The Role of Social Capital for Short-Video Platform Users’ Travel Intentions: SEM and Fsqca Findings

: This study examined the antecedents of travel intentions in the context of Chinese short-video platforms. Based on a review of the literature on travel intentions, we proposed an integrated model containing determinants of social capital and an elaboration likelihood model. In total, the data from 496 valid questionnaires were analyzed through structural equation modeling (SEM) and fuzzy-set qualitative comparative analysis (fsQCA). The SEM ﬁndings revealed that the determinants of social capital inﬂuenced the characteristics of tourist-generated content (TGC), which in turn a ﬀ ected users’ travel intentions. Homophily and interpersonal inﬂuence were direct antecedents of travel intentions, whereas tie strength non-signiﬁcantly a ﬀ ected travel intentions. The fsQCA results revealed four conﬁgurations of the determinants of social capital and TGC characteristics required to achieve a high level of travel intention. Source credibility was discovered to be a necessary but not su ﬃ cient condition for travel intention. These ﬁndings o ﬀ er insights for both academics and tourism marketers.


Introduction
Since its inception two decades ago, social media has revolutionized communication and enabled products and services to be marketed beyond geographical boundaries [1]. These changes have had a profound impact on the sustainable development of tourism industry [2][3][4]. Specifically, social media enables tourists to share their travel experiences at any time and place through text, pictures, and video [5]. Potential tourists can use social media to obtain destination information to aid their decision making [6]. The advent of social media has helped to develop the symbiotic relationship between former and potential tourists [5,7].
In the earlier stages of social media, tourists searched for travel-related information on either social networking sites (e.g., Twitter and Weibo) or travel agent platforms (e.g., ctrip.com, mafengwo.cn, and tuniu.com) before deciding on their holiday [8,9]. Information on the aforementioned platforms was typically in the format of text and photographs, and such information has attracted much scholarly attention [5,6,[10][11][12]. However, video platforms have become popular due to the widespread use of smartphones; of these platforms, short-video platforms have been especially popular in China since their explosive growth in 2016. Such platforms are valuable as they allow users to quickly and easily disseminate, consume and share a wealth of timely information without the need to obtain the original author's permission [13,14]. Recent industry research reports have indicated that short-video platforms had, up to 2018, been used by 501 million Chinese people [15]. TikTok and Kwai have approximately 200 million active users each and belong to the first echelon of platforms in the short-video industry. The next most popular platforms Watermelon video and Volcano video have 67 million and 50 million users, respectively [16].
Short-video platforms feature dynamic media that inspire users to capture and share memorable moments by creating videos lasting from a few seconds to a few minutes [13]. Sharing these videos enables users to maintain or expand their social network, as is the case with other social media platforms [17]. As the information contained in these videos approximates real-world stimuli [18], potential tourists can obtain a much greater telepresence from short-video platforms than from text and pictures in other types of social media. Consequently, both tourists and travel agents have recently developed communication strategies and content for short-video platforms; however, potential tourists perceive the content created by tourists as being more reliable than that created by travel agents [19]. Tourism organizations thus require a greater understanding of the tourist-generated content (TGC) displayed on short-video platforms if they are to develop more efficient strategies for the business sustainability.
Short-video platforms differ from other social media platforms in that the majority of their content is video; however, all types of social media platforms have the same essential function: to enable users to socialize with others and to obtain social capital. Social capital is considered an instantiated informal norm that induces cooperation between two or more individuals [20]. Everett [21] argued that the resources and power gained from social networks also constitute social capital. Although researchers have identified social capital as an indirect driver of travel intentions, through the common bonds and common identities forged in the context of social media [22], few studies have shed light on the influence of social capital on the characteristics of TGC and users' travel intentions. Accordingly, in the current study, we integrated the concept of social capital into the elaboration likelihood model (ELM) to investigate the antecedents of tourist travel intentions in the context of short-video platforms.
Rather than examining only the net effect of the determinants of social capital and TGC characteristics on travel intentions through structural equation modeling (SEM), we also employed qualitative comparative analysis. SEM is a commonly applied quantitative approach in the area of information systems (IS) research. In SEM, a symmetric relation between concepts is assumed, which leads to difficulty in determining how a combination of notions results in a particular outcome [23]. To overcome such limitations, Ragin [24,25] introduced the approach of fuzzy-set qualitative comparative analysis (fsQCA), which can provide combinations of variables and outcomes, and has attracted considerable attention from academics in the IS filed [26,27]. To the best of our knowledge, this research is the first to combine SEM and fsQCA for investigating individuals' travel intentions in the context of social media. Combining SEM and fsQCA, we revealed general patterns and several equifinal configurations resulting in travel intentions; we also determined the necessity or sufficiency of configurations if the travel intention is to be high. Hence, the following research questions were addressed: (1) Which determinants of social capital are associated with the characteristics of TGC and user travel intentions in the context of short-video platforms?
(2) How do the characteristics of TGC lead to user travel intentions in the context of short-video platforms?
(3) What combinations of determinants of social capital and TGC characteristics result in travel intentions in the context of short-video platforms?
The rest of this paper is organized as follows. In Section 2, we introduce the characteristics of TGC and the determinants of social capital and develop the conceptual model and associated research hypotheses. In Section 3, we detail the methodology of data collection and measurement. In Section 4, we present the SEM and fsQCA results. Finally, in Section 5, we discuss the research findings, the theoretical and practical implications, the limitations, and the future research directions.

Literature Review
Behavioral intention refers to an individual's attitudinal response to a target [28], and it has been viewed as a significant predictor of personal decisions and actions [29]. In the tourism context, an individual with a stronger intention to visit a destination is more likely to actually visit the destination [30]. Accordingly, understanding travel intention and its antecedents is crucial [31]. Travel intention is currently regarded as the subjective travel intention stimulated by external information and internal needs [32]. Figure 1 illustrates the conceptual model positing that short-video platform users' social capital-including their tie strength, homophily, and interpersonal influence-affects the argument quality and source credibility of received information, which, in turn, influences the travel intentions of these users.

Literature Review
Behavioral intention refers to an individual's attitudinal response to a target [28], and it has been viewed as a significant predictor of personal decisions and actions [29]. In the tourism context, an individual with a stronger intention to visit a destination is more likely to actually visit the destination [30]. Accordingly, understanding travel intention and its antecedents is crucial [31]. Travel intention is currently regarded as the subjective travel intention stimulated by external information and internal needs [32]. Figure 1 illustrates the conceptual model positing that short-video platform users' social capital-including their tie strength, homophily, and interpersonal influence-affects the argument quality and source credibility of received information, which, in turn, influences the travel intentions of these users.

Characteristics of Tourist-Generated Content
TGC is typically examined within the broader framework of user-generated content (UGC). Perez-Vega, Taheri, Farrington and O'Gorman [3] defined TGC as when "tourists share their own evaluations of (prior) destinations, tourism offerings, and services experience." Tourists generate content for various purposes; some do so to show a personal appreciation of an experience, service, or product, to feel socially accepted, to demonstrate personal knowledge, or to attain fame or recognition [33,34], whereas others share the content to inform potential tourists [5,35]. Due to the characteristics of TGC (being up-to-date, organic, reliable, impartial, and enjoyable), TGC has been found to more powerfully affect potential tourist attitudes than does firm-sponsored content [36][37][38]. Consequently, potential tourists tend to obtain detailed reviews of the experiences, travel agencies, destinations, and hotels from others when they make decisions [39,40].
Petty and Cacioppo [41] proposed the ELM to understand changes in personal attitudes. The ELM comprises peripheral and central routes, and has been widely used in the fields of social psychology, marketing, electronic commerce, and information technology [42,43]. The peripheral route relates to simple, heuristic-based cognition of the target behavior, whereas the central route relates to critical cognition, where task-related arguments are weighed based on their merits before a person forms a judgement on the target behavior. As reported by Mak, et al. [44], source credibility and argument quality constitute the major peripheral and central routes, respectively.

Source Credibility
Source credibility refers to "the extent to which [an] information source is perceived to be believable, competent, and trustworthy by information recipients" [45]. To obtain relevant information, individuals rely on online tourism reviews, namely airline and hotel reviews, from

Characteristics of Tourist-Generated Content
TGC is typically examined within the broader framework of user-generated content (UGC). Perez-Vega, Taheri, Farrington and O'Gorman [3] defined TGC as when "tourists share their own evaluations of (prior) destinations, tourism offerings, and services experience." Tourists generate content for various purposes; some do so to show a personal appreciation of an experience, service, or product, to feel socially accepted, to demonstrate personal knowledge, or to attain fame or recognition [33,34], whereas others share the content to inform potential tourists [5,35]. Due to the characteristics of TGC (being up-to-date, organic, reliable, impartial, and enjoyable), TGC has been found to more powerfully affect potential tourist attitudes than does firm-sponsored content [36][37][38]. Consequently, potential tourists tend to obtain detailed reviews of the experiences, travel agencies, destinations, and hotels from others when they make decisions [39,40].
Petty and Cacioppo [41] proposed the ELM to understand changes in personal attitudes. The ELM comprises peripheral and central routes, and has been widely used in the fields of social psychology, marketing, electronic commerce, and information technology [42,43]. The peripheral route relates to simple, heuristic-based cognition of the target behavior, whereas the central route relates to critical cognition, where task-related arguments are weighed based on their merits before a person forms a judgement on the target behavior. As reported by Mak, et al. [44], source credibility and argument quality constitute the major peripheral and central routes, respectively.

Source Credibility
Source credibility refers to "the extent to which [an] information source is perceived to be believable, competent, and trustworthy by information recipients" [45]. To obtain relevant information, individuals rely on online tourism reviews, namely airline and hotel reviews, from sources that they perceive to be credible [43]. As a peripheral cue, the source credibility affects the persuasiveness of the received information by influencing the message processing [46]. For individuals with a low elaboration likelihood or weak arguments, the message source credibility has a much greater explanatory power than does the information quality in explaining changes in personal attitude and related behavior [43,47,48]. Empirically, in the context of social media, when the source credibility is perceived to be high, users are more likely to consider reviews in their decision making [49][50][51][52].
In the context of tourism, source credibility enhances the perceived usefulness of suggestions and comments, driving individual behavioral intentions [42,53]. The direct effect of social relationships on the intention to use travel information is enhanced when the credibility of travel information sources in social media is high [54]. In this sense, the source credibility of information has been regarded an indirect driver of behavioral intentions among social media users. Short-video platforms, a form of social media, enable users to post and share personal travel experiences or forward others' shared experiences through short videos, and these videos have become a critical source of information when making travel plans. Based on the preceding analysis, we proposed the following hypothesis: Hypothesis 1 (H1). TGC source credibility positively affects travel intentions.

Argument Quality
Argument quality is viewed as "the persuasive strength of arguments embedded in an informational message" [45]. In the context of this study, argument quality refers to the accuracy, timeliness, comprehensiveness, and relevance of the TGC on short-video platforms [45,49,[55][56][57][58]. The accuracy of arguments and reviews has been demonstrated to be a positive determinant of customer decision making [59,60], and such accuracy reflects how timely the reviews are. Comprehensive information and reviews are also likely to beneficially impact the adoption of new technology services [61]. People tend to assess reviews and arguments more positively when they are more relevant to their needs [62,63].
On short-video platforms, TGC is dynamic and timely, with tourists updating their recent tourism-related experiences [64]. Arguments in TGC are users' firsthand experiences, which can, to a certain extent, guarantee information accuracy and result in potential tourists having a favorable attitude toward the destination or hotel mentioned in the TGC [65]. In addition, short-video platforms, similar to other social media, enable tourists to generate content freely in a manner of their choosing; such content on a typical destination or hotel is thus comprehensive [66]. Furthermore, as there are many users sharing their travel experiences on short-video platforms, the potential tourists are thus always likely to find relevant information regarding the places they wish to visit. Potential tourists using short-video platforms are more reliant on TGC when obtaining high quality information prior to decision making. Based on the preceding analysis, we hypothesized the following: Hypothesis 2 (H2). TGC argument quality positively affects travel intentions.

Social Capital
As UGC is a social phenomenon based on interpersonal communication between consumers, social interaction variables are also likely to play a key role in UGC [67]. Social capital is associated with relationships established through shared values in a social network [68,69]. Social capital is defined as "the sum of the actual and potential resources embedded within, and derived from, the network of relationships possessed by an individual or social unit" [70]. Lochner, et al. [71] viewed social relationship quality to be the essence of social capital; it is foundational to mutually beneficial relationships and a critical factor in driving individuals to collectively solve the problems that they face in common. Social capital is a resource for collective action, with implications for outcomes at the individual and group levels [72]. For short-video platform users, their degree of social capital relates to how much access they have to a reciprocal social relationship through which others' approval can be obtained. Social capital is a user's capacity to interact, organize, cooperate, and participate on a short-video platform.
Social capital can be classified into three essential types: relational, structural, and cognitive capital [70,73,74]. Relational capital is considered an intangible asset, such as interpersonal influence, that is rooted in interpersonal relationships forged through interpersonal interaction [75]. Structural capital refers to the characteristics of a social network, such as the strength of its constituent social ties [74,76]. Cognitive capital is regarded as the extent to which resources offer a widely accepted understanding, and this includes homophily [70]. In summary, relational capital manifests as interpersonal influence, structural capital manifests as tie strength, and cognitive capital manifests as homophily.

Tie Strength
According to Granovetter [77], "the strength of a tie is a combination of the amount of time, the emotional intensity, the intimacy, and the reciprocal services which characterize the tie." In the context of short-video platforms, the strength of a tie (between the sender and recipient) in the flow of information strongly influences information diffusion and persuasion in the communication network. The great influence of tie strength on customer preferences, particularly purchase intentions, has been widely recognized in the context of social networking [78]. Regarding business-to-business industries, Stanko et al. [79] empirically demonstrated that tie strength indirectly affected user behavioral intentions through the mediator of a commitment to the seller. Additionally, tie strength was confirmed to be a crucial motivation behind the intention to share knowledge [73] or transmit information [80]. Based on the preceding analysis, we hypothesized the following:

Hypothesis 3 (H3). Tie strength among short-video platform users positively affects travel intentions.
Many studies have investigated the relationship between tie strength and TGC. Lu and Yang [72] reported that social interaction ties were associated with information quality in the context of virtual communities. Trimble O'Connor [81] also demonstrated that individuals with stronger ties were more likely to be familiar with each other's qualifications, which are a source of credibility. Based on the preceding analysis, we hypothesized the following: Hypothesis 4 (H4). Tie strength among short-video platform users positively affects the TGC argument quality.

Hypothesis 5 (H5).
Tie strength among short-video platform users positively affects the TGC source credibility.

Interpersonal Influence
As a factor affecting social capital, interpersonal influence is a critical element affecting individual decision making [33,82,83]. Deutsch and Gerard [84] developed a dual-process theory and categorized interpersonal influence into two types based on how they affect individual judgement. The first type is informational influence, which leads one to accept information from the other person as true [85,86], and the other type is normative influence, which pressures individuals to conform to the positive expectations of others [86,87].
In online settings, behavior and opinion are associated with the satisfaction of a self-defining relationship's normative expectations, where the consumer desires to conform to others' expectations and image of the consumer as well as to group norms more generally [87][88][89]. However, consumers who are subject to informational influence tend to acquire product-related information by observing others or by seeking such information from knowledgeable contacts [85,89,90]. Several researchers have demonstrated consumer susceptibility to interpersonal influence when they are making product purchase decisions [91,92]. Chu and Kim [33] reported that a consumer's susceptibility to normative and informational influence is positively related to word-of-mouth recommendations on social media. Thus, the travel intention of a short-video platform user is higher if the user is more susceptible to informational and normative influences. Based on the preceding analysis, we hypothesized the following: Hypothesis 6 (H6). Informational influence among short-video platform users positively affects travel intentions.
The relationship between normative influence and TGC characteristics remains poorly understood. Individuals under a high level of normative influence tend to participate in activities to feel less lonely, seek others with similar interests, and gain approval and acceptance from others [83,89]. The use of social media is a form of exerting normative influence as users are likely to customize their personal pages to gain peer recognition through frequent interaction with others [83,93]. These actions also improve the stickiness among users, which in turn affects the guanxi [94]. Additionally, reliability, a typical feature of source credibility, can be improved through frequent communication [95]. In the context of short-video platforms, strong guanxi among users can enhance the argument quality and source credibility of the received information. Based on the preceding analysis, we hypothesized the following:

Hypothesis 8 (H8).
Normative influence among short-video platform users positively affects the TGC argument quality.

Hypothesis 9 (H9).
Normative influence among short-video platform users positively affects the TGC source credibility.
Few studies have empirically examined the association between informational influence and TGC characteristics. In the context of social media, individuals with a higher degree of informational influence were more likely to engage with and share content to obtain useful information from others [83]. Additionally, source credibility may be higher when a message comes from a user who always shares useful information [96]. Therefore, we hypothesized the following: Hypothesis 10 (H10). Informational influence among short-video platform users positively affects the TGC argument quality.

Hypothesis 11 (H11).
Informational influence among short-video platform users positively affects the TGC source credibility.

Homophily
In sociology, homophily refers to the degree to which individuals associate and bond with similar others [97], and this has been regarded as being critical to communication effectiveness [98]. The concept of similarity has been frequently adopted as a synonym or form of homophily [99,100]. That is, the homophily effect is consistent with the similarity effect in an online environment [101]. Similarity is defined as "a match in demographic characteristics or in psychographic traits" [102]. People with similar beliefs and traits (e.g., physical, cultural, and attitudinal characteristics) have more interactions with one another and feel more comfortable due to their similar background [103][104][105][106][107]. Accordingly, on social media, people prefer to interact with others who share similar sociodemographic, behavioral, and intrapersonal characteristics [108][109][110]. Therefore, information exchange occurs more easily among homophile users than among heterogeneous users [33,109].
Empirically, the attitudinal disposition [111] and demographic characteristic [112] aspects of homophily affect the communication effectiveness. As reported by Rogers et al. [113], individuals are more likely to decode the common meanings of exchanged information if the two communicators have a high level of homophily. That is, when a user perceives that an information provider is similar to them, the message is likely to be more persuasive to the receiver [114]. Filieri et al. [115] argued that homophily is a positive driver of the intention to purchase products recommended by others. Additionally, Zhang et al. [116] identified a significant positive effect of perceived homophily on electronic word-of-mouth recommendations and purchase intentions. Following this vein, users with a high level of homophily can be reasonably predicted to increase others' travel intentions in the context of short-video platforms. Hence, the following was postulated:

Hypothesis 12 (H12). Homophily among short-video platform users positively affects travel intentions.
Additionally, studies have suggested that homophily is a sufficient factor to decrease vulnerability, upgrade correct sentiments, and strengthen the health of interpersonal relationships, in turn improving the argument quality [49]. An argument has greater persuasive power if the message receiver perceives the message sender as being similar to them [100]. Kim, Kandampully and Bilgihan [109] argued that homophily is associated with source credibility in the social media context. Similarly, when the level of perceived homophily between the sender and receiver is high, the perceived credibility of the message is high [50,117,118]. From this perspective, the homophily among short-video platform users may significantly influence the argument quality and source credibility. Hence, we hypothesized the following: HHypothesis 13 (H13). Homophily among short-video platform users positively affects the TGC argument quality.

Data Collection and Sample Profile
An online self-administrated questionnaire was distributed through Sojump (www.sojump.com), a commonly used professional online survey platform in China. With convenience sampling adopted, the hyperlink to the survey was distributed on three social media platforms-namely, WeChat, QQ, and Weibo-over a 3-week period from October 5th to October 25th in 2019. Two screening questions were set at the beginning of the questionnaire: 1) "Have you used a short-video platform (i.e., TikTok, Kwai, or Huoshan) in recent weeks?" and 2) "Have your contacts on short-video platforms posted any information on travel?" Those who answered "no" to either question were told to stop the survey. Initially, it collected 826 respondents, but 37 and 293 respondents responded "no" for the first screening question and the second screening question respectively. Hence, a total of 496 usable responses were retained to the further data analysis. Table 1 details the sample profile and the respondents' general pattern of using short-video platforms. Male and female users accounted for 44.2% and 55.8% of the sample, respectively. The main age groups were 31-40 (41.3%), 18-25 (25.2%), and 26-30 (24.0%). The majority (67.7%) had a bachelor's degree. Thus, the respondents were primarily young and well educated. As for short-video platform usage, most of the participants were experienced users of short-video platforms, having used them for more than 3 years (36.7%). Specifically, 19.8% of the respondents browsed short videos for at least 2 h daily, whereas 54.8% did so for 1-2 h; 25.4% browsed short videos for less than 1 h. Most of the respondents watched short videos 4-9 times daily, whereas 40.1% did so 1-3 times daily. Approximately 47.8% of the respondents posted a short video 1-3 times daily. A surprisingly high proportion of 46.1% of the respondents reported visiting a destination that had been recommended on a short-video platform within the preceding year.

Measurements
The measures of the constructs were adapted from those in previous studies, with some minor adaptations made for this study's context of short-video platforms (Appendix A). The seven constructs were tie strength, homophily, normative influence, informational influence, source credibility, argument quality, and travel intention. Specifically, we used three items from Mittal et al. [119] to measure the tie strength; three items from Chu and Kim [33] to measure the homophily; three items from Chu and Kim [33] to measure normative influence and informational influence; four items from Wu and Shaffer [120] to measure the source credibility; and four items from Bailey and Pearson [121] to measure the argument quality. The scale for measuring the travel intentions developed by Chaulagain et al. [122] was employed. The items were scored on a 7-point Likert scale ranging from strongly disagree (1) to strongly agree (7).
The questionnaire was initially formulated in English and subsequently translated to Chinese using the back-translation method to ensure equivalence in meaning [123]. The Chinese questionnaire was discovered to have no major problems with understanding or wording in a pretest involving 30 short-video platform users. The questionnaire was revised only slightly based on their responses.

Results
The outliers must be detected and the normality must be evaluated prior to conducting the SEM. No outliers were found as the value of the Mahalanobis distance indicated acceptable and progressive distances from the centroid [124]. The skewness and kurtosis scores of each item were also acceptable [125]; Appendix B.

Measurement Model
Following the two-step approach suggested by Anderson and Gerbing [126], a maximum likelihood confirmatory factor analysis was performed with the raw data using AMOS 24.0. The model fit indices indicated that the measurement model was acceptable (χ2/degrees of freedom (df) = 1.031 (193.742/188), goodness-of-fit index (GFI) = 0.967, adjusted GFI (AGFI) = 0.955, root-mean-square error of approximation (RMSEA) = 0.008, Tucker-Lewis index (TLI] = 0.998, and confirmatory fit index (CFI) = 0.999). Table 2 presents the results for concept reliability. The unstandardized coefficients for all items were significant. As the standardized factor loading (SFL) of SC2 was relatively low, this item was removed. Two significant indicators of item reliability, SFL and the standardized multiple correlation coefficient (SMC), ranged from 0.612 to 0.852 and 0.375 to 0.726, respectively, and thus exceeded the requisite minimum values of 0.6 and 0.3, respectively [127]. The values of composite reliability (CR) considerably exceeded the requisite minimum of 0.7 [128]. Additionally, the values of average variance extracted (AVE) were between 0.491 and 0.653, exceeding the requisite minimum of 0.36 suggested by Chin [129]. These results demonstrated the high reliability and convergent validity of all constructs. The square roots of the AVE of the constructs were also higher than the coefficients of correlation between any two concepts (Table 3). Thus, the constructs had a satisfactory discriminant validity.

Common-Method Bias
When collecting data using a structured questionnaire, complete reliance on a single source can cause the common-method bias [130]. To address this potential problem, a Harman single-factor test was performed [131]. The results indicated a poor model fit (χ2/df = 11.639 (2432.624/209), GFI = 0.657, AGFI = 0.585, RMSEA = 0.147, TLI = 0.458, and CFI = 0.510). Thus, the common-method bias was not a cause for concern.

Reanalysis of the Data by Using fsQCA
The SEM analysis in the previous section concentrated on the net effects of the determinants of travel intention. Conventional statistical approaches, namely multiple regression analysis and SEM, have a limited ability to explain complex interaction effects on a particular outcome [24,132]. By contrast, fsQCA can accommodate complex complementarities and nonlinear relationships among variables [23,132].
fsQCA is an analytical technique that combines fuzzy-set theory with Boolean logic, which is commonly employed to analyze asymmetric data and determine those combinations of conditional variables that lead to an outcome [133]. The necessary and sufficient causal conditions of an outcome can be distinguished using fsQCA [133]. Accordingly, a configurational model of conditions affecting short-video platform users' travel intentions was analyzed using fsQCA.

Calibration
Six conditional variables (tie strength, homophily, normative influence, informational influence, argument quality, and source credibility) and one outcome concept (travel intention) were employed in our fsQCA analysis. As reported by Ragin [24], fsQCA requires the raw data to be transformed into values ranging from 0 to 1 (indicating full nonmembership to full membership in a fuzzy set). As all the concepts were unidimensional, the summed measures were constructed by summing the measurement items for all constructs [134]. The quartiles method was used to define the three anchors to be used for calibrating the fuzzy sets (Table 5). Note: Intention to travel (IT), argument quality (AQ), source credibility (SC), informational influence (II), normative influence (NI), homophily (HO), and tie strength (TS).

Configurations for the Existence of Travel Intention
Following the suggestion of Ragin [24], a truth table with 2k rows (k = number of conditions) was constructed using an fsQCA algorithm. The frequency threshold employed was 11, which was higher than the recommended value of 3 in situations with more than 150 responses [133]. Additionally, the minimum consistency was 0.8, exceeding the suggested value of 0.75 [135]. In addition, the intermediate solutions for travel intentions were analyzed because these solutions contained simplified assumptions [24]. Table 6 presents the fsQCA results for users with high travel intentions. The results suggested four equifinal configurations that were correlated with high travel intentions: tie strength × homophily × argument quality × source credibility + tie strength × normative influence × informational influence × argument quality × source credibility + homophily × normative influence × informational influence × argument quality × source credibility +~tie strength × homophily × normative influence ×~informational influence ×~argument quality × source credibility → travel intentions.
The consistency of each solution and the overall solution was higher than the suggested threshold of 0.75 [135]. The raw coverage for each solution ranged from 0.089 to 0.373, indicating that each combination could explain between 8.9% and 37.3% of the respondents' travel intention. In addition, the overall solution coverage was 0.467, indicating that 46.7% of travel intention was covered by the four configurations. Our examination of these four combinations revealed that source credibility was a necessary but not sufficient condition for travel intention. Hence, travel information from credible sources may not necessarily increase users' travel intention. Furthermore, no other variable was necessary or sufficient for the formation of travel intention.

Discussion and Implications
In the tourism industry, social media has become a critical channel through which potential tourists obtain travel information when making travel plans. The tourism industry cannot survive if it does not understand the process of how travel intentions are formed in the context of social media. The popularity of short-video platforms has skyrocketed in recent years, and these platforms offer a new channel through which users can share their travel experiences and seek information on potential destinations. Hence, investigation of the most major elements resulting in short-video platform users' travel intentions is essential for both academics and practitioners. Research on travel intention has largely failed to integrate the determinants of social capital and the characteristics of TGC. In the current study, we examined the theoretical linkage among the determinants of social capital, the characteristics of TGC, and short-video platform users' travel intentions using SEM and fsQCA.
The SEM findings revealed that four determinants of social capital (tie strength, homophily, informational influence, and normative influence) significantly affected the characteristics of TGC (argument quality and source credibility). Both argument quality and source credibility directly influenced short-video platform users' travel intentions. In terms of the direct relationship between the determinants of social capital and travel intentions, homophily, information influence, and normative influence, but not tie strength, were significant drivers of travel intention. Scholars have demonstrated that structural capital influences user behavioral intentions through the mediator of relational capital [136]. Hence, tie strength, as structural capital, may influence short-video platform users' travel intentions through the mediator of interpersonal influence.
The fsQCA results demonstrated that four configurations of causal conditions led to users' travel intentions in the context of short-video platforms. Source credibility was confirmed as a necessary but not sufficient condition for travel intention. The remaining conditions were neither necessary nor sufficient for a high travel intention. The results were consistent with the SEM finding that source credibility most significantly affected travel intention in short-video platform users. In addition, tie strength was a causal condition in two solutions (Configurations 1 and 2), which differed from the SEM result that tie strength did not significantly affect travel intentions. However, the SEM results also revealed that tie strength was a significant antecedent of argument quality and source credibility, which in turn affects travel intention. That is, tie strength indirectly affects travel intention through argument quality and source credibility. These findings have academic and practical implications.

Research Implications
The results make contributions for academics in the field of sustainable marketing. Consistent with other studies [137], the present study confirmed that elements of social capital were significant drivers of users' behavioral intentions. However, previous studies on travel intention have reported that social media users' social capital affects their travel intention through them having a shared bond and identity [22]. The current study extends the literature and argues that the TGC characteristics can be viewed as major drivers of travel intention. The determinants of social capital have been confirmed as being antecedents of the TGC characteristics. Additionally, studies on social media users' travel intention have largely been conducted in the context of social networking sites or professional tourism platforms [8,9]. The present study extended the range of research contexts to short-video platforms-which have rapidly developed in China-deepening our understanding of the behavioral intentions of short-video platform users.
The current study also demonstrated the usefulness of fsQCA for identifying combinations of causal conditions leading to travel intentions in the context of short-video platforms. This use of fsQCA stands in contrast to the common use of traditional statistical methods (such as SEM) in the literature to explore isolated antecedents of individual travel intention. fsQCA can capture causal complexity: It enables the analysis of configurations of determinants of social capital and TGC characteristics that create travel intentions, in the context of short-video platforms [138]. The fsQCA results revealed four equifinal causal solutions to short-video platform users' travel intentions; these are resulting that conventional methods cannot achieve. In addition, although the fsQCA findings are consistent with those of many SEM-based studies [54], this study offers a more comprehensive combination of causal conditions through its use of fsQCA.
The central role of source credibility in predicting individual behavioral intentions has been emphasized in the literature [43,47,48]; however, the fsQCA findings emphasize that source credibility is a necessary but insufficient condition for users' travel intention in the context of short-video platforms. Additionally, the current study revealed an asymmetry between source credibility and travel intentions. The other causal conditions included in the conceptual framework were neither necessary nor sufficient drivers of short-video platform users' travel intentions.

Practical Implications
Our results have optimistic implications on sustainable tourism for tourism marketers. Marketing Intelligence. We discovered that short-video platforms have become a highly essential network for travel information acquisition in China. Numerous posts and reposts on travel experiences on short-video platforms have greatly increased the flow of information to short-video platform users. Hence, the short-video platforms could be viewed as an innovative knowledge management system, the employees in tourism businesses can search for the knowledge from the client networks involved in short-video platforms [139]. Tourism marketers can obtain consumer insights through short-video applications to market their offerings and strengthen customer relations in China. Consumers can easily evaluate travel products through online exchanges of information, helping them reduce their uncertainty and secure lower prices. Despite the high level of consumer susceptibility in Chinese customers, which leads them to seek and pass on valuable and credible information, marketers should take care not to rely too much on short-video platform data when develop marketing strategies.
Strategic Marketing Segmentation. This study confirmed the significance of tie strength and homophily to user travel intentions in the context of short-video platforms. Although the homophily and tie strength effects may cause information to be transmitted around a limited social circle, the research findings suggest that tourism marketers can develop psychographics or lifestyle segmentation strategies as enterprises seek to identify homogenous groups of consumers, with stronger ties between the members, to which they market more efficiently. The sample profile indicated that the short-video platform users were relatively young, hence, the travel agents should fully consider the preference of youths when promoting tourism products in short-video platforms. For instance, young people prefer to take part in sports tourism [140], thus, the tourism organizations could focus on promoting this in the context of short-video platforms.
Electronic Word-of-Mouth Development. The analytical results of this study supported the view that both normative influence and informational influence are crucial antecedents of travel intentions among short-video platform users. In this case, Chinese tourism businesses should develop and maintain the relationship with opinion leaders in short-video platforms, as opinion leaders can generate impacts on the potential tourists from the aspects of norm and information. Additionally, the opinion leaders have a large number of followers, which means that their posts could reach greater audiences. Our findings also confirmed that short-video platform users spend a large amount of time daily browsing short videos. Of the 496 respondents, 46.1% had visited at least one destination, and shared their experiences on a short-video platform, in the preceding year. Accordingly, travel agents can encourage tourists to post the short videos for the destinations, which is an effective way to improve the followers' possibility of traveling to the same scenery spots.
Customer Relationship Management. The findings revealed that most of the short-video platform users browsing short videos 4-9 times per day (48.2%), which reflected the stickiness of short-video users is strong. Additionally, Internet users place great trust in companies that have social media accounts. Hence, the tourism business should establish their own short-video platform accounts [141], and maximize the unique potential of short-video platforms to manage online customer relationships. For instance, the employees could respond in a timely way to the users' comments and solve their issues, which could formulate superior relationships between the tourism enterprises and customers [142].

Limitations and Future Research Suggestions
The present study has several limitations that can be addressed in further research. First, convenience sampling was employed to collect data; thus, the sample may have been unrepresentative of the general population, limiting the generalizability of the research findings. Future research can overcome this limitation by using random sampling. Second, this study empirically investigated only Chinese short-video platforms; user behaviors and cultural practices may differ in other countries. In the future, a cross-cultural study can be conducted in different countries. Finally, although this study integrated the determinants of social capital and TGC characteristics to examine the antecedents of short-video platform user travel intentions, numerous potential moderators or control variables may have been overlooked. Future studies can establish a more comprehensive conceptual framework that includes moderators and control variables.

Appendix A. Measurement of Concepts
Argument quality AQ1: The travel information obtained from my contacts on short-video platforms is accurate. AQ2: The travel information obtained from my contacts on short-video platforms is relevant. AQ3: The travel information obtained from my contacts on short-video platforms is complete. AQ4: The travel information obtained from my contacts on short-video platforms is consistent.
Source credibility SC1: The travel information obtained from my contacts on short-video platforms is trustworthy. SC2: The travel information provider on short-video platforms is knowledgeable. SC3: The travel information obtained from short-video platforms is reliable. SC4: The travel information providers on short-video platforms are experts. II3: I frequently gather information from my contacts on short-video platforms about destination before I visit.