Cronbach’s Alpha and standard factor loading [61].

3.4. Data Collection and Sample

The snowball sampling methodology was chosen for the following two reasons. First, this methodology has been widely employed in tourism research [127,128], as well as with tourist behavior [129,130]. Second, due to occasional COVID-19 outbreaks across the country and implementations to prevent and control the pandemic, on-site data collection at tourist destinations became much more inconvenient. The application of snowball sampling can also be found in recent research [131]. As for the procedure of the snowball sampling technique, specifically, the questionnaire was delivered by the researcher to the invited informants in the researcher’s social network, who referred other informants to the researcher. The process continued by repeated these steps [132]. Referring to Qiu (2017)’s description of the snowball sampling methodology [133], respondents were selected according to the following standards: (1) they had to have visited the rural destination in the last month; and (2) they had to have a clear memory of the recent visit to this rural destination. Research assistants received formal training about the snowball sampling method to ensure the success of the survey. Initially, members of the research team found eight qualified participants in their social network (e.g., relatives, friends, and colleagues) to finish the questionnaire. Then, each participant invited 8 more participants at most for the second round. The same procedure was repeated for a third and final time. In total, 330 questionnaires conforming to the aforementioned standards were obtained as basic data, of which 285 were valid, resulting in an 86.36% response rate. The sample comprised 44.2% males, 55.8% females. Some 35.8% of the participants were under the age of 25, 30.5% between 25 and 34 years, while 33.7% were 35 years and older. In terms of educational level, 14.3% of the respondents were junior middle school graduates, and 25.8% graduates of technical secondary school, high school, or vocational high school. The majority (59.9%) had Bachelors’ degrees or above. As for rural destinations that participants visited, three types were included for the data collection, i.e., rural destinations within the city, not in the city but within the province, and outside the province, accounting for 66.3%, 23.2%, and 10.5%, respectively. Among them, rural destinations in Hangzhou City (Yuhang District and Tonglu County) and around the city (Deqing County and Anji County) are typical cases. There were 14 items in the questionnaire, corresponding to 285 valid responses. The sample size was over ten times the items, as Nunnally (1967) suggested, which met the requirement of the effective sample to explore variables in the model [134].

4. Data Analysis

4.1. Testing Common Method Variance

Multiple approaches to evaluate the problem of common method variance (CMV) were implemented because cross-sectional data were adopted in this current research [135]. The exploratory factor analysis results indicated that a multi-factor pattern explained 77.75% of the total variance, with the first factor accounting for 48.9% of the variance. It meets the requirement of the threshold of 50%. Additionally, the results implied that the common factor model was less suitable than the proposed measurement model (Δχ² = 1121.759, Δdf = 10, p < 0.001), which avoided the problem of CMV in this research [136].

4.2. Measurement Model Test

According to the recommendation of Anderson and Gerbing (1988), the two-step modeling method was applied in the current research [137]. The measurement model was estimated via confirmatory factor analysis (CFA). SEM was executed employing AMOS to assess the hypotheses subsequently. CFA was performed to test the reliability of the measurement model and its validity, resulting in an acceptable model fit (RMR = 0.028, χ²/df = 2.243, TLI = 0.948, RMSEA = 0.066, CFI = 0.957, SRMR = 0.0523) [112].

The composite reliability (CR) was varied from 0.822 to 0.925 for each construct (

Table 2. Results of measurement model.

Key Construct	Loading	t-Values	Composite Reliability	Average Variance Extracted
Destination image			0.822	0.607
DI1	0.821	11.938
DI2	0.803	11.791
DI3	0.708	—
Satisfaction			0.861	0.674
TS1	0.762	14.095
TS2	0.861	16.2
TS3	0.837	—
Destination trust			0.883	0.655
DT1	0.761	13.853
DT2	0.832	15.509
DT3	0.836	15.604
DT4	0.805	—
Private pro-environmental behavior			0.906	0.709
PRPEB1	0.91	14.621
PRPEB2	0.915	14.689
PRPEB3	0.818	13.228
PRPEB4	0.708	—
Public pro-environmental behavior			0.925	0.754
PUPEB1	0.831	18.34
PUPEB2	0.882	20.461
PUPEB3	0.883	20.521
PUPEB4	0.876	—

), surpassing the cut-off point of 0.70 [138]. The standardized factor loading’s value of each indicator was between 0.708 and 0.915, which indicated significance (p < 0.001). The average variance extracted (AVE) values, ranging from 0.607 to 0.754, were beyond the threshold of 0.50. This showed that acceptable convergent validity was identified [139]. For each construct, the corresponding square roots of the AVEs were compared with the correlation coefficient among pairs of latent variables to estimate the discriminant validity [140], which were supported by the results (

Table 3. Discriminant validity assessment.

Construct	DI	TS	DT	PRPEB	PUPEB
Destination image (DI)	[0.779]
Tourist satisfaction (TS)	0.622	[0.821]
Destination trust (DT)	0.627	0.684	[0.809]
Private pro-environmental behavior (PRPEB)	0.619	0.615	0.621	[0.842]
Public pro-environmental behavior (PUPEB)	0.414	0.541	0.636	0.520	[0.868]

). Overall, the reliability and validity were both established [141].

4.3. Testing Structural Model

SEM was adopted in the direct hypotheses test. Findings showed that the structural model had a receivable fit (χ²/df = 2.273, TLI = 0.946, RMR = 0.030, CFI = 0.956, RMSEA = 0.067, SRMR = 0.0564). The findings from

Table 4. Results of structural model.

Hypotheses	Path	Rural Destination Context
		Standardized Coefficient	t-Value	Results
H1	Destination image → Satisfaction	0.622 ***	8.376	Supported
H2	Destination image → Destination trust	0.33 ***	4.263	Supported
H3	Satisfaction → Destination trust	0.479 ***	6.216	Supported
H4	Satisfaction → Private pro-environmental behavior	0.248 **	3.012	Supported
H5	Destination trust → Private pro-environmental behavior	0.279 ***	3.389	Supported
H6	Satisfaction → Public pro-environmental behavior	0.215 *	2.488	Supported
H7	Destination trust → Public pro-environmental behavior	0.517 ***	5.756	Supported
H8	Destination image → Private pro-environmental behavior	0.289 ***	3.622	Supported
H9	Destination image → Public pro-environmental behavior	−0.034	−0.424	Not supported

Note: *** p < 0.001, ** p < 0.01, * p < 0.05.

demonstrate the eight direct relationships except for H9. Destination image had a direct and significant influence on trust (β = 0.33) and satisfaction (β = 0.622), supporting H1 and H2 accordingly.

Satisfaction produced a direct and significant role on trust (β = 0.479), private pro-environmental behavior (β = 0.248), and public pro-environmental behavior (β = 0.215), supporting H3, H4, and H6. Trust had a direct and significant effect on private pro-environmental behavior (β = 0.279) and public pro-environmental behavior (β = 0.517). H5 and H7 were both established. In addition, the direct and significant influence of destination image on private pro-environmental behavior was identified (β = 0.289), while its corresponding effect on public pro-environmental behavior was not supported (β = -0.034, p > 0.05). Accordingly, H8 was supported, but H9 was not supported.

The bootstrapping method in AMOS was conducted to test the mediating effects. The number of bootstrap samples was set to 5,000, using bias-corrected confidence intervals of 95% [61]. Findings from

Table 5. Specific mediating test.

Mediating Hypothesized Path	Indirect Effects	Lower	Upper	p-Value	Results
Destination image → Satisfaction → Private pro-environmental behavior	0.139	0.056	0.238	0.002	Supported
Destination image → Satisfaction →Destination trust → Private pro-environmental behavior	0.075	0.032	0.152	0.000	Supported
Destination image → Destination trust → Private pro-environmental behavior	0.083	0.035	0.159	0.000	Supported
Destination image → Satisfaction → Public pro-environmental behavior	0.184	0.029	0.369	0.019	Supported
Destination image → Satisfaction → Destination trust → Public pro-environmental behavior	0.212	0.122	0.364	0.000	Supported
Destination image → Destination trust → Public pro-environmental behavior	0.234	0.122	0.393	0.000	Supported

show that the mediating effect was verified for destination image on private pro-environmental behavior via the role of satisfaction (β = 0.139; CI = [0.056, 0.238]), supporting the destination image → satisfaction→ private pro-environmental behavior path. All other specific mediating effects were also identified.

4.4. Results of fsQCA

4.4.1. Applying fsQCA to Predict Private Pro-Environmental Behavior

(1)

Contrarian Case Analysis

Contrarian case analysis was conducted before the application of fsQCA to easily and rapidly examine the portion of instances in the collected sample that the main effects do not explain. As such, they would be excluded from the result of a normal variance-based method [142]. As highlighted by previous scholars, a common mistake made by researchers applying variable-level analysis is to ignore cases of association that are opposite to the main effect relationship [143]. Therefore, in order to examine possible positive, negative, or no relationships in the same data set, a contrarian case analysis is required [144].

Following the suggestion of Pappas and Woodside (2021) for the application of contrarian case analysis [145], the sample was divided by quintiles to investigate the relationship between the tested variables; then, cross-contingency analysis was performed on the quintiles. The result of a cross-contingency analysis of any two constructs is a 5 × 5 table showing every possible configuration at each quantile between the two variables in the sample. Among them, the cases in the upper left and lower right corners represent main effects, while the cases in the lower left and upper right corners cannot be explained by the main effects. If the cases in the lower left and upper right corners exist, it means that there are indeed contrarian cases in the sample.

Table 6. Cross-contingency table of destination image and private pro-environmental behavior.

Destination Image		Private Pro-Environmental Behavior					Total
Cramer’s V = 0.315, p < 0.001		1	2	3	4	5
1	Case number	25	13	6	0	2	46
	Percentage	54.3%	28.3%	13.0%	0.0%	4.3%	100.0%
2	Case number	18	35	3	6	2	64
	Percentage	28.1%	54.7%	4.7%	9.4%	3.1%	100.0%
3	Case number	7	24	12	9	19	71
	Percentage	9.9%	33.8%	16.9%	12.7%	26.8%	100.0%
4	Case number	4	9	8	4	11	36
	Percentage	11.1%	25.0%	22.2%	11.1%	30.6%	100.0%
5	Case number	1	15	7	19	26	68
	Percentage	1.5%	22.1%	10.3%	27.9%	38.2%	100.0%
Total	Case number	55	96	36	38	60	285
	Percentage	19.3%	33.7%	12.6%	13.3%	21.1%	100.0%

,

Table 7. Cross-contingency table of satisfaction and private pro-environmental behavior.

Satisfaction		Private Pro-Environmental Behavior					Total
Cramer’s V = 0.305, p < 0.001		1	2	3	4	5
1	Case number	25	33	5	0	0	63
	Percentage	39.7%	52.4%	7.9%	0.0%	0.0%	100.0%
2	Case number	11	12	6	2	1	32
	Percentage	34.4%	37.5%	18.8%	6.3%	3.1%	100.0%
3	Case number	14	31	7	9	25	86
	Percentage	16.3%	36.0%	8.1%	10.5%	29.1%	100.0%
4	Case number	5	13	11	10	15	54
	Percentage	9.3%	24.1%	20.4%	18.5%	27.8%	100.0%
5	Case number	0	7	7	17	19	50
	Percentage	0.0%	14.0%	14.0%	34.0%	38.0%	100.0%
Total	Case number	55	96	36	38	60	285
	Percentage	19.3%	33.7%	12.6%	13.3%	21.1%	100.0%

and

Table 8. Cross-contingency table of trust and private pro-environmental behavior.

Trust		Private Pro-Environmental Behavior					Total
Cramer’s V = 0.304, p < 0.001		1	2	3	4	5
1	Case number	27	32	4	2	0	65
	Percentage	41.5%	49.2%	6.2%	3.1%	0.0%	100.0%
2	Case number	18	31	7	2	10	68
	Percentage	26.5%	45.6%	10.3%	2.9%	14.7%	100.0%
3	Case number	6	12	11	7	13	49
	Percentage	12.2%	24.5%	22.4%	14.3%	26.5%	100.0%
4	Case number	4	5	4	8	12	33
	Percentage	12.1%	15.2%	12.1%	24.2%	36.4%	100.0%
5	Case number	0	16	10	19	25	70
	Percentage	0.0%	22.9%	14.3%	27.1%	35.7%	100.0%
Total	Case number	55	96	36	38	60	285
	Percentage	19.3%	33.7%	12.6%	13.3%	21.1%	100.0%

present the cross-contingency tables of destination image, satisfaction, trust, and private pro-environmental behavior. These tables show that there are contrarian cases in the sample. Consequently, fsQCA was performed for data analysis in order to incorporate counterfactual cases in the prediction of high-level private pro-environmental behavior.

(2)

Data Calibration

In fsQCA, each condition (destination image, satisfaction, trust) and outcome (private pro-environmental behavior) is treated as a separate set. When multiple items are used to measure a variable, each case in each construct needs to be assigned a value as an input value in fsQCA. The easiest way to do this is to enter a corresponding single value for each case by averaging all items [145]. On the basis of the criteria suggested by Calabuig Moreno et al. (2016), the calibration standard for full non-members for each variable was set to the 0.05th percentile, the calibration standard for the intersection was set to the 0.5th percentile, while the calibration standard for full members was set to the 0.95 percentile [146].

Table 9. Calibration of conditions and outcomes in the private pro-environmental behavioral model.

Category	Conditions and Outcomes	Calibration
		Full Member	Intersection	Full Non-Member
Outcome variable	Private pro-environmental behavior	5	4	3
Condition variable	Destination image	5	4	3
	Satisfaction	5	4	3
	Trust	4.75	4	2.75

presents a general description of the calibration information for each condition and outcome in the present research. Moreover, for all values after calibration, this study inputted values of 0.5 as 0.499 in the fsQCA software program [147].

(3)

Analysis of the necessary conditions of fsQCA

Before the conditional configuration analysis, the necessity of each condition needs to be checked individually [148]. The fsQCA software was used to test whether a single condition (including its non-set) forms a necessary condition for private pro-environmental behavior. In QCA analysis, when a certain condition always exists when the result occurs, then it becomes a necessary condition for the outcome [149]. Consistency is regarded as an important test of the necessary condition. A consistency of higher than 0.9 means that this condition is the necessary condition for the outcome [149]. The analytical results of the necessary conditions for high- and non-high-level private pro-environmental behavior are presented in

Table 10. Analysis of necessary conditions in private pro-environmental behavior model.

Condition Variable	Private Pro-Environmental Behavior		~Private Pro-Environmental Behavior
	Consistency	Coverage	Consistency	Coverage
Destination image	0.748	0.847	0.569	0.483
~Destination image	0.544	0.627	0.820	0.710
Satisfaction	0.737	0.850	0.558	0.483
~Satisfaction	0.552	0.624	0.826	0.702
Trust	0.725	0.848	0.525	0.462
~Trust	0.540	0.602	0.827	0.693

. The consistency for all conditions was below 0.9. Thus, there is no necessary condition for influencing high-level and non-high-level private pro-environmental behavior.

(4)

Sufficiency analysis of configuration conditions

As suggested by Fiss (2011), the consistency threshold was set to 0.8 in this study [150]; meanwhile, the PRI score threshold was set to greater than or equal to 0.67 in order to avoid simultaneous subset relations of attribute combinations in both the outcomes and the absence of the outcomes [151]. Accordingly, a PRI consistency threshold of 0.67 was set in this research, and the threshold for case frequency was set to 2. Through the above procedure, at least 80% of the sample was retained.

According to the configuration analysis process, the outcomes of each construct are shown in

Table 11. Configuration analysis of high-level private pro-environmental behavior.

Mode	Image-Relationship Quality Mode		Relationship Quality Mode
Condition configuration	Configuration2	Configuration3	Configuration1
Destination image	●	●
Satisfaction	●		●
Trust		●	●
Consistency	0.902	0.911	0.910
Raw coverage	0.616	0.620	0.615
Unique coverage	0.062	0.066	0.060
Overall consistency	0.871
Overall coverage	0.742

Note: ● or _• indicates the presence of a condition, ⊗ or _⨂ indicates its absence; ● or ⊗: core condition, _• or _⨂: peripheral condition. Blank space indicates “don’t care” condition.

. For the three configurations presented in this table, the consistency levels of both the single solution (configuration) and the overall solution were greater than the acceptable minimum standard of 0.75 [148], of which the consistency of the overall solution was 0.871, and the coverage of the overall solution was 0.742. The three configurations in Table 11 can be regarded as a sufficient combination of conditions for high-level private pro-environmental behavior.

After categorization, the antecedent configuration of private pro-environmental behavior is separated into the relationship quality mode and image–relationship quality mode. The relationship quality mode corresponds to configuration 1, while the image–relationship quality mode corresponds to configuration 2 and configuration 3.

Configuration 1 shows that the core elements of relationship quality together play a central role, which means that when satisfaction and trust coexist, other conditions are irrelevant for high-level private pro-environmental behavior. This indicates that, compared with other conditions, relationship quality is particularly essential for private pro-environmental behavior, because relationship quality alone can be a sufficient condition for interpreting outcomes. Thus, this study named this configuration as relationship quality. The consistency of this configuration was 0.910, the unique coverage was 0.060, and the raw coverage was 0.615. This path explained approximately 61.5% of the cases of private pro-environmental behavior. Figure 2 provides an explanation example of configuration 1.

In the image–relationship quality model, the core condition was the single component of destination image and relationship quality, which mainly included two sub-modes (configuration 2 and configuration 3). This means that the coexistence of the single component of destination image and relationship quality was particularly important for the private pro-environmental behavior. Figure 3 and Figure 4 provide the explanation examples of configuration 2 and configuration 3.

(5)

Robustness test

The robustness test was performed by adjusting the consistency threshold level. By adjusting the consistency threshold level from 0.8 to 0.85, this change did not lead to substantial changes in the number of configurations, configuration elements, or the fitting parameters of consistency and coverage. Consequently, the findings of this research are considered relatively reliable [152].

4.4.2. Applying fsQCA to Predict Public Pro-Environmental Behavior

(1)

Contrarian Case Analysis

Table 12. Cross-contingency table of destination image and public pro-environmental behavior.

Destination Image		Public Pro-Environmental Behavior					Total
Cramer’s V = 0.241, p < 0.001		1	2	3	4	5
1	Case number	23	8	8	1	6	46
	Percentage	50.0%	17.4%	17.4%	2.2%	13.0%	100.0%
2	Case number	24	7	20	8	5	64
	Percentage	37.5%	10.9%	31.3%	12.5%	7.8%	100.0%
3	Case number	9	11	23	11	17	71
	Percentage	12.7%	15.5%	32.4%	15.5%	23.9%	100.0%
4	Case number	1	5	9	4	17	36
	Percentage	2.8%	13.9%	25.0%	11.1%	47.2%	100.0%
5	Case number	9	9	19	4	27	68
	Percentage	13.2%	13.2%	27.9%	5.9%	39.7%	100.0%
Total	Case number	66	40	79	28	72	285
	Percentage	23.2%	14.0%	27.7%	9.8%	25.3%	100.0%

,

Table 13. Cross-contingency table of satisfaction and public pro-environmental behavior.

Satisfaction		Public Pro-Environmental Behavior					Total
Cramer’s V = 0.272, p < 0.001		1	2	3	4	5
1	Case number	27	13	17	1	5	63
	Percentage	42.9%	20.6%	27.0%	1.6%	7.9%	100.0%
2	Case number	11	3	11	6	1	32
	Percentage	34.4%	9.4%	34.4%	18.8%	3.1%	100.0%
3	Case number	23	14	25	10	14	86
	Percentage	26.7%	16.3%	29.1%	11.6%	16.3%	100.0%
4	Case number	1	6	15	7	25	54
	Percentage	1.9%	11.1%	27.8%	13.0%	46.3%	100.0%
5	Case number	4	4	11	4	27	50
	Percentage	8.0%	8.0%	22.0%	8.0%	54.0%	100.0%
Total	Case number	66	40	79	28	72	285
	Percentage	23.2%	14.0%	27.7%	9.8%	25.3%	100.0%

,

Table 14. Cross-contingency table of trust and public pro-environmental behavior.

Trust		Public Pro-Environmental Behavior					Total
Cramer’s V = 0.319, p < 0.001		1	2	3	4	5
1	Case number	35	11	16	2	1	65
	Percentage	53.8%	16.9%	24.6%	3.1%	1.5%	100.0%
2	Case number	22	11	22	8	5	68
	Percentage	32.4%	16.2%	32.4%	11.8%	7.4%	100.0%
3	Case number	5	9	17	6	12	49
	Percentage	10.2%	18.4%	34.7%	12.2%	24.5%	100.0%
4	Case number	2	2	11	2	16	33
	Percentage	6.1%	6.1%	33.3%	6.1%	48.5%	100.0%
5	Case number	2	7	13	10	38	70
	Percentage	2.9%	10.0%	18.6%	14.3%	54.3%	100.0%
Total	Case number	66	40	79	28	72	285
	Percentage	23.2%	14.0%	27.7%	9.8%	25.3%	100.0%

and

Table 15. Calibration of conditions and outcomes in the public pro-environmental behavior.

Category	Conditions and Outcomes	Calibration
		Full Member	Intersection	Full Non-Member
Outcome variable	Public pro-environmental behavior	5	4	2.325
Condition variable	Destination image	5	4	3
	Satisfaction	5	4	3
	Trust	4.75	4	2.75

include details of the cross-contingency table of destination image, satisfaction, trust, and public pro-environmental behavior. These tables all demonstrate that there are contrarian cases in the sample. Therefore, in order to incorporate contrarian cases into the prediction of high-level public pro-environmental behavior, data analysis was conducted with fsQCA.

(2)

Data calibration

Each condition (destination image, satisfaction, trust) and outcome (public pro-environmental behavior) in fsQCA was considered as a separate set. According to the criteria suggested by Calabuig Moreno et al. (2016), the calibration standard for full non-members for each variable was set to the 0.05th percentile, the calibration standard for the intersection the 0.5th percentile, while the calibration standard for full members was set to the 0.95 percentile [146]. Table 15 gives an overview of the calibration information for each condition and outcome in the present research. For all values after calibration, input values of 0.5 as 0.499 were inputted in the fsQCA software program [147].

(3)

Analysis of the necessary conditions of fsQCA

Table 16. Analysis of necessary conditions in pro-environmental behavior model.

Condition Variable	Public Pro-Environmental Behavior		~Public Pro-Environmental Behavior
	Consistency	Coverage	Consistency	Coverage
Destination image	0.763	0.718	0.596	0.621
~destination image	0.597	0.572	0.729	0.773
Satisfaction	0.777	0.745	0.567	0.602
~satisfaction	0.584	0.550	0.760	0.791
Trust	0.792	0.770	0.523	0.563
~trust	0.551	0.510	0.787	0.807

presents the test results of the necessary conditions for high- and non-high-level public pro-environmental behavior. Since the consistency for all conditions was below 0.9, no necessary condition for influencing high-level and non-high-level public pro-environmental behavior existed.

(4)

Sufficiency analysis of configuration conditions

According to the configuration analysis process,

Table 17. Configuration analysis of high-level public pro-environmental behavior.

Mode	Relationship Quality Mode
Condition configuration	Configuration 1
Destination image
Satisfaction	●
Trust	●
Consistency	0.831
Raw coverage	0.675
Unique coverage	0.675
Overall consistency	0.831
Overall coverage	0.675

Note: ● or _● indicates the presence of a condition, ⊗ or _⊗ indicates its absence; ● or ⊗: core condition, _● or _⊗: peripheral condition. Blank space indicates “don’t care” condition.

shows the outcomes of each construct. For the configuration in this table, the consistency levels of the single solution (configuration) and the overall solution were greater than the acceptable minimum standard of 0.75 [148], of which the consistency of the overall solution was 0.831, while the coverage of the overall solution was 0.675. The configuration encompassing satisfaction and trust in Table 17 can be regarded as a sufficient configuration of conditions for high-level public pro-environmental behavior.

Values in configuration 1 of Table 17 show that the core factors of relationship quality together play a key role. This means that when satisfaction and trust coexist, other conditions are irrelevant for high-level public pro-environmental behavior, which indicates that relationship quality is more important than other conditions for public pro-environmental behavior. The reason is that relationship quality alone can be a sufficient condition for interpreting outcomes. Thus, this configuration was named the relationship quality mode, with the consistency of 0.831, unique coverage of 0.675, and raw coverage of 0.675. This path explained approximately 67.5% of the cases of public pro-environmental behavior. Figure 5 provides an explanation example of configuration 1.

(5)

Robustness test

The consistency threshold level was adjusted from 0.8 to 0.85 for robustness testing. No substantial changes in the configuration number, configuration elements, or the fitting parameters of consistency and coverage were discovered, which confirmed the reliability of the present study [152].

5. Conclusions, Contributions, and Implications

5.1. Conclusions

The current study employed the SEM and fsQCA methods to examine the influence and configuration effects of pro-environmental behavior in rural tourism. The SEM method produced the following findings.

First, this finding aligns with prior studies [75], which confirmed the significant positive effect of destination image on relationship quality variables (satisfaction and trust). Most studies focused on the effect of image on satisfaction or trust [93,102], which severed the relationship between satisfaction and trust. This study verified the relationship between these two variables in the same setting. Moreover, positive destination image has been confirmed to help generate good relationship quality, which is in line with Choi’s finding [46]. It also verifies the validity of the appraisal theory of emotions in tourism research [74], offering an important theoretical perspective for destination image research. Moreover, it implies that directing attention towards the effects of destination image on relationship quality will benefit the sustainable development of rural tourism and rural land use.

Second, the results showed that relationship quality variables (satisfaction and trust) significantly and positively affect private and public pro-environmental behavior, supporting the viewpoint that relationship quality is a vital driver of pro-environmental behavior [8,40,41]. More importantly, the empirical results filled the lacuna through verifying the impact of relationship quality variables (satisfaction and trust) on sub-dimensions of pro-environmental behavior. In tourism research, they enrich the knowledge on the influence of relationship quality on pro-environmental behavior [41].

Third, this study showed that notable differences exist in the formation of pro-environmental behavior of various dimensions. Results of this research support the argument that destination image is an important antecedent of pro-environmental behavior [32]. Moreover, we specified the effects of destination image on two sub-dimensions of pro-environmental behavior, i.e., destination image has a significant direct impact on private pro-environmental behavior, but no significant direct effect on public pro-environmental behavior. It indicates that valuing destination image is of great benefit to improving private pro-environmental behavior, and its impact on public pro-environmental behavior should not be overlooked.

Fourth, relationship quality mediates the relationship between destination image and pro-environmental behavior in different ways. Specifically, relationship quality mediates the relationship between destination image and private pro-environmental behavior partially and significantly, and fully mediates the link between image and public pro-environmental behavior. The different mediating effects, however, do not change the important role of the relationship quality [46]. Findings of previous studies confirmed that more satisfaction and trust improved pro-environmental behavior [64]. On this basis, this research further found that, given a higher level of satisfaction and trust, tourists will be more inclined to adopt public pro-environmental behavior at a destination with a better image.

Additionally, the fsQCA presents the following results. First, none of the three factors (destination image, satisfaction, and trust) constituted a sufficient and necessary condition in predicting private pro-environmental behavior. Second, among the eight condition combinations generated from the aforementioned three condition variables, there were three configurations that met the requirements, with an overall coverage rate of 0.74. They constitute two modes: the relationship quality and destination image–relationship quality modes. Third, regarding the prediction of public pro-environmental behavior, neither destination image nor satisfaction nor trust formed a sufficient and necessary condition. Fourth, there was only one qualified configuration among the eight combinations from the above-mentioned three condition variables. The overall coverage rate of this configuration was 0.675, representing the relationship quality mode. Fifth, a further analysis of the configuration effects in predicting private and public pro-environmental behavior discovered that the configuration of relationship quality plays a vital role in both private and public pro-environmental behavior. As a useful complement of the SEM method, fsQCA is helpful for explaining the complexity of tourist pro-environmental behavior. The comparison of SEM and fsQCA demonstrated the commonalities and differences in their results, i.e., they both highlighted the significance of the relationship quality in predicting pro-environmental behavior; meanwhile, the role of destination image as an independent antecedent of pro-environmental behavior was only verified in the SEM analysis, and not in the fsQCA analysis. In detail, the results of the fsQCA showed that destination image was not a sufficient and necessary condition in predicting pro-environmental behavior and could not constitute a configuration. In this sense, the complementation of the two methods is critical in understanding both the linear and non-linear associations among factors leading to pro-environmental behavior.

5.2. Theoretical Contributions

This research contributes to the literature on pro-environmental behavior in several important ways.

First, the current research effectively confirms the framework of “destination image–relationship quality–pro-environmental behavior” in rural tourism. By applying the appraisal theory of emotions to investigate the influence of destination image on pro-environmental behavior, the efficacy of the appraisal theory of emotions in predicting pro-environmental behavior has been highlighted. This research also extends the traditional “appraisal theory of emotions” [43] to the framework of “destination image–relationship quality–behavior”, representing the usefulness of this theory in studying pro-environmental behavior.

Second, the present research tested the universal value of relationship quality as a predictor of pro-environmental behavior in the private and public dimensions. Following the literature on the driving role of relationship quality on pro-environmental behavior, this study furthered the research by subdividing the behavior into private and public pro-environmental behavior [115], and empirically tested the effect of relationship quality on these two types of behavior. The present research filled this void by verifying the importance of relationship quality, and complemented the study on the links between relationship quality and pro-environmental behavior. This also extends the understanding of the importance of the relationship marketing paradigm in explaining pro-environmental behavior.

Third, destination image presents obvious differences in the realization paths of pro-environmental behavior in the private and public domains. These empirical findings suggest that relationship quality exerts different mediating effects on types of pro-environmental behavior. Moreover, the analysis of image’s effect has not been confined to the single dimension of pro-environmental behavior, but that of two sub-dimensions, offering new evidence for the dissimilarities in image’s effects on private and public pro-environmental behavior [32,113,120]. This study found that this provides an important reference for the further exploration and analysis of two types of pro-environmental behavior decision-making mechanisms.

Fourth, methodologically, the existing literature in tourism research mainly employed the SEM method to explain linear associations among variables predicting tourist pro-environmental behavior [21,25,26,27,28,40,41,58], while fsQCA was scarcely utilized in this area. As an effective approach for revealing the non-linear configurational effects of variables, the fsQCA method can serve as a proper complementation to SEM. There have been studies in some fields that integrated the two methods for data analysis [146,153]. Given the complexity of tourist pro-environmental behavior, this research combined the symmetric approach (SEM) and asymmetric approach (fsQCA) to better understand the formation of tourist pro-environmental behavior, offering evidence for the application of this integration in rural tourism contexts.

5.3. Managerial Implications

In regard to rural destinations, the adequate integration of the sustainability of rural land use and rural tourism is needed [16]. A number of managerial implications also emerge from the current study in terms of rural tourism management and practice.

First, destination image is a key driver of relationship quality and outcomes important to pro-environmental behavior. The results of the SEM analysis show that there is still room for improvement in destination image, satisfaction, and trust to promote pro-environmental behavior and the sustainable development of destinations [154,155]. Tourism industry stakeholders should view the quality of products and services as an essential indicator that affects satisfaction and trust for destinations [76]. Despite the tremendous changes in rural land use brought about by rural tourism [17,18], the essential role of the rural landscape in rural tourism remains [156]. The balance between land conservation and rural tourism requires administrative authorities to properly change the pattern of land use. For example, farmland and forest land can be included in tourism planning, while local farmers and private sectors should be encouraged to participate in rural tourism development and the protection of the rural landscape through reasonable distribution [19]. These efforts are conducive to building the ecological tourism image and creating a rational, efficient, and intensive pattern of land use, which will contribute to the maximization of economic, social, and ecological benefits in rural land use and tourism development [5]. Moreover, in order to differentiate tourism products, regional coordination among rural destinations is necessary to explore a sustainable path for rural land use and rural tourism development [156]. From the aspect of tourists, destination managers are advised to raise tourist awareness of conserving the destination image.

Given that satisfaction is crucial to pro-environmental behavior, destination managers should provide satisfactory tourism experiences [41], such as fruit picking, traditional farming, and harvesting. In fact, living and working scenes of the rural residents play a vital role in fostering quality tourism experiences [157]. Based on the appraisal theory of emotions, positive assessments of destination environments and events from tourists lead to positive emotions, and increase levels of satisfaction [74]. Destination management also should encourage tourists to publicize their tourism experiences on social networks, because prior studies have noted that destination images on social networks influence the generation of positive emotions such as satisfaction [94,99].

Apart from tourist satisfaction, trust is another important predictor of pro-environmental behavior. Prior studies conclude that improvements in satisfaction increase levels of trust [75,107], and so do positive destination images [75,76]. Destination management thus should take measures to gain more trust through enhancing destination image and satisfaction. As previously mentioned, trust is closely linked with whether the destination is reliable or able to fulfill its commitment [58], while the enhancement of trust relies on service quality [70]. Administrative authorities should encourage enterprises in tourism to formulate industrial standards concerning service quality and track their service quality to reward or punish enterprises with outstanding or poor performance. The purpose is to improve the service quality of tourism enterprises and ensure that tourists can receive high-quality services [41]. Local governments should support the development of destination-related industries (e.g., hospitality, catering, transportation, travel services) through policy and tax preferences, so that they can offer quality products and services to increase trust for the destination [58].

Destination managers should employ professionals to enhance satisfaction and service quality [158], which is instrumental to the increase in trust [95]. Furthermore, publicizing and communicating public information is another way to increase trust toward the destination [31]. As the Organisation for Economic Co-operation and Development (2020) points out, promoting public awareness of the government and ensuring effectiveness need to be guided by the principles of transparency, integrity, accountability, and stakeholder engagement [159]. For example, through destination image promotion, environmental protection reminders will convince tourists to feel that local administrators attach great importance to destination image and environmental conservation, which, in turn, strengthens trust for the destination and the adoption of pro-environmental behavior. In addition, the meta-analysis demonstrates that changes in pro-environmental behavior affect rural land use and management [160]. Efforts should be made to encourage and advocate for tourist pro-environmental behavior when making rural land management policy, which will offer a new way to sustain rural land use and management.

Additionally, rural tourist destinations should enhance the integration of the dimensions of relationship quality and destination trust. The configuration analysis of fsQCA demonstrated that a single factor could not constitute the necessary and sufficient condition of predicting pro-environmental behavior. This study found that relationship quality, comprising satisfaction and trust, is an important configuration in predicting both private and public pro-environmental behavior. Consequently, rural destinations should improve the matching of different factors from multiple perspectives on the basis of relationship quality.

Finally, the coordinated integration of destination image and relationship quality should be emphasized at rural destinations. Though the combination of destination image and relationship quality cannot be a qualified configuration for predicting public pro-environmental behavior, it plays a vital role in predicting private pro-environmental behavior. Thus, rural destination managers should improve not only relationship quality but also destination image to achieve destination image–relationship quality coordination for enhanced private pro-environmental behavior.

6. Limitations and Future Research Directions

By employing SEM and fsQCA methods, this study constructed a conceptual model and empirically tested the model. However, the conclusions of this study should be interpreted cautiously due to the following reasons. First, the snowball sampling methodology was applied in this research. Due to the inconvenience brought by the COVID-19 pandemic, snowball sampling is a viable alternative for sampling that can be found in some recent studies [161,162]. Objectively, this sampling method has been criticized for lacking external validity and representativeness. Thus, future research should use more precise sampling methods at destinations when the conditions for such methods are available. Meanwhile, experimental research can be another option in the future [163]. Second, the integration of the appraisal theory of emotions and relationship marketing paradigm has been successfully executed in private and public pro-environmental behavior in rural tourism; however, such evidence is scarce in hospitality contexts. It provides a new opportunity to explore the sub-dimensions of pro-environmental behavior in the field of hospitality based on these two theories. Third, satisfaction and trust in this study were measured to investigate the relationship between destination image and pro-environmental behavior. There is still room for taking other dimensions of relationship quality into account, such as commitment and identification. Lastly, only domestic tourists were surveyed in this study. Therefore, in the future, the inclusion of other nationalities may offer a more complete picture.