The subjective judgement of the risks presented by COVID-19 to people's health is referred to as perceived risk of COVID-19. Most prior investigations have found that perceived risk can influence attitudes, which in turn influences behavioral intention (Quintal et al., 2010; Seow et al., 2017; Bae & Chang, 2021). Several authors have assessed the influence of risk perception as positive or negative regarding research objectives. Bae and Chang's (2020) study focused on the intention of limited-contact travel as a behavior to reduce risks stemming from the awareness of the COVID-19 pandemic, indicating that if people are aware of the risks, they will have a positive attitude toward contact-limiting travel behavior. Another recent study found that perceived risk affects subjective norms (Zhang et al., 2019). Perceived risk is a critical component that positively influences subjective norms (Bae and Chang, 2020; Xu et al., 2021). When visitors’ families, friends, and surrounding social interactions demonstrate a favorable attitude when engaging in vacation during the awareness of health risks caused by the COVID-19 pandemic, tourists are more likely to gain trust, which leads to the intention to engage in that behavior if exposure is limited. In recent research on the pandemic, several studies have examined the influence of perceived risk on perceived behavioral control (Zhang et al., 2019; Bae and Chang, 2020). According to Zhang et al. (2019), perceived risk reduces perceived behavioral control; however, this study was conducted considering different circumstances because it studied the behavioral intention to purchase poultry in the face of perceived danger during the earlier H7N9 epidemic. According to Xu et al. (2020), perceived risk influences perceived behavioral control and has been examined in the context of contact-limiting travel as a health-promoting behavior. Tourists who perceive a relatively lower concentration of infection risk will be more likely to expand their awareness of their potential, financial resources, expected duration, and available knowledge to prioritize active participation in tourist behavior accommodating their requirements, as suggested by Bae and Chang (2020).

In light of these arguments, we propose the following hypotheses.

• H1: High perceived risk negatively affects tourists’ attitudes toward travel.

• H2: High perceived risk positively affects perceived behavioral control of tourists.

• H3: High perceived risk positively affects tourist's subjective travelling norm.

Government trust refers to a person's belief that the political system and government policy will generate the desired outcomes (Miller & Listhaug, 1990). Nunkoo (2015) defined government trust as the determination of how individuals assess the effectiveness and performance of state policies. The level of government trust influences individual behavior. For example, people who have a high level of confidence in government policy are more inclined to support and adhere to policy requirements. People are more likely to travel if they trust the governments’ general strategies in response to the COVID-19 pandemic because they feel safer and believe the issue is under control (Shin et al., 2022). The concept of trust was used in this study to better understand tourists’ trust in government performance. Trust manifests itself in the implementation of government performance, where it aids community protection and strengthens national unity. If people trust the government to hasten the revival of the economy, particularly the tourism industry, they will respond more effectively to recovery initiatives. In the domestic tourism industry, people will travel if they trust appropriate epidemic management and information about the places they intend to visit. Therefore, we propose the following hypothesis:

• H4: High trust in government performance positively affects tourists’ attitude toward travel.

• H5: High trust in government performance positively affects perceived behavioral control of tourists.

• H6: High trust in government performance positively affects tourists’ subjective travelling norm.

The TPB framework can be used to explain travel behavior (Ajzen, 1985). The constructs of attitude, perceived behavioral control, and subjective norms comprise the TPB. TPB is founded on the theory of reasoned action (TRA), which states that attitudes and subjective norms predict behavior. Various fields utilize TPB, including psychology, health, marketing, physical education, and tourism (De Groot & Steg, 2007; Ferdous, 2010). However, researchers have sought to add more factors to the TPB to boost its explanatory power and improve its behavior prediction accuracy. Owing to the lack of explanatory power regarding behaviors influenced by several determinants, Ajzen (1985) later included a new variable, perceived behavioral control, in TPB. To better understand tourist behavior, other researchers have incorporated elements such as travel motivation, electronic word of mouth, destination image, destination familiarity (Bianchi, Milberg & Cúneo, 2017; Soliman, 2021), and climate change (Juschten, Jiricka-Pürrer, Unbehaun & Hössinger, 2019). For example, Bae and Chang (2021) recently introduced risk-related factors, indicating that cognitive and affective risk perceptions significantly impact behavioral intention. Bae and Chang investigated the impact of COVID-19 risk perception on untact tourist behavior in South Korea during the first wave of the pandemic (March 2020). The COVID-19 pandemic wreaked havoc, hurled current norms into disarray, and necessitated the development of a more robust model for predicting future visitor behavior (Bae & Chang, 2021). This study proposes an extended TPB model incorporating constructs, including perceived risk and trust in government performance, to predict tourists’ intentions. We propose the following hypotheses:

• H7: High attitudes toward travel positively affect tourists’ intention to travel.

• H8: High perceived behavioral control positively affects tourists’ intention to travel.

• H9: High subjective norms of travelling positively affect tourists’ intention to travel.

The survey was conducted from October 2021 to May 2022. This study employed quantitative research methods and was conducted through a questionnaire shared as a Google form that surveyed tourists. The results were used to test the theoretical model. The pre-test was a try-out of the questionnaire to assess and resolve design shortcomings before the actual survey was conducted. A group of 50 respondents, whose characteristics were reasonably similar to the survey population, were selected for the pretest. The questionnaire was then revised and finalized based on the pre-test results.

To maximize the value of the analysis and assess the complexity in the responses, we used two distinct forms of surveys in this study. The author used Google Forms to administer an online questionnaire. The author received an email for each questionnaire filled to eliminate the possibility of a participant filling the form more than once. To conclude, 469 responses were recorded.

A self-completed seven-point Likert scale questionnaire was used in this study. Its contents focused on six constructs including: (1) perceived risk (six items) adapted from Sebastian Zenker, Erik Braun, Szilvia Gyimóthy (2021); (2) trust in government performance (six items) borrowed from Lawrence Hoc Nang Fong, Rob Lawc, Ben Haobin Yed (2021); (3) Attitude to travel (three items) borrowed from Li, Thi Hong Hai Nguyen, J. Andres Coca-Stefaniak (2020); (4) Subjective norm (four items); (5) Perceived behavioral control (four items); Travel intentions (four items) adapted from So Young Bae & Po-Ju Chang (2020).

In this study, the investigated scales were first tested for reliability using Cronbach's alpha coefficients. A scale is considered reliable if the Cronbach's alpha coefficient exceeds 0.7, and the corrected total-item correlation of each item is greater than 0.3 (Nunnally & Bernstein, 1994). Then, the reliable scales were further analyzed by employing Exploratory Factor Analysis (EFA), which required: (1) eigenvalues ≥1; (2) total variance explained ≥50%; (3) KMO ≥ 0.5; (4) Sig. coefficient of the KMO test ≥0.05; (5) Factor loadings of all observed variables ≥ 0.5; and (6) Weight difference between the loadings of two factors > 0.3 (Hair, Black, Babin & Anderson, 2010).

Next, the data were analyzed using Confirmatory Factor Analysis (CFA) to further confirm the results obtained from the EFA. Finally, SEM was used to test the proposed model, which is considered appropriate if the significance value of the Chi-square test is no more than 5%, CMIN/df ≤ 2 (but in some cases, CMIN/df ≤ 3 is also acceptable), and GFI, TLI, CFI ≥ 0.9. Besides these criteria, recent researchers suggest that GFI should be greater than 0.8, RMSEA ≤ 0.08, with composite reliability > 0.6, and extracted variance > 0.5 (Hair, Black, Babin & Anderson, 2010).

In this study, 469 questionnaires were analyzed as valid responses. The results revealed gender differences among respondents, with 45.2% being male and 54.8% being female. In addition, there was a significant disparity in marital status, with 59% being single and 40.3% being married. The group accounting for the highest proportion of responses was between 20 and 39 years of age. The training level was mainly undergraduate or graduate. In terms of income, respondents earned VND10Mil - VND20Mil. One to three times was recorded as the highest number of travel time in 2019, at 56.7%. The demographic characteristics of the respondents are shown in Table 1.

Table 2 presents the results of the scale reliability tests, in which Cronbach's alpha coefficients of the scales were all greater than 0.7, and their corrected item-total correlations were all greater than 0.3. Therefore, the scales were deemed appropriate and acceptable.

The results of the EFA are briefly presented in Table 3, which shows that the analysis result of the four different factors was 61.480%. When evaluating the results of the factors, the value of each item was greater than 0.4, which is suitable for EFA (Hair, Black, Babin & Anderson, 2010). The results of Cronbach's alpha values of factors show that all values are satisfactory and meet the condition of being greater than 0.6 (Nunnally & Bernstein, 1994). The core items fully reflect the content of the factors. Most factor loading values were greater than 0.6, and significant at p <0.001 (Tabachnick & Fidell, 2013). However, one perceived risk item and one perceived behavioral control item were removed because of low loadings. The study model was appropriate for KMO = 0.863, chi-square = 7259.888, df = 351, and Sig = 0.000.

Fig. 1 demonstrates the results of the CFA model, with the aim of reconfirming the relevance of the research model compared to the data collected. Specific results are shown through Chi-square = 845.530, df = 309, p-value = 0.000, CMIN/df = 2.736 <3, GFI = 0.882, TLI = 0.914, CFI = 0.924, and RMSEA = 0.061, which are satisfactory considering the required criteria.

Fig. 1.

CFA results of the saturated model.

(0.27MB).

Fig. 1 confirms the one-way relationship that complies with the conditions related to the convergence value and unique value. Discriminant validity was used to evaluate the AVE values with squared correlations between coupling structures. All AVE estimates were higher than the squared interconstruct correlations. These structures are different from other factors (Fornell & Larcker, 1981; Hair, Black, Babin & Anderson, 2010). The results are presented in Table 4.

In terms of perceived risk, first, the results indicate that the higher the perceived risk due to COVID-19, the more negative the effects on tourists’ attitude toward travel because the results show coefficient β = −0.155 and significant p-value = 0.001 (Hypothesis 1). The results demonstrate that during the COVID-19 pandemic, travelers were concerned about being infected, which means that the greater the perception of risk when travelling, the more adversely it impacts their trip; fear of danger due to traveling, anxiety, and tension all rise sharply. However, when risk awareness is low and tourists feel safe travelling through the outbreak, this positively affects their travel attitude. This observation has been verified in previous studies such as Chew and Jahari (2014), Shim and You (2015), and Bea and Chang (2020).

When individuals are aware that they can control their behavior, it is easy for them to develop their intentions. However, perceiving a high risk of infection while travelling also greatly influences travel behavior control. Therefore, hypothesis 2 is supported owing to the result of coefficient β = −0.101 and significant p-value = 0.033, which show that the higher the perceived risk of COVID-19, the more negatively it will affect perceived behavioral control of tourists. This result is in accordance with the study of Sánchez-Cañizares et al. (2020).

Another finding indicates that risk perception is influenced by relationships with families, acquaintances, and colleagues. Individuals associated with the community tend to be anxious and nervous about infected relatives when travelling. Consequently, with coefficient β = −0.208 and significant p-value <0.01 (hypothesis 3), the higher the perceived risk from COVID-19, the more negatively it will affect the subjective norm of travelling, and similar results were reported by Murray and Schaller (2012) and Bea & Chang (2020).

During the COVID-19 outbreak in Vietnam, the tourism industry was severely and adversely affected, which influenced visitors’ perceptions. Following the outbreak, when the Vietnamese government successfully controlled it, Vietnamese people demonstrated a high level of trust in government decisions. The Vietnamese people supported government policies on quarantine periods and social exclusion, taking the government's prevention measures seriously. Consequently, the government's success affects tourists’ attitudes toward tourism, implying that the more they trust the government, the more likely they are to travel. This hypothesis is accepted because the result of Hypothesis 4 is coefficient β = 0.419 and significant p-value <0.01.

In addition, when the government makes the right decision regarding handling a pandemic and successfully prevents the spread of diseases, every citizen becomes cognizant of their ability to check their behavior. To comply with the government's disease prevention policies, each person should be fully aware of and check their behavior. Additionally, it contributes to successful disease prevention because of strong knowledge of executive control. This supports hypothesis 5 because higher levels of trust in government positively influence visitors’ perceived behavioral control, with a coefficient β = 0.427 and a significant p-value < 0.01.

In addition to the government's effective and correct actions, national solidarity is critical for assessing the effectiveness of the COVID-19 pandemic management in Vietnam (Sen Nguyen, 2020). This solidarity demonstrates that each person trusts and engages with the government in implementing healthy disease prevention measures. It further helps to promote same political goals and values that can be shared among friends, colleagues, and family members. Consequently, it will spread a positive attitude toward travel in the society. Therefore, Hypothesis 6 is accepted when as the results show that high trust in government performance positively affects the subjective norm of travelling among tourists, with the test result coefficient β = 0.144 and a significant p-value = 0.006.

Regarding travel intentions, Hypothesis 7 is accepted, which stated that attitude toward travel more positively affects intention to travel among tourists because the result is supported by coefficient β = 0.188 and a significant p-value <0.01. When a person has a positive attitude toward travel, tourists will have a positive intention to travel. Visitors are most likely to choose to travel as soon as the pandemic is under control and when they have the financial resources and time to do so. Other visitors believe that they would spend additional time ensuring healthcare protection.

Furthermore, as each person can choose the number of days they want to travel and is aware of considerations such as finances and health, this increases travel intention. People have control of their behaviors shows that Hypothesis 8 is accepted, which stated that high decisive perceived behavioral influence positively impacts visitors’ desire to visit. This result is support by coefficient β = 0.523 and a significant p-value <0.01.

Moreover, relationships with travelers who have a positive attitude toward travel can positively affect an individual's travel intention. This suggests that people are more likely to travel if their friends, families, and relatives help them, and can support the view that travelling at this time is safe and supportive for them. With a coefficient of β = 0.122 and a positive p-value <0.009, Hypothesis 9 was accepted. Consequently, the higher the subjective travel norm, the more favorable the impact on tourists’ intentions to travel. Furthermore, the results reported through Hypotheses 7, 8, and 9 have been found in previous research, such as Huang et al. (2020), Lam and Hsu (2006), Bea and Chang (2020), and Sandra M. Sánchez-Cañizares et al. (2020).