The Mechanism of Word-of-Mouth for Tourist Destinations in Crisis

In an era of mobile internet, word-of-mouth marketing has become a powerful tool for optimizing tourist destination marketing. Extensive research studies on the mechanism of word-of-mouth have been carried out. However, there is little research on the mechanism of word-of-mouth concerning tourist destinations in crisis. This article focuses on the influence mechanism of word-of-mouth in the relationship between perceived images and behavioral intentions. A structural equation model was established, with “perceived image” as the independent variable, “word-of-mouth” and “psychological distance” as mediating variables, and “behavioral intention” as the dependent variable. Our study is based on a survey in Southern Xinjiang. The final results show that word-of-mouth plays a mediating role in the relationship between perceived image and behavioral intentions. The moderating effect of word-of-mouth plays two roles in the mechanism of the perceived image on the behavioral intention: the promotion mechanism and the repression mechanism. Tourists’ sense of psychological distance significantly mediates the relationship between perceived image and behavioral intention. The findings enrich word-of-mouth and the marketing theories for the destinations in crisis, which provide enlightening insights for the sustainable development of the destinations in crisis.


Introduction
Word-of-mouth (WOM) marketing has an important impact on consumer behavior (Bughin et al., 2010;Huete-Alcocer, 2017;Litvin et al., 2018;Wilson et al., 2017;Yoon et al., 2017). In the age of Omnimedia, managers have paid increasing attention to the role of WOM marketing, an effective and efficient marketing tool (Gong et al., 2018;Hornik et al., 2015;. For example, numerous travel agencies have established online comment systems on websites to encourage consumers to spread information via their personal social media (Yan et al., 2018), which helps the agencies attract potential consumers (El Ouardighi et al., 2018; J. J. Kim et al., 2019). Extensive research studies on the mechanism of WOM have been carried out. Based on the expectation difference model, Shi et al. (2016) found that WOM could affect not only customers' expectations and satisfaction regarding a certain product or service but also the perceived quality of that product or service, which ultimately affects their purchase intentions. Liang et al. (2018) indicated that WOM can effectively improve consumers' perceived value and reduce their perceived risk, thus improving their repurchase intentions. Manes and Tchetchik discovered that WOM can reduce the uncertainty caused by information asymmetry and thus increase hotel reservations (Manes & Tchetchik, 2018). Most scholars focus on its effect in increasing value perceptions and reducing uncertainty, when consumers are buying general products or services.
In this article, Southern Xinjiang is selected as the study area. The image of tourism safety in Southern Xinjiang was severely damaged because of several terrorist incidents, resulting in tourists' bias about the safety problems in Southern Xinjiang . Kuto and Groves (2004) indicated that the consequence of terrorist incidents in these destinations is serious, and it is very hard to recover from. The tourism market in Southern Xinjiang has been sluggish since the terrorist incidents, even though there have been no terrorist attacks since 2014. Hence, it is necessary for Southern Xinjiang to find out measures to eliminate such cognitive bias.
This article focuses on the mechanism of WOM in tourists' decision-making. A structural equation model was established, taking tourists' perceived image as the independent variable, WOM and psychological distance as the mediated variables, and behavioral intention as the dependent variable. The article tests the mediating effects of WOM between perceived images and behavioral intentions. The article also tests the moderating effects of WOM on the relationship between perceived images and psychological distances, and the relationship between psychological distances and behavioral intentions. Several conclusions are drawn: WOM played a mediating role in the relationship between perceived images and behavioral intentions. As a moderating variable, WOM played an important role in the promotion mechanism and repression mechanism. Tourists' sense of psychological distance significantly mediated the relationship between perceived images and behavioral intentions. The findings in this article provide enlightening insights for the sustainable development of marketing activities in Southern Xinjiang.
The contributions of this article are as follows. First, the findings in this article enrich WOM theories. That is to say, WOM can act as a mediating variable and a moderating variable, not just as a result variable. As a moderating variable, WOM played an important role in the promotion mechanism and repression mechanism. Second, the findings also enrich the marketing theories for the destinations in crisis. The value of tourists' WOM behavior cannot be overemphasized, which should be the focus of future research in the era of self-media.
The rest of this article is structured as follows. Section "Literature Review" presents a literature review on perceived image, WOM, psychological distance, and behavioral intention. Section "Research Method" provides an introduction about the study area, questionnaire design, and sampling. Section "Results" includes the results of several analysis methods. The results were discussed in section "Discussion." The conclusions, implications, and limitations and further research are drawn in section "Conclusion and Implications."

Perceived Image
The perceived image was proposed by Hunt from the perspective of tourists in the 1970s (Hunt, 1975). He believed that the image of a tourist destination, as a purely subjective concept, is the impression that people hold on the status of nonresidence. Crompton (1979) continued and further enriched its definition of perceived image as a combination of tourist beliefs, thoughts, impressions, and emotions (Lai & Li, 2016;Marine-Roig, 2015). Later, scholars carried out multidimensional subdivision for advancing in-depth research (Gartner, 1994;Marine-Roig & Ferrer-Rosell, 2018;Martín-Santana et al., 2017;Michaelidou et al., 2013). Among them, the two-dimensional division proposed by Baloglu and Brinberg is widely recognized (Baloglu & Brinberg, 1997;Baloglu & McCleary, 1999). They believed that the perceived image was formed through the interpretation of the rational and emotional of the tourists, which included two dimensions: cognitive image and emotional image. In concrete terms, cognitive image is the understanding of tourists' objective conditional attributes (Bornhorst et al., 2010;Žabkar et al., 2010). Emotional image is the subjective emotional response of tourists to destinations (Beerli & Martin, 2004). However, in the later research, the effect of the emotional dimension on the perceived image of the destination is often ignored (Chew & Jahari, 2014). On this basis, scholars proposed a general term that included cognitive and emotional components-the overall perceived image (Fu et al., 2016;Stylidis et al., 2017), which contained the gains (Souiden et al., 2017), the cost (Özdemir & Şimşek, 2015), and the overall evaluation (Da Silva et al., 2018;Sharma & Nayak, 2018;Whang et al., 2016). Hence, we adopted these indicators to measure the perceived image.

WOM
The study of WOM began in the 1960s. According to Arndt (1967), WOM refers to verbal interpersonal communication between the giver and the receiver, and the receiver believes that the giver's WOM about a brand, product, or service is noncommercial. The later scholar further refined the definition and considered that it included both positive and negative views. In the 1990s, WOM in the field of tourism began to increase with the formation of the service economy and service marketing theories, research on WOM in the field of tourism (Bristor, 1990). In previous papers, givers, receivers, and disseminated information are the three focuses of scholars' interest (Ahmad & Laroche, 2017;Le et al., 2018;Stern, 1994;Yang & Mattila, 2017;M. Zhang et al., 2017). In the measurement of WOM, Harrison-Walker (2001) first proposed a two-component WOM measurement method, which includes WOM activity (frequency of mentioning) and praise (message valence) from the perspective of the giver. Sweeney et al. (2012) deemed that we should also examine specific message details, including cognitive content, the richness of content, and the strength of delivery.

Psychological Distance
The concept of psychological distance stems from the temporal construal theory Liberman & Trope, 1998). It refers to a subjective distance perception of actors and events within the perpetrator's psychological space (Mei et al., 2018;Trope et al., 2007), which is a psychological representation of events. It was originally composed of a single time dimension (Trope & Liberman, 2010). Henderson et al. argued that spatial distances and temporal distances had similar effects on people's reasoning and judgment (Henderson et al., 2006). Bar-Anan et al. (2007) thought that social distance was also one of the dimensions of psychological distance (Bar-Anan et al., 2006). On this basis, Wakslak et al. (2006) increased the dimension of probability. Then, Fiedler (2007) introduced information distance to previous theories. At this point, the psychological distance has become a cross-disciplinary category. Psychological distance could affect tourists' judgment and decision-making (K. Kim et al., 2008). This is the core idea of the theory of constructive level that the response to social events depends on the social cognition of the psychological representation of events (Liberman et al., 2002). Under the high level of construction, people tend to abstract the distant future event and will not consider the specific situation. It is easy to form a large deviation compared with the actual situation. Under low-level construction, people specifically characterize recent events and make judgments in actual situations. The farther the psychological distance is, the more abstract the psychological representation of the object characteristics, and the more inclined it is to construct at a high level. In terms of measurement, we used the three items: tourists' understanding of the social situation about destinations before traveling, the probability of travel behavior, and tourism information familiarity to measure psychological distance (Bar-Anan et al., 2007).

Behavioral Intention
Behavioral intentions are the possibility that individuals will produce certain behaviors based on their subjective tendencies (Zeithaml et al., 1996), which is a prelude to real behavior (Kozak, 2001;Moon & Han, 2018). Increasing research on it is important for innovative destination marketing and customer management practices (Loi et al., 2017;Petrick, 2004). Previous scholars have divided behavioral intentions into positive and negative categories (Parasuraman et al., 1988). Among them, positive behavioral intentions mean that customers are willing to pay high prices for services or products, recommend products or services to others, and make oral publicity for the enterprises (Oh, 1999). Negative behavioral intentions refer to unfavorable verbal propaganda to the firms, complaining about the company's products and services, complaining about the company, and reducing the purchase of products and services (Zeithaml et al., 1996). In terms of measurement, the loyalty intention and recommendation intention are the two important dimensions (Prayag et al., 2017;Woodside & Lysonski, 1989), and the premium tolerance can also speculate behavioral intention (Graciola et al., 2018;Gronholdt et al., 2000;W. G. Kim et al., 2017). Hence, in this article, the behavioral intentions of tourists were measured by the three dimensions (Kock et al., 2016;C. R. Liu et al., 2012).

Perceived Image, WOM, and Psychological Distance
The perceived image can reduce the perceived risk brought by its intangible service, maintain the trust of customers, and thus influence their purchasing decisions, and ultimately prompt customers to improve their loyalty (Gray & Balmer, 1998;Nguyen & Leblanc, 2001). Shankar et al. (2003) pointed out that a better image means that the customers are willing to recommend the service provider to others. This is consistent with the study by Abbas et al. (2018) that perceived image is an important predictor of WOM communication (Ifie et al., 2018). Papadimitriou et al. (2018) further used data analysis to confirm that the behavior of visitors participating in WOM communication is dependent on their overall perception of the destination.  empirically found that both tourists' cognitive images and emotional image perceptions of destinations have significant negative effects on psychological distances. Therefore, this article proposes the following hypotheses:

Hypothesis 1 (H1):
Tourists' perceived image has a significant positive impact on their WOM activities. Hypothesis 2 (H2): Tourists' perceived image has a significant negative impact on their sense of psychological distance.

WOM, Psychological Distance, and Behavioral Intention
In terms of the relationship between positive WOM and behavioral intentions, Hossain and Kim (2018) proposed that WOM is an important antecedent of behavioral intentions, which indicates a sustainable will. The distance is a binding condition that affects tourists' behavioral intention (Walmsley & Jenkins, 1992). A large number of scholars paid attention to the relationship between psychological distance and behavior (M. Thomas & Tsai, 2011;Zhao & Xie, 2011), which indicated that psychological distances could affect behavioral intentions. For example, Hallab and Kim (2011) found that destination preferences of inbound tourists showed a distance attenuation law. Through experimental research, Tasci (2009) found that visual information affects the social distance of destinations. That is, social distance is an important cause of tourists' behavioral intentions (Nyaupane et al., 2015). Summarizing the previous research, we can find that tourists have a high probability of revisiting the destination if the perceived psychological distance is smaller. Therefore, this article proposes the following hypotheses: Hypothesis 3 (H3): Tourists' WOM has a significant positive impact on their behavioral intentions. Hypothesis 4 (H4): Tourists' sense of psychological distance has a significant negative effect on their travel intentions.

The Mediating Effect of WOM, Psychological Distance, and the Moderating Effect of WOM
Tourists' perception of the destination's image has an important influence on their travel decision-making (Bigne et al., 2001;Court & Lupton, 1997;Milman & Pizam, 1995). Some scholars studied the relationship between them and then found that the perceived image could affect their behavioral intention through the role of many mediating variables (Cham et al., 2014;Ryu et al., 2008). This article creatively discusses the mediating effect of tourists' WOM between perceived images and behavioral intentions in Southern Xinjiang. Psychological distance plays a mediating role between perception and behavior (Darke et al., 2016). Therefore, this article believes that psychological distance mediates the relationship between perceived images and behavioral intentions. Numerous papers found that WOM is a significant moderating variable (De Matos & Rossi, 2008;Kuo et al., 2013;Lin & Lu, 2010), which demonstrated the moderating effect of positive WOM on the relationship between trust and consumers' purchase intentions, consumer inertia, and repeated purchases. Later studies focused on the moderating effect of WOM on the relationship between brand images and brand equity (Ansary & Hashim, 2018). V. L. Thomas et al. (2017) found that encouraging customers to spread WOM information can alleviate the psychological discomfort caused by self-threat. On the basis of this, this article will discuss the moderating effect of WOM on the relationship between perceived images and behavioral intentions. Therefore, the following hypotheses are proposed: Hypothesis 5 (H5): Tourists' WOM significantly mediates the relationship between perceived images and behavioral intentions. Hypothesis 6 (H6): Tourists' sense of psychological distance significantly mediates the relationship between perceived images and behavioral intentions. Hypothesis 7 (H7): Tourists' WOM significantly moderates the relationship between perceived images and psychological distances. Hypothesis 8 (H8): Tourists' WOM significantly moderates the relationship between psychological distances and behavioral intentions.
The study framework of this article is shown in Figure 1:

Study Area
This article selects Kashgar, with rich tourism resources, a unique geographical location (see Figure 2) and a diverse ethnic composition, as the research area. As the saying goes, "You have not been to Xinjiang until you have been to Kashgar" . Kashgar, the core of Southern Xinjiang, is the most representative region of Xinjiang (Xu et al., 2019). Kashgar is located in the northwest of China and the southwest of the Xinjiang Uygur autonomous region. Islam was first introduced to Kashgar because of its location (T. Liu & Yuan, 2019). It contains the ancient city of Kashgar (the only ancient city with a history of more than 1,000 years on the ancient Silk Road), the Id Kah Mosque, the tomb of the fragrant imperial concubine, and other cultural and tourist attractions. Due to the influence of geographical distances and emergencies, the psychological distance of tourists from the eastern region is relatively great, resulting in a low rate of visiting intentions.

Questionnaire Design
The questionnaire in this study is divided into two parts. The first part uses the five-point Likert-type scale to measure perceived images, WOM, psychological distances, and behavioral intentions, from "strongly disagree" (1) to "strongly agree" (5). To reduce the possibility of common method biases, reverse questions were set in the questionnaire to exclude the sequence effect of items and minimize the similarity between questions. At the same time, a neutral expression is adopted in the design of the questionnaire items and prompts. The second part concerns the demographic information of the interviewees. The measurement items of the research variables were all from the mature scales of wellknown scholars. After consulting relevant experts in the tourism research field of Southern Xinjiang, we modified the items according to their feedback. Before the formal investigation, this study conducted a presurvey from 1 May to 15 May 2016. Of the 100 questionnaires that were distributed, 86 were recovered, with an effective recovery rate of 86.0%. The outcomes showed that the questionnaires are well organized. The final questionnaire was developed after the presurvey. Therefore, the questionnaire used in this study has a good content validity.
The three items to measure perceived images are adapted from the scale of Assaker et al. (2011). The items were as follows. PEI1: I gained a lot from the tour in Southern Xinjiang; PEI2: The whole trip was worth the cost; and PEI3: For me, traveling in Southern Xinjiang is a good choice. Four items used to measure WOM are adapted from the scales developed by Bansal and Voyer (2000) and Park and Lee (2009). The items were as follows. WOM1: I often tell people about my travel experiences in Southern Xinjiang; WOM2: Compared with other tourist destinations, I prefer to tell others about my experiences in Southern Xinjiang; WOM3: I will seize every opportunity to tell others about my travel experience in Southern Xinjiang; and WOM4: When I mentioned my travel experience in Southern Xinjiang to others, I often went into details. Three items used to measure psychological distances are adapted from the scale of Bar-Anan et al. (2007). The items were as follows. PSD1: I had a certain understanding of the social situation in Southern Xinjiang before traveling there; PSD2: I am highly likely to travel to Southern Xinjiang again; and PSD3: Before traveling to Southern Xinjiang, I fully mastered the tourism information relating to Southern Xinjiang. Three items used to measure behavioral intentions were adapted from scales developed by Parasuraman et al. (1993) and Gronholdt et al. (2000). The items were as follows. INT1: I would recommend Southern Xinjiang to my relatives or friends; INT2: If I had the chance, I would be very happy to visit Southern Xinjiang again; and INT3: I am willing to pay more for a trip to Southern Xinjiang. The second part of the questionnaire includes seven items. I1: How many times do you travel each year? I2: Which city do you live in? I3: What is your gender? I4: What is your highest educational degree? I5: How old are you? I6: What is your occupation? I7: What is your monthly income range?

Sampling
The final questionnaire was handed out by three professors and seven graduate students from 22 June to 5 October 2017. To ensure random sampling, the questionnaires were randomly handed out in multiple locations, using the convenient sampling method. The questionnaires were handed out in Kashgar ancient city, Id Kah Mosque, the youth hostel, Kashgar airport, and Hotan Prefecture in Southern Xinjiang. Any personal preference was avoided when filling in the questionnaires for the sake of objectivity. In this study, a total of 400 questionnaires were distributed, and 335 valid questionnaires were obtained, with a recovery rate of 83.75%. To establish a valid structural equation model, the sample size should be more than 10 times the number of observed variables (Bentler & Chou, 1987). Meanwhile, the sample size should also be more than 200 to obtain stable estimation results (Loehlin, 1992). Therefore, the questionnaires obtained in this study can meet the needs of the data analysis. The specific sample conditions are shown in Table 1.

Analysis of Normality and Common Method Bias
In this study, kurtosis and skewness tests were used to analyze the normality of the data, and the results showed that there was no deviation value, which proved that the data collected from the questionnaire have a normal distribution. At the same time, we tested several methods, including Harman's single-factor test, the partial correlation method, and the multitrait-multimethod model, to avoid common method biases (Podsakoff et al., 2003). All the results showed that there was no common method bias in the data, meaning the data were suitable for further analysis.

Exploratory Factor Analysis
In this study, exploratory factor analysis was used to test the applicability of each measurement item in the scale (Kolar & Zabkar, 2010;. Both the principal component method and the maximum variance rotation method were used to obtain the loading factors. The outcome of the Kaiser-Meyer-Olkin (KMO) test is 0.843, which means that the data are suitable for factor analysis. The final results are shown in Table 2. The factor loading values of all items were greater than 0.6 after factor rotation, and the items for each  variable were all gathered together (see Table 2), indicating that the data had a good construct validity.

Reliability and Validity Analysis
In the reliability analysis, Cronbach's α was selected in this article (Lee et al., 2017). The results showed that Cronbach's α values of all dimensions were greater than .7 (see Table 3), indicating good data consistency, which could be further analyzed.
In the validity analysis, the convergent validity and discriminant validity were mainly tested (Bagozzi & Yi, 1988;Fornell & Larcker, 1981;Xu & Li, 2016). The convergent validity was tested by measuring the standardized factor loading, composite reliability (CR), and average variance extracted (AVE). The results showed that the standardized factor loading of each measurement variable was greater than 0.7, the CR value was greater than .6, and the AVE was greater than 0.5 (Table 3), which conform to the standards proposed by Hair et al. (1998). Fornell and Larcker stated that if the root mean square of AVE value of a variable is greater than its correlation coefficients with other variables, the variable has a good discrimination validity (Fornell & Larcker, 1981). As shown in Table 4, the discrimination validity of each variable conforms to the criterion.

Model Fit Analysis
In this study, the AMOS17.0 (International Business Machines Corporation, New York, NY, USA) was used to test the model fit. The results are shown in Table 5. All  indicators meet the criteria, which shows that the model fit is good enough for further analysis.

Path Analysis
The results of path analysis are shown in Table 6. The standardized path coefficients between the perceived image and WOM was 0.428 (p = .000). This indicated that the tourists' perceived images of Southern Xinjiang had a significantly positive influence on their WOM concerning Southern Xinjiang. Hence, the result supports H1. The standardized path coefficient between the perceived image and psychological distance was −0.302 (p = .000). This indicated that the tourists' perceived images of Southern Xinjiang had a significantly negative influence on their psychological distance from Southern Xinjiang. Hence, the result supports H2.
The standardized path coefficients between WOM and the behavioral intention was 0.256 (p = .000). This indicated that the tourists' WOM concerning Southern Xinjiang had a significantly positive influence on their behavioral intention. Hence, the result supports H3. The standardized path coefficients between the psychological distance and behavioral intention was −0.560 (p = .000). This indicated that the tourists' psychological distances from Southern Xinjiang had a significantly negative influence on their behavioral intentions. Hence, the result supports H4. The results of path analysis are shown in Figure 3.

Mediating Effect Analysis
In this study, the method of the product of coefficients and bootstrapping (Hayes, 2009) were used to jointly test the  Note. The value on the diagonal represents the root mean square of AVE, and the correlation coefficients between variables are below the diagonal. PEI = perceived image; WOM = word-of-mouth; PSD = psychological distance; INT = behavioral intention; AVE = average variance extracted. Note. CMIN/DF = minimum discrepancy divided by its degrees of freedom; GFI = goodness of fit index; RMR = root mean square residual; RMSEA = root mean square error of approximation; TLI = Tucker-Lewis coefficient.
mediating effect of WOM and psychological distance (see Table 7). The product of coefficients judges the significance of the mediating effect by comparing the absolute value of the Z value with 1.96. The bootstrapping method determines the significance of the mediating effect by judging whether the confidence interval of the mediation effect includes zero. The results show that the absolute value of the mediating effect of WOM is greater than 1.96. In addition, the confidence interval does not include zero. This indicated that tourists' WOM concerning Southern Xinjiang significantly mediated the relationship between the perceived image and behavioral intention. Hence, the result supports H5. The absolute value of the mediating effect of the psychological distance is greater than 1.96. In addition, the confidence interval does not include zero. This indicated that tourists' sense of psychological distance from Southern Xinjiang significantly mediated the relationship between the perceived image and behavioral intention. Hence, the result supports H6.

Moderating Effect Analysis
The article used the structural equation model, with product terms, to test the moderating effects (Wan et al., 2018). The results showed that the standardized path coefficient of the interaction term in the relationship between the perceived image and WOM was 0.171 (p = .001). This indicated that tourists' WOM concerning Southern Xinjiang significantly moderated the relationship between the perceived image and psychological distance. Hence, the result supports H7. The standardized path coefficient of the interaction term in the relationship between the psychological distance and behavioral intention was 0.094, (p = .046). This indicated that tourists' WOM concerning Southern Xinjiang significantly moderated the relationship between the psychological distance and behavioral intention. Hence, the result supports H8.

Discussion
WOM plays a mediating role in the influence mechanism of the perceived image in tourists' behavioral intention.
Tourists' WOM concerning Southern Xinjiang significantly mediated the relationship between the perceived image and behavioral intention. First, the result shows that the perceived image is an important predictor of tourists' WOM behavior, which verified the results of many previous studies (Anggraeni, 2015;Lban et al., 2015;Wang et al., 2018). As far as this study is concerned, the perceived image is the subjective memory that tourists have already formed, which has a direct effect on WOM. This means that managers should be committed to improving the image of Southern Xinjiang. Specifically, it is necessary to pay attention to creating a good social environment to enhance the emotional image that matches the self-image of tourists (H. Zhang, 2018). Meanwhile, the result shows that WOM is an  important predictor of tourists' behavioral intentions, which is consistent with previous studies (Cham et al., 2016;Hossain & Kim, 2018;Jalilvand et al., 2013). This means that WOM behavior has an important impact on premium behavior and willingness to revisit, which suggests that incentive WOM marketing should be employed to promote tourists' willingness to revisit. In addition, WOM plays a mediating role in the relationship between the perceived image and behavioral intention. This article creatively demonstrates the feasibility of using WOM as a mediating variable, which enriches WOM theories that, in previous papers, only considered WOM behavior as a result variable (Anderson, 1998;Berger & Schwartz, 2011;Cheung & Lee, 2012;Chu et al., 2019;Karimi Alavijeh et al., 2018). This suggests that we can conduct a holistic study of the antecedents and consequences of tourists' WOM behavior, which contributes to the improvement of destination marketing and management. WOM moderates the relationship between the perceived image and psychological distance, and the relationship between the psychological distance and tourists' behavioral intention. This article creatively studies the role of WOM as a moderating variable. Specifically, the moderating effect of WOM plays two roles in the mechanism of the perceived image on the behavioral intention: the promotion mechanism and the repression mechanism, which extended existing WOM models (Anderson, 1998;Chu et al., 2019;Karimi Alavijeh et al., 2018) and found a new mechanism of WOM. The first is the promotion mechanism, which means that tourists' WOM concerning Southern Xinjiang significantly and positively moderates the relationship between the perceived image and psychological distance. That is to say, WOM can promote the influence degree of the perceived image in reducing tourists' psychological distance from Southern Xinjiang. This means that frequent and detailed WOM behavior can significantly reduce tourists' psychological distance. The second is the repression mechanism, which means that tourists' WOM concerning Southern Xinjiang significantly and positively moderates the relationship between the psychological distance and behavioral intention. That is to say, WOM can suppress the expansion of the negative impact of the psychological distance. This means that, although tourists have a certain degree of psychological distance from the destination, their behavioral intention can still be improved by WOM. For Southern Xinjiang, such a stigmatized crisis tourist destination, the promotion mechanism and repression mechanism effectively guide travel agencies and destination managers in grasping the opportunity to encourage WOM behavior. Ultimately, this may minimize the negative impact of the perceived psychological distance. This is a new finding which is an important supplement for previous research, such as  and Xu et al. (2019).
Psychological distance plays a mediating role in the influence mechanism of the perceived image on tourists' behavioral intention. Tourists' psychological distance significantly mediates the relationship between the perceived image and behavioral intention, which is consistent with previous studies Zhou et al., 2015). This means that, after forming an image of a destination, tourists will first form a psychological distance judgment, which, in turn, influences their decision-making. This further confirms the important role of psychological distance in a crisis tourist destination, such as Southern Xinjiang . First, the perceived image is an important antecedent variable relating to tourists' psychological distance, which is consistent with previous studies . Specifically, tourists' perceived images of Southern Xinjiang have a significant negative impact on their sense of psychological distance from Southern Xinjiang, which means that, as the perceived image of the tourist destination is improved, the tourists' sense of psychological distance will gradually decrease. Second, the result shows that tourists' sense of psychological distance from Southern Xinjiang has a significant negative effect on their travel intentions relating to Southern Xinjiang, which is consistent with previous studies (Hamilton, 2014;Tan & Chang, 2015). Finally, we should keep it in mind that the mediating role of the psychological distance is moderated by WOM. That is to say, encouraging WOM behavior can maximize the mediating role of the psychological distance.

Conclusion
In this article, we explored the mechanism of tourists' perceived images on their decisions in relation to traveling to Southern Xinjiang. The final conclusions were obtained via common method bias analysis, exploratory factor analysis, a reliability test, a validity test, model fit analysis, and path analysis. We found that WOM played a mediating role in the relationship between the perceived image and behavioral intention. As a moderating variable, WOM played an important role in the promotion mechanism and repression mechanism. Tourists' sense of psychological distance significantly mediated the relationship between the perceived image and behavioral intention. The findings in this article provide enlightening insights for the sustainable development of marketing activities in Southern Xinjiang.

Theoretical Implications
The findings in this article enrich WOM theories. That is to say, WOM can act as a mediating variable and a moderating variable, not just as a result variable. As a mediating variable, WOM played a mediating role in the relationship between the perceived image and behavioral intention. As a moderating variable, WOM played an important role in the promotion mechanism and repression mechanism. The findings also enrich the marketing theories for the destinations in crisis. The value of tourists' WOM behavior cannot be overemphasized, which should be the focus of future research in the era of self-media. Meanwhile, the findings in this article illustrate the mediating effect of the psychological distance, which is moderated by WOM. This expands the research perspective relating to WOM and psychological distance.

Practical Implications
The empirical results of this article show that WOM plays an important role in crisis tourist destinations. That is to say, WOM marketing is a good way for the destinations in crisis to restore their image. It may shed light on the marketing strategies and tactics for these destinations. The tourism marketing authorities could adopt a new marketing idea. Therefore, the following policy recommendations are made. First, a variety of marketing strategies should be launched to promote tourists' WOM behavior. Second, the travel agencies in Southern Xinjiang can establish tourist files to strengthen the management of tourist relations and keep in touch with tourists through SMS and email. Third, tourism operators can provide bonuses, gifts, coupons, and other material incentives to encourage tourists to spread WOM, like the activity "buy one ticket, gain a lifetime of free tours." Fourth, in the context of highly developed social media, tourist destination managers and operators should actively use the virtual communities formed by people to improve the popularity of tourists' WOM. For example, in cooperation with "influencers," a series of topic discussions could be launched under the organization of opinion leaders to guide tourists to conduct WOM behavior. Fifth, various measures should be taken to reduce the tourists' sense of psychological distance from Southern Xinjiang. The managers and travel agencies in Southern Xinjiang should adopt various measures to increase interaction with tourists and reduce their psychological distance. The tourism marketing authorities should keep up with the new internet and self-media marketing trends. They can use Tik Tok, Micro-Blog, WeChat, and other means to increase the information disclosure of historical traditions and cultural practices in Southern Xinjiang. In addition, they can provide training for tourism practitioners to enhance their language ability and service consciousness.

Limitations and Further Research
The study still has some limitations. First, the convenient sampling method was adopted in this article because Southern Xinjiang covers an area of more than 1,080,000 km 2 . Hence, the findings need further verification. Second, this article found the promotion mechanism and repression mechanism of WOM. However, its robustness in relation to other tourist destinations in crisis, similar to Southern Xinjiang, still needs to be tested. In the future, we will select other destinations in crisis to verify the robustness of this model and compare the difference between them. And we will find out the specific antecedents and consequences of tourists' WOM behavior. We will also test whether other potential tourists could form a strong willingness to travel under the influence of such WOM behavior. Then, which marketing incentives can be used to motivate these potential tourists to turn their willingness to travel behavior.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Social Science Foundation of China (NSSFC), grant number 15BMZ052; Shandong University Multidisciplinary Research and Innovation Team of Young Scholars, grant number 2020QNQT019.