Research on the Influencing Mechanism via Which Security Perception of Personal Information Affects Tourist Happiness: A Moderated Mediation Model

Abstract

Due to the continuous advancement of smart tourism and the improvement of tourists’ security perception of personal information (SPPI), SPPI has become an important factor affecting tourist happiness (TH). However, few studies have been conducted on the influencing mechanism via which SPPI affects TH. According to the risk perception theory and subjective wellbeing theory, the following conclusions were drawn from mixed methods research: (1) SPPI has a significant direct effect on TH; (2) SPPI has a significant indirect effect on TH through the perception of smart tourism application (PSTA); (3) the perception of smart tourism technology (PSTT) does not have a significant moderating effect between SPPI and PSTA or between PSTA and TH, but has a significant moderating effect between SPPI and TH. The discussion of this mechanism can not only promote the combined development of the two theories but also remind destinations or scenic spots to focus on improving tourists’ SPPI when developing smart tourism.

1. Introduction

The progress of smart tourism technology is driving the development of smart tourism applications, offering convenience to tourists. However, this progress has also brought about personal information security problems. Tourists can now plan travel itineraries, make transportation arrangements, make hotel reservations, and book scenic spot tickets in advance [1,2,3,4]. This is particularly true now that the COVID-19 pandemic has become normalized; contactless electronic tickets for scenic spots can effectively prevent the spread of COVID-19 during travel [5]. However, the personal information security issues brought about by smart tourism technology and smart tourism applications have become increasingly evident. Personal information leakage not only affects the reputation of institutions, such as the credibility of the government [6], but may also create fraud problems and affect the safety of consumers’ personal property [7].

The importance of ensuring the security of personal information is particularly evident in tourism-related applications [8,9,10] and payment methods [9,11], such as when booking hotels and tickets for scenic spots [12]. During the use of tourism-related applications and payments, tourists use personal information to log into an APP or enter a password; they may have concerns about whether their personal information could be inappropriately disclosed. Therefore, it is necessary to assess the level and degree of personal information security risk [13], as well as to encrypt personal information with information technology (such as four-qubit entangled state) [14], and to protect the security of personal information with effective laws [7]. Even so, personal data leakage occasionally occurs on online tourism platforms, and tourists’ concerns about security perception of personal information (SPPI) affect their willingness to use smart tourism applications [15]. At the same time, this also shows the importance of smart tourism technology in smart tourism application. The level of smart tourism technology can directly affect the willingness to use smart tourism applications.

Tourist happiness (TH) is an important reason to travel. Studies using the Q methodology have found that the tourists’ SPPI affects TH [16], and quantitative research has found that tourists’ SPPI indirectly affects TH [17,18]. However, there is a lack of quantitative research exploring the direct impact of SPPI on TH. Therefore, this study employs quantitative research to explore whether SPPI directly affects TH. In practice, the perception of smart tourism application (PSTA) increases tourists’ TH by booking transportation tickets, accommodation, and scenic spot tickets in advance [1,2,3,4]. Therefore, under the condition that personal information security is threatened, discussing whether SPPI affects TH through PSTA has certain research value. At the same time, smart tourism technology is the basis for the development of smart tourism [17,19,20]. The perception of smart tourism technology (PSTT) actually refers to the perception of the development of smart tourism technology, and PSTA is tourists’ cognition of the development degree of smart tourism technology. Thus, whether different levels of PSTT have different effects on the pairwise relationship of SPPI, PSTA, and TH is still unknown. In addition, tourists’ SPPI is not only related to TH and to tourists’ use of smart tourism-related applications, but also related to smart tourism destinations and the benign development of the whole field of smart tourism. These issues must be given great attention.

Therefore, this paper studies the influencing mechanism linking tourists’ SPPI and TH. Specifically, with SPPI, PSTA, PSTT, and TH as the major variables, data were collected through questionnaires, and the following problems were addressed: 1. Does SPPI directly affect TH? 2. Can SPPI affect TH through PSTA? 3. Does PSTT have a moderating effect on a relationship between any two of SPPI, PSTA, and TH? With regard to the above problems, this paper uses a mixed research method [21,22,23,24] and discusses an influencing mechanism via which SPPI affects TH according to perceived value theory and subjective wellbeing theory. Data on relationships among SPPI, PSTA, and TH were collected through small-sample in-depth interviews [25,26,27], and it was found that the height of PSTT affects the relationship between two of the previous three in different ways. The results of the qualitative study were used as the basis for a quantitative analysis of a large sample. Such a combination of qualitative and quantitative research not only allows the exploration of the possible relationships between different variables from small samples but also validates the mechanism of the influence of SPPI on TH through large samples [21,28]. The remainder of this paper is organized as follows: Part 1 is a literature review on SPPI, PSTA, PSTT, and TH, and a discussion of the main concepts and theories in this paper. Part 2 is divided into two phases, namely, qualitative research and quantitative research, with qualitative research as a supplement and quantitative research as the main focus. The first phase, qualitative research, includes qualitative data collection and qualitative research results. The second phase, quantitative research, includes a quantitative research study model and theoretical hypothesis, research design, and results. Part 3 includes five segments, namely, the discussion, conclusion, research implications, research limitations, and outlook.

2. Literature Review

2.1. Security Perception of Personal Information (SPPI)

Personal information security specifically relates to the content of consumers’ personal information data and whether those data are secure after being collected [29]. With the rise and development of Internet platforms, the problem of personal information security has become increasingly obvious, and these security issues exist in all aspects related to the Internet, including payment security [11], personal health data security [30,31], and network access security [32]. The disclosure of personal information may lead to the occurrence of Internet crimes, and the individual whose information is leaked becomes the victim [33]. Therefore, different methods must be used to prevent personal information from being leaked. Strengthening government legislation to protect personal information would be one such method [34,35,36]. Enhancing the software encryption function to increase the security of message transmission would be another [14]. Furthermore, the life cycle management of personal information is also necessary [37]. Even so, the disclosure of personal information, which has been reported from time to time, is harming consumers’ personal interests in different ways.

SPPI, as discussed in this paper, emphasizes tourists’ perception of whether their personal information is secure while touring; this includes issues such as payment security perception [11,12,38,39], information query security perception [40], and sharing security perception [41,42]. Personal information security may involve tourists’ personal information [29,43] and vital interests, or even threaten their property security. Personal information security is, therefore, vital in the travel industry. Tourists’ SPPI is involved in the whole touring process, and this may directly affect tourists’ use of smart tourism applications. According to the study by Pai et al. [18], SPPI affects TH through the mediation of tourist satisfaction. Thus, whether tourists’ SPPI can affect TH directly is a topic worth discussing. Tourists’ SPPI is reflected through the use of PSTA (such as making reservations [12]) in travel. Therefore, the fact that PSTA can become a mediating variable of SPPI and TH must be discussed. At the same time, PSTT plays an important role in SPPI; thus, it is also necessary to explore the moderating effect of PSTT on the SPPI, PSTA, and TH.

2.2. Perception of Smart Tourism Technology (PSTT)

Smart tourism technology is the synthesis of different technologies in the tourism industry, which includes three categories. The first category is information technology. Information technology (such as 5G networks) is the basis for the development of smart tourism [44,45]. With the development of information technology, different applications can be developed, and better services can be provided for tourists. In addition, targeted marketing can be carried out for tourists, and the tourism industry can be managed in a scientific and orderly manner [46]. The second category comprises the technologies related to immersive experiences; these are mainly augmented reality (AR) and virtual reality (AV) [47,48,49]. In practice, VR and AR technologies allow visitors to immerse themselves in the characteristics of the scenic area and even intuitively experience the local history and culture. Unlike traditional sightseeing, this is a technological and intuitive experience. The third category comprises technologies related to artificial intelligence (AI). The application of AI technology in tourism mainly takes two forms; intelligent robots provide services for hotel guests, and intelligent services are provided on tourism platforms [50,51,52]. Smart tourism technology is needed for the development of smart tourism. Although different tourists have different attitudes toward smart tourism technology, technology is everywhere in the current era. In order to acquire destination information in advance and to make travel arrangements, smart tourism technology must be used and experienced. The level of smart tourism technology, such as the ability to support smart tourism applications and the ability to protect personal information, will indeed affect people’s tourism experience.

In practice, PSTT is tourists’ perception of smart tourism technology applications of the above three types of tourism-related technology, such as analysis of social context mobile marketing [53], AR technology [49,54], and artificial intelligence [16]. Lee et al. built a model based on smart tourism technology experience and perception to evaluate TH [18]. Pai et al. found that PSTT is related to SPPI and influences TH by affecting tourist satisfaction. However, the use of smart tourism applications will inevitably bring about many problems. Personal information disclosure is a common problem, as tourists want their personal information not to be leaked or misused [16]. Excessive PSTT can also bring negative experiences to tourists at the time, forcing them to reduce negative experiences by limiting the use of smart tourism technology [55]. Moreover, excessive PSTT may also reduce TH. Therefore, it remains uncertain whether PSTT will affect the relationship between any two of SPPI, PSTA, and TH. However, these are very crucial problems for smart tourism development.

2.3. Perception of Smart Tourism Application (PSTA)

In practice, PSTA is tourists’ perception of the application of smart tourism technology in the tourism industry [56,57]. Smart tourism applications have undergone a progressive development process. Tourists’ PSTA has become increasingly evident with the increasing popularity of smartphones [39,58] and the development of information technology [45,59]. The smart tourism applications most often experienced by tourists involve aspects of “food, shelter, transportation, tours, shopping, and entertainment” [42,54,60,61], as well as the whole travel process (before, during, and after travel). Tourists must book a reservation before travel. They must also collect destination information [4,62], book air or train tickets [12,63], book hotels [64], and book tickets to visit scenic spots [12]. These reservations and tickets are all obtained from tourism-related websites or apps (such as Ctrip and elong). For example, the process of booking a hotel may be influenced by the hotel’s ‘smart information’ [65]. The location of the hotel [66], whether the hotel can be booked [67], the price at the time of booking, reviews of the authenticity of the hotel, and whether the reservation can be cancelled after the reservation are all part of tourists’ perceptions of the smart tourism experience before they travel. In tourism, PSTA involves activities from the beginning of the travel process to the end, such as navigation [68], smart experiences in scenic spots [50,69,70], and even the hotel check-in experience [20]. In this process, tourists’ PSTA has the most significant influence on the tourists’ sense of the tourism experience. After-travel PSTA focuses on the sharing of travel experiences on personal social platforms or travel-related platforms (such as Ctrip and elong) [42]. Whether the review of the booking can be displayed on the relevant platform [71], and whether the review (especially a bad review) will be blocked [72] are all important factors affecting tourists’ post-travel PSTA. For these precise reasons, tourists’ PSTA (before, during, and after traveling) brings convenience to tourism activities and can affect their experience in the whole process of travel. Furthermore, some studies have shown that tourists’ PSTA can affect tourist satisfaction and TH [17,69]. Meanwhile, PSTA is likely to involve tourists’ SPPI [73]. Thus, whether PSTA can play a mediating role between SPPI and TH is a question that should be explored.

2.4. Tourist Happiness (TH)

TH reflects subjective wellbeing with regard to tourism [74,75], and TH is influenced by many factors. The factors affecting TH are roughly divided into two types. One type is the subjective factors of tourists, such as psychological state [76,77,78,79], quality of life [80], and gender [81]. The factors related to tourists are the emotions or states generated by the tourists themselves, and how those emotions affect the tourists’ experience. Different emotions may be produced before traveling and may also be produced during traveling. The SPPI discussed in this study is generated in tourism activities, which is also an important factor affecting the whole spectrum of tourism activities [50]. Meanwhile, the other factors related to tourism involve the level of PSTT and the level of PSTA, such as destination image [62,74,82], hotels [83,84,85], transportation [86,87], and scenic spots [70,75]. Other factors related to tourism are also the objective reasons that affect TH. The destination’s image [62], the price and quality of the destination hotel [20,83], the degree and quality of transportation convenience [3,63], and the degree of knowledge of the scenic spot [4] are all important aspects of PSTA, which in turn affect the TH. Meanwhile, a low SPPI in the process of tourism will create a feeling of the risk of personal information leakage, bring property risks, and reduce TH. Therefore, this paper mainly discusses the influence of tourists’ subjective factors (SPPI) on TH through the objective factors related to tourism (PSTA), as well as discusses the role of SPTT in travel.

2.5. Risk Perception Theory and Subjective Wellbeing Theory

Risk perception theory can be traced back to the risk comparison method [88]. Langford et al. stated that risk not only includes benefits but also takes into account people’s subjective judgements [89]. Fischhoff et al. proposed a conceptual model that includes changes in risk perception [90] (p. 204). Risk factors can be identified through the risk perception theory; for example, the risk factors related to COVID-19 could be identified [91]. Risk perception mainly involves consideration of physical, psychological, and financial risks in tourism [92,93,94,95]. For example, risk perception theory is used to identify the risks perceived by tourists in shared accommodation, as well as to promote the development of shared accommodation [96]. In this study, risk perception mainly focuses on tourists’ SPPI in smart tourism applications, psychological risk perception, and psychological risk perception, as well as how such SPPI affects tourists’ TH.

Subjective wellbeing theory mainly refers to happiness in life [97,98] and identifies the things that affect subjective wellbeing [99,100], in order to increase happiness in life. Subjective wellbeing theory is a theory commonly used to explain TH [101], which mainly includes aspects of life satisfaction, positive affect, and negative affect [102]. Research has found that tourists’ subjective wellbeing is significantly enhanced after traveling [101]. Through subjective wellbeing theory, the factors affecting tourism happiness are identified [103,104], thus helping to promote the healthy development of the tourism industry.

Risk perception theory and tourism happiness theory are two theories with low correlation. However, the SPPI discussed in this paper can be classified as psychological risk in the risk perception theory, while the TH in this paper belongs to the positive emotion in the tourism happiness theory [105,106,107]. Therefore, risk perception theory and subjective wellbeing theory are suitable for use in explaining the influencing mechanism of SPPI on TH. Moreover, the combination of risk perception theory and tourism happiness theory can promote the common development of the two theories in tourism.

3. Research Methodology

This paper employs a mixed-methods research approach using an exploratory design with quantitative research as the main focus and qualitative research as a supplement [21,22,24]. The use of an exploratory design approach can bring more insight into the mechanisms that influence the SPPI on TH [108,109]. Few studies to date have discussed the moderating effects of PSTT on SPPI, PSTA, and TH. Therefore, a qualitative study is required to explore the interviews about SPPI with PSTA and TH. In particular, the high and low PSTT influencing relationships of the previous variables need to be examined. The mixed-methods research approach of this paper is shown in Figure 1. The first phase is to conduct qualitative research about the content related to SPPI, PSTA, PSTT, and TH [26,27]. The second phase is to conduct quantitative data collection, analysis, and processing, on the basis of the qualitative results [21,22].

4. Qualitative Study

4.1. Qualitative Data Collection

In the first phase, in-depth interviews were used to explore the impact of SPPI on TH. In this paper, semi-structured interviews were used to conduct an exploratory study. The study adopted a convenience sampling method [110] (pp. 107–119) and selected 13 interviewees from the more frequent travelers involved in tourism activities. The interviews were conducted in two ways: face-to-face interviews and telephone interviews.

Table 1. Profile of the interviewees.

Interviewee Code	Gender	Age	Education	Occupation
No. 1	Female	38	Postgraduate degree	Employee of government agencies and public institutions
No. 2	Female	34	Postgraduate degree	Company employee
No. 3	Female	31	Postgraduate degree	Student
No. 4	Male	25	Undergraduate	Student
No. 5	Female	22	Undergraduate	Company employee
No. 6	Male	53	Undergraduate	Employees of government agencies and public institutions
No. 7	Male	36	Associate degree	Self-employed
No. 8	Female	36	Undergraduate	Company employee
No. 9	Male	30	Postgraduate degree	Employees of government agencies and public institutions
No. 10	Female	32	Postgraduate degree	Freelancer
No. 11	Female	30	Undergraduate	Employees of government agencies and public institutions
No. 12	Male	26	Associate degree	Company employee
No. 13	Female	26	Associate degree	Self-employed

shows the basic information of the respondents, including gender, age, and occupation. The semi-structured interviews included questions regarding whether the respondents adopted smart tourism applications during their travels, how they perceived these smart tourism application technologies, whether personal information security was threatened, and the relationship between the previous three variables and TH. The interviews were terminated when data saturation was reached.

4.2. Qualitative Data Reanalysis

Qualitative data analysis was conducted in two steps: data processing and data analysis. Step 1, data processing: the interview recordings were converted to text and analyzed by NVivo. Step 2, data analysis: a three-stage coding procedure including open coding, axial coding, and selective coding was performed in this study [111] (pp. 222–302). In qualitative research, the objectivity of the study was ensured mainly by controlling the research process [112] (pp. 115–120). To ensure the objectivity of the coding, researchers who have been engaged in smart tourism research for a long time were selected for primary and secondary coding, and the categories of the final selective coding were determined after consulting with experts, increasing the confidence of the coding. Four main codes were finally obtained, which were SPPI, PSTT, PSTA, and TH (

Table 2. Interview data coding.

Selective Coding	Axial Coding	Open Coding
Perception of smart tourism application (PSTA)	Smart information	Contents, channels, strategies
	Smart food and accommodation	Food, accommodation
	Smart transportation	Public transportation, self-driving, navigation
	Smart scenic spot	Book tickets, smart explained, factors affecting smart explanation, recreational facilities
	Booking channels	WeChat public, Hubei tourism App, 12306, Ebooking, Airbnb, Ctrip, Flying pig, Meituan, Mafengwo tourism, Popular review, Smart train ticket
Security perception of personal information (SPPI)	Consequences of personal information leakage	Sales promotion, fraud, other potential risks
	Handling of personal information leakage	Personal handling, official handing
Perception of smart tourism technology (PSTT)	High	Reliable technology, funding protection, self-judgment
	Low	Login or register, preference leakage, location leakage
	Do not care	Do not care
Tourist happiness (TH)	Positive impact	Convenience, personalized events, information, plans, discounted price, smart services
	Negative impact	App usage problems, traditional booking, contrast, lack of autonomous plans, nontransparent information
	No impact	No impact

).

4.3. Qualitative Study Results

The analysis of the interview data led to two results. On the one hand, the results were related to the relationship among SPPI, PSTA, and TH. Firstly, PSTA affects TH, and this finding is in line with previous studies on the relationship between the two [1,70,113]. From the analysis of the interviews, it was found that PSTA is able to improve TH. Transportation tickets can be booked in advance: “Ctrip (app that allows you to book transportation, accommodation, etc.) booked (airplane) tickets (No. 11)”. It is also possible to book a hotel in advance: “Booking tickets and booking a hotel with Flying Pig, generally, we also use Ctrip” (No. 3). “We stayed at the Sheraton Hotel; we checked in advance. The Sheraton, in that small period; the room rate is also cheaper (No. 10)”. Secondly, the SPPI affects the perception of smart tourism applications, a finding which is in line with the study of Afolabi et al. [73]. The level of SPPI affects the PSTA, and the SPPI is different for different platforms of smart tourism applications: “Meituan, Ctrip, Yilong: these (applications) are still more reliable, I think. Therefore, when there is a personal information login problem with the applet, I will be more cautious about it” (No. 2). Thirdly, the findings regarding the effect of SPPI on TH are consistent with those of Pai et al. [18]. Trust in large software programs or platforms allows travelers to use them with peace of mind, bringing variables to travel. “Meituan, Ctrip, Yilong: these (applications) are still more reliable, I think. These kinds of applications that everyone uses, usually protect my personal information” (No. 2).

On the other hand, there is more disagreement about the effect of high or low PSTT on the relationship among SPPI, PSTA, and TH. Previous studies have focused on the effect of PSTT on SPPI [114], PSTA [20], and TH [17,55], with less attention given to moderating variables to discuss the effect on the relationship among the three. First, the degree of PSTT varies across smart tourism applications. Higher SPPI increases the PSTA when respondents have a higher PSTT; higher SPPI also increases TH: “Technology is somewhat guaranteed” (No. 9). “Meituan, Ctrip, Yilong: these (applications) are still more reliable, right?” (No. 2). “It also does bring convenience to my travel and life” (No. 1). When interviewees’ PSTT is low, lower SPPI would reduce the PSTA and TH, while lower PSTA would reduce TH. “Many other times, when I need to scan the QR code, I will be cautious and not use it if I can” (No. 6). “The news reports many frauds (being committed) through phone numbers or WeChat” (No. 1), and “It brings trouble to work and life” (No. 2). Second, different smart travel programs have the same PSTT, and interviewees felt that all of the programs were risky, but they still had to use them. “Probably the most used ones are the ones with problems like cell phone number registration and WeChat login” (No. 8). “I don’t care, and I just use them (these smart tourism applications).” (No. 13). “I have to use it. We do not want to buy tickets at the train station” (No. 9). Therefore, it is necessary to focus on whether PSTT as a whole, when no distinction is made between smart tourism applications, has a moderating effect on the relationship among SPPI, PSTA, and TH.

5. Quantitative Study

The second phase was a quantitative study [21,24]. On the basis of the results of the above literature analysis and qualitative analysis, hypotheses and arguments were made in order to verify whether the influence of SPPI on PSTA exists, whether the influence of SPPI on TH exists, and whether the influence of PSTA on TH remains. The objective was to verify whether PSTT has a moderating effect on SPPI, PSTA, and TH, and whether SPPI can influence TH through PSTA.

5.1. Research Models and Theoretical Hypotheses

5.1.1. Security Perception of Personal Information (SPPI) and Perception of Smart Tourism Application (PSTA)

SPPI is tourists’ perception of whether their personal information is secure before, during, or after touring [18] and is especially affected by the perception of PSTA. SPPI includes payment security perception [11,12,38,39], perception of information sharing security on social networking sites [41,42], and security perception of reservation applications [12,63]. Personal information security is related to tourists’ vital interests, especially their economic interests. Therefore, tourists’ SPPI will affect their behaviors, such as purchasing behavior [115], making this an issue of interest to scholars [9]. PSTA inevitably involves the various methods that can bring convenience to touring, with the most common methods including tour reservation [12,38,62], navigation [62,116], AR experience [48,49,54], and wisdom experience in scenic spots [4,64]. When tourists use smart tourism applications, there may be problems involving login with personal information [73], which relates to tourists’ SPPI. Therefore, in this paper, we hypothesized the following:

Hypothesis 1a (H1a). 

SPPI affects PSTA.

5.1.2. Security Perception of Personal Information (SPPI) and Tourist Happiness (TH)

SPPI affects TH. SPPI mainly comprises payment security perception [11,12,38,39] and application security perception [12,63,117], all of which are related to the status of tourists’ SPPI before, during, and after touring. If the security of personal information is threatened and personal information may be leaked, the security of the property (and other properties of tourists) will be potentially damaged during or after the tourism event [9,15]. Tourists’ positive perception of their experiences will be reduced, and their happiness will be affected. Furthermore, TH develops from subjective wellbeing [75,118,119], which exists during the entire touring process [120,121,122], as well as in a period of time after the tour ends [123]. If tourists’ SPPI decreases, they will have an unpleasant experience, and TH will also decrease [124]. Therefore, the SPPI may directly affect the TH. Moreover, SPPI can affect TH through tourist satisfaction [18]. We, thus, hypothesize the following:

Hypothesis 1b (H1b). 

SPPI affects TH.

5.1.3. Perception of Smart Tourism Application (PSTA) and Tourist Happiness (TH)

PSTA affects TH. PSTA is related to information query [2,125], reservation before touring [54,68], navigation during touring [4,48,49], and other behaviors. The PSTA not only includes the perception of the content of a smart tourism application, but also includes other problems associated with smart tourism applications. Will smart tourism applications be affected by network signals [45]? Can smart tourism applications be used on smart phones [67,126]? Will there be risks of personal information leakage [127]? PSTA is the tourists’ overall perception of smart tourism applications and the subjective feeling of smart tourism. TH is a subjective experience produced by tourism [75,81,121]. There are many factors that affect TH, such as destination image [74,128], smart tourism technology [17,18], and food [120]. PSTA is an important factor influencing TH [1,48]. Therefore, we hypothesize the following:

Hypothesis 2a (H2a). 

PSTA affects TH.

5.1.4. The Mediating Role of Perception of Smart Tourism Application (PSTA)

PSTA is a mediating variable in the relationship between SPPI and TH. SPPI and PSTA are interrelated and inseparable parts of tourists’ touring process [73], especially with regard to reservation and navigation [113,127]. Moreover, SPPI is reflected through PSTA. SPPI refers to sensing whether the security of personal information may be compromised in the process of using smart tourism [114]. Therefore, SPPI will affect PSTA, and PSTA brings convenience to tourism activities. Knowing more tourist destination information in advance reduces “tourism blindness” [2,116]. People book tickets online to reduce the hassle of going to train stations or airports [54,68]. They book accommodation in advance to avoid getting ripped off at their destination to gain a sense of security [64,67]. Meanwhile, PSTA is an important factor affecting TH [1,48]. Thus, in this paper, we hypothesize the following:

Hypothesis 2 (H2). 

PSTA plays a mediating role between SPPI and TH.

5.1.5. The Moderating Effect of Perception of Smart Tourism Technology (PSTT)

PSTT moderates the relationship among SPPI, PSTA, and TH. SPPI plays a significant role in PSTA [73,129]. The findings of the qualitative study showed that, when the level of PSTT is high, the role of SPPI on PSTA is more obvious; when the level of PSTT decreases, the effect of SPPI on PSTA may not be significant. Therefore, this study proposes that PSTT has a moderating effect on SPPI and PSTA. Similarly, PSTA plays a significant role in TH [48,69]. The results of the qualitative study showed that, when the level of PSTT is high, the effect of PSTA on TH is more significant; when the level of PSTT is low, the effect of PSTA on TH may not be significant. Therefore, PSTT has a moderating effect on PSTA and TH. Similarly, SPPI plays a significant role in TH [18,124]. The results of the qualitative study showed that, when the level of PSTT is high, the effect of SPPI on TH is more significant; when the level of PSTT is low, the effect of SPPI on TH may not be significant. Therefore, PSTT has a moderating effect on SPPI and TH. Therefore, H3a, H3b, and H3c were posited on the basis of the above discussion. Moreover, a diagram of the research model in this paper is shown in Figure 2.

Hypothesis 3a (H3a). 

PSTT has a moderating effect on the link between SPPI and PSTA.

Hypothesis 3b (H3b). 

PSTT has a moderating effect on the link between SPPI and TH.

Hypothesis 3c (H3c). 

PSTT has a moderating effect on the link between PSTA and TH.

5.2. Study Design

5.2.1. Measurement of Variables

This study mainly concerned four dimensions: SPPI, PSTA, PSTT, and TH. The scale provided by Pai et al. was used for SPPI items [18], those provided by Tavitiyaman et al. and Wang et al. were used for PSTA items [113,127], those provided by Huang et al. and Um and Chung were used for PSTT items [130,131], and those provided by Huta and Ryan and Liu et al. were used for TH items [120,132]. The questionnaire was divided into two parts: the first measured the variables and was scored on a five-point Likert scale from 1 = strongly disagree to 5 = strongly agree, while the second concerned demographic information, which was completed by the participants according to their actual situations (The complete questionnaire is in the Supplementary Materials).

5.2.2. Sample Selection

In this paper, the period for data collection was from April to May 2022. The questionnaires were collected online and offline. Because China’s COVID-19 control policies are strict, different local governments have adjusted their policies to limit population movement, specifically in response to the epidemic situation in their respective provinces. This made it difficult for the authors of this study to enter scenic spots or other public places to collect questionnaires. Ultimately, online snowball sampling had to be used to collect questionnaires. At the same time, China’s tourism has undergone a trend of popularization. Tourism APPs, the Internet, and official accounts are the main ways to obtain tourism information. Tourists have direct or indirect experience of smart tourism, and even online snowball sampling is representative. Not many questionnaires are collected offline; random sampling in university cafeterias and gymnasiums is mainly used. Due to the COVID-19 epidemic prevention policy (zero-contact principle), even questionnaires collected offline were completed online (https://www.wjx.cn/, accessed on 18 October 2022).

Ultimately, 523 questionnaires were collected, of which 361 valid questionnaires were obtained after those with the same option for all items—except the personal information section (answers with the same option indicated that the respondents may not have filled them in carefully)—or those that were completed in too short a time were excluded. According to three investigators, the shortest valid completion time was 48 s; thus, the questionnaires with a completion time of less than 48 s were excluded). This gave an effective rate of 69.02%.

5.2.3. Data Analysis

SPSS and AMOS were combined to analyze the data. In particular, the plug-in Process 3.5 was used to analyze the moderating effect; SPSS was used for descriptive statistical analysis (basic population information statistics) and exploratory analysis (testing the construct validity of the scale and eliminating invalid items). In addition, SPSS and AMOS were used for reliability and discriminant validity analyses, and AMOS was used for hypothesized mediating effect analysis. Process 3.5, a plug-in of SPSS, was also used to analyze the moderating effect [133]. The steps taken in the moderating effect analysis in SPSS software are relatively complicated, and only piecewise tests can be carried out for multiple models. Therefore, this paper employed Hayes scholars to develop a plug-in Process 3.5 that can be set for SPSS, to analyze the moderating effect of PSTT between SPPI, PSTA, and TH.

5.3. Results

5.3.1. Descriptive Statistics

Through questionnaire analysis, several results were obtained, as shown in

Table 3. Characteristics of participants in the study.

Variables (N = 361)		Frequency	Percentage
Gender	Female	189	52.4%
	Male	172	47.6%
Age	18–30	166	46.0%
	31–40	109	30.2%
	41–50	74	20.5%
	51–60	9	2.5%
	>61	3	0.8%
Level of education	High school degree or less	19	5.3%
	Associate degree	64	17.7%
	Undergraduate	182	50.4%
	Postgraduate degree	96	26.6%
Work	Employees of government agencies and public institutions	82	22.7%
	Self-employed	34	9.4%
	Company employee	100	27.7%
	Freelancer	36	10.0%
	Student	87	24.1%
	Other	22	6.1%
Monthly income (CNY)	<3000	79	21.9%
	3001–5000	87	24.1%
	5001–8000	110	30.5%
	8001–10,000	40	11.1%
	>10,001	45	12.5%

. The proportion of females (52.4%) was higher than that of males (47.6%); 71.3% of respondents were in the range of 18–40 years old. In the collected questionnaires, the participants with associate degrees (17.7%), undergraduate degrees (50.4%), and postgraduate degrees (26.6%) accounted for 94.7%, and those with high school degrees or less accounted for a lower proportion (5.3%). In terms of work, company employees (27.7%) and employees of government agencies and public institutions (22.7%) jointly accounted for a relatively high proportion, 50.4% in total. The distribution of monthly income did not vary greatly, with no significant differences presented in the proportions of <3000 CNY (21.9%), 3001–5000 CNY (24.1%), 5001–8000 CNY (30.5%), and above 8001 CNY (23.6%). In short, the descriptive statistics of the samples were ideal, which laid a good foundation for later analyses.

5.3.2. Exploratory Factor Analysis (EFA)

SPSS (version number: SV26; creators: Norman H. Nie, C. Hadlai (Tex) Hull, and Dale H. Bent; company location: New York, America) was used for EFA analysis, to test the construct validity of the scale and exclude items. Principal component analysis and the maximum variance method were used to extract principal components [134] (pp. 612–680), and the eigenvalues were required to be greater than 1. The KMO value for the scale was 0.893, and the significance level of Bartlett’s test was p = 0.000 (

Table 4. Exploratory factor analysis.

Items	Factors	Factor Loading	Characteristic Root	Cumulative Explained Variance
PSTA	PSTA1	0.624	6.349	37.349%
	PSTA2	0.788
	PSTA3	0.825
	PSTA4	0.817
	PSTA5	0.730
SPPI	SPPI1	0.842	3.698	21.754%
	SPPI2	0.920
	SPPI3	0.918
	SPPI4	0.910
	SPPI5	0.924
PSTT	PSTT1	0.793	1.716	10.093%
	PSTT2	0.829
	PSTT3	0.848
	PSTT4	0.735
TH	TH1	0.864	1.091	6.419%
	TH2	0.823
	TH3	0.817

Notes: (1) Total variance explained: 75.614%. Extraction method: principal component analysis. (2) Rotation method: oblimin with Kaiser normalization. Rotation converged in five iterations. (3) Security perception of personal information (SPPI), perception of smart tourism application (PSTA), perception of smart tourism technology (PSTT), and tourist happiness (TH).

), indicating that the scale was suitable for factor analysis. The items with a factor loading higher than 0.4 were valid [134] (pp. 612–680). Therefore, 19 items were retained, involving four dimensions: PSTA, SPPI, PSTT, and TH. The eigenvalues were all greater than 1, and the cumulative variance explanation rate was 75.614%, indicating that the scale had good construct validity and was suitable for subsequent analysis.

5.3.3. Reliability and Validity Test

SPSS 26.0 and Amos were used to conduct reliability analysis on the scale (

Table 5. Reliability and validity test of the scale.

Items	Factors	uStd	SE	Z-Value	p	Std	Cronbach’s α	CR	AVE
PSTA	PSTA1	1.000				0.518	0.853	0.858	0.555
	PSTA2	1.496	0.153	9.762	***	0.779
	PSTA3	1.624	0.161	10.109	***	0.843
	PSTA4	1.705	0.165	10.319	***	0.894
	PSTA5	1.180	0.136	8.685	***	0.626
SPPI	SPPI1	1.000				0.783	0.947	0.948	0.786
	SPPI2	1.129	0.055	20.461	***	0.921
	SPPI3	1.151	0.058	19.771	***	0.899
	SPPI4	1.148	0.059	19.525	***	0.890
	SPPI5	1.139	0.055	20.757	***	0.931
PSTT	PSTT1	1.000				0.772	0.903	0.908	0.712
	PSTT2	1.188	0.064	18.648	***	0.895
	PSTT3	1.215	0.063	19.174	***	0.919
	PSTT4	1.184	0.075	15.768	***	0.780
TH	TH1	1.000				0.921	0.855	0.872	0.695
	TH2	0.878	0.045	19.333	***	0.831
	TH3	1.013	0.062	16.474	***	0.739

Notes: uStd = unstandardized coefficient, Std = standardized coefficient, CR = composite reliability, AVE = average variance extracted; *** p < 0.001.

). The results indicate that the values for Cronbach’s α of PSTA, SPPI, PSTT, and TH were all in the range of 0.853 to 0.947 and, thus, all greater than 0.7, indicating that the scale had good reliability and met the requirements of internal consistency [135] (p. 816). The standardized coefficients were all in the range of 0.5 to 0.95, which was of practical significance [135] (p. 816). The CR values ranged from 0.858 to 0.948, all greater than 0.7 [135,136] (p. 816). The values of AVE for PSTA were in the range of 0.555 to 0.786, all greater than 0.5 [137], meeting the conditions of data analysis. It is generally believed that, when the correlation coefficients between latent variables and other latent variables are less than the square root of the AVE, they have good discriminant validity. It can be seen from

Table 6. Results of the discriminant validity test.

	PSTA	SPPI	PSTT	TH
PSTA	0.745
SPPI	0.166	0.886
PSTT	0.616	0.153	0.844
TH	0.612	0.230	0.422	0.834

Note: The bold italic figures on the diagonal represent the square roots of the average variances extracted (AVE).

that the square roots of AVE (the bold figures in the diagonal) were all larger than the correlation coefficients between latent variables, indicating that they had good discriminant validity and met the analysis requirements [136].

5.3.4. Model Fitting and Structural Model Verification

The data were analyzed using the maximum likelihood method with Amos, and the following fitting indices were obtained: χ² = 96.011, df = 62, χ²/df = 1.549, SRMR = 0.028, RMSEA = 0.039, CFI = 0.990, and TLI = 0.987. The main fitting indices were all greater than 0.9, indicating that the model fit the data well [135] (p. 816). According to data analysis (

Table 7. Results of the structural equation model.

Hypothesis	Path	Estimate	T-Value	p	Results
H1a	SPPI → PSTA	0.070	2.609	**	Yes
H2a	PSTA → TH	0.624	6.027	***	Yes
H1b	SPPI → TH	0.122	3.183	**	Yes

Notes: *** p < 0.001, ** p < 0.01.

), H1b (λ = 0.122, p < 0.01) was supported, indicating that SPPI had a significant positive direct impact on TH. Therefore, the model in this paper was a partial mediation model. H1a (λ = 0.070, p < 0.01) was supported, indicating that SPPI had a positive impact on PSTA. H2a (λ = 0.624, p < 0.001) was supported, indicating that PSTA affects TH. Figure 3 presents all the hypothesis results. The hypothesis path connection lines that passed the test are revealed, and the hypothesis numbers are in bold; the hypothesis path lines that failed are dashed, and the numbered fonts are normal.

5.3.5. Mediating Effect Test

In this study, the bootstrap method was used to test the mediating effect of PSTA between SPPI and TH [138] (p. 488). According to the suggestion of Hayes, the standard for bootstrap sample size was 1000, and a confidence level of 95% was used for testing [139]. For the total effect, indirect effect, and direct effect, if the confidence interval of the bias-corrected method and that of the percentile method did not contain 0 at the 95% confidence level, the total effect, indirect effect, and direct effect were significant [139].

Table 8. Mediating effect tests.

Path Relationship	Point Estimate	Product of Coefficient		Bootstrapping 1000 Times 95%
				Bias-Corrected		Percentile
		SE	Z	Lower	Upper	Lower	Upper
Indirect effects
SPPI → TH	0.044	0.021	2.095	0.009	0.091	0.008	0.087
Direct effects
SPPI → TH	0.122	0.039	3.128	0.051	0.210	0.050	0.209
Total effects
SPPI → TH	0.166	0.043	3.860	0.081	0.253	0.084	0.255

Note: SE = standard error.

shows that the confidence interval of the bias-corrected method and that of the percentile method did not contain 0 for the direct effect between SPPI and TH at the 95% confidence level; thus, the direct effect was significant (H1b was supported), and the mediation model was a partial mediation model. Similarly, it can be seen from Table 8 that the mediation effect was significant (H2 was supported), and the total effect was also significant (as shown in Figure 3).

5.3.6. Moderating Effect Test

The plug-in Process 3.5 of SPSS was used to test for moderating effects; results are shown in

Table 9. Moderating effect results test.

	Model 1	Model 2
Constant	0.009 (0.044)	0.031 (0.047)
PSTT	0.554 *** (0.045)	0.461 *** (0.055)
SPPI	0.055 (0.046)	0.134 ** (0.046)
PSTA	——	0.058 ** (0.055)
PSTT × SPPI	−0.050 (0.044)	−0.103 * (0.047)
PSTT × PSTA	——	−0.023 (0.034)
Overall Coefficient	R2 = 0.332 *** (p < 0.001)	R2 = 0.331 *** (p < 0.001)
Interaction Coefficient	△R2 = 0.002 (p = 0.258)	△R2 = 0.009 * (p = 0.029) (PSTT × SPPI)	△R2 = 0.001 (p = 0.500) (PSTT × PQSTA)
Result	No	Yes	No

Notes: *** p < 0.001; ** p < 0.01; * p < 0.05.

and Figure 3. The data for Model 1 indicate that the interaction term (SPPI × PSTT) of PSTT and SPPI had no significant effect on PSTA (coefficient: 0.050; SE: 0.044; p > 0.05); thus, H3a was not supported. The data for Model 2 were divided into two parts. The interaction terms of PSTT and SPPI had a significant negative effect on TH (coefficient: −0.103; SE: 0.047; p < 0.05); thus, H3b was supported. The interaction terms of PSTT and PSTA had no significant effect on TH (coefficient: −0.023; SE: 0.034; p > 0.05); thus, H3c was not supported.

The data in Table 9 indicate that PSTT only had a moderating effect on SPPI and TH. To intuitively show the moderating effect of PSTT on SPPI and TH, the PSTT data higher than the sample mean plus one standard deviation were defined as high PSTT, and while the data lower than the sample mean minus one standard deviation were defined as low PSTT [140] (pp. 259–253), to obtain the moderating effects of SPPI and TH under different levels of PSTT (Figure 4a). As can be seen from Figure 4a, compared with high PSTT, low PSTT had a larger slope, and SPPI had a stronger relationship with TH.

The Johnson–Neyman method was used to analyze the moderating effects through a simple slope test. By first setting the critical point (p = 0.05) for whether the regression coefficients p for the independent variable and the dependent variable were significant or not, the value from which the moderator variable started to have moderating effects, such that the effect of the independent variable on the dependent variable became significant, was determined [141]. The results shown in Figure 4b indicate that the effect was significant when the confidence interval for the conditional effect of SPPI on TH was above 0, and the corresponding value for PSTT was less than −2.453 in this statistical area, indicating that PSTT had a significant negative effect on SPPI and TH. When the value for PSTT was greater than −2.453, the interval of the conditional effect included 0, indicating that PSTT had no significant effect on SPPI and TH in this statistical area. In short, the increase in SPPI could reduce TH only when the value for PSTT was within the confidence interval lower than the critical value; when the value for PSTT was within the interval higher than the critical value, the increase in SPPI had no significant effect on TH.

6. Discussion and Conclusions

6.1. Discussion

Four main conclusions can be drawn from this study. First, the quantitative study verified that PSTT had no significant moderating effect between SPPI and PSTA or between PSTA and TH, but PSTT did have a significant moderating effect between SPPI and TH. Previous studies have indicated that PSTT is related to SPPI [73,114], PSTA [20,142], and TH [55]; however, both the quantitative study and the qualitative study in this paper showed that PSTT only had a moderating effect on the relationship between SPPI and TH.

There was a difference in the moderating effect of PSTT on SPPI and TH in the qualitative and quantitative studies. The reason for the difference may be that the qualitative study investigated fewer subjects limiting the conclusions drawn, while the quantitative study investigated a wider range of subjects and reached statistical significance; hence, the results may be more representative. The quantitative study suggested that there may or may not have been a positive moderating effect of PSTT on SPPI and TH. Quantitative studies showed that PSTT had a negative moderating effect on SPPI and TH. When the value for tourists’ PSTT was higher than the critical value (−2.453), the confidence interval for the moderating effect contained 0, and the moderating effect of PSTT on SPPI and TH was, thus, not significant [141]. When the value for PSTT was lower than the critical value (−2.453), the confidence interval for the moderating effect did not contain 0, and the moderating effect of PSTT on SPPI and TH was significant. When the value of PSTT was lower than the critical value, the SPPI played a significant role in TH. When the PSTT’s value was higher than the critical value, the SPPI played an insignificant role in TH. This may have been caused by tourists’ doubts about whether smart tourism technology can protect their personal information [9,38]. PSTT brings about more PSTA; more places need to input personal information, and the risk of personal information leakage is greater [33,63]. In addition, different humans have different acceptance levels with regard to information technology and the application of that technology [143,144]. Too many applications may reduce users’ willingness to use them. Therefore, when the PSTT is higher than the critical value, the role of SPPI on TH is not significant.

Second, the qualitative study found that SPPI affects TH; the quantitative study verified that SPPI was found to directly affect TH, and it could also affect TH through PSTA. SPPI indirectly affects TH, a finding which is consistent with the research of Pai et al. [18]. Pai et al.’s research found that SPPI can affect TH through tourism satisfaction [18], while this paper found that SPPI affects TH through PSTA. Previous studies on SPPI have tended to investigate only one aspect at a time, such as payment security [9,11,115] or reservation security [12,63], and studies have only rarely been conducted in the field of tourism. However, these problems inevitably occur in tourism. Therefore, this paper explored the influencing mechanism for SPPI and TH, indicating that tourists’ SPPI can affect TH both directly and indirectly.

Third, the qualitative study showed that the PSTA affects TH. The quantitative study verified that PSTA was found to not only directly affect TH, but also act as an intermediary variable via which SPPI affected TH. Prior studies have explored PSTA from various perspectives, such as the impact of PSTA on the management of tourist destinations [62] and the role of mobile applications in smart tourism [61]. There have also been studies exploring the relationship between PSTA and TH; for example, a VR application increased TH by producing flow experience [145]. This paper explored not only the impact of PSTA on TH, but also the impact of PSTA as an intermediary variable in the effect of SPPI on TH, thereby pushing our understanding of PSTA one step forward.

Lastly, the qualitative study and quantitative study together validated that, as an outcome variable, TH was found to be related to SPPI, PSTA, and PSTT. Consistent with previous studies, this paper found that SPPI is related to TH [18,124], PSTA affects TH [54,69,146], and PSTT also affects TH [17]. However, this paper not only revealed the impact of a single factor on TH, but also highlighted the positive impacts of SPPI and smart tourism-related factors on TH, except for the impact of negative factors of TH and life satisfaction on TH.

6.2. Conclusions

Using mixed methods, in combination with perceived value theory and subjective wellbeing theory, this paper established a model with SPPI as an independent variable, PSTA as a mediating variable, PSTT as a moderator variable, and TH as a dependent variable. Through data collection and analysis, the following conclusions were drawn:

• SPPI has a significant positive effect on TH (λ = 0.070, p < 0.01). The data analysis results showed that SPPI had a significant direct effect on TH, and the degree of tourists’ SPPI could directly affect the degree of their TH. Meanwhile, the mediating effect was supported; hence, the model in this paper was a partial mediation model.
• SPPI affects TH through PSTA (as evidenced by the fact that neither the confidence interval of the bias-corrected method nor the confidence interval of the percentile method contained 0). The data analysis in this paper showed that SPPI could have an effect on TH through the mediating effect of PSTA, and this was an influencing mechanism for the effect of SPPI on TH.
• PSTT has no significant moderating effect on SPPI and PSTA (coefficient: 0.050; SE: 0.044; p > 0.05) or on PSTA and TH (coefficient: −0.023; SE: 0.034; p > 0.05), but it has significant moderating effects on SPPI and TH (coefficient: −0.103; SE: 0.047; p < 0.05). In the interval with significant moderating effects, PSTT was found to have negative moderating effects on SPPI and TH.7.

7. Research Implications, Research Limitations, and Prospects

7.1. Research Implications

To explore the relationship between different variables, SPPI was applied to the tourism environment to explore the mechanism via which SPPI affects TH. SPPI can not only directly affect TH, but also affect TH with PSTA as an intermediary, pushing our ability to apply SPPI in the tourism industry one step forward. This study also found that PSTT only has a moderating effect on the relationship between SPPI and TH, and the effect is negative. In other words, the impact of SPPI on TH is significant with low PSTT.

Theoretical innovation is conducted and achieved by combining the two theories. Combining risk perception theory and subjective wellbeing theory to explain the action mechanism for SPPI and TH is a way to advance both theories. SPPI belongs in the category of psychological risk perception in risk perception theory and is a very important aspect in risk perception theory. Meanwhile, SPPI is one of the important factors affecting TH. Risk perception theory was used to explain the phenomenon of SPPI; subjective wellbeing theory was used to explain the phenomenon of TH [101]. The two theories were combined to explain the roles of SPPI and TH, and the combination enhances our ability to apply risk perception theory in the tourism industry.

The mixed-methods research approach was used to explore the mechanism of the influence of SPPI on TH. In the first phase, an understanding of the relationship among SPPI, PSTA, and TH was generally achieved through the method of qualitative research. This study found that high and low PSTT may have different effects on the relationships among SPPI, PSTA, and TH. In the second phase, the relationships among the three were explored through a quantitative analysis, and the study examined whether there is a moderating effect of PSTT. The mechanism of the influence of SPPI on TH was explored by discovering ideas and verifying them.

In the process of smart tourism development, destinations or scenic spots should pay attention to protecting tourists’ personal information security and improving tourists’ SPPI. Tourists’ SPPI will affect their willingness to use smart tourism. Therefore, when a smart tourism product is developed, whether for making a reservation, a payment method, or smart scenic spots APP login with personal information, it is necessary to take into account tourists’ personal interests and improve their SPPI. In the long run, improving tourists’ SPPI can bring a stronger sense of TH and promote the benign development of smart tourism.

7.2. Research Limitations and Prospects

This study had some limitations. Firstly, both quantitative and qualitative data analyses were limited by only using Chinese people as the survey group. There is a huge difference in the degree of development of smart tourism applications in different countries, such as Western countries and African countries. With people of different countries as the surveyed group, the moderating effects of PSTT on the SPPI and the PSTA, as well as the two on TH, are not the same. The survey group can be extended to increase the broader applicability between variables. Secondly, the questionnaires collected in this paper were all completed online. The questionnaires completed in this way were prone to the possibility that a respondent might not have filled out the questionnaire seriously, resulting in a low effective rate of the data in this paper. However, to comply with the COVID-19 epidemic prevention policy and reduce the probability of infection, the questionnaires had to be completed online. In subsequent studies, data can be collected in various forms, and the proportion of paper questionnaires can be increased to improve the validity of the collected data. Thirdly, in this paper, the model was designed and analyzed with SPPI as the independent variable, PSTA as the intermediary variable, PSTT as the moderating variable, and TH as the dependent variable, while some scholars took SPPI as the second-order variable for PSTT for analysis [18]. With SPPI and PSTT taken as two different dimensions for analysis, this paper focused on the effect of SPPI on TH. In subsequent research, SPPI can also be used as a second-order variable to analyze the effects of PSTT and PSTA on TH.