**A Study on the E**ff**ects of Crowdfunding Values on the Intention to Visit Local Festivals: Focusing on Mediating E**ff**ects of Perceived Risk and e-WOM**

#### **Hun Kim and Byenghee Chang \***

Department of Journalism and Mass Communication, Sungkyunkwan University, Seoul 03063, Korea; kimhun6301@naver.com

**\*** Correspondence: mediaboy@skku.edu; Tel.: +82-2-760-0687

Received: 6 March 2020; Accepted: 15 April 2020; Published: 17 April 2020

**Abstract:** Information technology is recognized as an important means of expanding the sustainability of local festivals, but most research and practices only focus on existing information technologies such as websites and social network services. This study examines the potential of crowdfunding platforms to ensure the success of local festivals and assesses how emerging information technologies impact the sustainability of the tourism industry. This study proposed four values based on the value theory that is frequently applied in consumer research. We also applied inner innovativeness as a personal characteristic and examined the effects of economic, emotional, social, altruistic, and inner innovativeness regarding film festival crowdfunding on the intention to visit the film festival. We applied perceived risk and the intention to use electronic word of mouth (e-WOM) as mediating variables. As a result, emotional, social, and altruistic values were found to significantly affect the intention to visit film festivals by mediating perceived risk. In addition, the social value was found to have positive effects on the dependent variable through the intention to use e-WOM. The results show that crowdfunding platforms are considered an important tool for promoting the festival. It is also important to develop value in favor of the festival by increasing value through marketing strategies.

**Keywords:** crowdfunding; consumption value; inner innovativeness; perceived risk; e-WOM; the intention to visit festival

#### **1. Introduction**

The sustainability of local festivals is important because they provide benefits to the community. Communities can generate economic income by discovering unique local content [1], and festivals provide an opportunity to externally promote this content [2]. However, festivals held by small cities often face budgeting issues, so it is hard to be active in marketing. In this situation, advances in information technology (IT) may act as another opportunity to promote the sustainability of the local festival industry.

During the period in which the Internet became popular, local tourism websites were used as the main platform to provide travelers with up-to-date tourist information in the region [3]. Many travelers have recently been able to get or share information about local festivals through various IT platforms. Social network services (SNS), which emerged around 2010, are regarded as one of the megatrends influencing tourism systems [4]. These platforms not only allow travelers to search for information [5] but also help them make decisions about visiting tourist attractions or festivals [6]. SNS is also rated as having the ability for travelers to engage with potential customers by sharing comments, videos, photos, and more [4]. Tourists upload their local photos to SNS, and potential customers who view this content through SNS are attracted to the region. Given that potential tourists primarily rely on the

opinions of others, such as word of mouth (WOM), for decision-making [7], SNS are the best method to encourage them to participate in local festivals. From the festival organizers' point of view, SNS help to effectively promote local festivals [8].

The emerging crowdfunding platform is also a valuable tool to promote local festivals. Businesspeople have said that they use crowdfunding to market their projects or communicate with fans or supporters [9]. They can also increase their brand awareness through crowdfunding [10]. Many arts and cultural projects, for example, make heavy use of crowdfunding because in addition to attracting funds, they can also promote funding simultaneously [11]. A notable example is Big Issue Korea, which raised funds through crowdfunding for the Nutcracker, a play in which homeless people played a major role. The press recognized the crowdfunding by Big Issue Korea, which saw great results from the promotion [12].

Crowdfunding can be linked with SNS to maximize the promotion of local festivals [13]. Most crowdfunding sites provide the SNS information of individual operators and include the function to share crowdfunding information. Potential participants can use SNS information to determine various indicators of various crowdfunding, which can have a social impact on participants' support decisions [14]. Individuals spread crowdfunding information to others, and those exposed to it can receive festival and crowdfunding information. Therefore, it is meaningful to consider the effects of electronic word of mouth (e-WOM) and crowdfunding factors on social network services.

While crowdfunding platforms are being assessed as a means of sustainability for local festivals, previous research has focused primarily on the effect of social network services [4,5]. However, these works omit the impact of associations between one or more online services and the intention to visit a tourist destination. Online marketing tools show an important diversity, and consumers are also able to receive tourism information through a range of various online channels. This indicates the importance of investigating the effects of multiple marketing tools simultaneously. Therefore, this study constitutes the framework of study considering not only (1) e-WOM, which is the main effect of social network services, but also (2) the role of a new medium—crowdfunding.

With regard to crowdfunding factors, this study examines the influence of perceived economical, emotional, social, and altruistic values about film festival crowdfunding on the intention to visit the region of film festivals. In addition, the perceived risk of film festival crowdfunding is applied as a mediating variable. Perceived risk is considered a useful variable in exploring the user's perception of uncertainty in accepting new IT services. Therefore, it is a meaningful approach to look at the effects of perceived risks on film festival crowdfunding. Regarding the role of e-WOM, it applies the intention to use electronic word of mouth (e-WOM) as mediating variables. E-WOM was proven to be an effective way to promote tourist destinations at a low cost [15]. The intention to use e-WOM is thus a meaningful variable to examine the effects of crowdfunding as a tool for disseminating information about festivals. In addition, the framework includes inner innovativeness as one of the independent variables. This approach is to consider not only the marketing platform but also the personal characteristic as an explanatory variable.

This research model draws on the frameworks of several studies that examine factors influencing the intention to visit tourist destinations [16,17]. These studies assume that measures of online consumer behavior, such as e-WOM [16] and blog usage [17], mediate explanatory variables relating to the intention to visit tourist destinations. This study proposes an extended framework by applying the explanatory variables and some mediating variables suggested above considering the characteristics of the subject and factors. Thus, this framework examines tourists' perceptions, intentions, and actions as a whole. Given the continuing increase in the impact of information technology services in the tourism industry, it will help to increase sustainability in regional tourism industries with unstable conditions.

#### **2. Literature Review**

#### *2.1. Local Festival and Crowdfunding*

Local festivals provide tourists with a variety of events that utilize cultural elements within the city for a certain period of time. During the festival, the community can introduce the community's local traditions, intangible heritage, natural scenery, and ethnic backgrounds to the tourists [18]. Thus, festivals are meaningful because the community can discover local cultural elements and use them as highly sustainable tourism resources. Local festivals also contribute to increasing the city's demand for travel [19]. Tourists visit the festival area and pay a cost to use various amenities such as transportation, accommodation, and restaurants. Thus, local festivals are also recognized as a means of promoting the local tourism industry and activating the economy [20]. In addition, local festivals not only expose a city as a tourist destination but also influence the re-creation of a city's image [21]. From a city marketing point of view, festivals are an effective approach to strengthening the identity of a city and reversing the negative image of a city [22]. In summary, local festivals are important tourist resources that generate economic benefits and enhance a city's image.

In recent years, local film festivals have become popular as one of the exciting local events. Major cities around the world, such as Cannes, Venice, and Berlin, have held international film festivals for a long time. For example, Cannes, a local small town, established the world's third film festival in 1939, following the Venice International Film Festival (1932) and Moscow International Film Festival (1935). Currently, hosting the most popular international film festival worldwide is the most powerful element defining the identity of Cannes [23]. The Cannes International Film Festival attracts approximately 200,000 tourists, including audiences and professionals, each year, generating 200 million euros' worth of economic benefits [24].

The Cannes International Film Festival attracts audiences of all ages by offering movies and popular entertainment content. The general public can take an interest in international film festivals because movies from different countries can be seen in one place. Professional groups such as directors, actors, and students may visit the city to participate in special sessions or workshops at the festival [25]. Therefore, film festivals are not just entertaining events but also culturally valuable events that can be inspired by various academic topics, such as socio-cultural, aesthetic, geopolitical, and economic issues [26].

The success of film festivals is often related to proper promotion. Promotions, in particular, are an essential element in promoting the sustainability of start-up film festivals in the locale. Marketers at local festivals, including film festivals, are actively using IT as a promotional tool. Websites and SNS are representative IT platforms for festival promotion. Recently, they have also used crowdfunding platforms as a means of promoting festivals.

Crowdfunding can be classified into four types based on investment method: lending-based, equity-based, donation-based, and reward-based [27]. Equity-based is a way of investing in the idea of a startup, and lending-based is a way of investing in a project through a small loan. Donation-based means that investors will purely support the donation with no rewards. Finally, reward-based refers to offering products instead of financial rewards. Cultural and art events such as film festivals are mainly associated with reward-based crowdfunding. For instance, 154 and 32 crowdfunding projects on festivals were posted on Tumblbug [28] and Wadiz [29], respectively, which are Korea's leading reward-based crowdfunding platforms. Although the main purpose of crowdfunding is to fund projects by recruiting small investments of many people through an online platform [30], crowdfunding also generates promotional benefits for tourism projects [31]. In particular, crowdfunding is also used as a marketing tool to implement word-of-mouth strategies on social media. Crowdfunding platforms offer the chance for participants to pitch an idea to their social network [32]. This means that crowdfunding and SNS are highly compatible.

#### *2.2. Consumption Value of Crowdfunding*

Consumption value is an estimate of the utility of the product that a consumer receives when compared to the price and effort paid [33,34]. This explains whether a consumer purchases a particular product or why the consumer chooses one product over other ones [35]. Consumption value is thus a key determinant of consumer shopping behavior and product choice [36]. Previous studies segmented specific factors of value based on the concept of consumption value. For example, Park, Jaworski, and Maclnnis (1986) proposed functional, symbolic, and experiential values based on the concept of consumer needs [37]. Sheth, Newman, and Gross (1991) confirmed functional, conditional, social, emotional, and epistemic values as factors that influence consumer behavior [35]. Sweeney and Soutar (2001) proposed a multiple-item scale that includes a price value to measure the cost value of a product as well as the quality, emotional, and social values suggested in previous studies [38].

However, Smith and Colgate (2007) argued that value factors fall into four broad categories: (1) functional/instrumental, (2) experiential/hedonic, (3) symbolic/expressive, and (4) cost/sacrifice values [39]. The functional/instrumental value is related to the usefulness of a product or service, or the extent to which they perform a required function, and it encompasses concepts of functional, practical, and useful value. The experiential/hedonic value refers to the level at which a product or service creates an emotional value for a consumer. This value became an emotional or sensory value in some studies. The symbolic/expressive value is related to the degree to which consumers give psychological meaning to a product and includes the meaning of intrinsic value or possession value. Finally, the cost/sacrifice value represents the level of economic costs that consumers perceive during purchasing activities and includes factors such as economic and psychological costs. The consumption value factors presented by Smith and Colgate (2007) have academic implications because they have proposed categories in consideration of many previous studies [39].

Perceived values have been applied to describe consumers' behavior about various services and products. Some studies investigated the effect of consumption value on crowdfunding [40–42]. Schulz et al. (2015) proposed a hedonic value as a predictor of crowdfunding success, and it was confirmed that crowdfunding projects that show a high level of hedonic value are more successful than others [42]. Studies by Ho, Lin, and Lu (2014) and Moysidou and Spaeth (2016) introduced various consumption values as explanatory variables in the crowdfunding arena [40,41]. Moysidou and Spaeth (2016) categorized investor types into loan-based, equity-based, and presales-based projects based on crowdfunding investment methods and examined the factors that influence their willingness to support the project [41]. Financial, functional, informational, emotional, social, aesthetic, and novelty values were considered as independent variables. Consequently, financial and informational values were identified as factors affecting the willingness to support equity-based projects. Additionally, financial, emotional, and informational values were predictors for loan-based projects, and functional, emotional, and informational values were predictors in presales-based projects. In addition, Ho, Lin, and Lu (2014) also found that emotional, social, and function values have a positive effect on behavioral intentions for crowdfunding of cultural goods [40].

Previous studies commonly supported that the influence of financial, emotional, social, functional, and informational values presented by traditional value theories also influence behavioral intentions on crowdfunding. These values reflect the nature of crowdfunding as an investment platform but do not consider the characteristics of donation-based crowdfunding. Individuals are likely to participate in crowdfunding to satisfy their social approval desire rather than for monetary compensation in the reward or donation-based crowdfunding, where outputs are not monetary benefits [43]. As other-oriented factors, altruistic values are also important in the area of consumption [44].

Yet, another limitation of previous studies was that they did not consider the mediating effects between value factors and user behavior. Crowdfunding is an investment of personal money, so participants are likely to perceive financial risks. Generally, perceived risk is negatively correlated with a preference for financial instruments such as crowdfunding [45]. When crowdfunding includes high levels of value and credibility, participants perceive a low level of risk, which might positively affect behavior intentions [46]. In light of previous studies, this study applies economic, hedonic, social, and altruistic values as independent variables and assumes that these variables negatively affect perceived risk.

**Hypothesis 1-1 (H1-1).** *Economic value will negatively a*ff*ect perceived risk.*

**Hypothesis 1-2 (H1-2).** *Emotional value will negatively a*ff*ect perceived risk.*

**Hypothesis 1-3 (H1-3).** *Social values will negatively a*ff*ect perceived risk.*

**Hypothesis 1-4 (H1-4).** *Altruistic values will negatively a*ff*ect perceived risk.*

This study assumes that four values influence crowdfunding's e-WOM. If consumers perceive that one service provides more value than another one, they will voluntarily recommend it to others. In general, the intention to use e-WOM is determined by how a consumer evaluates a service or product. Specifically, consumers will be more likely to use word of mouth when a supplier provides value satisfaction compared to other suppliers [47].

Previous studies reported that the perceived value of consumers positively affects intention to use WOM [48,49]. Anwar and Gulzar (2011) found that perceived food value affects WOM endorsements through the mediating effects of consumer satisfaction [49]. By contrast, Abdolvand and Norouzi (2012) and Hansen et al. (2008) reported that perceived values have a direct effect on WOM without mediating effects [48,50]. Furthermore, Hansen et al. (2008) argued that the WOM behavior originated from the intention to repay the benefits provided by suppliers [50]. The relationship between perceived value and WOM is also important in emerging markets such as crowdfunding services. Customers in emerging markets determine their WOM behavior by considering how much value they have gained [51]. Potential investors are unlikely to know how much value they can return in crowdfunding before the experiences. Furthermore, WOM is likely to affect potential consumers' indirect experience of crowdfunding, and they might perceive the value. Although it is difficult to find a study that applies four types of value, the relationship between perceived values and e-WOM is supported by previous studies. Therefore, this study suggests that the perceived value of crowdfunding positively affects e-WOM.

**Hypothesis 2-1 (H2-1).** *Economic value will positively a*ff*ect e-WOM.*

**Hypothesis 2-2 (H2-2).** *Emotional value will positively a*ff*ect e-WOM.*

**Hypothesis 2-3 (H2-3).** *Social values will positively a*ff*ect e-WOM.*

**Hypothesis 2-4 (H2-4).** *Altruistic values will positively a*ff*ect e-WOM.*

#### *2.3. Innovativeness*

Inner innovativeness is defined as the degree to which an individual accepts a new idea or makes a decision about a new one [52]. The concept of inner innovativeness, based on innovation diffusion theory, adequately describes how to recognize new subjects and further adopts real targets according to the level of individual innovation [53,54]. Perceptions and behaviors regarding new technologies or services, such as crowdfunding, are likely to be determined by an individual's personality [53]. Specifically, when having a high level of inner innovativeness, the individual is better able to receive new technologies [55]. These individual tendencies have been consistently observed in the areas of media technologies such as personal computers, online games, and mobile technology [54–56].

Prior to the 1990s, studies of new technologies and services were primarily focused on the individual's perception of the subject instead of that individual's personal characteristics. After that, several studies examined the effects of individual characteristics such as self-efficacy [57] and intrinsic involvement [58]. For instance, Chang et al. (2006) identified the impact of internal innovation among the personal characteristics supporting online game adoption and suggested that highly innovative game players are more likely to adopt game playing [54]. In addition, several studies suggested that innovativeness leads to adoption behaviors [55,59]. In summary, previous research has focused primarily on examining the relationship between inner innovativeness and adoption behaviors.

However, several perceptual factors are proposed as well as adoption intentions as outcomes of inner innovativeness. First, inner innovativeness is related to perceived risk factors. In general, individuals face risks due to uncertainties about outcome behavior regarding new products or services [60,61]. In this situation, highly innovative individuals are willing to accept the uncertainty [53]. These risk-taking behaviors are considered typical features of highly innovative individuals [62]. Previous studies proved the negative relationships between inner innovativeness and perceived risks [60,63]. In summary, individuals with high levels of innovation are less likely to perceive risks and, as a result, are more likely to act aggressively on new services or products. The link between inner innovativeness and perceived risk has also been found in areas of financial transactions such as crowdfunding. Users cannot be sure of the quality of their financial transactions and, as a result, experience uncertainty [60]. Highly innovative individuals perceive lower risks than those who do not, resulting in more financial transactions.

Another outcome of inner innovativeness is the intention to use e-WOM. Traditional opinion leader research suggests that the higher the individual's inclination to innovate, the more likely they are to be opinion leaders [64]. Opinion leaders are considered active persons with a strong willingness to spread information about a particular object to others [65]. Given this, inner innovativeness can also have a positive effect on an individual's e-WOM intentions. Yoo, Jin, and Sanders (2013) classified consumer innovativeness into social, functional, hedonic, and cognitive [66]. Of these, social and cognitive innovativeness were found to affect intentions to use e-WOM. Sun et al. (2006) proposes online opinion leadership and online opinion seeking as two components of the e-WOM [67]. Online opinion leadership was identified as an outcome of inner innovativeness. In particular, Sun et al. (2006) provided significant results, showing that the effects of inner innovativeness on e-WOM are also found online [67]. In the study of tourism and experience goods, direct relationships between inner innovativeness and intention to use e-WOM are hard to find. It has been suggested that personal characteristics could affect the intention to use e-WOM in tourism [68]. Because inner innovativeness is a personal characteristic whose relationship with intention to use e-WOM has been verified in marketing studies, it can be expected that inner innovativeness will affect the intention to use e-WOM in tourism.

From a marketing point of view, an e-WOM strategy via SNS is a powerful tool that can leverage a range of features, such as the hashtag, to expand the reach of potential customers [69]. Therefore, inner innovativeness seems to affect e-WOM on SNS more clearly.

**Hypothesis 3 (H3).** *Inner innovativeness will negatively a*ff*ect perceived risk.*

**Hypothesis 4 (H4).** *Inner innovativeness will positively a*ff*ect intention to use e-WOM.*

#### *2.4. Perceived Risk*

Perceived risk has been applied consistently in analyzing consumer behavior because it has a decisive effect on behavior. Perceived risk is defined as the degree of uncertainty about the outcome of a decision [70] and anxiety about the loss of expected value [71]. In particular, individuals are more likely to perceive risk in an online environment where uncertain information is relatively high. High risk perception means that consumers believe themselves more likely to suffer loss as a result of using a product or service. This potential is expected to diminish consumer intentions or behaviors to buy products or services, especially in the online context [72]. Perceived risk thus leads to a negative causal relationship between consumption intentions. Individuals experience risk in an online environment for a variety of reasons, including technical complexity [73], privacy exposure [74], and uncertainty in the non-face-to-face communication environment. Previous studies on online services adopted perceived risk as the main variable [75,76]. The online shopping study classified risk types into economic, social, functional, personal, and privacy risks [76], and the research on e-services suggested, overall, that financial, psychological, privacy, time, and performance risks are perceived risk types. [75]. In summary, it was found that the type of risk adopted depends on the type of service or product covered by each study, but most of the previous studies consistently applied at least economic risk.

Economic risk is defined as "the potential monetary outlay associated with the initial purchase price as well as the subsequent maintenance cost of the product" [77] (p. 146). Featherman and Pavlou (2003) argued that economic risk has the highest explanatory power over other risk types [75]. Economic risk is considered as an important factor in crowdfunding research [78,79]. Kim and Jeon (2017) checked how economic risk negatively affects the intention of participating in crowdfunding [78]. Zhao et al. (2017) also found that perceived economic risk has a negative correlation with backers' funding intention in crowdfunding [79]. Therefore, if crowdfunding participants perceive that the project has high economic risks, they may be able to reconsider their next actions rather than actively investing or participating.

This study also considers perceived economic risk as a mediator. Although most festival crowdfunding is for donation-based or reward-based crowdfunding, this relates strictly to monetary spending. Thus, regardless of whether the rewards provided by crowdfunding managers are psychological or material, individuals can compare monetary expenditures and rewards. If the value of the perceived reward of a donation is lower than the monetary expenditure, perceived risk is likely to appear [71]. Thus, we consider that economic risk is a major variable because such risk is triggered by individual monetary spending.

Previous studies suggested that the intention to participate in crowdfunding was an outcome of perceived economic risks [46,78], but this study assumes that perceived economic risks for festival crowdfunding services negatively affect the intention to visit festivals. The crowdfunding platform is not only an investment platform but also an information resource for the festival. Individuals are more likely to visit film festivals based on the information. Particularly, considering that many festivals' crowdfunding offers movie tickets as a reward, which must be used on-site, participants are likely to visit the area.

#### **Hypothesis 5 (H5).** *Perceived risk will negatively a*ff*ect the intention to visit.*

#### *2.5. Intention to Use e-WOM*

In the traditional sense, WOM is the interpersonal communication that occurs between the sender and the receiver, and it also describes the process of changing the recipient's behavior or attitude [38]. Recently, many studies have examined the effect of e-WOM. E-WOM is defined as "all informal communications directed at consumers through Internet-based technology related to the usage or characteristics of particular goods and services" [79] (p. 461). As such, online services are considered the most effective means of interpersonal communication [80]. As IT advances and various communication platforms such as social networks increase, individuals can exchange information seamlessly with each other. The information produced on the major platforms for e-WOM influences individual planning and decision-making [81]. In fact, e-WOM was estimated to secure 30 times more customers than traditional channels [82].

Given some characteristics of tourism, travelers are more likely to be influenced by e-WOM. First, tourism is an experience good [83]. Individuals try to lower the uncertainty they perceive in decision-making by considering an indirect experience regarding tourism information. Second, tourism is characterized by the information-intensive industry [84]. Tourists might try to get reliable information from many of the communications generated through e-WOM. Given these characteristics, individuals use e-WOM as an easy way to find information about travel destinations [85].

Several studies examined how e-WOM influences attitudes and behaviors to visit tourist areas [15,17]. Wang (2015) found that the intention to use e-WOM positively affected tourist visits [17]. Previous studies argued that WOM behavior includes a positive attitude and loyalty to the target [17,44]. Abubakar et al. (2017) reported that e-WOM influences the intention to re-visit the medical tourism industry [86]. Previous studies have found that those who share a positive message on the elements of a tourist experience with others are more likely to be immersed in tourism information or to have a positive understanding of it. In this sharing process, positive images can be created that arouse intention to visit [16]. It can be expected that this mechanism will appear here as it has elsewhere. Additionally, it is inferred that people with a high level of e-WOM intention are more likely to show the actual behavior, like visiting a tourist destination. According to this view, the sharing of film festival crowdfunding information is based on positive royalties for the festival. We assume that intention to use e-WOM based on the loyalty toward festival crowdfunding affects the intention to visit.

**Hypothesis 6 (H6).** *E-WOM will positively a*ff*ect the intention to visit.*

#### **3. Research Methods**

#### *3.1. Sampling Procedure and Sample*

This study collected survey data from February 7 to 11, 2020, by requesting data from Marcromill Embrain, an online survey company. Korean survey respondents included those who had visited an international film festival in Korea at least once. International film festivals included Busan International Film Festival, Jeonju International Film Festival, Bucheon International Fantastic Film Festival, Jecheon International Music and Film Festival, Ulju Mountain Film Festival, DMZ International Documentary Film Festival, PyeongChang International Peace Film Festival, Muju Film Festival, Animal Film Festival in Suncheonman, Jeju Film Festival, Seoul Independent Film Festival, and Seoul International Women's Film Festival. Except for 15 samples that were unreliable responses, 447 samples were considered as sample data for the research.

To help understand film festival crowdfunding, participants reviewed the crowdfunding content uploaded to Tumblbug, which is a leading crowdfunding platform in Korea. Specifically, before responding to the questionnaire, participants read information about film festival crowdfunding, fundraising, and benefits from participating in the crowdfunding.

The quota sampling was conducted by considering the proportion of age and gender in the census population. As shown in Table 1, 221 (49.4%) of the survey respondents were male and 226 (50.6%) were female. Furthermore, 108 (24.2%) of the respondents were aged 20–29, followed by those aged 30–39 (112, 25.1%), 40–49 (115, 25.7%), and 50–59 (112, 25.1%). Most respondents, 314 participants (70.2%), had a university degree followed by those who had completed graduate school or higher (59, 13.2%), those who had a high school diploma or lower (36, 8.1%), and those who were university students (37, 8.3%). Regarding income, 23% of respondents had a monthly personal income between 2,000,000–2,999,999 Korean Won (approximately US \$1688–US \$2532), followed by 3,000,000–3,999,999 (18.1%), 4,000,000–4,999,999 (11.4%), and 5,000,000–5,999,999 (11.0%). Finally, 10.3% of respondents had a personal income of less than 1,000,000 won (See Table 1).


**Table 1.** Demographic information for the sample. KRW: Korean Won.

#### *3.2. Measurement and Analysis Method*

The research model includes the following eight variables: (1) economic value; (2) emotional value; (3) social value; (4) altruistic value; (5) inner innovativeness; (6) perceived risk; (7) e-WOM; and (8) intention to visit film festival. We set four value factors and one personal characteristic factor as independent variables that affected perceived risk (Figure 1).

**Figure 1.** Research model.

The economic value describes how far it is perceived that film festival crowdfunding can provide monetary value. The emotional value describes how far it is perceived that the film festival crowdfunding can lead to a positive emotional state. The social value describes how far one expects to be recognized by social contacts for participating in the film festival crowdfunding. Economic, emotional, and social value were revised based on the Sweeney and Soutar (2001) study on shopping value [38]. The altruistic value describes how far one believes that film festival crowdfunding participation will help others. To measure the altruistic value, we considered items from Holbrook's (2006)

study of the applicability of subjective personal introspection factors in consumption experience [87]. Inner innovativeness is "the degree to which an individual is receptive to new ideas and makes innovation decisions independently of the communicated experience of others" [52] (p. 236). Items of inner innovativeness were revised based on Chang, Lee, and Kim's (2006) study, which examined factors that influence online game adoption [54]. In an e-service adoption study, Featherman and Pavlou (2003) proposed seven types of risks, such as performance, financial, time, psychological, social, privacy, and overall risks [75]. Among them, the items of financial risk by Featherman and Pavlou (2003) were used in the study [75]. Perceived risk describes how much one expects to lose in relation to money for participating in film festival crowdfunding. Online word-of-mouth behavior is divided into opinion giving, opinion passing, and opinion seeking [88]. In this study, opinion passing was considered as observed e-WOM. E-WOM describes the willingness to share festival crowdfunding information with others on SNS. We used items based on research from Sun et al. (2006), Chu and Choi (2012), and Lee and Lee (2013) [67,89,90]. Finally, intention to visit the film festival describes the willingness to visit a local film festival in the near future. To measure this, we used the items of Perugini and Bagozzi (2001), which were based on the theory of planned behavior [91]. All questions in this study consisted of a Likert five-point scale (1 = strongly disagree and 5 = strongly agree). Three items were contained for each variable (See Table 2).


**Table 2.** Items and scales used in questionnaire.

The general characteristics of the samples were identified using SPSS18 statistical software. In particular, frequency analysis and composite reliability were performed to verify the reliability of the measurement questions. Next, this study utilized the structural equation model (SEM). The statistical software Amos 20 was used to carry out the calculations. SEM enacts multi-equation system procedures, determining multiple indicators of concepts, continuous latent variables, errors in equations, errors of measurement, and continuous latent variables [92]. This is useful for the comprehensive examination of relationships among multiple variables, as needed in this study. First, confirmatory factor analysis (CFA) was performed to check the latent variables and ensure the absence of measurement errors by CFA. Next, the relationship between the variables was examined through the structural model.

#### **4. Results**

#### *4.1. Measurement Model*

In this study, we applied the coefficient of Cronbach's α to verify the composite reliability between the measured items. In general, reliability is satisfied when Cronbach's α is greater than 0.7 [93]. Each variable includes three observed variables, and all the items in the model met the reliability criteria (See Table 3). Convergent validity is desirable when the average variance extracted (AVE) is greater than 0.5 [94]. As a result of the verification, all values of AVE were found to meet the criteria. To verify the indicator reliability, this study applies the criterion that every loading must be greater than 0.4 [95]. Of all the loading values, two observed variables were above 0.6 (AV\_2 = 0.621, II\_3 = 0.610), but most of them were greater than 0.7. In addition, all observed variables were verified to be statistically significant (p = 0.00). Therefore, good indicator reliability was confirmed. Regarding discriminant validity, AVE must be greater than all constructs of the square root [94]. As shown in Table 4, it was found that AVE exceeds the square of the correlation.


**Table 3.** Construct reliability and validity.

a. CR: composite reliability coefficient; b. AVE: average variance extracted; e-WOM: electronic word of mouth.

**Table 4.** Test of discriminant validity.


#### *4.2. Hypothesis Testing*

To verify the fit of the structural model, absolute fit index, parsimonious fit index, and incremental fit index were considered. Scholars applied different criteria to assess model fit [92], but many previous studies usually applied χ2/df, RMSEA (Root Mean Square Error of Approximation), AGFI (Adjusted Goodness of Fit Index), CFI (Comparative Fit Index), TLI (Tucker Lewis Index), and NFI (Normed Fit Index) as criteria for the model fit. As a result of the model fit verification, all of the fitness indices (χ2/df = 2.578, RMSEA = 0.059, AGFI = 0.867, CFI = 0.948, TLI = 0.938, NFI = 0.919) signaled a good model fit (See the Table 5).


**Table 5.** Model fit statistics.

The purpose of this study was to analyze the effects of value factors and inner innovativeness on the intention to visit film festivals through perceived risk and e-WOM. The results of the analysis are shown in Table 6. With regard to the value factor, all variables, except for EV (economic value), showed significant effects on perceived risk. EmV (emotional value) (B = −0.410, ρ < 0.044) had a negatively significant effect on PR (perceived risk). It was also confirmed that AV (altruistic value) (B = −0.306, ρ < 0.025) negatively affected PR. However, SV (social value) (B = 0.528, ρ < 0.000), unlike the hypothesis, positively affected PR. Therefore, H1-2 and H1-4 were supported. Regarding H2-1 to H2-4, SV (B = 0.278, ρ < 0.000) had statistically significant effects on e-WOM, but EV, HV, and AV had no effect on e-WOM. Thus, only H2-3 was supported and H2-1, H2-2, and H2-4 were not supported. Next, InI (inner innovativeness) did not affect both PR and e-WOM. Therefore, H3 and H4 were not supported. Regarding the effects of PR and e-WOM on the dependent variable, PR (B = −0.099, ρ < 0.005) negatively affected IVFF (intention to visit film festival). The influence of e-WOM (B = 0.521, ρ < 0.000) was positively significant on IVFF. Thus, H5 and H6 were supported.

**Table 6.** Hypotheses test for structural models by the general maximum likelihood structural equation model (SEM).


\* *p* < 0.05, \*\* *p* < 0.001.

#### **5. Discussion and Conclusions**

This study aims to empirically examine how the perceived value of crowdfunding and inner innovativeness affect the intention to visit the film festival. For this, perceived risk and e-WOM were applied as mediating variables. Structural equation modeling was applied, and data were collected through an online survey. As a result, 6 of 12 research hypotheses were supported. Accordingly, we confirmed through its effects that crowdfunding is an important platform for promoting sustainability at local film festivals.

First of all, emotional values (H1-2) and altruistic values (H1-4) negatively affected perceived risk. This supports the findings of previous studies, which indicate that individuals perceive low levels of risk when they have a sense of worth regarding a given service or product [46]. These variables are commonly considered to be related to emotional characteristics rather than to cognitive ones. These results indicate that consumers are more likely to obtain emotional value as a result of the crowdfunding of cultural products, such as festival crowdfunding. Another notable aspect is that the path coefficient for emotional value (B = −0.401) was higher than the altruistic value (B = −0.306). This implies that emotional value has greater explanatory power. Future studies should develop this framework by developing detailed factors for emotional value. Interestingly, economic value effects were not confirmed. It is expected that the effects of economic value vary in relation to the type of crowdfunding. Moysidou and Spaeth's (2016) study, which examined the impact of economic value on behavioral intention by project type, found an influence of economic value in revenue-based crowdfunding, such as equity-based and loan-based projects [41]. Individuals may not take economic value into account for donation-based or reward-based crowdfunding, including cultural goods. Economic relates to the amount of monetary benefit that an individual can receive. Individuals do not generally seek financial benefits when they participate in donation and reward-based crowdfunding. Rather, they seek to meet non-monetary needs, such as social approval or self-realization [43]. In particular, individuals' non-monetary motivations are likely to be more prominent in cultural goods, such as in relation to local film festival crowdfunding, which mainly has non-commercial purposes.

For hypotheses 2-1 to 2-4, only social values (H2-3) positively affected the intention to use e-WOM. Social values and intention to use e-WOM both rest on the concept of social relationships. This result means that perception of the value of interacting with others through participation in festival crowdfunding can affect social interactions online. Prior studies have found that perception of quality in the general product area, such as food, can affect WOM [49]. On the other hand, social value effects were identified in this study. It is unclear whether these results can be attributed to the nature of the experience or to crowdfunding type. Further research is needed.

Next, inner innovativeness was found not to affect perceived risk and e-WOM intentions. These results indicate that the consumer's inner innovativeness is not related to the perception of festival crowdfunding. Regarding perceived risks, Foxall (1988) proposed that risk perceptions vary according to the type of users [96]. This means that early adopters underestimate the risk compared to late adopters. Accordingly, future studies should identify the audience characteristics for film festival crowdfunding and examine the relationship between inner innovativeness and perceived risk for each type. Regarding e-WOM, previous studies on innovation have shown a different view of claims. For example, Summers (1970) argued that highly innovative people tend to give opinions to people [64], but Rogers (1995) argued that innovation is related to information seeking [53]. The two studies each focused on the production and acceptance of information. By contrast, Sun et al. (2006) verified that innovativeness has a significant impact on both online information leadership, which includes the concept of e-WOM, and information seeking [67]. Given the inconsistent claims of the previous studies, further research should examine the relationship between innovativeness and e-WOM.

Third, perceived risk negatively affects the intention to visit the film festival (H5). This result is consistent with the findings of previous studies, which indicated that perceived risk for crowdfunding affects actual behavior [46,78]. This also means that, if an individual underestimates the risk of festival crowdfunding, they can visit the festival. Last, the intention to use e-WOM also had a positive effect on the intention to visit the film festival (H6). This result is consistent with previous studies that have suggested that e-WOM has a positive effect on travel intention [15,17]. The action to share the usefulness of festival crowdfunding with friends is believed to be based on trust or interest in the film festival. This means that positive perceptions of the film festival can lead to real action, such as visiting a festival. However, there is also the possibility that control variables may have some bearing on the intention to use e-WOM and the intention to visit a film festival. For example, actual behavior may vary depending on perceived involvement in film festivals or crowdfunding platforms. Future research will be needed to expand the research model by exploring moderating variables.

As for H5 and H6, earlier study of tourism crowdfunding used crowdfunding behavior as a dependent variable. Our study, however, set the intention to visit a festival as an outcome variable and not a crowdfunding behavior, and found that it is affected by explanatory variables. This means that the effects on the service platform can affect consumers' behavior in relation to the topic of crowdfunding beyond the platform. If individuals have a positive perception of crowdfunding services, this may indirectly indicate that they have a positive perception of crowdfunding as well as of the service platform. Therefore, crowdfunding goes beyond investment to acting as a marketing tool, attracting individual visits to local film festivals and further enhancing the sustainability of local film festivals.

This study has theoretical implications. First, previous work has examined factors affecting the intention to participate in festival crowdfunding [97,98], but overall, too little work has been done on the effects of crowdfunding in tourism. Additionally, from a marketing perspective, few analyses or in-depth discussions of festival crowdfunding have been done. This study is thus an important examination of the influence of tourism crowdfunding. Second, this study proposed a framework to examine the marketing effects of crowdfunding in tourism. It has been confirmed that the influence of certain variables varies depending on the type of crowdfunding being considered. As noted, we found that emotional factors should be taken seriously in the study of donation-based and reward-based crowdfunding. If this framework is properly modified in response to the results of this study, festival-visiting behavior can be explained in greater detail. Third, to supplement existing research trends, structural modeling is proposed in relation to perceived risk and e-WOM as mediating variables between value factors and behavioral intentions. This approach can provide a useful aid to studies dealing with online communication environments and platforms with high uncertainty, such as crowdfunding platforms.

This study provides practical guidelines to help with festival sustainability. First, this study confirmed that the crowdfunding platform can be used as a marketing tool. It was found that factors related to crowdfunding platforms, such as perceived risk and e-WOM, drive intentions to visit the festival. Therefore, even if the film festival does not need financial help, marketers should open a crowdfunding page to provide an opportunity to access the film festival crowdfunding and obtain information. Second, marketers should build their value strategy differently depending on the type of crowdfunding. Marketers encourage consumers to perceive economic value for equity-based and loan-based crowdfunding, whereas, for non-revenue types such as donation-based or reward-based, marketers should provide an opportunity to gain emotional value or the need for approval. Finally, marketers can understand consumers' e-WOM behavior and strategically use it. Crowdfunding, for example, is compatible with SNS. Moreover, users can share information about crowdfunding via SNS. Therefore, this study is expected to provide a guide for developing and implementing a marketing strategy that simultaneously utilizes SNS and crowdfunding platforms to enhance the e-WOM effect on film festivals.

Regarding limitations, in this study, film festival crowdfunding is still an emerging market, so it was difficult to conduct online surveys for actual users. Respondents to this study experienced indirect festival crowdfunding only through the experimental treatment provided by the researcher. Future researches may conduct online surveys in collaboration with film festival crowdfunding providers, who can directly contact actual users. Next, follow-up studies should take a closer look at the effects of other factors on the intention to visit a film festival. Potential consumers will recognize not only the risks directly related to the festival but also the risks of the crowdfunding investment in their decision to visit the festival. Thus, in addition to the perceived risk of participation in film festival crowdfunding, risks associated with the consumption behaviors that are seen during participation in

the festival should also be considered. Finally, future studies should be conducted that investigate multi-dimensional inner innovativeness. Certain studies have also reviewed how well individuals accept brands or images related to innovation [54], but these have mainly involved emotional indicators, so to examine the effects of inner innovativeness, both cognitive and emotional perspectives should be considered.

**Author Contributions:** H.K. and B.C. conceived and designed the experiments; B.C. performed the experiments; H.K. analyzed the data; B.C. contributed reagents/materials/analysis tools; H.K. and B.C. wrote the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Global Research Network program through the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2017S1A2A2041908).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Relationship between Viewing Motivation, Presence, Viewing Satisfaction, and Attitude toward Tourism Destinations Based on TV Travel Reality Variety Programs**

#### **Bo-Kyeong Kim \* and Kyoung-Ok Kim**

Department of Tourism Management, Pukyong National University; 45 Yongso-ro, Busan 608737, Korea; okson334@hanmail.net

**\*** Correspondence: ynnij@hanmail.net

Received: 29 April 2020; Accepted: 27 May 2020; Published: 5 June 2020

**Abstract:** The purpose of this study was to clarify the relationship between viewing motivation for reality programs and the viewing experience (presence) during watching, viewing satisfaction after watching, and attitudes toward presented tourism destinations. While this study notes that the empirical verification of travel reality variety programs is insufficient, various discussions are presented with regard to the grip of reality variety program fever in Korea. Notably, viewers are interested in the emotional experience related to characters and tourism destinations in reality variety programs. Therefore, we asked the following questions: (1) What kind of motivation encourages viewers to watch reality variety programs? (2) How does motivation for viewing a reality variety program affect viewing satisfaction through a certain approach (presence)? (3) How does viewing satisfaction affect one's attitude toward a program's tourism destinations? To answer these questions, we conducted a survey of 358 viewers of travel reality variety programs. The results of this study are as follows: (1) viewing motivation for travel reality variety programs consists of five factors: vicarious gratification, entertainment, information-seeking, habitual time-spending, and socializing; (2) it was confirmed that the effect of these five factors on satisfaction was mediated by presence (viewing experience); and (3) viewing satisfaction through presence affected the subsequent attitude toward presented tourism destinations.

**Keywords:** travel reality variety program; viewing motivation; viewing satisfaction; presence; attitude toward tourism destination

#### **1. Introduction**

In the past, most programs about travel were mainly documentaries focused on information delivery. However, the recent trend involves reality variety programs, which combine variety, fiction, and reality; travel programs combine entertainment and seeking fun, transforming into a new genre called "travel entertainment" [1]. It is said that this change in travel programs is leading to corresponding changes in travel pursuit and tourism trends [2]. In addition, after travel reality variety programs are broadcast, credit card usage for travel and leisure increases. This suggests that consumers' or viewers' desire to travel is indirectly connected to travel programs.

Reality variety programs are continuously increasing in terms of attractive content due to accelerated competition among media, as well as the advantage of securing viewer ratings at relatively low production costs [3–6].

Due to the format of reality travel programs, tourist destinations are shown on the screen frequently and for longer periods of time compared to drama programs and movies. Therefore, viewers are more likely to be more immersed in the presented situation than they would be with dramas and movies, They also form images of tourist destinations in the programs and make meaningful connections to the selection of a travel destination [7].

As such, mass media plays an important role in the travel industry, and the reality variety format is particularly important for theoretical research and has managerial aspects that can be key in the travel industry.

In particular, travel programs in the reality format are increasing and the formats are diversifying, but related previous studies are limited to the exposure effects on the tourist destinations, factors influencing the choice of destinations, and airline ticket price analysis for travel programs.

Recently, there have been several studies on how viewing motivation for reality travel programs influences future viewer behavior. However, they are limited to discussing the general relationship between viewing motivation and viewing satisfaction.

Therefore, this study aimed to investigate the effects of viewing motivation, viewing satisfaction, and attitudes toward tourism destinations, and to empirically reveal the mechanism of presence (watching experience) in a sample of men and women over the age of 18 who live in Busan and Gyeongnam provinces. To achieve this purpose, based on previous research on viewing motivation for broadcast programs, it is classified into five factors related to reality travel programs, and the examination of the effect of the five factors on satisfaction is mediated by presence (viewing experience). Lastly, viewing satisfaction through presence is confirmed with regard to its influence on subsequent attitudes toward presented tourism destinations.

#### **2. Literature Review**

#### *2.1. Viewing Motivation for Travel Reality Variety Programs*

Previous studies on viewing motivation divided it into social reward and psychological motivation [8], including information, entertainment, establishing a relationship, and recreation [9]; and entertainment, time-spending, and social learning [10,11]. Viewing motivation is based on uses and gratification theory [12,13], which posits that viewers use or watch TV programs to satisfy their desires and needs. Unlike traditional research about the effect of mass media asking questions such as, What does media do to people? this theory conversely asks, What do people do with media? Thus, this theory is used to understand viewers' media experience and to identify their motivation in selecting and using specific media in comparison with other media.

Research on uses and gratification has been extended to studies ranging from categorizing viewing motivation for TV programs to describing viewer motivation according to the genre of programs.

According to Lim (2008) [14], college students' motivations for viewing American dramas were entertainment and rest, information, and environment/companion, while those for Korean dramas also included drama characteristics (feature) and habitual free time-spending.

Ban and Park (2014) [15] studied viewing motivation for reality dating programs and categorized it into virtual avoidance, indirect experience, pleasure, social interaction, and free time-spending. For a certain type of reality travel program, such as "Daddy! Where Are You Going?", there are four main factors of viewing motivation: interest, the attractiveness of the children, alleviating loneliness, and expressing everyday emotions [16,17]. "Hyori's B & B," which aired in 2017 and 2018, was an observational reality travel program that recently became an entertainment trend, with entertainment, empathy, interaction, aesthetics, information, and indirect experiences selected as viewing motivations [2]. It was found that only aesthetics, information, and indirect experiences had significant effects on viewer satisfaction and intention to visit.

The results of these studies show differences in viewing motivation according to program format or genre, and that generally, but not always, it affects viewer satisfaction.

Therefore, the motivation for viewing travel reality variety programs does not only involve acquiring information about tourism destinations, but there is also a psychological motivation in comparison with other travel programs [16]. Accordingly, this study proposes vicarious gratification, entertainment, information-seeking, habitual time-spending, and socializing as factors of viewing motivation for travel reality variety programs based on previous studies.

#### *2.2. Presence*

While the definition of presence varies according to the researcher [18], it generally refers to the psychological and subjective experience of a person's feeling of being directly in the mediating environment outside their surrounding physical environment [19–21]. To be specific, there are six subtypes of this dimensionality: (1) social richness that feels the media intimately, (2) realism based on realistic expression, (3) transportation that seems to be going to a virtual media environment, (4) immersion in a virtual reality, (5) a sense of reality as a social actor within the medium who interacts with the mediated character, and (6) feelings toward the medium itself as a social actor [20]. The study of this type of presence has been extended to its role as a mediator between viewing motivation and satisfaction, as well as various media environments. According to Kim and Biocca (1997) [19] and Lombard and Ditton(1997) [22], while there are differences in the types and scopes of media, presence can be applicable to all media. In addition, presence is established by the format, genre, content, and characteristics of the media, and can be sufficiently experienced even in a traditional medium where the mediated environment is relatively unrealistic. In particular, in the case of TV media, the amount and precision of sensory channels mobilized by users, i.e., the "sensory output", creates a higher reality experience that is called presence in TV but not radio, which is only auditory [23].

Therefore, it can be said that it is difficult to explain the genre of reality travel programs only by the relationship between viewing motivation and satisfaction in terms of the realistic expression (presence as reality) of TV media [8,24]. In particular, in the case of reality travel programs, with travel and storytelling, viewers become more immersed in the tourism destination, which is the background of the program, making them more aware of the reality of the place depending on the level of perceived similarity and wishful identification with the characters [1]. In accordance with this view, it is critical to confirm that presence plays a mediating role in the relationship between viewing motivation and satisfaction [25].

Based on the above discussion, this paper examines the mediating role of presence in accounting for the relationship between viewing motivation and viewing satisfaction for travel reality variety programs. We accordingly set out to prove the significance of presence as a parameter that affects viewing satisfaction.

#### *2.3. Viewing Satisfaction*

Satisfaction with viewing media and programs starts with expectations for content. Thus, since satisfaction with a program is shaped when the viewer's expectations for that program are met, or the viewer evaluates them positively through a process of assessing the cognitive and emotional dimensions of the content, their consequent satisfaction reflects the fulfillment of their desires. Therefore, satisfaction after watching a program is a very critical parameter for understanding the viewer's media experience. Satisfaction with television viewing is formed by a combination of various factors, such as motivation and viewing experience. In fact, there has been extensive research on viewing satisfaction and viewing experience regarding television and specific programs [11,24,26].

According to the expectancy value model of Palmgreen and Rayburn (1984) [27], media consumption is driven by satisfaction, or the seeking of gratification, and perceived gratification obtained afterwards. Beliefs and evaluations are modified, thereby affecting the repeated seeking of gratification. In other words, if gratification sought and obtained is consistent, viewing satisfaction increases, which leads to continued viewing. This model explains the cyclical relationship between viewing motivation, viewing behavior, and satisfaction (fulfillment) through various media.

Indeed, on travel reality programs, tourism destinations are realistically described from the tourists' perspective, and viewers' expectations and desires are satisfied by being providing rich information on tourism destinations [10,28]. In addition, detailed information about the region/place is provided while famous tourists travel directly in the program, unlike in drama and movies [6,29]. In short, as it is a place that is not merely an image, such as in a film or drama, but is directly exposed to viewers, it can be expected to arouse their expectations and intentions for future behavior.

Therefore, this study intends to expand viewing satisfaction theoretically through the specific parameter that the motivation to view travel reality variety programs affects viewing satisfaction, based on previous studies.

#### *2.4. Attitude toward Tourism Destinations*

Attitude is an ongoing assessment of beliefs and emotions about an object [30,31]. Attitudes are also individuals' thoughts, emotions, and tendencies regarding particular subjects, and their future behavior will vary depending on how they establish attitudes toward particular subjects [32]. In addition, it is not necessary for awareness, influence, and behavior to simultaneously exist in order to form or express an attitude; rather, such can be established when one or more components are combined [33]. Fishbein and Ajzen (1975) [31] proposed the theory of reasoned action by adding subjective norms to attitudes and behavioral intentions to clarify the human decision-making process. Ajzen (1991) [34] then developed the theory of extended planned behavior by adding perceived behavior control considering the external environment. In the tourism industry, it is important to understand tourists' decision-making, so research has been performed to identify behavioral intent through attitude [28,35–37]. In particular, decision-makers include many criteria when selecting tourism destinations, determining their preferences and behaviors concerning future destinations by comparing and evaluating the attributes of tourist sites [38].

Thus, tourists can be said to have different behaviors depending on their prior preferences [39]. Tourists are exposed to diverse mass media, actively use media to gather information about tourism destinations, and make various decisions regarding tourism, such as selecting destinations, through such information. The media sector that has attracted the most attention recently in the field of tourism is entertainment, which is indirectly connected to viewers (potential tourists) through movies and TV programs, and directly connected to the tourism industry [40]. Based on previous research, it is necessary to extend the research area into the effect of viewing satisfaction on attitudes toward tourist destinations according to the motivation for viewing travel reality variety programs.

#### **3. Research Model and Methodology**

#### *3.1. Research Model and Hypotheses Development*

Viewing motivation is a significant factor in understanding how to meet viewer expectations for a program as well as influencing attitudes and future behaviors. Taking a look at previous studies, there was a difference in viewing motivation depending on the viewer or the program format or genre. In particular, the effect on viewing satisfaction or other behavioral variables can be different through the realistic expression (presence as reality) of the place shown in a travel reality program, unlike in films and dramas [7,40,41]. Therefore, presence is a psychological factor of the viewer and can be said to be a weighty parameter that influences satisfaction with the use of TV or media [5]. For example, viewers motivated to pursue information on travel will have a higher level of interest in tourism destinations that are in the background of the program. Particularly, while moving along with performers, they can experience presence in the places they see on the screen, and this shows high satisfaction with the program [15].

Therefore, the following hypothesis was proposed:

**Hypothesis 1 (H1).** *Viewing motivation for a travel reality variety program positively a*ff*ects presence.*

**Hypothesis 1-1 (H1-1).** *Vicarious gratification positively a*ff*ects presence.*

**Hypothesis 1-2 (H1-2).** *Entertainment positively a*ff*ects presence.*

**Hypothesis 1-3 (H1-3).** *Information-seeking positively a*ff*ects presence.*

**Hypothesis 1-4 (H1-4).** *Habitual time-spending positively a*ff*ects presence.*

**Hypothesis 1-5 (H1-5).** *Socializing positively a*ff*ects presence.*

Previous studies on viewing motivation and satisfaction found a positive relationship between these factors through the mediation of parasocial interaction and presence on the program "Cookbang" [15]. This means that high levels of involvement in, attention to, and engagement with the reality program by viewers showed a positive effect on their viewing satisfaction according to their watching experience (presence). Therefore, this reality presence can increase enjoyment and engagement even when watching a travel reality variety program, which can increase viewing satisfaction. Hence, the following hypothesis was proposed:

**Hypothesis 2 (H2).** *Presence positively a*ff*ects viewing satisfaction with a travel reality variety program.*

Viewing satisfaction means that a program exceeds the viewer's expectations, and viewers are more satisfied with a program if their expectations are exceeded, and dissatisfied if their expectations are not met [11]. Satisfaction with TV viewing is influenced by the viewer's attitude and behavioral intention, because the viewing motivation or expectation for the program is satisfied or the viewer's behavior is closely related to the program [42]. Therefore, the following hypothesis was proposed:

**Hypothesis 3 (H3).** *Viewing satisfaction with a travel reality variety program positively a*ff*ects the viewer's attitude toward the tourism destination.*

Based on the above hypotheses, this study's research model is shown in Figure 1.

**Figure 1.** Research model.

#### *3.2. Operational Definition*

Viewing motivation was defined as "the viewer's personal viewing desire/need for the media" [8]. Next, based on the research of McQuail (2005) [36], Ebersole and Woods (2007) [43], and Jang and Kim (2016) [44], five motivational factors were derived: vicarious gratification, entertainment, information-seeking, habitual time-spending, and socializing. With slight modification, a seven-point Likert scale was used to measure these factors (1 = strongly disagree to 7 = strongly agree). Presence

was defined as "feelings/emotions of immersing into the environment created by television programs without being able to perceive the reality of television" based on the research of Bracken (2009) [23] and Pope and Wansink (2015) [11]. With slight modification, a seven-point Likert scale was used to measure four items (1 = strongly disagree to 7 = strongly agree). Viewing satisfaction was defined as follows: "the viewer uses media to satisfy his or her various needs/desire and establish satisfaction," from the uses and gratification theory of Kartz (1959) [9]. Four items were extracted and slightly modified from Kim and Rubin (1997) [5] and An and Han (2018) [16], and measured by a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). Lastly, attitudes toward tourism destinations were measured by using a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree) for the four items used in the survey on the impact of the series "Better Late Than Never" of the Korea Tourism Organization-International Travel Trend Report (2014).

#### *3.3. Data Collection and Analysis Method*

For this study, the survey was conducted via an Internet questionnaire, first confirming the experience of watching reality variety programs. Next, the authors set as samples those living in Busan and Gyeongnam provinces who had experience watching travel reality variety programs from 1–31 March 2020. The convenience sampling method was applied. Among the 382 total respondents, 358 valid questionnaires were received and used for analysis, excluding (1) nonresponse, (2) overlapping response, and (3) unfinished questionnaires. Because of the high response rate, analysis of nonrespondents was not conducted, since a 94.7% response rate was considered adequate [45]. SPSS 25.0 and AMOS 23.0 were used for the analysis.

The statistical analysis methods and procedures applied in the empirical data analysis process of this study were as follows:

First, frequency analysis was performed to examine the general demographic characteristics of the sample. Second, exploratory factor analysis was conducted to identify the factor structure, reliability, and validity of the measurement tool, and confirmatory factor analysis and discriminant validity analysis were performed to verify the fit of the factor structure. Third, Structural Equations Models (SEMs) were examined to investigate the hypotheses.

#### **4. Results**

#### *4.1. Profile of the Sample*

The demographics of the 358 participants were as follows: 188 men (52.7%) and 170 women (47.3%). Regarding the age groups, 103 people (29.0%) were in their 20s, 113 people (31.7%) were in their 30s, 88 people (24.6%) were in their 40s, and 38 people (10.7%) were in their 50s. In terms of marital status, 230 (64.5%) were married and 128 (35.5%) were single. Regarding the distribution of occupations, 51 were homemakers (14.3%), 84 had specialized jobs (23.7%), 193 (54.0%) were office workers, and 30 (8.0%) had other occupations. Regarding the distribution of the amount of travel in a year, 119 people (33.5%) traveled once, and 142 people (39.7%) traveled two to three times a year. Regarding the number of reality variety programs watched in a week, 54 people (14.7%) watched once per week, 177 people (49.6%) watched two to three times, and 76 people (21.4%) watched four to five times. Considering distribution of monthly income (in million won), 198 people (55.4%) earned 3 to 4.99 million won and 76 people (21.0%) earned more than 5 million won. The demographic characteristics of this study are shown in Table 1.


**Table 1.** Demographic variables (n = 358).

#### *4.2. Reliability and Validity*

This study confirmed the reliability and validity of the variables before hypothesis testing. Regarding the reliability of constructs, internal consistency was found with all measuring factors with Cronbach's α over 0.7 [46]. The validity of the variables was verified through confirmatory factor analysis. The results are presented in Table 2. First, it was concluded that the standardized factor loading value was over 0.6, the significance level was less than 0.1%, and the AVE (average variance extracted) value was over 0.5, indicating that there was convergent validity [47]. Comparing the AVE value with the square of the correlation coefficient (based on Table 3) showed that the AVE value of latent variables was greater than the square of correlation value. Thus, the validity of discriminant validity was assured [47].

**Table 2.** Results of confirmatory factor analysis of the measurement model.



#### **Table 2.** *Cont.*


Note: Values in parentheses are AVE, and the lower values of the diagonal indicate the factor correlation: \* *p* < 0.05, \*\* *p* < 0.01.

#### *4.3. Verification of Hypotheses*

Given an acceptable measurement-model fit, SEM was carried out in AMOS to investigate the overall fit of the structural model and hypotheses. The fit of the proposed model was tested using fit indices. The results (χ2/df = 958.581/483, CMIN/df = 1.984, RMR = 0.028, GFI = 0.900, AGFI = 0.902, CFI = 0.908, NFI = 0.901, IFI = 0.909, RMSEA = 0.063) proved that the model was a good fit and appropriate for the sample data. Therefore, the analysis was performed without any model modification. The results of the hypotheses in this study are summarized in Table 4.


**Table 4.** Results of hypothesis testing.

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

Specifically, in terms of Hypothesis 1, *Viewing motivation of reality travel variety programs positively a*ff*ects presence*, Hypothesis 1-1, which assumes that vicarious gratification positively affects presence, was positive and significant at the 5% level. Hypothesis 1-2, which assumes that entertainment positively affects presence, was positively related to presence and significant at the 5% level. Hypothesis 1-3, which assumes that information-seeking positively affects presence, was also positive and significant at the 5% level. Hypotheses 1-4 and 1-5, which posit that habitual time-spending and socializing positively affect presence, were not significant and not supported. Regarding Hypothesis 2, which assumes that presence positively affects viewing satisfaction of reality travel variety programs, was positive and significant at the 5% level, thus supporting this hypothesis. Hypothesis 3, which assumes

that viewing satisfaction of reality travel variety programs positively affects attitudes toward tourism destinations, was positive and significant at the 5% level, supporting this hypothesis as well.

#### **5. Discussion and Conclusions**

#### *5.1. Discussion*

This study was conducted with viewers who have experience watching travel reality variety programs; the program format is travel reality show, which is a mixture of travel and reality rather than a movie or drama. This aspect seems to contribute in a way that is different from previous research. In addition, previous studies investigated only the relationship between motivation for viewing a reality program and satisfaction in various aspects, but few studies have analyzed the psychological effect on viewing motivation and satisfaction by presence, by which viewers vividly and realistically recognize the places in the programs. Thus, it can be considered that the attempt to identify presence in travel reality variety programs is a contribution to the scalability of the theoretical application utilized in this study. The results of this study are as follows.

First, among five motivational factors, vicarious gratification, entertainment, and information-seeking have a positive influence on presence. This is because of the nature of travel reality variety programs, which provide destination information and entertainment at the same time, provide accurate information on the locations, then have fun with the content itself, and the vicarious gratification provided by the indirect experience with this kind of program gives viewers a vivid sense of aliveness or presence about the place, which leads to tourists expectations toward the place. This result is meaningful, because it proves that the effect of presence is mediated in travel reality variety programs as well as reality programs [10,48]. Second, it was confirmed that presence had a positive effect on the satisfaction of viewing travel reality variety programs, similar to the presence of reality programs such as "Cookbang" and "Mukbang" [15,24]. It can be concluded that viewers can experience more presence in the place due to the liveliness, which is one of the features of travel variety programs, and this presence has a positive influence on viewing satisfaction. Therefore, presence can be said to be a significant parameter that can influence the relationship between viewing motivation and satisfaction. Third, a positive attitude toward the tourism destination in the program was found when the viewing satisfaction was higher in this relationship. This supports previous research on satisfaction and attitude [49]. Ultimately, in travel reality variety programs, like other reality programs, viewing satisfaction causes the viewers to have good image and expectation for the place, and finally, viewing satisfaction can lead to a positive attitude toward tourism destinations [50].

#### *5.2. Implications*

Above all, we note the theoretical implications of this study. First, this study divided viewing motivation into five variables based on a literature study on previous media-viewing motivation, and it is meaningful that this research considered media diversity by applying reality travel variety programs. In particular, among five motivational factors, considering the characteristics of the travel reality program, adding the variable of "vicarious gratification" as an indirect experience through the screen can also be said to be a meaningful theoretical extension from the previous research.

In addition, presence, the realistic feeling given by a reality travel variety program, was verified empirically as a mediating role between viewing motivation and viewing satisfaction. This is an attractive factor for viewers watching reality programs, so it can be applied to various genres of media in the future. This is also an empirical evidence for the claim that presence is applicable to all media, although there are differences in the type and exposure degree of media [12], and it was confirmed that the experience of presence is a highly significant variable in the travel reality program.

Lastly, it is meaningful that we reconfirmed (based on existing research) the satisfaction of viewing these TV programs and attitudes toward their tourism destinations.

In addition, this study suggests the following practical implications. People often indirectly experience tourism destinations through mass media without actually visiting them. This indirect experience will have a great influence on future behavioral intentions [12,13]. In Korea, due to the diversification of broadcasting channels, each broadcasting company produces programs that include indirect advertisements. Indeed, since the tvN "Over Flowers Series" travel program was broadcast, the number of Korean tourists who visited the broadcast country increased by an average of 52.7% [51]. In this respect, therefore, the results of this study can provide a guideline for mapping strategies to attract tourists to certain places in terms of marketing the tourism destination itself. In particular, the exposure of places in travel reality programs is more realistic and specific than in dramas or movies. In the end, this evokes expectations of a tourism destination [16], and since it approaches the viewer with easy and comfortable images, the effect of enhancing the attractiveness of the space will be strong. Therefore, tourism destinations exposed through travel reality programs are imprinted with a positive image for viewers or potential tourists, and can also affect actual behaviors such as visiting certain locations.

#### *5.3. Limitations and Future Research*

Finally, we present the limitations of this study and the direction of future research. In this study, even though the questionnaire survey was conducted for respondents who had watched the programs, their memories may have been distorted over time. Therefore, subsequent research should be conducted on programs that are being broadcast to obtain more accurate results. In addition, recent travel reality variety programs may also include entertainers of specific genders or ages. In further studies, it will be necessary to consider the ages and genders of the entertainers on various programs. Lastly, the sample for the survey was limited to adults residing in Busan and Gyeongsangnam provinces. Therefore, further research should consider all regions in Korea. In future research, it is expected that more specific and diverse studies on travel reality programs will be conducted based on this study. The results of this study can then be supplemented with meaningful facts, which will enable a wider understanding of the results and generalization of previously discovered theories.

**Author Contributions:** Conceptualization, B.-K.K.; methodology, K.-O.K.; data curation, K.-O.K.; writing original draft preparation, B.-K.K.; writing—review and editing, B.-K.K.; project administration, B.-K.K.; funding acquisition, K.-O.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRFS1A5B5A07XXXXXX).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Topic Modeling of Online Accommodation Reviews via Latent Dirichlet Allocation**

#### **Ian Sutherland, Youngseok Sim, Seul Ki Lee, Jaemun Byun and Kiattipoom Kiatkawsin \***

Department of Hospitality and Tourism Management, Tourism Industry Data Analytics Lab (TIDAL), Sejong University, Seoul 05006; ianosutherland@gmail.com (I.S.), iamssys@gmail.com (Y.S.), seulkilee@sejong.ac.kr (S.K.L.), jaemun.byun@sejong.ac.kr (J.B.)

**\*** Correspondence: kiatkawsin@gmail.com

Received: 24 January 2020; Accepted: 27 February 2020; Published: 28 February 2020

**Abstract:** There is a lot of attention given to the determinants of guest satisfaction and consumer behavior in the tourism literature. While much extant literature uses a deductive approach for identifying guest satisfaction dimensions, we apply an inductive approach by utilizing large unstructured text data of 104,161 online reviews of Korean accommodation customers to frame which topics of interest guests find important. Using latent Dirichlet allocation, a generative, Bayesian, hierarchical statistical model, we extract and validate topics of interest in the dataset. The results corroborate extant literature in that dimensions, such as location and service quality, are important. However, we extend existing dimensions of importance by more precisely distinguishing aspects of location and service quality. Furthermore, by comparing the characteristics of the accommodations in terms of metropolitan versus rural and the type of accommodation, we reveal differences in topics of importance between different characteristics of the accommodations. Specifically, we find a higher importance for points of competition and points of uniqueness among the accommodation characteristics. This has implications for how managers can improve customer satisfaction and how researchers can more precisely measure customer satisfaction in the hospitality industry.

**Keywords:** topic modeling; latent Dirichlet allocation; tourism 4.0; online travel agency; online review; text analytics; improve customer satisfaction; inductive approach; dimensions of interest; era of big data

#### **1. Introduction**

Customer satisfaction is "a person's feelings of pleasure or disappointment that results from comparing a product's perceived performance or outcome with his/her expectations" [1]. Prior research in service industries has shown that customer satisfaction has a strong relationship with future repurchase intention, dubbed customer loyalty [2–4]. This link, in turn, is also suggested to have a strong impact on firm performance in service industries [5,6]. In the tourism industry specifically, research has shown that the quality of visitors' relationship with a destination has a positive impact on their environmentally responsible behavior as well [7]. However, the subjective nature of meeting customer expectations makes measuring customer satisfaction notoriously difficult in the service sector since many dimensions exist to shape customer expectations.

A new paradigm in the form of Industry 4.0 potentially lends itself to holding many of the answers to the barriers that such difficult problems posed in the past [8,9]. Industry 4.0 combines "smart objects" (such as products and machines) with internet technology, allowing for economies of scale but for small-batch-sized productions [8]. This concept has been applied to service industries as well, and in the tourism industry it has been dubbed "Tourism 4.0" among some academics and industry professionals [10–13]. Due to the recency of the Tourism 4.0 paradigm, industry and governments have adopted the concept, for example a government project by Slovenia uses the term [10,13], though

there is little academic research on the concept as of yet. Two of the key information technologies of Tourism 4.0, artificial intelligence and big data [10], are suggested by this study to be robust for identifying customers' topics of discussion by utilizing online user-generated content. As part of the greater Industry 4.0 paradigm, Tourism 4.0 is defined by an "individualization on demand" wherein a shift to a buyer's market rather than a seller's market is a defining characteristic [8] (p. 239), thereby emphasizing the customers' perspective over the sellers'.

We submit that customer satisfaction should be approached from the customer perspective. As such, we argue that online reviews offer a richer source of data than traditional survey methods for better understanding the holistic experiences of customers. Utilizing a topic modeling technique called latent Dirichlet allocation (LDA), we extract topics of discussion in online reviews because they reflect the topics that customers find important enough to discuss, but are not confined to only the service encounter itself nor confined by existing theoretical constructs of service quality. Comparing the topics between different types of accommodations and locations, a better understanding of customer satisfaction can be achieved relative to traditional, deductive approaches. Nevertheless, extant constructs of determinants of customers' experiences in hospitality literature, are critical in the validation of useful information extracted from online reviews, and therefore are presented in greater detail in Section 2.

#### **2. Literature Review**

#### *2.1. Measurement Scales of Customer Satisfaction in Hospitality*

In 1986, Parasuraman, Zeithaml, and Berry published a multi-item survey to gauge the perceptions of service quality in four service sectors and called it SERVQUAL. Subsequent research critiqued SERVQUAL on that it does not adequately cover all service settings [14], and was found not to be optimal for application in the hospitality industry [15–17]. Several industry-specific modifications of the SERVQUAL scale have been introduced to suit the specific needs of customers within the context of the hospitality industry, such as LODGQUAL [15] and HOLSERV [17]. However, these measures focus mainly on hotel customers rather than customers in other accommodation types. Some researchers argue that SERVQUAL, along with similar measures, overemphasizes the provider and fails to represent a modern view of customer satisfaction wherein not only the rational functional product/service benefits, but the irrational emotional benefits affect purchasing decisions [18–20].

As such, the Customer Experience Quality (EXQ) construct was proposed and validated to encompass a more holistic perspective of the customers' experiences [20,21]. However, implementation of EXQ has seen mixed results in the hospitality and tourism literature, such as a recent study in the tourism context that found that the structure of EXQ's four dimensions were not confirmed via confirmatory factor analysis [22]. The shortcomings of currently used scales suggest that more research into customer satisfaction dimensions is still needed.

#### *2.2. Dimensions of Customer Satisfaction in Hospitality*

Measurement of customer satisfaction relies on sufficiently measuring all relevant dimensions of the customer experience. As the most commonly utilized instrument, SERVQUAL measures the perception of quality of the services provided via 22 items in five subscales, which include Tangibility (physical characteristics of the service encounter), Reliability (proper performance of services), Responsiveness (employees' willingness to help), Assurance (employees' knowledge and courtesy), and Empathy (employees' ability to care) [23].

LODGQUAL identified three dimensions as important: Tangibility, Reliability, and a new dimension, Contact (representing the contact experience with employees) [15]. Similarly, HOLSERV found three dimensions: Employees (willingness to help, polite, knowledgeable, etc.), Tangibles (modern-looking equipment, fixtures and fittings, etc.), and Reliability (accurate and timely service, safe and secure stay, etc.) [17].

More recently, research has continued to develop using a variety of dimensions and questionnaire items for specific usage in different contexts within the hospitality industry. One research paper into Jordanian hotels used a modified SERVQUAL questionnaire called SERVPERF to evaluate hotel guests [24]. Research in Serbian 3-star hotels used SERVQUAL [25], and in Taiwan hotel guests were surveyed using another modified version of SERVQUAL [26].

#### *2.3. Satisfaction and Accommodation Modalities*

While a significantly large amount of research exists surrounding hotels and other traditional accommodation types, a relatively smaller amount of research has been done on specialist accommodation. While numerous studies have tried to define the concept of specialist accommodation in various ways, such as 'supplementary accommodation' [27], 'boutique accommodation' [28], and 'specialist accommodation' [29,30], we adhere to the terminology of 'commercial home enterprises' referring to accommodation in which guests may stay in private homes or other facilities that often have shared living areas with a relatively high level of interaction with owners, hosts or other guests, such as B&Bs, homestays, guesthouses, etc. [31–33]. In commercial home enterprises (CHEs), customers have different expectations since the products and services offered differ from traditional hotels [30]. Owners of commercial homes "prioritize personal values, quality of life, and personal interactions above economic gains" and it is reflected in the dimensions valued by CHE guests [34] (p. 115).

A study of bed and breakfasts in the United States found that important dimensions included personalized service, homeliness feel, bed comfort, privacy, cleanliness, and the helpfulness of advice about nearby restaurants and other attractions by the owner [35]. In South Africa, business travelers were observed as finding importance in several other dimensions for bed and breakfasts, such as professional and functional skills and abilities, amenities in general and those provided only for the room, ambiance, secure parking, and staff professionalism [36]. For bed and breakfasts in Taiwan, the facilities, service, environment, management, and emotive aspects were found to be important dimensions [37], while another Taiwanese bed and breakfast study found that the physical environment (e.g., equipment, décor, design ambiance), and personal interaction (e.g., staff expertise and problem solving skills) were important factors [38].

For guesthouses in China, a home atmosphere, facilities in the room and on the premises, cleanliness, service quality, value for money, and the location were found to be important dimensions in an analysis of online user-generated content (UGC) [39]. Similarly, another study of guesthouses in China found that sanitation, service and climate, room facilities, bedding, and shower were important dimensions [30]. It was found that for guesthouses in Ghana, atmospherics, enjoyment, entertainment, feeling of escape from daily life, efficiency, excellence, and economic value were important dimensions [40].

In Korea, an endemic, traditional type of CHE accommodation exists called a "hanok", and it is popular for both domestic and international tourists. Hanok "literally means Korean house and it refers to an abode designed according to the traditional principles of Korean architectural culture" [34]. The traditional principles are known as pungsu-jiri (i.e., Feng Shui), and they direct how the traditional hanok accommodations are constructed [41]. As such, guests have very different expectations of the product and services offered at such establishments. While research is minimal on the subject of customer satisfaction dimensions for Korean hanoks, a study showed that hanok guests value staff and facilities (e.g., staff was helpful, bathrooms were clean, etc.), emotional value (e.g., pleasant stay experience), and functional value (e.g., value for price), wherein all dimensions had either direct or indirect impacts on revisit intention and word-of-mouth [42].

#### *2.4. Extracting Dimensions of Interest in the Era of Big Data*

Research suggests that big data, such as online reviews, is key to solving problems of the 21st century, such as by helping to reach sustainability development goals [43] or by giving insight into consumer experiences and behaviors [44]. Online reviews offer great insight into customer experiences for most services and products, and recent improvements in computing along with advances in methodologies have allowed for the analysis of online reviews at scale. There is evidence showing that online product ratings do not significantly impact sales [45,46]. On the other hand, many studies show a correlation between sales of products and customer's online reviews [47–50]. Thus, implying that the content of online text reviews can be more important than numeric ratings. One study found that recommendations of book reviews were significant with book sales using 58,566 reviews of 610 books via multiple regression [47]. Similarly, another study determined a significant correlation exists between posted reviews and box office sales via pooled regression of 12,136 reviews of 40 movies [48]. Using the differences-indifferences method to analyze 134,904 reviews of 1636 books and 176,112 reviews of 2387 books, respectively, at different time points, it was demonstrated that improvement in book reviews can lead to an increase in relative book sales [49]. A logistic regression analysis of 31,522 reviews of 148 digital cameras indicates that reviews of products on review sites is an effective indicator of sales for high-involvement products [50]. Tobit regression on 1587 reviews of six products showed that while review depth is correlated with helpfulness, review extremity is less helpful particularly for experience goods [45]. Furthermore, no significant impact on box office revenues was found from the ratings of online reviews in a study of 95,867 reviews of 71 movies via three-stage least squares [46].

A comprehensive review of tourism literature for service quality showed that the most prominent issue in the literature from 2008 to 2019 was the impact on customer behavior from service quality [51]. In tourism and service sectors, such as the hospitality industry, studies show that the qualitative content in text of online reviews are influential to customers and offer rich content regarding customer experiences. One study used logistic regression on 5090 restaurant reviews to discover that the qualitative aspects of online reviews were the most influential factor in determining whether the reviews were useful or not [48]. Furthermore, another study used 5090 reviews of 45 restaurants in New York and London to find that the valence of the reviews has a U-shaped effect on usefulness and enjoyment, wherein negative ratings are more useful and positive ratings are associated with more enjoyment [52]. Using content analysis of 42,886 reviews of 774 Beijing hotels, customer satisfaction determinants were extracted through analyzing online reviews [53]. More recently, latent Dirichlet allocation was used to analyze 266,544 reviews from 39,287 unique reviewers of 25,670 hotels in 16 countries in order to extract the relative importance of customer experience dimensions [54]. A total of 29 dimensions were identified, of which, room experience and service quality were deemed most important [54]. In this study, we utilized the same technique to extract the dimensions of customer satisfaction within various accommodation types of South Korea in order to extract and compare the accommodation types without the confounding effect of country-specific differences.

#### **3. Materials and Methods**

#### *3.1. Data Collection and Screening*

This study applied latent Dirichlet allocation to Online Travel Agencies' (OTAs) review corpora. Using scraping techniques applied through Java, the OTA reviews of accommodations across South Korea were extracted from Booking.com, Hotels.com, and Agoda.com over the course of one year, from January 1, 2018 to December 31, 2018. Subsetting the data for only English reviews, the 120,875 observations in the English subset were then cleaned to exclude duplicates and short texts of four words or less, resulting in a final dataset for analysis of 104,161 reviews. Therefore, the final dataset utilized for analysis included text reviews from 104,161 customers and several other corresponding variables including the date of review, data source, name of accommodation, location of accommodation, review rating, and type of accommodation.

Reviews were categorized by both location and accommodation type. For location types, reviews were dubbed as urban (*n* = 88,692) for accommodations located in seven Korean metropolitan cities: Seoul, Busan, Incheon, Daegu, Daejeon, Gwangju, and Ulsan. Otherwise, reviews were classified as rural (*n* = 15,469) for accommodations located outside metropolitan cities within six provinces: Gyeongi, Gyeongsang, Chungcheon, Jeolla, Gangwon, and Jeju. For accommodation types, hotels and motels were dubbed hotels (*n* = 22,011), Korean hanok-style accommodations in which guests usually sleep on the floor in a traditional-style Korean house were classified as hanok (*n* = 966), and guesthouses, bed and breakfasts, and hostels were named commercial home enterprises (CHEs) (*n* = 8784).

#### *3.2. Analysis Tools and Techniques*

Javascript was used for the scraping of data, whereas all data handling, screening, processing, and analysis was achieved through R versions 3.5 and 3.6. Key R packages used were *stringr*, *tidyverse*, and *tm* for handling data and preprocessing. Dictionary stemming was executed with the hunspell stemming dictionary via the *hunspell* package. For assessing the fit statistics of different levels of *k*, the *ldatuning* package was used. Training of the LDA model was accomplished through the *topicmodels* package. Base R coding was favored wherever applicable.

Preprocessing the text prior to modeling is a critical component of successful topic modeling. The first step in preprocessing textual data is the removing of stopwords. A list of common stopwords, such as 'the', 'a', 'and', etc. was used, as well as an extended list of stopwords specific to the dataset, such as 'lol', 'haha', etc. Tokenization of relevant word chunks and phrases was useful to avoid losing meaning in certain cases, such as 'bus stop', 'taxi stop', 'metro station', etc. Dictionary word stemming was selected over algorithmic stemming in order to minimize the occurrence of eliminating useful words, for example by avoiding short words like 'bed' being stemmed to 'b' due to the ending of 'ed'. Dictionary stemming also proved to be more successful than algorithmic stemmers by removing proper nouns, which confounded the creation of useful topics, for example it creates topics focused on location names in reviews rather than the content of the reviews.

#### **4. Results**

#### *4.1. Topic Extraction*

The number of latent topics is an important decision in topic modeling. The optimal number of topics was assessed using several metrics [55–58]. The number of topics was set at 14 (*k* = 14) since it is the optimal number of topics that maximizes information divergence between all pairs of topics [58]. The information divergence criterion was chosen for its sensitivity to changes in the number of latent topics which was found to yield the most practical results in exploratory tests.

After extracting 14 topics through latent Dirichlet allocation, the topics were validated in a four-step procedure. First, a panel of four professors of hospitality and tourism management and one industry practitioner individually assessed the keywords with the highest beta values for each of the 14 topics. Second, the panel was given the topics in context using the top reviews with the highest proportions of each topic. In the third step, each topic was named accordingly by each panel member individually. Fourth, the topics were discussed and voted on until the panel reached a unanimous decision on the most suitable names for each of the 14 topics.

The reviews in the dataset were given probabilities by topic via the posterior distribution of the LDA model. Each review was classified into one of the 14 topics based on the topic that comprised the highest proportion of that review. The classifications were then analyzed by region and accommodation type.

#### *4.2. Topic Identification*

The 14 topics that were extracted via the latent Dirichlet allocation model are summarized in Table 1. Keywords were determined by those words with the highest beta value within the topic, i.e., the words with the highest relative probability of belonging to the given topic and were organized in order from most important to least important keyword per topic. As LDA is an admixture model, the same words can belong to more than one topic. Below, we briefly introduce each topic and the meaning of each.


**Table 1.** Latent topics with relevant keywords.

General guest satisfaction topics include the overall evaluation and the value of the stay. The evaluation topic encompasses general guest evaluations in the veins of return intention (e.g., 'back', 'plan', 'future', 'regret'), meeting/exceeding expectations (e.g., 'expectations', 'satisfy', 'exceed') and recommending to others (e.g., 'recommend', 'mention'). The value topic relates to the price–quality relationship. Keywords related to the accommodation's relative quality (e.g., 'choice', 'quality', 'valuable', 'competitive', 'comparison') signify a quality comparison with competitors in relation to keywords related to price (e.g., 'price', 'money', 'cheap', 'expensive', 'afford').

Service-related topics include reception, staff professionalism, hospitality, and amenities. The topic of reception includes the tasks associated with reception such as checking in and out (e.g., 'check', 'booking', 'arrival', 'leave'), payment (e.g., 'pay', 'charge', 'quote'), and others (e.g., 'card', 'morning', 'call', 'request'), as well as includes the reception staff (e.g., 'receptionist', 'lady', 'guy', 'rude'). The staff professionalism topic encompasses staff service quality (e.g., 'friendly', 'helpful', 'service', 'polite', 'professional'), and communication ability (e.g., 'speak', 'communicate', 'fluent', 'language'). The hospitality topic emphasizes the emotional warmth of the experience (e.g., 'feel', 'experience', 'warm', 'atmosphere') and the hospitality of the owner or staff (e.g., 'owner', 'people', 'kind', 'accommodate', 'care'). The topic of amenities consists of bathroom amenities (e.g., 'towel', 'toiletries', 'shampoo'), room amenities (e.g., 'laundry', 'machine', 'fridge', 'kitchen'), free mini-bar items ('free', 'included', 'provided', 'coffee', 'tea', 'water' ), and breakfast (e.g., 'breakfast', 'toast', 'egg', 'cereal', 'fresh').

Location-related topics include accessibility, mobility, and the surrounding neighborhood. The topic of accessibility relates to transportation (e.g., 'airport', 'terminal', 'international', 'stop', 'transfer') to and from the accommodation from a long distance (e.g., 'overnight', 'express', 'hour', 'layover'). Specifically, modes of transportation often associated with travel to and from the airport are observed (e.g., 'shuttle', 'limousine', 'bus', 'taxi', 'flight'). The topic of mobility relates to transportation (e.g., 'station', 'street', 'distance', 'road', 'exit', 'connect') to and around the accommodation locally (e.g., 'close', 'convenient', 'minute', 'nearby'). Specifically, modes of transportation often associated with travel within a localized area or city are observed (e.g., 'subway', 'metro', 'train', 'walk'). The neighborhood topic encompasses the local attractions (e.g., 'restaurant', 'shopping', 'café', 'market', 'mall', 'bar', 'sightseeing'), centrality (e.g., 'nearby', 'local', 'centrality', 'district', 'heart', 'middle'), and local atmosphere (e.g., 'quiet', 'busy', 'crowded', 'safe') of the location.

Topics related to the physical condition of the room include room size, room condition and décor. The room size topic relates specifically to the size of the room (e.g., 'room, 'small, 'big, 'size', 'space', 'floor', 'large', 'bigger'), type of room ('family', 'double', 'single', 'person', 'upgrade', 'standard'), and changing of rooms along with moving of luggage (e.g., 'luggage', 'upgrade', 'move', 'lift', 'carry', 'stairs', 'elevator'). The room condition topic relates to the physical aspects of the room (e.g., 'bathroom', 'shower', 'toilet', 'floor', 'pillow', 'sheet', 'sink', 'wall', 'door') in relation to the condition of those physical aspects in terms of cleanliness (e.g., 'smell', 'dirty', 'change', 'hair', 'smoke', 'stain', 'wet', 'mold', 'housekeeping'), and comfort (e.g., 'hard', 'privacy', 'broken', 'soft'). The decor topic refers to the interior decoration and design of the room (e.g., 'decorate', 'interior', 'modern', 'chic', 'stylish', 'furnishing', 'minimalist', 'decor', 'elegant', 'stiff') as well as the general comfort and atmosphere (e.g., 'clean', 'comfortable', 'cozy', 'spacious', 'neat', 'environment', 'roomy').

Topics related to physical condition of the environment of the accommodation include the facilities and ambiance. The topic dubbed facilities relates to the facilities of the accommodation (e.g., 'facility', 'pool', 'rooftop', 'gym', 'bar', 'spa', 'lounge', 'sauna', 'balcony', 'garden', 'terrace'), and, to a lesser extent, the view of the surrounding area (e.g., 'view', 'beautiful', 'ocean', 'beach', 'sea', 'park', 'mountain'). The ambiance topic mainly encompasses three distinct ambient qualities of the room including sound (e.g., 'noisy', 'hear', 'sound', 'loud'), light (e.g., 'light', 'dark'), and air (e.g., 'hot', 'cold', 'heat', 'stuffy', 'ventilation'). Things that directly affect these ambient qualities of the room are also included in the topic (e.g., 'AC', 'air', 'conditioner', 'heater', 'floor', 'window', 'fan', 'thin', 'wall', 'construction', 'season', 'winter', 'summer').

#### *4.3. Topic Validation*

In addition to keyword analysis of words with the highest beta values in each topic, validation of the topics was achieved by analyzing the review text with the highest proportion of the given topic. For example, the review with the highest proportion of the accessibility topic starts off with, "Bad memory about Airport shuttle bus to hotel. We catched a Limousine Airport bus to [the] hotel ... but I didn't know the bus stop at [the] hotel was the opposite road, we lost our way to Myeongdong ... ", and continues discussing the author's difficulties with accessibility from the airport to the hotel. Validation of topics was achieved by analyzing the top 10 reviews for each topic by each member of the panel, in addition to the keywords, and then voting until unanimity of naming the topics was achieved by the panel. For the sake of length, the full validation of topics is not expanded upon in this paper, however the authors strongly urge the importance of validating the topics in LDA by analyzing not only the top keywords, but analyzing the words in context of the reviews by evaluating the most prominent review for each topic.

#### *4.4. Comparisons of Accommodation Characteristics*

The number of reviews for each topic by the characteristics of urbanization of location, and accommodation types are reported in Table 2, while Table 3 reports the proportions of reviews as a percentage of each trait. Due to the relatively small sample size of hanok reviews, a larger discrepancy

#### *Sustainability* **2020**, *12*, 1821

is required for statistical significance, so differences in proportion sizes should be inferred with caution in the hanok category.


**Table 2.** Latent topic review frequencies by accommodation characteristic.

**Table 3.** Latent topic review proportions by accommodation characteristic.


Note: statistical significance levels are shown as 0.10 ≥ *p*-value > 0.05 (•), 0.05 ≥ *p*-value > 0.01 (\*), 0.01 ≥ *p*-value > 0.001 (\*\*), *p*-value ≤ 0.001 (\*\*\*). Significance levels shown indicate statistical significance in two-proportions *z*-tests between the proportion of a give topic in a subset and proportion of the same topic in all other subsets.

Chi-square goodness of fit tests were applied to assess whether the distributions of topics within each trait (e.g., Urban, Rural, Hanok, etc.) were distributed equally. Wilcoxon Signed Rank tests were applied to assess whether there is a difference between the distribution of topics in one trait with all corresponding traits within the same characteristic (i.e., Urbanization, or Accommodation Type). Two proportions *z*-tests were applied to test whether a statistically significant difference exists between the proportion of each topic for each trait compared to the proportion of each topic in the corresponding traits.

#### 4.4.1. Urbanization Comparison

The chi-square goodness of fit tests reveal that topics are not equally distributed in reviews of urban locations (χ<sup>2</sup> = 3339.40, *p*-value < 0.001) and rural locations (χ<sup>2</sup> = 591.92, *p*-value < 0.001), thereby showing a difference in the importance of topics among customers staying in both urban and rural locations. Wilcoxon Signed Rank tests demonstrate that the distribution of the prominence of topics between urban versus rural locations are not the same (*p*-value < 0.001). Specifically, the most prominent topics for urban accommodations are evaluation, staff professionalism, and reception. Whereas, rural accommodations observe the highest proportion of reviews regarding both value and evaluation, followed by staff professionalism and hospitality. Comparisons between urban and rural accommodations via the proportions for each topic are visualized in Figure 1.

**Figure 1.** Proportions of topics by urbanization.

Two proportions *z*-tests determine whether a statistically significant difference exists between the proportion of a given topic between the urban and rural locations and their statistical significance is reported in Table 3. Reviews of rural accommodations relative to urban accommodations have a statistically higher focus on value, hospitality, ambiance, mobility, and neighborhood. We submit that the emphasis on warm hospitality in a location with easy access to the local area and sights is a response of guests wanting to connect with people and the local environment at a low price. The relatively lower emphasis on topics such as reception, room size, staff professionalism, facilities, room condition, and décor reinforce the image of customers staying in the countryside in order to escape the higher luxury and social rigidity of the metropolitan areas, in favor of stronger connections with the local people and the area.

Likewise, the higher focus on reception, room size, staff professionalism, facilities, room condition, and décor in urban accommodations relative to rural locations supports the idea that metropolitan guests tend towards professional service and chic, comfortable facilities with greater access to the airport compared to their rural counterparts. The lower emphasis on value indicates that the metropolitan

clientele are less concerned about the price and more concerned with their overall satisfaction from higher service quality compared to overnight travelers to countryside towns.

#### 4.4.2. Accommodation Type Comparison

The chi-square goodness of fit tests reveal that topics are not equally distributed in reviews of hotels (χ<sup>2</sup> = 901.02, *p*-value < 0.001), commercial home enterprises (χ<sup>2</sup> = 296.12, *p*-value < 0.001), and hanoks (χ<sup>2</sup> = 80.00, *p*-value < 0.001), thereby showing that some topics are more prominent than other topics for each accommodation type. The rank order of topics was compared using dependent two-group Wilcoxon signed rank tests between all pairs of accommodation types. All comparisons show a statistically different rank order between accommodation types at *p* < 0.001, meaning that the importance of topics differs between customers staying overnight at hotels, CHEs, and hanok accommodation. Specifically, hotels experience most reviews for evaluation, value, and professionalism. The most prominent topics for CHEs are reception, evaluation, and amenities. While similarly, hanok reviews are most prominent regarding reception, evaluation, and mobility.

Finally, two proportions *z*-tests were performed for each topic within each accommodation type and significance is reported in Table 3 as whether the topic within the type of accommodation is significantly different from the proportion of the same topic in all other accommodation types. The comparative proportions for each topic by accommodation type are visualized in Figure 2.

**Figure 2.** Proportions of topics by accommodation type.

Points of competitiveness and uniqueness tend to be discussed topics with significantly higher proportions, while points that are often standardized or that have less variation are observed with much lower emphasis between accommodation types. A point of particular interest is the high proportion of ambiance in the traditional Korean hanok reviews relative to the lower proportions in CHEs and even lower in hotels. Hanoks are unique in that the building design virtually always has a heated 'ondol' floor system and guests often sleep on mats on the floor, which, from foreign travelers' perspectives are a very unique point to overnight stays in hanoks. This ondol system is sometimes used in commercial home enterprises as well, but is much rarer in hotels. In fact, ambiance is one of the lowest ranking topics in hotels, which may be due to precisely the opposite effect—standardized

control of air temperature and other ambient factors in hotels make the topic much less variable, and therefore less prominent as a topic of discussion for customers.

Another notable point of competitiveness is high price competition in hotels. The topic of value indicates that a relatively high proportion of reviews are dedicated to discussing the value of the hotel stay, while lesser so in CHEs. Commercial home enterprises, while also competitive on price, are much less so, since CHEs are comparatively less variable in price and already relatively cheap. We expect overnight travelers staying in hanoks are more interested in the experience than the price and are therefore less price sensitive.

While hanoks and hotels sometimes offer breakfast, it is not a considerably strong selling point. However, in CHEs, breakfast and other amenities are a strong selling point since many travelers staying in guesthouses, bed and breakfasts, and hostels are traveling on a budget and most CHEs offer some breakfast service, whether paid or otherwise. As might be expected, due to the competitive nature of amenities in guesthouses, it is of higher concern for reviewers, while much less so in both hotels and hanoks.

What makes each accommodation unique relative to other accommodation types is also a source for statistical differences in topic prevalence. Hotels highly prioritize professional service and, therefore, staff professionalism is notably more prevalent in hotel reviews. However, both hanoks and CHEs are considerably smaller than hotels, so without as many resources for staff service training, hanoks and CHEs instead focus on creating a more intimate and less formal relationship with customers. As such, we see a higher prominence of reception (and while not a significant difference, it may be notable that hospitality is also higher), rather than staff professionalism in CHEs and hanoks relative to hotels.

Another difference between accommodation types is the importance of location. While there are no significant differences between the prominence of the topic of neighborhood, there do exist differences in getting to and from the accommodation locations. Specifically, accessibility to the hotel from long distances is of more importance than accessibility to CHEs and hanoks. We propose that the reason is based on the purpose of stay. While access to the airport is often an important trait for hotels, the main reason of stay in a commercial home enterprise is generally price or connection with the locals, while hanok guests are generally most interested in the experience. Similarly, while mobility to and from the accommodation to the local area is also important in hotels, it is less important in CHEs, since the purpose of stay in guesthouses is price and also connecting with other people. For hanoks, however, mobility is important since hanoks are often built in clusters in close proximity to other hanoks in old neighborhoods. As such, traveling out of the old neighborhoods into the city centers and to other attractions might be an important concern.

#### **5. Discussion**

#### *5.1. Conclusions*

Latent Dirichlet allocation provides an inductive, data-driven approach that validates and extends current theory regarding the dimensions that affect customers in the hospitality industry by utilizing large, unstructured text directly coming from the consumers themselves. More precisely, important dimensions seen in literature such as service quality, location, value, etc., are corroborated by the LDA approach. However, LDA extends the theory by offering more precise distinctions between the dimensions. Whereas extant literature identifies the location as important [39,54], this study extracted three dimensions related to the location of the accommodation—its accessibility to and from out-of-town, the mobility in the local area, and the attractions and centrality of the neighborhood. LDA contributes further by distinguishing a statistical difference between different aspects of service quality in the topics of staff professionalism, hospitality, and reception as separate topics, as opposed to different items within the same dimension [24,30,37,39,40,42].

Evaluating the extracted topics by their proportionality within different characteristics of the accommodations reveals that points of competitiveness and uniqueness are the topics of higher relative prominence. Competitive points such as value in hotels, and staff professionalism in metropolitan cities are of statistically more prominence relative to the same topics in other accommodation types or rural towns. Points of uniqueness also see higher relative prominence, such as the ambient temperature in hanoks which are heated by traditional floor heating systems, or the neighborhood attractions in rural areas which generally offer more unique attractions than those in metropolitan areas.

#### *5.2. Implications*

Similarly to other studies using LDA [54], this study identifies more dimensions than traditional studies have identified. Since topics are determined statistically and inductively based on the consumers' perspectives, the results should guide future researchers in developing more precise measurement tools that consider a holistic approach to customer determinants of satisfaction. The results also demonstrate that the importance of topics differs by accommodation characteristics, showing which topics are more important by urbanization of the city or accommodation type. We recommend researchers consider such factors when developing measurement tools, in that, all topics do not have equal weighting for shaping customer experience and those weightings differ for accommodations with different characteristics.

Results indicate topic areas that customers willingly choose to discuss and therefore the results bring the attention of industry practitioners to the topics that are relevant. However, customer reviews should be viewed with caution since they do not indicate valence. Therefore, it is of utmost importance to take the individual characteristics and preferences of the customers into account but to use the recommendations of this research to focus on which areas of interest those customers may focus on. This study helps hotel managers develop those areas of concern within their operations from the consumers perspective, not the providers perspective.

#### *5.3. Limitations and Future Research*

As the digital world, along with technologies of Tourism 4.0 such as big data and artificial intelligence, are being harnessed more efficiently and effectively in the hospitality industry [12], areas of future research are rapidly opening up. The new technologies in this paradigm shift are fundamental toward better understandings and predictions of tourist and consumer behaviors. Particularly, we hope to see future research applying topic modeling and other techniques in order to better serve the individualized preferences of customers and progress towards more sustainable business practices in the tourism industry. For example, research shows that better prediction and understanding of tourists' behaviors can better help inform decisions on how to mitigate environmental impacts from tourism, such as cruise tourism [59], or with tourists renting cars [60]. We expect that better understanding of visitors' behavior in the hospitality industry will better inform policymakers and managers for better environmental practices as well as business practices.

As with all research, several important limitations should be noted in the study. Using a sample of English reviews of Korean accommodations allows for a more precise understanding of the differences between accommodation types and locations since it is not confounded by the different characteristics that exist between countries within the same accommodation types. It also allows for a clearer distinction between countryside and metropolitan accommodations since, for example, there is more commonality between Korean metropolitan cities than there is between Korean metropolitan cities and other countries' metropolitan cities. However, the inherent weakness of a single-country analysis means a lack of generalizability of the results to other countries. Therefore, we urge that further research be carried out to reveal the distinctions between accommodation characteristics in other countries as well.

Secondly, the use of online reviews may be biased in terms of the topics of interest being discussed. We expect selection bias in terms of which types of customers post reviews may exist, particularly in terms of the extremity of reviews. For example, customers with a particularly good or extremely bad experience are more likely to post an online review to recommend or warn others about the

accommodation. While this would likely help to distinguish the topics of interest, it may bias the prominence of which topics are represented in terms of tending toward extremity.

Finally, due to limitations in the data available, many accommodation types are unlisted, and some are relatively broad categorizations, so more precision should reveal more distinctions between accommodation type categories. While star-rated hotels were used for the hotel categorization, lower rated hotels are often dubbed motels, and were not distinguished in the data used. Furthermore, different types of commercial home enterprises, such as hostels, guesthouses, pensions, and bed and breakfasts could also not be distinguished between. We expect more differences exist within the three broad categories of accommodation types that were used in this study, and hope that further research will shed light on those differences in the future.

**Author Contributions:** Conceptualization, I.S., K.K., S.K.L. and Y.S.; data collection, Y.S. and S.K.L., analysis, I.S., K.K., S.K.L. and J.B.; writing, I.S.; editing, I.S., K.K., S.K.L., J.B. and Y.S.; supervision, S.K.L. and J.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding

**Acknowledgments:** We would like to thank the Tourism Industry Data Analytics Lab members who helped support the research with helpful discussion—Youngjoo Ahn, Namho Lee, Young Sun Baek, Nahyoung Kim, and Cheolwoo Park.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Shared Short-Term Rentals for Sustainable Tourism in the Social-Network Age: The Impact of Online Reviews on Users' Purchase Decisions**

#### **Jie Zhao \* and Zhixiang Peng**

School of Business, Anhui University, Hefei 230601, China **\*** Correspondence: zhaojie@ahu.edu.cn

Received: 9 July 2019; Accepted: 26 July 2019; Published: 28 July 2019

**Abstract:** With the development of social networks and the Internet-based sharing economy, shared short-term rentals are emerging as a new kind of service that provides a convenient way for people to buy short-term rental services in cities through social-network-enabled platforms. However, like other social-network-based services, shared short-term rental is also likely to be impacted by online reviews. This paper aims to investigate the impact of online reviews on users' purchase decisions toward shared short-term rentals, and further to provide optimization suggestions for the future advance of shared short-term rentals. The contributions of this paper are many-fold. First, we introduce the Stimuli-Organism-Response (SOR) model into the study and propose new variables for the model, including stimulus variables, organism variables, response variable, and moderating variables. Second, we propose eight hypotheses to evaluate the impact of online reviews on users' purchase decisions toward shared short-term rentals. Finally, we collect data through a questionnaire survey and present comprehensive results on many aspects. Based on the data analysis, we find out that the quality of online reviews impacts users' perceived value and perceived risk, which in turn impacts users' purchase decisions toward shared short-term rentals. In addition, the cognitive needs of users can adjust the impact of online reviews on the perceived risk of users but have no explicit adjusting effect for users' perceived value. Further, we present some research implications as well as suggestions for rental platforms to advance shared short-term rentals in the Internet age.

**Keywords:** shared short-term rental; sustainable tourism; online reviews; purchase decisions; social networks

#### **1. Introduction**

The popularity of the Internet and social networks have made the sharing economy develop rapidly in modern city life [1]. Internet-based sharing economy, such as shared rentals, shared cars, and shared bicycles, offers many kinds of smart services to cities, becoming an intrinsic feature of smart cities [2,3]. Among all sharing economy products, shared short-term rentals have become a recent focus, because of the rapid development of the sustainable-tourism industry [4,5]. An Internet-based platform for shared short-term rentals provides a convenient way for people to sell or buy short-term rental services in cities through the Internet. This is especially helpful for tourists. As a result, many people consider shared short-term rentals as their first choice of accommodation during a travel.

Shared short-term rental is a product of sharing economy in the field of hotel accommodation [6]. It means that the owners of houses can share their rooms or houses to get economic returns within a certain time period. This new type of leasing mode develops rapidly due to its diversified leasing style, simple operation, and low cost. However, due to the inequality of information, consumers usually do not trust shared rentals and are eager to know necessary information about shared rentals before they buy the service [7]. Therefore, online reviews, probably provided by social-network-enabled rentals

platforms, become a major information source for people to make decisions [8]. Online reviews are popular in social networks; they can reflect the real experiences of consumers, which are more reliable than the advertisements or other information offered by service providers. At the same time, online reviews cover a wide range of contents and diverse forms of expressions, which can effectively stimulate consumers' psychological perception and influence users' decisions on shared short-term rentals.

So far, the related research on shared short-term rentals mainly focused on the operation modes of shared short-term rentals, factors influencing shared short-term rentals, and the impact of shared short-term rentals on traditional hotel industries [7,9,10]. Factors affecting shared short-term rental decision include the owner's features, reputation, listing status and location, and rental rules. However, most of the existing explorations of influencing factors are based on websites or tenants, and the results of such explorations cannot well explain the impact of shared customer reviews on consumer decision on shared short-term rentals. At present, there is little work focusing on the internal influencing mechanism of online reviews on users' decisions toward shared short-term rentals. Thus, this paper builds a research model to study the impact of online reviews of shared short-term rentals on users' purchase decisions under the shared economic environment. In particular, this study aims to answer the following questions:


Briefly, we make the following contributions in this paper:


The remainder of this paper is organized as follows: Section 2 reviews related work. Section 3 presents the theoretical research model and the research hypotheses. Section 4 details the data collection and analysis, and Section 5 discusses research implications and proposes some suggestions for the future development of shared short-tern rentals. Finally, Section 6 concludes the whole paper.

#### **2. Background and Related Work**

#### *2.1. Sharing Economy and Shared Short-Term Rentals*

Felson and Speath first proposed the concept of sharing economy [11]. They argued that sharing economy is an information-technology-based marketplace for platforms created by third parties. Individuals can achieve point-to-point direct commodity and service transactions through third-party market platforms. Sharing economy offers users the opportunity to enjoy products without ownership. The widespread use of mobile Internet and online marketing has made sharing economy a sustainable profitable area [1]. Generally, the sharing economy can be divided into three stages. The first stage is code sharing, initiated by Linux. The second stage is content sharing on YouTube or life event sharing on Facebook, which is boosted by Web 2.0 and social networks. The third stage is the sharing of offline

resources [12]. Houses as a kind of offline resource began to be a sustainable and profitable resource with the rapid development of the sharing economy, because the sharing economy offers house owners opportunities to get economic benefits.

Shared short-term rentals developed with the sharing moving of offline resources. Zervas et al. took the American Airbnb company as an example to study the impact of the short-term rentals on the tourism and accommodation industry [9], which indicated that Airbnb's short-term rentals highly impacted the local hotel industry. As a result, it decreased the revenue of the local traditional hotel industry by 8 to 10%. In order to exploit users' willingness to accept shared short-term rentals, Botsman et al. conducted a study from the perspective of economic motivation and social motivation [12]. They found that the appearing of shared short-term rentals was highly associated with the global economic recession, which caused economic and social pressure and made people consider the cost-effective shared short-term rentals for accommodation. In addition, Liu et al. empirically studied the strategy of online advertising [13]. They found that the attractiveness of shared short-term rental advertisements and the sense of user power interacted with the willingness to click and the willingness to purchase. In recent years, scholars have been keen to explore the impact of landlord photos on users' decisions on shared short-term rentals: Ert et al. used the experimental method to study Airbnb and found that compared with the landlord's reputation, the landlord's photos can influence consumers' shared short-term rental behavior [14]. Zhang et al. and others studied the impact of the source and quality of landlord's photos on the use of housing resources [15]. They found that the platform professionals took more convincing pictures, and high-quality pictures would also increase landlord's income. In addition, they argued the influence of the source and quality of the landlord on the use of the property. They also found that the high-quality images taken by the platform professionals were more convincing which could increase the income of the landlord.

With the rapid development of the Internet, more and more users start to experience online hotel booking on Internet-based hotel reservation platforms such as Booking.com and Ctrip.com. According to a recent study [16], a ten-percent increasing of online reviews on hotel reservation platforms would lead to five-percent increasing of online booking orders. This shows that online reviews are influential to hotel booking. Moreover, a study of Swiss hotels revealed that the number of positive online reviews had a significant positive impact on hotel sales revenue [16,17]. Other researchers [18,19] considered the users' professionalism when quantifying the impact of online reviews on users' decision-making toward hotel booking. In addition, other features like user characteristics that might impact hotel-booking intention were analyzed [20]. In summary, previous studies have shown that online reviews have a significant impact on the sales of the hotel industry. However, differing from existing studies that focused on hotel booking, this paper concentrates on shared short-term rentals. Most of short-term rentals are private apartments. As the features of apartments surely differ from those of hotels, online reviews toward shared short-rentals are also quite different from those toward hotels. To this end, it is reasonable that users purchasing shared short-rentals have varying needs compared to users in the traditional hotel-booking market.

#### *2.2. Perceived Value and Perceived Risk*

The concept of perceived value was formally proposed by Porter [21]. Perceived value refers to the competitive advantage of a company compared with other competitors. Since then, perceived value has been applied to the research of e-commerce, network shopping, and market development [22,23]. Zeithaml et al. studied perceptual value from a psychological perspective [24]. He believed that perceived value was the overall evaluation of a customer toward a service or a product of a particular company. He also claimed that value was superior to other factors. Further, value can be perceived and was influenced by personal knowledge. To this end, perceived value is subjective.

Perceived value was divided into functional value and emotional value [25]. Later in the field of brand research, Sweeney et al. measured the perceived value of customers of high-end durable goods from the four dimensions of emotion, society, quality, and currency [26]. Based on Sweeney's work, Petrick et al. further studied the measurement of perceived value and proposed five dimensions, including quality perception, emotional value perception, currency value, industry cost, and reputation [27].

Another factor that has been extensively studied in the field of e-commerce is perceived risk. Since Bauer proposed the concept of perceived risk of consumers [28], many scholars have discussed the issue and have presented numerous extended definitions. Perceived risk is a possibly previous measure of consumer perceived usefulness and perceived ease of use towards purchasing on the internet. Bauer believed that human behavior was highly uncertain; thus so people's behavior was risking. Due to the inequality of information, users' decisions on shared short-term rentals are also highly uncertain. Cox et al. believed that perceived risk can be divided into two dimensions [29]: one is that consumers may be suspicious of bad consequences before making consumption decisions; the other is the extent to which an accident affects them. People are used to reducing perceived risk before making decisions. For sharing this emerging thing and model, the user's uncertainty may be higher compared to the traditional consumption model, and perceived risk is more valuable for studying user behavior.

So far, there have been a lot of studies toward perceived risk. Peter and Ryan pointed out that people will experience perceived risks from a social, economic and functional perspective [30]. Jacoby et al. argued that the perspective of perception should include five aspects [31]: body (whether user behavior will cause physical damage), property (whether user behavior will cause loss of property), society (whether user behavior will threaten user position), function (whether user behavior will satisfy user's own needs), and psychology (whether use behavior will impact user's emotion). Peter and Tarpey proposed the dimension of time based on the above research [32]. In the shopping environment of the Internet era, Jarvenpaa et al. argued that consumers' perceived risk needs further exploration [33]. They recalibrated the six dimensions to five, which include efficacy level, economic level, security level, privacy level and social level. The level of privacy is based on the new environment. He points out that when consumers consume in the network environment, their personal data will be leaked and used. This perspective is also the product of the new environment. In short, the uncertainty of the consequences of user behavior creates a hesitant state of mind that affects the generation of behavior. Although there are a few existing research on shared services and economy [34], consumer uncertainty may be more serious in the Internet age. The high quality information provided by online reviews may affect the perceived risk of consumers, and further affect users' decisions on shared short-term rentals.

#### *2.3. Users' Purchase Decisions*

Users' purchase decisions are always an important issue in the field of user behavior [35]. Online reviews are also regarded as a stimulus for consumers to buy share short-term rentals. Positive online reviews will make consumers feel that they can benefit from share short-term rentals, which will guide the purchase behavior of users. User behavior is an act or activity. Nicosica et al. believed that user behavior is a purchase, not a resale [36]. Other researchers claimed that users' purchase-decision behavior is the activity of consumers to choose, use, and process products or services [37]. This includes not only purchase, but also ideas, advice, and actions related to the purchase behavior of users. In this paper, users' decisions on shared short-term rentals are regarded as one kind of purchase behavior. Feeny argued that user behavior is an activity in which users choose, use, or process products or services [38]. Users have temporary use rights for idle houses by booking shared short-term rentals; this behavioral user decision on shared short-term rentals can be regarded as a process. In addition, user behavior can also be a process that encompasses all activities for a product or service [39]. The essence of user behavior is an activity in which consumers satisfy their needs by purchasing products or services. In this paper, the main concern is whether users are engaged in bookings when they are shared short-term rentals.

Users' purchase decisions refer to the optimal purchase plan that is analyzed, evaluated, selected, and implemented among two or more purchase plans that are available for selection in order to meet a

certain demand under the control of certain purchase motivation as well as post-purchase evaluation activities. User purchase decision has been mostly studied in the area of online shopping and social networks [40]. Olbrich et al. analyzed certain social commerce data, and found that products with labels and praise had better sales [41]. In addition, Moody et al. revealed that the seller's reputation, size, amount of information, transaction security, communication, and word-of-mouth had a positive impact on user information and purchase intention [42]. Previous research results showed that the collection and processing of information was an important factor affecting users' purchase intension. In the field of online shopping, users' online reviews posted on platforms are often one of the main sources for user to get information about products. Godes et al. used online reviews as antecedents for decision-making based on users' past experiences [43]. Online review has turned traditional interpersonal communication into online communication. Online reviews spread more widely and faster than traditional product information, and have a deeper impact on sales. In short, there are many factors affecting user behavior, and online review has an important impact on users' purchase intension [44]. In this study, we attempt to reveal the impact as well as the internal mechanism of online reviews on users' decisions on shared short-term rentals.

#### **3. Research Model and Hypotheses**

#### *3.1. Research Model*

In 1974, Mehrabian and Russwell proposed the Stimuli-Organism-Response (SOR) model based on environmental psychology [45]. The model is used to explain the relationship between the physical environment and human behavior, consisting of an antecedent variable (Stimulus), a mediator variable (Organism), and a result variable (Response). In the SOR model, S represents the external environmental stimulus that causes individual response, O represents the evaluation of the stimulus by the organism or the subject of the response, and R represents the behavioral response after the evaluation of the stimulus. The SOR model implies that external stimuli affect an individual's behavioral response by affecting the individual's mental state. Eroglu et al. (2003) improved the SOR model for external online shopping scenarios by using online product display as an external stimulus to explore customer online buying behavior from the perspective of online user sentiment and cognitive changes [46].

In this study, we adopt the SOR model as the fundamental research model. We believe that online review on the shared short-term rentals platforms is the external environmental stimulus that causes individual reactions, which affects the individual's psychological perception, i.e., the internal evaluation of shared short-term rentals. Then, users can make purchase decisions. We also use online review as an independent variable to explore the internal mechanism of users' decision behavior of shared short-term rentals through the mediating role of perceived risk and perceived value of users. In addition, this paper attempts to explore the moderating effect of online reviews on users' psychological perception.

Figure 1 shows the research model of this study, which mainly measures the relationship between the quality of online reviews and users' decisions on shared short-term rentals. The higher the quality of online reviews, the more complete the information provided, and the more effective it is to stimulate users' behavior. For the intermediary variables, we mainly measure two aspects of user perception, namely perceived value and perceived risk. Shared short-term rentals, when compared with traditional city hotels, have higher uncertainty because of the less understanding of users on shared short-term rentals.

We also introduce a new moderating variable called cognitive needs of users (shown as cognitive need in Figure 1). That is because the personality characteristics of users are the main adjustment factors to shared short-term rentals. In general, if the information obtained by users is more complete, the perceived value is higher and the perceived risk is lower. However, this depends on users' personal cognitive needs.

**Figure 1.** The research model.

Cognitive needs refer to the process in which users participate in thinking and enjoying thinking in the process of understanding things. Users with high cognitive needs are eager to get more information about product attributes, which can help them make decisions. Table 1 summarizes the variables of this research model.

**Table 1.** Variables in the research model.


#### *3.2. Research Hypothesis*

Users' decisions on shared short-term rentals have a major impact on the development of shared short-term rental platforms. Therefore, it is necessary to explore the factors that influence users' decisions on shared short-term rentals. Online reviews are one of the important factors that affect users' online shopping behavior. Previous work has demonstrated the impact of online reviews on hotel accommodation sales and users' decision-making [47,48].

According to the research model shown in Figure 1, the quality of online reviews as an independent variable affects users' decisions on shared short-term rentals by affecting users' perceived value and perceived risk. In order to explain the internal mechanism of online reviews on the impact of users' decisions on shared short-term rentals, we first set up the following questions, as shown in Table 2. The Q1 question is to find out the influential relationship between independent variables and the mediating variable. The Q2 question is to find out the influential relationship between the mediating variables and the dependent variable. The Q3 question aims to find out the intermediate relationship among all the factors in the research model. Finally, the Q4 question is to find out what effect the moderating variable has on the relationship between independent variables and mediating variables.


**Table 2.** Research questions and corresponding hypotheses.

Previous studies have pointed out that perceived value is the ratio between the total revenue of a product or service users get from their purchase and the total cost they pay for it [24]. The results showed that the high information quality provided by websites leads to lowness of users' perception on costs as well as the increasing of the perceived value of users. According to these studies, in the sharing economy environment, the high quality of online reviews displaying on shared short-term rentals platforms indicates that platforms can offer high quality information, which can therefore lower the perceived cost of users and increase the perceived value of users.

On the other hand, perceived risk is also the subjective feeling of users, referring to the uncertainties and negative consequences that users perceive subjectively in purchase activities. Murray et al. showed that when potential users perceived high risk in the shopping process [49], they tended to use Internet word-of-mouth to search for relevant products or services. Through summarizing the research literature on the influencing factors of perceived risk of online shopping at home and abroad, we found that perceived risk of consumers mainly comes from three aspects: individual factors (gender, involvement degree, etc.), product factors (product type, brand, etc.) and environmental factors (economy, culture, etc.) Since online reviews reflect users' experiences in the real world, we can reasonably assume that online reviews are more reliable than the information provided by shared short-term rentals platforms. Thus, users are likely to refer to online reviews when they try to make a decision.

In order to answer these questions, we set up several hypotheses:

**Hypothesis** (H1a): *In shared short-term rentals, the high quality of online reviews will increase the perceived value of users.*

**Hypothesis** (H1b): *In shared short-term rentals, the high quality of online reviews will lower the perceived risk of users*.

Users' decisions on shared short-term rentals also follow the general process of user behavior, meaning that users undertake complex psychological activities under the influence of external stimuli and finally make behavioral decisions. In the study of the relationship between online reviews and user behavior, some scholars believed that online reviews had an impact on user behavior. What really work are not online reviews themselves, but the subjective psychological perception that users form after reading online reviews, including perceived value and perceived risk. Specifically, when choosing to use shared short-term rentals, users' psychological perception directly affects users' decisions on shared short-term rentals. Thus, we make the following hypotheses.

#### **Hypothesis** (H2a): *User perceived value has a significant impact on users' decisions on shared short-term rentals.*

#### **Hypothesis** (H2b): *User perceived risk has a significant impact on users' decisions on shared short-term rentals.*

In addition, Sussman et al. integrated the elaboration likelihood model (ELM) and the technology acceptance model (TAP) and proposed an online review information adoption model [50]. This model mainly analyzes the influence of users' adoption of online review information on their consumption decision-making behavior in the network environment. They also proposed to use perceptual value as a mediating factor in the relationship between online review quality and user behavior of travel decision-making. In addition, Miyazaki et al. pointed out that the perceived risk is negatively correlated with user consumption behavior through the study of perceived risk antecedent variables [51]. Users' psychological perception is the subjective attitude of people. When reading the same online reviews, users' psychological perceptions may be different, which leads to different decision-making behavior. In other words, in the context of sharing economy, users' psychological perception may play a mediating role in the relationship between online reviews and users' decisions on shared short-term rentals. Thus, we make the following hypotheses:

**Hypothesis** (H3a): *The perceived value of users plays a mediating role between online reviews and users' decisions on shared short-term rentals.*

**Hypothesis** (H3b): *The perceived risk of users plays a mediating role between online reviews and users' decisions on shared short-term rentals.*

Next, we consider that the influence of online reviews on users' psychological perception. We guess that users' psychological perception depends not only on the characteristics of online reviews, but also on the characteristics of recipients. This is because users may have different personality characteristics and attitudes towards online reviews, which may lead to different user purchase decision. Cognitive needs refer to the process in which an individual participates in and enjoys thinking in the process of recognizing things. Users with high cognitive needs are eager to get more information about product attributes that can help them make decisions. Cacioppo et al. carried out experiments to measure the level of cognitive needs [52]. They found that people can be divided into two categories: a group of people with low cognitive needs, who think that the situation should be orderly and regular, and another group of people with high cognitive needs who think that the situation is not clear and they will analyze the situation according to personal experience to understand the situation. Cacioppo et al. also pointed out that people with high cognitive needs prefer to think and form a brand attitude through thinking. On the contrary, people with low cognitive needs are susceptible to the side-effects of advertising. That is to say, people with different cognitive needs have different ways of processing information. In general, the information about the products with high-quality online reviews is more about post-consumer experiences and feelings, which can meet the needs of users with low cognitive needs and fasten decision-making processes. In summary, because of high-quality online reviews, users with low cognitive needs may get high perceived value and lower perceived risk. Thus, we make the following hypotheses:

**Hypothesis** (H4a): *Cognitive needs play a moderating role during the process where online reviews impact perceived value*.

**Hypothesis** (H4b): *Cognitive needs play a moderating role during the process where online reviews impact perceived risk.*

#### **4. Data Collection and Analysis**

#### *4.1. Questionnaire Design*

We used a questionnaire to test the hypotheses. Specially, we design the questionnaire according to the general principles and steps of the literature. Firstly, the indicators of each variable were extracted through a great number of literature studies, and the designed questionnaires were tested on a small scale among 28 college students who had rented a house on the Internet shared short-term rental platforms. Some of the items were revised according to the results of the small sample analysis, and the formal research questionnaire was finally formed through repeated revisions. The questionnaire consists of three parts:


The questionnaire was designed according to the Likert five-point scale, and the respondent was asked to choose a score ranging from 1 (totally disagree) to 5 (totally agree) to answer each question. The details of the questionnaire are shown in Table 3.

#### *4.2. Data Collection*

The data collection of this study was conducted by an online questionnaire, mainly through the WeChat circle of friends. The product introduction of the shared short-term rental platforms is detailed on the homepage of the questionnaire. The participants are required to have an Internet accommodation booking experience. After screening the questionnaire and eliminating the items that do not have online accommodation booking experiences, we finally obtained 232 valid questionnaires. Baumgartner and Homburg proposed that the number of samples required for a structural equation analysis is at least five times the estimated parameter [60]. Since the research model in this study involves 18 parameters, we need at least 90 samples. Thus, the number of collected samples can satisfy the data analysis. The descriptive statistics of the samples are as follows: 121 males and 111 females, whose age is mostly within the range from 20 to 25 years (85.7%). In addition, 77.5% of participants hold a college degree.


#### **Table 3.** Indicators of the questionnaire.

#### *4.3. Factor Analysis*

In this section, we use the confirmatory factor analysis (CFA) to analyze the reliability and validity of factors. This paper first uses the SPSS 22.0 software (IBM, Armonk, NY, USA) to conduct the reliability analysis. The results are shown in Table 4. The value of KMO (Kaiser-Meyer-Olkin) is 0.874, and the Bartlett spheroid test results are significant (sig. = 0.000). The Bahrain alpha coefficient (Cronbach's α) and the composite reliability are mostly greater than 0.8, indicating that the scale has high reliability.

The normalized factor load of the seven explicit variables of the model is higher than 0.8 and reaches a significant level. The CR value is greater than 0.7, and the AVE rate is greater than 0.5. Overall, the model meets the requirements for convergence and validity.

Next, we perform the discriminant validity test. Discriminant validity is proposed to examine whether a measurement is a reflection of any other measurement. As shown in Table 5, the correlation coefficient of each variable is smaller than the square root of the mean variance extraction rate of its corresponding variable, indicating that the discriminant validity is satisfied.


#### **Table 5.** Discriminant validity test.


Note: each value along the diagonal line is the square root of the corresponding AVE.

#### *4.4. Structural Model Evaluation*

The structural model measures the relationship between latent variables. The path coefficients between the latent variables in the study model were verified using AMOS 23.0 (IBM, Armonk, NY, USA). The path coefficient values between the variables are shown in Figure 2.

When users subscribe to shared short-term rentals, the quality of online reviews significantly affects users' perceived value (β = 0.84, *p* < 0.01), assuming H1a is verified. This shows that the quality of online reviews can significantly improve the perceived value of users when users subscribe to shared short-term rentals. When users subscribe to shared short-term rentals, the quality of online reviews significantly affects visitors' perceived risk (β = −0.55, *p* < 0.01), assuming H2b is verified. This shows that the quality of online reviews can reduce the perceived risk of users. This may be due to the fact that when visitors receive higher quality online reviews, users get more information with more value, less uncertainty, perceived cost of decline, and thus higher perceived value and perceived risk.

**Figure 2.** The path coefficient values between the variables for structural model evaluation.

The perceived value of users has a significant impact on users' decisions on shared short-term rentals (β = 0.81, *p* < 0.01), assuming H2a is verified. This shows that the perceived value of users is positively related to users' decisions on shared short-term rentals. In addition, the perceived risk of users has a significant impact on users' decisions on shared short-term rentals (β = −0.23, *p* < 0.01). This shows that users' perceived risk is negatively correlated with users' decisions on shared short-term rentals. Perceived value and perceived risk are subjective feelings of people, which means that users may be affected by various stimulating factors, resulting in the change of perceived value and perceived risk, and in turn the decision behavior of users.

In addition, the permissible coefficients of perceived value and perceived risk are 0.71 and 0.30, respectively. This indicates that the quality of online reviews explains the variance variation of 71% of the perceived value generated by users' online bookings for shared short-term rentals. The quality of online reviews explains the perceived risk by 30% variance variation. As a result, the model explains the 84% variance variation of users' decisions on shared short-term rentals, showing that the degree of interpretation is satisfied. Table 6 shows the model fitting results. The descriptions of the symbols shown on the first row are provided in Table 7.


**Table 6.** Model fitting results.


**Table 7.** Explanations of the symbols used in Table 6.

#### *4.5. Mediating E*ff*ect Test*

In order to verify the intrinsic mechanism of the impact of online-review quality on users' decisions on shared short-term rentals, we use the bootstrap method (Bootstrap) for mediation analysis (sample size is set to 2000) [61]. Thus, it is judged whether the mediating effects of perceived value and perceived risk exist. The results are shown in Table 8. We can see that under the 95% confidence interval (CI), the bias-corrected, percentile, and PRODCLIN2 testing results show that perceive value and perceived risk have a mediating effect in the research model.


**Table 8.** Coefficient products and bootstrapping checking results.

Note: All the bootstrapping values are non-standardized values.

In summary, there is a complete mediation between perceived value and perceived risk in the decision-making path of online-review quality affecting users' decisions on shared short-term rentals. This shows that the quality of online reviews mainly affects the perceived value and perceived risk of users' psychology, and further affect users' decisions on shared short-term rentals. Both perceived value and perceived risk have a dual intermediary role between online-review quality and users' decisions on shared short-term rentals, demonstrating the validity of H3a and H3b.

#### *4.6. Moderating E*ff*ect Test*

In this section, we test the moderating effect of the moderating variable. Here, SPSS 20 is used to conduct the test by using the hierarchical regression method (HRM). The results show that the variance inflation factors (VIF) of the related variables are all less than 2, indicating that there is less information overlap between explanatory variables and there is no serious multicollinearity problem between variables.

Table 9 shows the moderating effect of user cognitive needs. We can see that the moderating effect of users' cognitive needs between online-review quality and perceived value does not exist. Meanwhile, user cognitive need has a moderating effect between online-review quality and perceived risk. Therefore, the hypothesis H4a is not valid, but the hypothesis H4b is verified. For people with different cognitive needs, the valuable information provided is the same, so the change of perceived value is not significant. On the other hand, the quality of online reviews indirectly proves the authenticity of information provided by platforms, which can reduce the perception of users' uncertainty. The reduction of users' uncertainty leads to the reduction of users' perceived risk. Compared with users with low cognitive needs who are eager to make decisions as soon as possible, users with high cognitive needs still need more information to help them verify the authenticity of their information and make decisions. Therefore, the perceived risk of users with high cognitive needs decreases at a slow rate with the improvement of online-review quality, which means that cognitive needs have a significant moderating effect between online-review quality and perceived risk.


**Table 9.** Moderating effect testing results of cognitive needs.

#### *4.7. Summary of Hypotheses Validation*

Based on the results described in Sections 4.1–4.6, we present the summary of the hypotheses validation in Table 10. The validation-result column in Table 9 shows the final validation results of each hypothesis. The hypotheses H1a, H1b, H2a, and H2b are all established. These assumptions correspond to our Q1 and Q2. This means that the quality of online reviews, which is the independent variable in our research model, has a positive impact on users' decisions on shared short-term rentals. H4b has also been verified, and H4a has not been verified, which corresponds to our Q3. This means that the personality characteristics of users can influence users' internal perception of online reviews. The main impact is the perceived risk of users, and its moderating effect on the perceived value is not significant. Platforms should pay more attention to the personality characteristics of users and adopt relevant marketing strategies to reduce users' perceived risks. Regarding the hypotheses of H3a, H3b, which aim to answer the research question of Q4, the results show that both perceived value and perceived risk play a full intermediary role between online-review quality and users' decisions on shared short-term rentals. Therefore, platforms should focus on user psychological perception and promote users' decisions on shared short-term rentals.


**Table 10.** Summary of hypotheses validation.

#### **5. Discussion**

#### *5.1. Research Implications*

First, in this study, we employ the SOR model to explore the internal mechanism of online reviews on users' decisions on shared short-term rentals. This model expands the application scope of the SOR model and provides referential values for other related research. In addition, we also presented empirical results that reflecting the impact of online reviews on users' decisions on shared short-term rentals, which can provide a new perspective for promoting shared short-term rentals and new marketing mechanisms.

Second, this paper studied the factors impacting shared short-term rental decisions, and provided referential values for enterprises to increase users' willingness to be engaged in today's mobile internet enabled sharing economy. Our study reveals that online reviews can inspire users' decisions on shared short-term rentals.

Third, we used perceived value and perceived risk as a mediator between online reviews and users' decisions on shared short-term rentals. The empirical results show that perceived value and perceived risk can influence users' decisions on shared short-term rentals. In addition, both perceived value and perceived risk play a full intermediary role between online reviews and users' decisions on shared short-term rentals. Therefore, the sharing of short-term rental platforms needs to pay attention to the psychological perception of users. Moreover, platforms need to reduce the perceived risk of users and increase the perceived value of users.

Fourth, regarding the measurement of users' personality characteristics, this paper mainly focused on users' cognitive needs. The results show that cognitive needs play a moderating role in the relationship between online reviews and perceived risk. Users with high cognitive needs need better information to reduce their perceived risk. Thus, users need more information from the Internet to reduce their uncertainty. Therefore, shared short-term rental platforms should provide users with more opportunities to communicate on the basis of correctly guiding online reviews.

Fifth, the data used in this paper are mainly from surveys of young people. Our research shows that young people can quickly accept a new type of city accommodation such as shared short-term rentals. Therefore, how to improve young people's choice of shared short-term rentals is what a shared short-term rental platform should consider. Many people are struggling to find a hotel around their destination when they have a temporarily traveling. Due to geographical locations and other issues, the traffic around the hotel may be inconvenient. Based on the wide range of sources of tenants on the shared short-term rental platform, when users post accommodation requirements or choose accommodation, they can easily choose the right place to stay. To this end, shared short-term rental platforms still have spaces and potentials for further development.

In summary, the main conclusions of this study, i.e., online reviews are highly influential to users' decision-making behavior, are coherent with previous works [17,19]. However, in the traditional hotel industry, the studies about online reviews and users' purchasing intention mainly focused on the characteristics of hotels and users. On the contrary, this paper is mainly toward online reviews on shared short-term rentals, which typically refer to private apartments advertised by owners on Internet-based rental platforms. Apartments have different features from hotels, and users engaged in shared short-term rentals are also much different from those choosing hotel booking. For example, many young students are likely to try shared short-term rentals, while most business people prefer high-standard hotels. Based on these differences, we set up new variables for the SOR research model. Specially, we focus on the mediating effect of users' perceived value and perceived risk, because users choosing shared short-term rentals have to balance their perceived values and possible risks. On the other hand, although shared short-term rentals as a new kind of shared economy have received much attention in recent years, there are few previous studies that focus on the impact of online reviews on users' purchasing decision on shared short-term rentals. We believe that the main conclusions of this paper can provide referential values for further studies in this field.

#### *5.2. Suggestions*

Based on the empirical analysis of this study, we provide the following suggestions for better implementation of shared short-term rental platforms.

(1) Guide users to provide effective online reviews.

The quality of online reviews significantly affects users' decisions on shared short-term rentals. On the one hand, the higher the quality of online reviews, the richer the review content, the higher the perceived value of users, and the lower the perceived risk. On the other hand, platforms should appropriately encourage purchased users to comment on the shared rentals, so that users with low cognitive needs can make up useful suggestive strategies. At the same time, platforms should guide users to provide product information that the website has not yet paid attention to, so as to help users with high cognitive needs to make decisions.

(2) Create topics and provide material rewards for online reviews.

By creating topics and rewarding material prizes, platforms can effectively guide the review content, direction, and quantity of online reviews. According to previous research, the number of online reviews positively affects users' willingness to purchase. More online reviews imply that more potential users will be able to learn more product-related information and be able to better perceive products. If there are many online reviews with high quality, shared rental platforms will attract more attention of users. In order to do so, platforms are suggested rewarding users with little gifts to encourage users to share short-term rental photos or experiences, so that the quality and quantity of online reviews can be improved.

(3) Provide better qualified rental products and services.

We suggest that shared short-term rental platforms provide qualified rental products and services to ensure that users can experience shared short-term rentals with their psychological expectations. The products and services provided by platforms can directly affect the perceived value of users. This is because shared rental industry is actually a service-oriented industry. So shared rental platforms should not only provide shared short-term housing, but also provide services related to the products, such as route planning, traffic information, and information about surrounding sightseeing places. This will increase the perceived value of users and has a significant positive impact on users' decisions on shared short-term rentals.

#### **6. Conclusions and Future Work**

In this paper, we empirically investigated the impact of online reviews on users' decisions toward shared short-term rentals in the context of social networks, sustainable tourism, and sharing economy. Our study found that the quality of online reviews significantly affects users' perceived value, and negatively affects perceived risk. In addition, both perceived value and perceived risk have a dual intermediary role between online-review quality and users' decisions on shared short-term rentals. We also studied the moderating effect of the cognitive needs of users. We found that the moderating effect of user cognitive needs between online-review quality and perceived value does not exist. Meanwhile, user cognitive need has a moderating effect between online-review quality and perceived risk.

Some limitations of this study can be summarized as follows. First, although hundreds of valid questionnaires are theoretically enough to conduct data analysis, in this big data era, this number is relatively small to draw reliable and robust conclusions. Second, the results of this study have shown that online reviews have positive impact on users' decisions on shared short-term rentals. However, the inherent theoretical basis has not been revealed yet.

Thus, in the future there are some research issues that are worth further investigating. First, in terms of the impact of review quality on the purchase intention of short-term rentals, secondary data collected from home sharing platforms like Airbnb could be a better choice. One possible way is to conduct text mining of the reviews of each property to measure review quality monthly, and to use the review volume to predict the purchase intention of each property. Second, because users' decision-making behavior could be impacted by other factors, future work can be focused on other possible factors, such as users' income and education background, review volume or variance, and users' online experience. Finally, it is also worth to study other research models [62] on user acceptance and consider online reviews within the community context [63].

**Author Contributions:** J.Z.: Conceptualization, Project administration, Supervision, and Writing–original draft, Z.P.: Data curation, Formal analysis, Investigation, and Software.

**Funding:** This paper is partially supported by the National Science Foundation of China (No. 71273010) and the Innovation Team Construction Project of Anhui University.

**Acknowledgments:** We would like to thank the editors and anonymous reviewers for their suggestions and comments to improve the quality of the paper.

**Conflicts of Interest:** The authors to declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **An Analysis of the Sustainable Tourism Value of Gra**ffi**ti Tours through Social Media: Focusing on TripAdvisor Reviews of Gra**ffi**ti Tours in Bogota, Colombia**

#### **Hwayoon Seok, Yeajin Joo and Yoonjae Nam \***

Department of Culture, Tourism and Content, Kyung Hee University, South 130-701, Korea; vipshy5734@khu.ac.kr (H.S.); jyj6241@khu.ac.kr (Y.J.)

**\*** Correspondence: ynam@khu.ac.kr

Received: 1 May 2020; Accepted: 26 May 2020; Published: 29 May 2020

**Abstract:** This study attempts to analyze the value of graffiti tours from the perspective of sustainable tourism by examining actual reviews by social media users using text mining and social network analysis. The text mining technique indicates that "artist," "history," "political," "culture," "social," "city," "background," "great," "recommend," "excellent", and "worth," are frequently used keywords for the reviews. When comparing word frequencies per review between 2013–2016 and 2017–2019, the results show that the words, "history," "political," "culture," and "social," have been more frequently used over time. Furthermore, the network visualization shows that those words connoting socio-cultural sustainability are mutually connected. Therefore, the study suggests that graffiti tours can have potential for assuming the role of sustainable tourism, since the keywords from reviews are associated with the perspective of socio-cultural sustainability.

**Keywords:** sustainable tourism; graffiti; social media; text mining; social network analysis

#### **1. Introduction and Research Backgrounds**

Visual image is a medium containing a clear reality or truth. This visual communication is an important theoretical concept as it emphasizes various characteristics of the cultural process [1]. The mural, which was the first visual communication, is an example of a work that consists not only of an artistic aspect, but also depicts the culture and way of life at the time it was created. Graffiti has existed as a social phenomenon since the beginning of cave paintings. Graffiti is defined as writing or drawing on walls or other surfaces of public places by uneven scratching or spraying [2]. This rather unregulated form of artwork is often misinterpreted as an expression of crime and disorder, since "in its purest form," street art is "artwork created without authorization, usually illegally, on either private or public property" [3]. In tackling vandalism, graffiti has been the subject of discussion among policymakers around the world. However, there is a certain complexity to the culture of graffiti; it is not a simple matter of acceptance of art or vandalism [4]. Recent urban policies, along with increasing discourse on the 'creative city,' have led to changes by encouraging communication among people to promote urban development [5]. In the same context, the importance of creativity in economies after industrialization has pushed urban managers to reevaluate graffiti [6]. Likewise, graffiti arts in some cities including Bogota, Colombia has been regarded as a tool for tourism development, heritage conservation, socio-economic regeneration and cultural empowerment.

For example, in 2007, graffiti was legalized in Colombia's capital, Bogota, by the power of civil society. The city has since changed through the cooperation of the government, graffiti artists, and citizens. The first project involving graffiti was the 26th road environment improvement project, which extended to Bogota's buildings, shops, and homes. Such efforts contribute to the transformation of the image of Colombia as a crime-ridden and depressed country [7]. Jose Andres Duarte, the head of the Research Institution of Local Tourism of Bogota, stated, "We have improved the competitiveness of Bogota through the development of tourism wherein local residents are put first and improvement of environment of internet connection." This was in line with its policy of prioritizing local residents, which it sought to implement by creating a new environment for the city and boosting tourism According to the UNWTO's (United Nations World Tourism Organization) [8] country-specific inbound tourism data, Colombia had 4,282,000 people enter the country in 2018. The average increase rate of inbound tourists over the past 10 years was 9.7%; this is significantly higher than the 5.4% average growth rate of other 101 countries. Furthermore, in the Travel and Tourism Competitiveness Index of 136 countries in 2017, Colombia scored 3.37 in the Cultural Resources and Business Travel sector (average of 2.32 points in the whole country) [9]. In urban planning, resilience has become a complex interaction between society, nature, land use management, and policy. This dynamic ensures meaningful participation as the different arms prepare for the collapse of socio-economic change, thus achieving equality and meaningful stakeholder involvement. A policy-adaptable social infrastructure must be built to ensure participation of these sectors [10,11]. In other words, urban regeneration combines sustainability with economic, social, cultural, and environmental perspectives. Under the goal of urban regeneration, graffiti tours that interact with tourists through works containing cultural messages, along with the cooperation of local residents, are found to be linked to sustainability.

In a situation where concepts of sustainable and responsible tourism are gaining more attention, it is necessary to understand the role of graffiti tours as sustainable tourism. Studies on graffiti and graffiti tours mainly present graffiti's various social phenomena, roles, and utilization plans through case studies, spatial analysis, and in-depth interviews with local residents, graffiti artists, and policy-related individuals [4,6,12–15]. However, still there have been only few researches on the perspectives of tourists who have actually experienced graffiti tours and evaluated them. In particular, tourism is a major source of information for many tourists. Many tend to search online-based reviews of various users in evaluating experiential products [16]. These opinions spread to potential tourists through various social media platforms; as such, these data also serve as a tool for understanding the opinions of numerous tourists from diverse backgrounds.

Thus, this study assumes that graffiti tourists' online reviews may reflect economic, socio-cultural, and environmental aspects of sustainability beyond showing their satisfactions from travel experiences. Therefore, this study examines how sustainable issues have been discussed among actual tourists by examining their reviews on social media. In 2017, the United Nations General Assembly declared 2017 to be "the International Year of Sustainable Tourism for Development," and it was also the year that the UNWTO proclaimed the "2030 Agenda for Sustainable Development." For these reasons, the year 2017 is a pivotal period for sustainable tourism and was determined to be a reference point when analyzing the shift in views on graffiti tours.

Social media platforms are an important resource of big data, as they have become diversified by generalized use. They are suitable for text mining analysis because they consist mainly of unstructured texts [17–19]. For the above purposes, the study employed the technique of text mining in order to extract useful information in big data to analyze and interpret texts posted on social media. In addition, the structure of the network of words was examined based on the results of the analysis for the social network analysis. Through this analysis, we confirm the role of sustainable tourism in graffiti tours from a tourist's perspective, instead of in the indiscriminate development of graffiti tours, and draw implications accordingly.

#### *1.1. Function of Sustainable Tourism*

The concept of sustainability originates in the report "Our Common Future," published by the WCED (World Commission on Environment and Development) in 1987. It defines sustainable development as the ability of humanity to make development sustainable to ensure that it meets

the needs of the present without compromising the ability of future generations to meet their own needs [20]. Ever since the report was published, sustainable development has received extensive public attention and provided many challenges worldwide [21]. There are several factors that explain why the tourism industry shows such an interest in sustainable development: Tourism has great economic significance, as it creates employment and generates income, it impacts various industries, and it depends on the cultural heritage and landscape of a place. Tourism also has negative and positive effects on the natural environment and tourist locations [22]. The UNEP (United Nations Environment Program) and UNWTO [23] also mentioned two relationships between tourism and sustainable development. First, tourism is a dynamic, growing industry that makes key contributions to many countries. Second, tourism is a set of activities that includes special relations among visitors, industries, the environment, and communities. Therefore, sustainable tourism is defined as tourism that protects and promotes the opportunities of future generations, while simultaneously meeting the present needs of tourists and communities by not only preserving and maintaining culture, essential ecological processes, biodiversity, and life support systems but also managing resources to meet economic, social, and aesthetic needs.

Donyadide [24] presented "The Magic Pentagon of Sustainable Tourism," which describes five elements of sustainable tourism, from an ethical standpoint, in order to maintain a high level of balance with tourists in terms of the social and ecological characteristics of travel sites. The five elements are: economic health, the satisfaction of guests, a healthy culture, protection of resources, and the well-being of locals. Recently, the approach has been subdivided into more categories to emphasize the relevance to tourism of social values, such as population, peace, prosperity, pollution, and protection [25]. Tosun [26] mentioned that the operation of sustainable tourism development principles requires tough political and economic choices and decisions based on complicated socioeconomic and environmental transactions. Crouch and Ritchie [27] argue that not considering sustainability in discussing the competitiveness of travel destinations is an unsubstantial concept, and that it is necessary to reflect not only on the economic resources but also on the ecological, social, cultural, and political resources. Tourism competitiveness is the capacity to accumulate and create value-added products that make the value of resources sustainable, while maintaining the current market position [28]. Sustainable tourism is in line with tourism competitiveness—it creates and integrates product values and maintains resources.

As numbers of tourists increase each year, so do issues such as greenhouse gas emissions, financial leakage, resource management, or effects on communities and cultural heritage, which require strong partnership among all stakeholders of tourism in implementing critical measures through the "2030 Agenda for Sustainable Development" [29]. These efforts for sustainable tourism require an integrated policy that addresses multiple aspects. It is desirable to understand sustainable tourism not as a substitute for conventional mass tourism, but as a process to improve it and solve the current issues tourism faces, while meeting the needs of all the stakeholders of sustainable tourism, such as governments, local governments, residents, communities, tourism business operators, and tourists [30]. In other words, sustainable tourism is a concept that considers and promotes positive interactions among all residents, tourists, organizations, and business operations related to tourist sites. From the relationship-based perspective, the key components of the paradigm for sustainable tourism are enhancing the competencies of a community and building relationships and alliances to steer the regional economy in a sustainable direction that is beneficial for the environment [28]. In other words, sustainable tourism requires development in consideration of both the community and tourists. The indicators of sustainable tourism, according to the three perspectives proposed by UNEP and UNWTO [23], are presented in Table 1.


**Table 1.** Sustainable development of tourism [23].

Richins [31] and Choi and Turk [32] summarized the details of each subcategory of sustainability in addition to the contents of previous research and viewpoints of international organizations. This information is shown in Table 2.


**Table 2.** Sub-categories according to sustainable tourism perspective [31,32].

**Table 2.** *Cont.*


In this study, economic, socio-cultural, and environmental sustainability, along with the subcategories of sustainability, form the framework for the analysis of the perspectives of Bogota graffiti tours that are reflected in sustainability as values of sustainable tourism

#### *1.2. Gra*ffi*ti Tour*

Practical visual communication like the Altamira and Lascaux cave paintings are able to show the culture of an era. Their main purpose is the outcome of the image, which was the first visualization work [33]. Cave paintings were the first attempt to construct an effective visual layout system through the arrangement of visual elements and replace abstract concepts with visual forms. In the same vein, Jordan's 5th century BC mural includes geometric designs and hierarchical, complex social depictions, which help viewers understand the role of art in early visual communication systems and preliterate societies [34]. This allows art not to be appreciated simply as an aesthetic element, but also as an innovative creation that maximizes the use of natural objects and explains art from a semiotic perspective. In modern times, the visual communication of various artefacts, designs, social media platforms, etc., is ingrained in our lives and spreads different ideologies by controlling the time and space they create [35,36]. Visual image is a medium containing a clear reality or truth, and visual communication is an important theoretical concept as it emphasizes various characteristics of cultural processes [1].

The mural—the first visual communication and the origin of art—again emerged in the present and developed into a genre of art that conveys social messages and expresses the artist's values and personality. Graffiti, which can be seen as a representative model of cave murals, is defined as a writing or painting that is irregularly scratched or sprayed upon walls or other surfaces in a public place [2]. Graffiti has existed for as long as mankind. Cave paintings can be considered to be an example of graffiti [2]. In the 1970s, graffiti was derived from hip-hop culture, which played a significant part in representing the culture of the young generation through the expression of a new form of music, dance, and art [6]. From the 1970s to the 1980s, graffiti was regarded negatively, as it was characterized as a key factor in profit loss, depreciation, and decrease in retail sales in certain regions or communities [37]. However, recent urban policies, along with increasing discourse on the "creative city," have led to changes by encouraging communication among people to promote urban development [5]. In the same context, the importance of creativity in economies after industrialization has pushed urban managers to reevaluate graffiti [6]. The urban environment has countless surfaces and objects for expression and, at the same time, provides an audience with an opportunity to see these expressions. The aesthetic image of graffiti is being magnified in galleries, on the internet, and via social media [38]. This suggests that the aforementioned visual communication is directly connected to graffiti.

Gomez [39] distinguished art in the form of graffiti as "graffiti art" and stated that it is characterized as "high art" and "folk art" in which the desire for artistic creation has become its motivation. This art also plays an important role in tourism. All of the cultural elements experienced by tourists are included in the category of cultural tourism, which is expanding in an artistic sense. As seen in Table 3, art tourism refers to a set of tourism-related activities of consuming contemporary culture. These include participation in the fine arts, performing and exhibiting arts, artistic events, festivals, and entertainment.


**Table 3.** Definition of art tourism.

From the commercial aspect, art invigorates tourism products, increases market attractiveness, provides new value to a region, and performs an important role in sales and promotion, while also improving product quality and appeal [41]. Graffiti has become an art tourism space that contributes not only to the creation of new tourism assets, but also to its revitalization [15]. For example, through the promotion of graffiti tours, graffiti has now come to be perceived as valuable rather than negative in terms of urban aesthetics, and as contributing to place-making by performing a key role that gives an identity to a city [4,38]. Graffiti can be linked to urban morphologies and the social identity of a place, [13] leading to it ultimately becoming a tourist attraction. In other words, a city's graffiti can be promoted as a new tourist spot and become a communication tool that leads to interaction between residents and tourists.

#### *1.3. Gra*ffi*ti Tours in Bogota, Colombia*

Initially, graffiti art in Bogota was not recognized as art and was seen as spoiling the cityscape. Following an action brought by a civic organization to legalize graffiti-art activities in 2007, the city began to view graffiti art as an important activity of art and culture. According to the "Bogota graffiti tour", a representative cooperative and commercial site related to Bogota graffiti tours [44], graffiti tours in Bogota began with Austrian and Canadian graffiti artists who wanted to show the unique and diverse aspects of the work of local artists to international tourists. Since then, graffiti art in Bogota has been growing, and art sites have become famous sightseeing spots. Graffiti art painted on commercial places or on the walls of private houses covers culture, history, and messages related to society, such as the indigenous people of Colombia, female aborigines, racial discrimination, preservation of life, and feminism. The Bogota Graffiti Tour provides tourists with an introduction to graffiti tours in Bogota, free or private tours led by guides, bike tours, city tours, and workshops. The Bogota Graffiti Tour is guided by a graffiti promoter, manager, artist, anthropologist, and designer. They guide tourists throughout the city, providing noteworthy information and key facts about the Bogota graffiti scene to facilitate appreciation of the artwork. The bike tour is significant as it can help reduce environmental issues related to traffic. Graffiti artists and associated creators have become members of a team that plans the best alternatives to help artists' projects. This differentiated approach to the cultural development of Bogota is "Responsible Graffiti Practice." The "Responsible Graffiti Practice" is a participative process aiming to develop new cultural policies for graffiti art, wherein graffiti is recognized as a form of art in which the participation of citizens in making decisions on public spaces is encouraged. Such efforts contribute to the transformation of the image of Colombia as a crime-ridden and depressed country [7]. As examined earlier in the theoretical background, sustainable tourism coincides with the process of solving problems in current tourism with satisfying the needs of relevant stakeholders. As such, graffiti in Bogota incorporates the virtuous cycle.

According to UNWTO statistics [8] on the number of inbound tourists to countries each year, the rate of increase in the number of tourists to Colombia from 2008 to 2018 (4,282,000) exhibited a 9.7% annual average increase, whereas the annual average in the other 112 countries over the same period was 5.4%. In particular, the number of inbound tourists to Colombia in 2010 and 2011 were 1,719,000 and 3,324,000—the rate of increase was 93.4%. According to UNWTO's October 2010 report, tourism in

Colombia showed a trend of recovery through new investment and active promotion, despite disasters, such as flooding from the end of 2010 to the middle of 2011. Tourism was concentrated in business tours, and domestic sightseeing in Bogotá, Cali, Medellin, and Cartagena increased [45]. Figure 1 presents graffiti in Bogota, Colombia.

**Figure 1.** Graffiti in Bogota, Colombia. Source: Photo: Hwayoon Seok.

#### *1.4. Value of Sustainability of Gra*ffi*ti Tours*

In general, the behaviors associated with environmental conservation are frequently counted as those of sustainability. Therefore, a survey of tourists' perception on sustainable tourism in the Netherlands and the differences between the types of tourists and the perception of sustainability in certain places were analyzed. Particular importance was given to the ecological view of sustainability, followed by socio-cultural and economic views [46]. The ecological perspective is considered an environmental perspective, which measures the perception of sustainable tourism by classifying it into three categories of sustainability. In general, it can be assumed that the tourists' perception of sustainability focuses on the environment. The economic perception is given the lowest weight because it affects the locals more than the tourists. Currently, it is the environment that inhabits the recognition of consumers and takes precedence to everything else, along with the efforts made to secure sustainability in diverse industries. Thus, as for sustainable tourism, alternative tourism comprising ecotourism, responsible tourism, and fair tourism, is often characterized by recognition of environment [47]. It is the form of positive tourism insisting conservation of current sites of tourism for the sustainable tourism. However, there exist aspects wherein the educational element rather than entertainment ones are spotlighted by tourist conscious of morality and regulations on tourism. From the standpoint of traditional socio-tourism, tourists experience the 'communitas' wherein the morality and regulations set forth in the space of ordinary life are ignored through migration into sites of tourism [48,49]. Undoubtedly, though the introspection and moral norm are taken seriously in contemporary tourism along with the changes in ages, the evidence implies that sustainable tourism needs newness and to be interesting if it is to be presented to tourists.

As examined previously, sustainability comprises environmental sustainability together with economic and socio-cultural sustainability. By taking this viewpoint into account, the preservation of socio-cultural resources, based on the conservation of environmental resources, posits the economic value thereof. In the field of tourism, the concept of sustainability was defined by Inskeep [50] as the sustainment of a life conservation system comprising cultural preservation, the sustainment of necessary ecological processes, and biodiversity together with all resources that can fulfill economic, social, and aesthetic needs to improve opportunities for future tourism. This is compatible with the thinking intending to protect resources for tourism and contemporary tourists. UNEP and UNWTO [23] also mentioned sustainable tourism as the promotion and protection of the opportunities of tourism for future generations, balanced with the fulfillment of the needs of contemporary tourists and local communities to maintain cultural integrity, ecological processes, biodiversity, and systems for sustainable life, as well as to control all resources for the fulfillment of economic, social, and aesthetic needs. Graffiti tours, which construct a new environment by creating mural paintings of stories of society and culture, can be regarded as a rearrangement of resources and the composition of a new environment to attract the interest of tourists. Moreover, culture is the way that members respond to specific political situations facing the city [7]. The proliferation of cultural planning and public art policies, along with

the right of a large city to promote creativity, provides an opportunity to symbolize graffiti as a productive creative practice [6]. In the current trend of increasing tourism, the balanced development of cities can offer a solution to the concentration of tourists visiting famous sights. The importance of the effectiveness of local tourism and its sustainable competitiveness, through improving attitudes to accommodate tourists, has been growing gradually [28,51]. Graffiti tours align with sustainable tourism in how they develop socio-cultural functions based on their aesthetic functions, environmental functions by relocating the urban environment, and the revitalization of local economic functions due to job creation and tourism through the cooperation of citizens and graffiti artists. Therefore, it is necessary to conduct an empirical analysis to confirm whether graffiti tours reflect the value of sustainability from the perspective of sustainable tourism. Recently, with the increasing popularity of social media, reviews by consumers enable them to exchange product information more easily and be affected by other people's choices [52–54]. As tourism products—experiential products—cannot be evaluated before the actual experience, consumers find useful information about tourist spots, activities, and experiences on a real-time basis through online reviews, which influence decision-making and behaviors when traveling [55,56]. For these reasons, many previous studies [57–61] used online reviews of hotels, Airbnb, restaurants, and other tourist attractions to examine customer satisfaction attributes, causes, perception, and destination characteristics. Therefore, we intend to measure the sustainability of graffiti tours by analyzing the online reviews on social media based on the three perspectives of sustainable tourism. In 2017, the United Nations General Assembly declared the international year of sustainable tourism for development, and UNWTO promulgated the '2030 Agenda' for sustainable development. Given that this is a major starting point for the development of sustainable tourism, it is necessary to look at the changing issues by cycle. Based on this theoretical background, the following research questions were identified:

Research question 1: How are perspectives on sustainability reflected in the online reviews about graffiti tours in Bogota?

Research question 1-1: Which keywords in the online reviews about graffiti tours in Bogota are related to perspectives on sustainability?

Research question 1-2: How have perspectives on sustainability in the online reviews about graffiti tours changed over time?

Research question 1-3: What are the structural characteristics, meanings, and roles of the key words network from the online reviews about graffiti tours for reflecting sustainability over time?

Research question 1-4: How are keywords in the online reviews about graffiti tours associated with the perspectives of sustainability?

#### **2. Data and Methods**

#### *2.1. Data*

Online travel communities that share tourist information include TripAdvisor, Lonely Planet, Travellerspoint, VirtualTourist, and Yahoo! Travel. Many studies have analyzed the utility of tourist information by reviewers and its effects on the decision-making of other tourists to assess the reliability of information on the largest online travel community, TripAdvisor. The results proved that such information had significant effects, indicating that there is high utility value in terms of information delivery [62,63]. TripAdvisor provides countless reviews and feedback on accommodation, restaurants, experiences, airline tickets, and cruises, and has the ability to influence approximately 500 million travelers. Travelers compare prices of hotels, flights, and cruises; book popular tours and attractions; and make restaurant reservations. Currently, TripAdvisor offers its services across 49 markets in 28 languages [64].

Accordingly, this study examines 1991 online reviews in English for "The Original Bogota Graffiti Tour" on tripadvisor.com to analyze the direct perception of tourists who experienced this tour. As there were only seven reviews in 2012, this study analyzed a total of 1984 reviews from January 2013 to July 2019. Moreover, to examine changes in the issues of each period, an analysis was conducted on online reviews from 2013 to 2016 and from 2017 to 2019 on the basis of the year 2017, which was declared the International Year of Sustainable Tourism for Development by the United Nations General Assembly, and in which the "2030 Agenda" was introduced for sustainable development by UNWTO.

For traveler ratings, TripAdvisor's "The Original Bogota Graffiti Tour" was rated excellent in 1843 reviews (92.9%) and terrible in six reviews (0.3%). On the basis of the date of experience, there were 764 reviews from January 2013 to December 2016 and 1220 reviews from January 2017 to July 2019. Examined by month, most travel experiences were in December, followed by similar numbers in February, March, July, and August. December is the month in which the dry season begins in Colombia, with most Colombians taking a vacation between December and mid-January (see Figures 2 and 3).

**Figure 2.** Bogota Graffiti Tour annual trend (No. of reviews based on experience date).

**Figure 3.** Bogota Graffiti Tour monthly average (No. of reviews based on experience date in 2013–2018).

#### *2.2. Analysis*

Recently, big data has been employed as a means of analyzing trends and customers, predicting sales, and developing marketing strategy in diverse industries. Social media has contributed to the rapid emergence of big data [17]. Social media platforms are an important resource of big data, as they have become diversified by generalized use. They are suitable for text mining analysis because they consist mainly of unstructured texts [17–19]. Text mining was introduced initially as a concept of knowledge discovery in a textual database [65]. It is a process of extracting relevant information automatically through the interaction of interest between language (text) and a mechanical algorithm (analysis) for the purposes of finding useful patterns and knowledge based on the technology of natural language processing (NLP) from text-based data. It can, therefore, be seen as an extraction of interesting and non-ordinary knowledge from unstructured language [66–68]. Thus, text mining is employed as a theoretical approach and a corresponding method in diverse fields, as well as a means of searching for information in disciplines associated with information retrieval [69].

Therefore, this study analyzed the online reviews available on TripAdvisor using text mining to address the research questions established earlier. In the present study, the program R (Version 3.5.0, RStudio Inc., Boston, MA, USA) was used to analyze source codes of the homepage TripAdvisor to examine online reviews posted there, by text mining. Upon acquisition of the data of 'online reviews' in 'TripAdvisor,' the following jobs were carried out: the preprocessing of data, the extraction of keywords, the identification of the relationship between the extracted words, and the visualization

of the extracted words based on frequencies through word cloud [69]. Then, the structure of the network of words was examined based on the results of the analysis for the social network analysis. In addition, unnecessary or inappropriate words for this study were removed through the general refinement procedures like tokenization, normalization, stop words, strip whitespace, tolower, and stemming. By those procedures, special symbols, spaces, numbers, English punctuation marks, and stop words were eliminated; capital and lower letters were differentiated; word roots were extracted from phrases. Furthermore, some words which could imply graffiti tours, such as graffiti, mural, tour, trip, travel, and art were excluded, since the issue of 'nested patterns' could occur during the text mining procedure. Next, the structure of the network of words was examined based on the results of social network analysis. To analyze the network structure between words, a co-occurrence matrix based on the frequency of occurrence between words was generated through text mining. Each word is a node in the network based on N × N matrix structure (S). From this network, degree centrality is calculated. Here, each cell, Sij, indicates the intensity between word i and word j [70]. Therefore, the word network structure based on the frequency of co-occurrence between words was presented as visualized data. Such an approach to network analysis can be regarded as appropriate to examine and compare relationships inside a network [71]. These matrices were utilized as inputs for the social network analysis software package UCINET (Version 6.624, Analytic Technologies, Lexington, KY, USA). The word networks were analyzed and visualized by NetDraw (Version 2.160. Analytic Technologies, Lexington, KY, USA).

#### **3. Results**

RQ1-1. The Table 4 shows the results of keywords analysis of "The Original Bogota Graffiti Tour" reviews.


**Table 4.** Review data set word frequencies results (2013~2019).

The top 50 words from the keywords analysis show that "artist" had the highest frequency with 1024 appearances, followed by "great." Excluding sentiment words, words such as "history," "city," "political," "culture," "work," "story," "social," and "background" showed high frequency.

Moreover, people used words related to knowledge acquisition and influence, such as "informative," "learn," "know," "insight," "worth," "understand," and "meaning." The results showed that these words played an important role in tourists' perceptions of the Bogota Graffiti Tour. Figure 4 presents a word cloud of word frequencies results.

**Figure 4.** Online review word frequencies results in word cloud format (2013~2019).

RQ1-2. In 2017, the United Nations General Assembly declared "the International Year of Sustainable Tourism for Development," and UNWTO promulgated the "2030 Agenda for Sustainable Development." On the basis of date of experience, there are 764 reviews from January 2013 to December 2016 and 1220 reviews from January 2017 to July 2019. Then, frequency of each word per the number of total reviews were calculated, and the differences of those scores between the two time periods of 2013–2016 and 2017–2019 were presented in Table 5. In a comparison of the results of the analyses done between 2013–2016 and 2017–2019, words such as "artist," "great," "knowledge," and "city" appeared with decreased frequency per number of total reviews. However, the words that show an increase, excluding sentiment ones, are "history," "political," "culture," "behind," "story," "social," "background," "context," "artwork," and "community." Considering the rank of word frequency, there are new words ranked the top 50 for 2017 to 2019, such as "context," "artwork," and "community." New words also appeared, such as "fantastic," "wonderful," "super," and "perfect" signifying more intense, positive feelings.




**Table 5.** *Cont.*

RQ1-3. Figures 5 and 6 show the results of the structural characteristics of the network formed by the words' co-occurrence. These figures are visualized images of a network based on words with a stronger connection. "Artist," which was ranked the highest, was strongly connected to "great," "history," "culture," "city," "recommend," and "political," indicating that they frequently appeared together. This shows a common feature in each period. In other words, people tend to refer to "artist," while associating it with history, politics, culture, and society, and they feel that it is "great," or "amazing," so they "recommend" it to others. This result is similar to that of the previously conducted keywords analysis. The words with the second-highest connectivity vary across the different periods. While words such as "style," "technique," "amazing," and "suggest" are observed between 2013 and 2016, "community," "context," "mean," and "fantastic" are observed between 2017 and 2019. That is, these words appeared with the words that play a key role in understanding graffiti in line with concepts such as society, community, and context.

**Figure 5.** Word co-occurrence network (2013–2016).

**Figure 6.** Word co-occurrence network (2017–2019).

RQ1-4. The words of high rank, such as "artist," "history," "city," "political," "culture," "work," "story," "social," and "background," which are associated with societal and cultural aspects were found from the keywords. There are also words associated with satisfaction and an experience of significance felt by tourists, including, "great," "recommend," "interesting," "informative," "learn," "amazing," "insight," "worth," "excellent," "incredible," and "awesome" etc. The significance of these words can be understood in the context associated with socio-cultural sustainability among viewpoints of economic, socio-cultural, and environmental sustainability. The actual reviews from tourists that include the words that show increased frequencies per total reviews between 2017 and 2019 are highlighted in Table 6.


**Table 6.** Samples of tourist's reviews.

The highlight of tourist's reviews is "The art work may seem just pretty but if you take a closer look, it's very political with Strong messages," "The artists styles, their political views, plus history of the native Colombians and their struggle to keep their identity," "The art is impressive and the stories behind it are captivating," "The graffiti tour was an excellent way to explore Bogota's neighborhoods through the artistic and cultural lense of street art," and "Good knowledge shown of the street art and context social and political and the importance of this art in Bogota reflecting some of the indigenous people and political struggles as well as the legalities." This can be interpreted as socio-cultural sustainability in the Bogota graffiti tour.

#### **4. Discussion and Conclusions**

The purpose of this study was to analyze the value of graffiti tours from the perspective of sustainable tourism by examining actual reviews by social media users using text mining and social network analysis. The text mining technique indicated that "artist," "history," "political," "culture," "social," "city," "background," "great," "recommend," "excellent", and "worth," were frequently used keywords for the reviews. When comparing word frequencies per review between 2013–2016 and 2017–2019, the results showed that the words, "history," "political," "culture," and "social," had been more frequently used over time. Furthermore, the network visualization showed that those words were mutually connected. From the results of contents analysis of reviews, it is suggested that tourists have gradually focused on not only artistic aspects, but also social, cultural, historical and political aspects of graffiti in Bogota. In addition, the results imply that tourists have been aware of the function of graffiti for establishing and preserving the identity of the community. Thus, this study may suggest that graffiti tours can have potential for assuming the role of sustainable tourism, since the keywords from reviews are associated with the perspective of socio-cultural sustainability, which comprises the appreciation of socio-cultural authenticity, conservation of cultural heritage and traditional values, and intercultural understanding and tolerance [23]. Based on these results, three more implications can be made.

First, the mentions of history, politics, culture, and society in graffiti tours are connected with positive emotions, interest, and the insight of tourists. The results of the network analysis on words, and their frequency of use, comprise words like the above. The factors that have the most influence over tourists' positive recognition of sightseeing spots are complex ones, comprising history, politics, culture, and society, together with the artistic perspective of graffiti. As discussed earlier, Zeppel [41] stated art invigorates tourism products, increases market attractiveness, provides new value to a region, and performs an important role in sales and promotion, while also improving product quality and appeal. Graffiti art can be a way to inspire tourists' sympathy through content associated with politics, culture, and society. This should be considered in the development or improvement of graffiti tours.

Second, tourists' viewpoint of the sustainability of graffiti tours is expanding over time. In a comparison of the results of the analyses done between 2013–2016 and 2017–2019, words such as "history," "political," "culture," "behind," "story," "social," "background," "context," "artwork," and "community" appeared with increased frequency per number of total reviews. This suggests a gradual increase in the recognition of the importance of sustainability in the consciousness of tourists confronted with graffiti tours. Expressions of more intensified positive feelings demonstrate that consequent satisfaction of tourists growing. This is in accordance with the study by Parker et al. [38] and Haworth et al. [4] that states graffiti has now come to be perceived as valuable rather than negative in terms of urban aesthetics, and as contributing to place-making by performing a key role that gives an identity to a city. Moreover, Dovey et al. [13] stated graffiti can be linked to urban morphologies and the social identity of a place. Therefore, areas where graffiti appears should be encouraged to be evolving spaces for the art through the creation of additional space to reflect contemporary issues.

Third, observing the keywords of the review, there was more content mainly related to socio-cultural sustainability, which also proves that tourists mainly perceive graffiti tours through the socio-cultural sustainability lens. This also supports Richins [31] and Choi and Turk's [32] study that explained how socio-cultural sustainability includes preserving cultural experiences, history, and traditional values, including a variety of architecture, street scenes, and art. In addition, socio-cultural sustainability's subcategory consists of tourist satisfaction/attitude toward tourist attractions, strengthening identity by developing and enhancing a vibrant community identity, and integrating and connecting brand and cultural identity. Given that the arts are at the center of elaborate visual discourse on neoliberalism, democracy, and the battle against public space, researchers can continue to examine how street artists inscribe social justice in, on, and around the streets [72]. The frequent mention of socio-cultural sustainability in the 'online review,' compared to economic or environmental sustainability—which is more relevant to local residents—is attributable to the

viewpoints of tourists. However, greater recognition of the significance of environmental sustainability is necessary, as it is important for both tourists and local residents. This also supports the idea presented in the literature review that Cottrell et al. [46] stated environmental sustainability is the most important perspective in tourists' perception of sustainable tourism. Therefore, the stories of the role of graffiti art in environmental improvement and sustainability, as well as socio-cultural sustainability in the places in which graffiti appears, should be researched.

#### **5. Limitations and Future Studies**

This study assumes that graffiti tourists' online reviews may reflect the economic, socio-cultural, and environmental aspects of sustainability, beyond showing their satisfactions from travel experiences. Nevertheless, since only the case of Bogota graffiti tours was analyzed in this study, it would be prudent to gain a deeper understanding of the topic through a comparative analysis of various cases. Through this, it would be possible to draw the direction and policy implications for graffiti tours as sustainable tourism. This study analyzed only English reviews, despite the subject of the study being a Hispanic country. By analyzing reviews in other languages such as Spanish, the diversity and practicality of opinion would be better represented. Furthermore, the meaning and significance of the relationship between each word in diverse contexts can be concretized through emotional analysis on the results of network analysis.

**Author Contributions:** Data curation, Y.J.; Project administration, Y.N.; Supervision, Y.N.; Writing—original draft, H.S.; Writing—review and editing, Y.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a grant from KyungHee University in 2019 (KHU-20191216).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
