Tourists’ preventive travel behaviour during COVID-19: the mediating role of attitudes towards applying non-pharmaceutical interventions (NPIs) while travelling

ABSTRACT Previous studies that extended the Theory of Planned Behaviour (TPB) with the Health Belief Model (HBM) did not integrate all constructs. To address this research gap, we combined those two theories and the subscale on Domain-Specific Risk-Taking (DOSPERT) to predict tourists’ willingness to apply NPIs against COVID-19 while travelling. The proposed hypotheses on the mediating role of attitudes were tested using structural equation modelling based on a random sampling of the Swiss population (n = 1683). The results indicate that attitudes are the strongest predictor of behavioural intentions to apply NPIs while travelling. Attitude also acts as a mediator between health beliefs and the willingness to apply protective measures voluntary for the next tourism trip. The results permit some managerial implications to be suggested for supporting preventive travel behaviour.


Introduction
To promote safe travel during a pandemic, it is crucial to understand how tourists cope with coronavirus disease (COVID-19)-related risks (Zheng et al., 2021).Yet, the COVID-19 outbreak has prompted a vast amount of literature in tourism studies on how COVID-19-related risk perceptions and perceived travel risks influence travel intentions and avoidance (e.g.Abraham et al., 2021;Agyeiwaah et al., 2021;Neuburger & Egger, 2021).Various studies have extended the Theory of Planned Behaviour (eTPB; Bae & Chang, 2021) with COVID-19-related risk perceptions to predict tourists' travel behaviour during and after COVID-19 (for example Liu et al., 2021b;Pahrudin et al., 2021;Rahmafitria et al., 2021;Sánchez-Cañizares et al., 2021).One drawback of these eTPB-studies is that they rely on different concepts and theories of risk.Thus, different operationalizations of the risk construct can be found across the relevant literature, which makes joint basic research difficult.
Another stream of literature stresses the importance of non-pharmaceutical interventions (NPIs) and its relation to tourism (for example Castañeda-García et al., 2022;Kim et al., 2022a).A number of these studies focus on individual tourists' trust in NPIs and their adoption of NPIs prior to and during travel and its impact on tourism behaviour (Chung et al., 2021;Das & Tiwari, 2021;Lee et al., 2012).Bae and Chang (2021) as well as Kim et al. (2022a) have used tourists' perceptions of and willingness to adopt NPIs before and during travel as an independent variable in predicting travel intentions.But only a few studies use behavioural intentions to apply NPIs while travelling as a dependent variable in order to increase travel safety.
Against the background of this stream of literature on eTPB and NPIs in the COVID-19 context, a more complete conceptual model is needed that includes all the explanatory dimensions of suitable theories.Thus, the aim of the present study is to address this research gap by presenting a theorybased integrated conceptual model for identifying factors that influence tourist's behavioural intentions to apply NPIs while travelling during the COVID-19 pandemic.The novelty of our work is that we apply a new conceptual framework based on well-grounded theories that has been validated for the Swiss population based on a random sampling technique.
In the remainder of this paper, we provide our theoretical framework, set out our hypotheses and link our major findings from the related literature in order to justify them.This is followed by a presentation of our data, the modelling approach we used to test our conceptual model, the results and a discussion section, including the limitations of this study, its practical implications and need for future research.

Theory and conceptual research model
The theory of planned behaviour (TPB) The TPB (Ajzen, 1991) was formed to explain behaviour that is volitional.The theory postulates that the trinity of (1) attitudes toward the behaviour, (2) subjective norms and (3) perceived behavioural control exert an influence on behavioural intentions, that in turn are the main motivational predictors of behaviour.Attitudes refer to salient beliefs about consequences that lead to favourable and unfavourable assessments of the behaviour and its alternatives.Subjective norms is linked to the social conformity of aligning oneself with significant others in performing the behaviour in question.Perceived behavioural control relates to the abilities and capacities to perform the behaviour.The increase of favourable attitudes and subjective norms and abilities to perform the behaviour result in higher behavioural intentions (Ajzen & Albarracín, 2007).Against the background of our research endeavour, three hypotheses are centre-stage: H1: There is a positive relationship among attitudes towards applying NPIs while travelling and behavioural intentions to apply NPIs while travelling.
H2: There is a positive relationship among subjective norms of applying NPIs while travelling and behavioural intentions to apply NPIs while travelling.
H3: There is positive relationship among perceived behavioural control of applying NPIs while travelling and behavioural intentions to apply NPIs while travelling.
The postulate of the hypothesis can be justified with the aid of related approaches in the field.Bae and Chang (2021) discovered a positive relationship that is significant concerning attitudes, subjective norms and perceived behavioural control over intentions for the NPI of minimizing contact with other people when travelling.Rahmafitria et al. (2021) revealed that favourable attitudes toward physical distancing, favourable subjective norms not travelling in a pandemic and increased perceived difficulties of travelling during a pandemic all had a negative relationship with one's travel intentions.
However, Bae and Chang (2021) added COVID-19-related cognitive and affective risk perceptions in the TPB.Rahmafitria et al. (2021) combined the TBP with the constructs of knowledge, social concerns and risk perceptions with respect to COVID-19.Because of the similarities of the COVID-19related risk perceptions with the constructs of HBM of perceived susceptibility and severity, perceptions of risk regarding the health domain will be best captured by combining the TPB with the HBM (see also Bae & Chang, 2021;Huang et al., 2020;Suess et al., 2022).

The health belief model (HBM)
The HBM is often utilized in health behaviour studies and intervention design (Champion & Skinner, 2008;Suess et al., 2022).According to the model, health-related behaviour depends on how the individual perceives susceptibility, severity, benefits and barriers (Rosenstock, 1974).Perceived susceptibility is understood as beliefs regarding the risks of contracting an illness or another negative health situation, while the perceived benefits is linked to beliefs about the efficacy of preventive measures or actions to avoid a negative health condition (Champion & Skinner, 2008).Additional constructs are perceived severity, that is understood as beliefs about the probability of contracting an illness and includes beliefs regarding both the medical and social consequences (Champion & Skinner, 2008;Rosenstock, 1960).The perceived barriers relate to beliefs about the adverse quality aspects associated with the preventive health behaviour that can act as impediment, such as cost, time and side-effects (Champion & Skinner, 2008;Rosenstock, 1966;Rosenstock et al., 1988).With regard to our study field, four more hypotheses are postulated: H4: There is a positive relationship among the perceived susceptibility to COVID-19 while travelling along with behavioural intentions to apply NPIs while travelling.
H5: There is a positive relationship among the perceived severity of COVID-19 and behavioural intentions to apply NPIs while travelling.
H6: There is a positive relationship among the perceived benefits of NPIs while travelling and behavioural intentions to apply NPIs while travelling.
H7: There is a negative relationship between the perceived barriers to NPIs while travelling and behavioural intentions to apply NPIs while travelling.
Again, these hypotheses can be justified by related previous studies.Perceived susceptibility was commonly found to be positively associated with preventive travel behaviours (Cahyanto et al., 2016;Huang et al., 2020;Kim et al., 2022b;Zheng et al., 2021), whereas higher perceived barriers to travelling during COVID-19 were associated with lower travel avoidance (Kim et al., 2022b).Another study by Suess et al. (2022) revealed that the perceived benefits of vaccines are positively correlated along with one's willingness to vaccinate against COVID-19 before travelling.Likewise, Zheng et al. (2021) work supports that perceived severity leads to higher motivations to protect against COVID-19.
However, Seddig et al. (2022) show that the influence of COVID-19-related fears on vaccination intentions is mediated through attitudes.Based on Seddig et al. (2022), there are good theoretical reasons for investigating the effect of COVID-19-related perceptions on behavioural beliefs (i.e.background factors).However, whether the background factors have an influence on behavioural beliefs or not has to be empirically analysed (Ajzen & Albarracín, 2007).
(Health) beliefs as a predictor for attitudes towards the behaviour serving as a mediator for travel intentions Based on TPB, behavioural attitudes are affected by salient behavioural beliefs regarding the behaviour in question.The relationship between beliefs and attitudes is incorporated in the Expectancy-Value Model of attitude formation.In fact, attitudes toward a behaviour are formed on the basis of evaluations of the outcome of a certain behaviour and the strength of the belief that the behaviour in question will influence the associated outcome (Ajzen & Albarracín, 2007).
For example, people may believe that implementing NPIs while travelling will be likely to decrease risks of contracting COVID-19, while at the same time finding the implementation cumbersome.The aggregation of salient (health) beliefs regarding the outcomes associated with the behaviour forms the basis both favourable and unfavourable attitudes toward the behaviour (Fishbein & Ajzen, 1975;Seddig et al., 2022).
Moreover, the perception of risk is a construct that involves beliefs about potential losses (Quintal et al., 2010).Tourism research has shown that tourists' perceptions of health risks influence protective travel behaviour (e.g, Chien et al., 2017;Laver et al., 2001).Hence, from a theoretical perspective, it is likely that the perception of risk has an influence that is negative on beliefs and thus on attitudes toward the risky behaviour (see Huang et al., 2020;Zhang et al., 2018).This leads to the following two hypotheses, with attitude as a dependent variable: H8: There is a positive relationship among perceived susceptibility to COVID-19 and attitudes regarding to apply NPIs while travelling.
H9: There is positive relationship among the perceived severity of COVID-19 and attitudes toward applying NPIs while travelling.
These theoretically derived hypotheses can be justified in addition by the empirical literature.Previous studies in tourism and leisure have observed that the perceived risk is negatively linked to attitudes toward the risky behaviour (Choi et al., 2013;Quintal et al., 2010;Zhang et al., 2018), positively associated with attitudes toward preventive health behaviour in the COVID-19 period (Rahmafitria et al., 2021), and negatively associated with attitudes toward travelling (Sánchez-Cañizares et al., 2021).This is consistent with this reasoning and with the Expectancy-Value Model.More specifically, health beliefs about the susceptibility and severity associated with COVID-19 can be seen as expectations about the potential losses and risks associated with a particular behaviour (Huang et al., 2020).
Two studies conducted in a tourism context found that perceived susceptibility have an increasing effect on attitudes toward preventive health behaviour (Bae & Chang, 2021;Huang et al., 2020).While Bae and Chang (2021) conceptualized susceptibility as cognitive risk perception and specified the preventive health behaviour as a form of physical distancing, Huang et al. (2020) did measure preventive behaviour using a single item by asking respondents how much they are doing to protect themselves from the risks.
The precondition for performing preventive behaviour depends on beliefs about its effectiveness in reducing the threat.Likewise, the perceived barriers, such as costs, side-effects and inconvenience of the preventive measures, need to be taken into account (see Rosenstock, 1966Rosenstock, , 1974;;Rosenstock et al., 1988).This discussion results into the following hypotheses: H10: There is positive relationship between the perceived benefits of NPIs while travelling and attitudes toward applying NPIs while travelling.
H11: There is a negative relationship between the perceived barriers to NPIs while travelling and attitudes toward applying NPIs while travelling.
Again, we have theoretical and empirical justifications for these hypotheses.Previous research demonstrates that the perceived effectiveness and barriers were major determinants of attitudes toward NPIs and vaccinations during COVID-19 (Song et al., 2022).Similar findings are found in other contexts as well.For instance, the perceived benefits and barriers of health tourism had a positive and negative effect respectively on attitudes toward medical tourism (Chaulagain et al., 2021).
Whereas the elements of the HBM focus on a specific behavioural domain (e.g.safe travel), a more general indicator is needed that measures the individual disposition for risk-taking by a person in the COVID-19 period.Investigating the interrelations among risk-taking attitudes and the risky behaviours of tourists has been scarce so far (Farnham et al., 2018).

Scale on domain-specific risk-taking (DOSPERT)
The DOSPERT scale is a psychometric scale.It can evaluate individual risk-taking attitudes in many domains, including, for example, finance, health and safety, and recreation (Blais & Weber, 2006;Weber et al., 2002).Based on our conceptual framework, two hypotheses are proposed: H12: There is a negative relationship among risk-taking attitudes in the domain of tourism and recreation, and behavioural intentions to apply NPIs while travelling.
H13: There is a negative relationship among risk-taking attitudes in the domain of tourism and recreation, and attitudes toward applying NPIs while travelling.
These hypotheses can also be justified by the following literature.Few studies have adopted the recreational subscale of DOSPERT to predict attitudes and behavioural intentions toward compliance with COVID-19 protective measures (see Keinan et al., 2021;Konc et al., 2022).For example, DOS-PERT's recreational subscale was positively associated with active coronavirus risk-taking behaviour (i.e.actively shaking hands) and negatively associated with passive risk-taking behaviour (i.e.not using hand disinfectant) (Keinan et al., 2021).

Conceptual research model
The elements of our conceptual research model were not only justified from a theoretical perspective, they were also based on related literature setting out empirical findings on the specific relations.Though similar studies have combined the TPB with the HBM (see Bae & Chang, 2021;Huang et al., 2020), this research is the first to include all the factors of these two theories.The model presented in Figure 1 will be empirically tested in the remaining parts of this paper.

Sampling procedure
A total of 4530 people from the Swiss resident population aged 18 and over were contacted for this research.The sampling frame came from the Swiss Federal Statistical Office (FSO) and included language region (German, French, Italian) and gender as strata.A probabilistic sampling procedure was applied by random selection of the addresses within those strata.The cross-sectional survey was conducted between March and May 2021.Respondents had the choice of completing and returning the questionnaire in writing (paper and pencil survey) or by completing the questionnaire online using a QR code to navigate to the survey in the browser of their own computers or mobile phones.Additionally, a pre-paid return envelope was included in the package they received.Reminders were sent to them after two weeks.Alongside this, a telephone hotline and an e-mail address for supporting the respondents were set up.Of the 4530 persons contacted by mail, 164 could not be reached (mainly due to their having moved).Of the persons reached, 390 answered the questionnaire online and 1293 did so in writing and returned it by mail.The total of 1683 valid responses leading to a response rate of 39%.The drawing up of the sample was based on the aim of representativeness regarding Swiss language regions and socio-demographic variables, including sex, age and education.Table 1 shows the proportions of the strata of our sample in comparison to the Swiss population census of 2021 (FSO, 2021a(FSO, , 2021b)).Our random selection matched the census data to a high degree.However, there is a slight bias towards those with higher education in our sample.

Measures
Prior to the main study in 2021, an item pool based former studies (e.g.Cahyanto et al., 2016;Lee et al., 2012;Montanaro & Bryan, 2014) was collected, and items were carefully selected and pre-tested with two representative quota samples of the Swiss population following the recommendations of Matsunaga (2010).Based on this pilot study, first, items were selected through Principal Component Analysis (PCA) (n = 333), then we tested the validity of the selected items using Principal Axis Analysis (PAA) and Cronbachs' Alpha on a fresh sample (n = 332).
Attitudes toward applying NPIs while travelling were measured using three semantic differentials with negative and positive evaluations on the left and right respectively, as proposed by Ajzen (1991).Subjective norms of applying NPIs while travelling and perceived behavioural control of applying NPIs while travelling were both measured with three items on a Likert scale with fivepoints (1 = does not apply at all to 5 = applies fully).
The perceived susceptibility, the perceived severity of COVID-19, the perceived benefits of NPIs while travelling and the perceived barriers of NPIs while travelling were each measured with three items on a five-point Likert scale (1 = fully disagree to 5 = fully agree).
Risk-taking attitudes in the domain of tourism and recreation were operationalized using three items taken from Blais and Weber (2006) by asking the respondents how possible it is that they would engage in a certain risky behaviour on a five-point Likert scale (1 = very unlikely to 5 = very likely).
Finally, behavioural intentions to apply NPIs while travelling were also measured with three items on a five-point Likert scale (1 = does not apply at all to 5 = applies fully).
All items can be found in Table 2.

Confirmatory factor analysis (CFA)
Our measurement model is tested based on the lavaan package in the R statistical software (see Rosseel, 2012).Standard maximum likelihood estimation is applied, which produces robust standard errors and robust test statistics with respect to violations of multivariate normality (Seddig et al., 2022).Given that there are missing patterns in the data (n = 87), they are handled by full information maximum likelihood (see Enders & Bandalos, 2001).Our measurement model reveals a good fit (χ 2 = 823.331,df = 314, p < 0.001, RMSEA = 0.035, TLI = 0.978, SRMR = 0.034, CFI = 0.982).
Both Cronbach's alpha and the composite reliability both exceed the generally accepted lower limit of 0.70 (see Hair et al., 2017).The Average Variance Extracted (AVE) as an indicator of convergent and discriminant validity is higher than the threshold of 0.50 for all latent constructs (Fornell & Larcker, 1981).The results are summarized in Table 2.
Furthermore, discriminant validity was assessed using a method proposed by Henseler et al. (2015).Table 3 shows the implied correlations of the model and the latent constructs.Table 4 depicts the Heterotrait-Monotrait Ratio (HTMT) of all possible correlations.All values are far below the conservative level of 0.85, indicating that discriminant validity prevails between latent variables (Henseler et al., 2015).
Regarding the recreational subscale of DOSPERT, the data show that higher risk-taking behaviour (i.e.risk-taking attitudes) in recreation have a significant negative effect on behavioural intentions to apply NPIs while travelling (β = −0.088,SE = 0.026, z = −3.363,p < 0.01) and a negative effect on attitudes toward applying NPIs while travelling (β = −0.094,SE = 0.028, z = −3.135,p < 0.01), thus supporting H12-H13.The results can be found in Table 5.
Mediation effects we also tested and found a positive indirect effect of the perceived susceptibility to COVID-19 while travelling (β = 0.096, SE = 0.013, z = 6.803, p < 0.001), the perceived severity of COVID-19 (β = 0.063, SE = 0.012, z = 4.833, p < 0.001) and the perceived benefits of NPIs while There is currently a high risk of infection from the coronavirus when travelling.

sus_4:
There is currently a high risk of passing on the coronavirus when travelling.

sus_5:
There is currently a high risk of coming into contact with the coronavirus when travelling.travelling (β = 0.081, SE = 0.016, z = 5.845, p < 0.001) on behavioural intentions to apply NPIs while travelling through attitudes toward applying NPIs while travelling.Furthermore, there is a negative indirect effect of the perceived barriers to NPIs while travelling (β = −0.081,SE = 0.011, z = −6.331,p < 0.001) and risk-taking-behaviour regarding the recreation domain (β = −0.030,SE = 0.010, z = −2.913,p < 0.01) on behavioural intentions to apply NPIs while travelling through attitudes toward applying NPIs while travelling.The indirect effects stemming from perceived severity, perceived barriers as well as risk-taking behaviour regarding behavioural intentions are partially mediated through attitudes, whereas the indirect effect of perceived susceptibility and benefits are fully mediated through attitudes.The total and indirect effects are summarized in Table 6.

Discussion
Our key findings highlight the fact that attitudes toward applying NPIs while travelling were by far the strongest predictor of behavioural intentions to adopt NPIs.This indicates that the behavioural beliefs that form the attitude play a major role when it comes to protective motivations and is coherent to research by Seddig et al. (2022), who showed that, regarding the TPB constructs, attitudes toward vaccination had the strongest impact on vaccination intentions.Subjective norms and perceived behavioural control over compliance with NPIs against the spread of COVID-19 also have a significant positive direct effect on tourists' behavioural intention, as predicted by the TPB and supported by previous studies of COVID-19-related risk perceptions and preventive behaviours in tourism (e.g.Bae & Chang, 2021).
Regarding the effects of the HBM constructs, we found a direct positive effect of the perceived severity of contracting COVID-19 on tourists' behavioural intentions to undertake non-pharmaceutical measures, which is comparable to prior findings indicating that persons who report higher perceptions of severity are engaging more strongly in COVID-19 preventive actions (Sánchez-Arenas et al., 2021).The negative direct effect of the perceived barriers to practicing NPIs against COVID-19 while travelling on tourists' behavioural intentions to adopt these protective measures also supports prior research (e.g.Chaulagain et al., 2021).
Interestingly, we found no direct effects of perceived susceptibility of the risk of contracting a COVID-19 infection and perceived benefits of applying NPIs on behavioural intentions, which at first seems to contradict prior research (Kim et al., 2022b;Suess et al., 2022).However, it is noteworthy that the effects of perceived susceptibility and the benefits were fully mediated through tourists' attitudes toward applying NPIs, which in turn is comparable to prior research, showing that attitudes toward the behaviour (partially) mediate the HBM constructs on behavioural intentions and behaviours (Huang et al., 2020;Park & Oh, 2022).The direct effects of the HBM constructs on attitudes and their indirect effect through the attitudes found in this study were also supported by prior research (Huang et al., 2020;Park & Oh, 2022).

Theoretical implications
From a theoretical TPB perspective, all the background factors are mediated through attitudes on behavioural intentions (Ajzen & Albarracín, 2007).Seddig et al. (2022), for example, confirmed this empirically about vaccination intentions.From a theoretical standpoint, it is noteworthy that the empirical literature is not uniform in this respect.Although attitudes were shown to mediate the impact of perceived susceptibility and benefits on behavioural intentions (e.g.Choi et al., 2013;Huang et al., 2020;Zhang et al., 2018), attitudes did not mediate the impact of perceived severity on behavioural intentions in the study by Huang et al. (2020), while the influence of perceptions about COVID-19 on behavioural intentions, which included perceived severity, was mediated through attitudes in the study by Anaki and Sergay (2021).Future research should, therefore, clarify the role of the HBM factors, whether they act as background factors by indirectly influencing behavioural intentions through behavioural beliefs, or whether they can be considered to be additional independent factors influencing behavioural intentions.This is especially true for perceived susceptibility and the perceived benefits, since in the present study these two effects were fully mediated through attitudes towards the behaviour.

Managerial implications
Formulating pointers to interventions to reduce the COVID-19 pandemic is dependent on tourists' willingness to accept protective measures (Ohnmacht et al., 2022).Likewise, NPIs effect tourists' travel intentions during a pandemic (Thao et al., 2022).For the formulation of practical implications, we follow Aiken's (2011) two-step approach that starts by formulating the social-psychological model and is followed by a multicomponent intervention that addresses the relevant determinants.It is argued that, in the formulation of effective protective measures, a general understanding of the interwoven factors of socio-psychological factors is needed.
Given that attitudes toward NPIs while travelling are the strongest predictor of behavioural intentions to adopt NPIs while travelling and fully mediate between both the perceived susceptibility of COVID-19 and the benefits of NPIs, decision-makers and destination managers should create information-based interventions targeting the perceived susceptibility to, for example, long-COVID and the perceived benefits of NPIs in order to build psychological resilience and to increase the perceived efficacy of NPIs in reducing the circulation of disease (Kim et al., 2022a;Zheng et al., 2021).Marketing and open communications can be important strategies driving tourism demand both during (Hystad & Keller, 2008) and after COVID-19 (Yeh, 2021).TPB-based (information) interventions have been shown to be especially effective when focusing on groups and conducted in public (Steinmetz et al., 2016) like tourist hotspots.Thus, a destination communicating compliance with NPIs as a confidence-building measure is likely to boost acceptance of the latter (Castañeda-García et al., 2022;Chung et al., 2021).
With regard to the negative impacts of the perceived barriers on behavioural intentions, providing free disinfectants and facial masks at popular tourist attractions, airports and railway stations might be beneficial in reducing the barriers by enhancing availability and accessibility.Tourists should be informed about the available test locations when developing symptoms of COVID-19.Tourism practitioners like hotels, cruises and travel agencies can bundle (health) insurance arrangements that cover the accommodation costs of self-isolation for positively tested tourists while travelling.Furthermore, according to Castañeda-García et al. (2022), tourist control (i.e.COVID certificates) is a measure that destination managers regard as potentially having a positive impact on tourist demand and is not perceived as a barrier by domestic tourists.
Finally, the increased use of social peers (e.g.celebrity marketing) can help boost the uptake of NPIs to reduce COVID-19 infections (Haverila et al., 2022;Liu et al., 2021a), as this induces belief in the perceived benefits of the NPIs (Haverila et al., 2022).

Limitations and future research
Our analysis is only representative of Switzerland which limits our study.We assume that the tendency of our results can be generalized to other populations.However, this must be tested in future research.Moreover, due to the dynamic situation of the pandemic and the study data for the first half of 2021, i.e. at the peak of the pandemic in Switzerland, tourists' perceptions may have changed in the meanwhile.However, while the strength of the effects may have changed, the positive or negative directions may not have done.
Future research can apply the general results of this analysis to investigate specific NPIs, such as the role of vaccination passports, surgical masks and quarantining, to investigate measures and intervention design using an experimental design approach.Individual differences should be analysed in this context.In this way it could be proved that specific measures increase the acceptance of support for safe tourism under hazardous conditions.

Conclusion
This research provides a holistic view of the interconnectedness between antecedents and mediating factors and the protective behaviour in the tourism domain validated by a representative sample from Switzerland.Our study adds to the current understanding of the factors influencing Swiss tourist's behavioural intentions to adopt NPIs against COVID-19 while travelling.Our conceptual framework has been empirically proved and acts as theory for explaining the socio-psychological process whereby individual travellers apply NPIs in the field of tourism.The most important background factors influencing behavioural beliefs are the perceived susceptibility of contracting COVID-19 while travelling and the perceived benefits regarding NPIs in reducing its spread.

Table 2 .
Summary of results for the measurement model.

Table 3 .
Model-implied correlations of the latent variables.

Table 5 .
Estimates predicting behavioural intentions and attitudes toward the behaviour.

Table 6 .
Total effects and indirect effects through attitudes toward the behaviour.