Psychometric Assessment of the Italian Version of the Vaccination Attitudes Examination (VAX) Scale and Exploration of Its Link With Policy Endorsement

Vaccination hesitancy, the patient-level reluctance to receive vaccines, is considered one of the top ten threats to global health by the World Health Organization in 2019 (World Health Organization, 2019). The resistance to joining vaccination campaigns led to resurgence and outbreaks of preventable diseases (Atwell & Salmon, 2014; Cherry, 2012; Jansen et al., 2003; Pearce et al., 2008; Phadke et al., 2016) as vaccination levels in the population remain too low.

For instance, in the UK, as of March 2017, only 91.8% of the target population of 2-year-old children received vaccinations for haemophilus influenzae type b and meningitis c according to Public Health England, while the World Health Organization (WHO) in 2019 aimed at an immunity target of at least 95% to eradicate influenza-like infectious diseases.

This resistance is even more concerning in light of the Coronavirus Disease (COVID-19) pandemic and the associated severe acute respiratory disease severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which lacks a definitive treatment for infected patients and has shown potential long-term effects (Orrù et al., 2021) including neurological complications (Orrù et al., 2020). The danger is due to the fact that joining vaccination campaigns, the main prevention tool available, is a totally voluntary act. Therefore, efforts in containing the pandemic are limited by individuals’ vaccination resistance.

Psychological research can target the vaccination hesitancy issue by relying on the existing literature on attitudes. Indeed, several theoretical models link behaviors to related attitudes (Ajzen, 2001; Ajzen & Fishbein, 2000; Nisson & Earl, 2020). These theoretical models are general in nature, they suggest a link between attitude and behavior. In theory, such a generic link can be extended by connecting the attitude toward vaccination to behaviors such as joining vaccination campaigns. Specifically, the attitude-behavior link has been tested in the domain of health behaviors suggesting that attitude toward vaccination is related to behavioral intentions (Catalano et al., 2017; Zhou et al., 2018). Crucially, the finding was replicated in scenarios involving pandemic diseases such as H1N1 influenza (Yang, 2015), and confirmed by meta-analyses on the topic (Glasman & Albarracín, 2006; Kraus, 1995).

These data and findings are driving the demand for scientifically validated tools to properly assess vaccination hesitancy and resistance. Indeed, several scales are available which, however, have two major drawbacks: first, most of these scales focus on specific diseases such as the case of HIV Vaccine Attitudes Scale (Lee et al., 2014) or the Carolina HPV Immunization Attitudes and Beliefs Scale (McRee et al., 2010); second, these scales such as the Attitudes and Behaviors Regarding Vaccination Decisions (Kennedy et al., 2011), the Parent Attitudes about Childhood Vaccines survey (Opel et al., 2011), or the Vaccine Hesitancy Scale (Larson et al., 2015), focus on parental decisions about their children’s vaccinations instead of focusing on individual vaccination hesitancy. An assessment tool that overcomes such drawbacks is the Vaccination Attitudes Examination (VAX) Scale (Martin & Petrie, 2017), a 12-item scale that measures individual resistance toward vaccination in a non-vaccine-specific way. This scale is composed of four factors related to mistrust toward vaccination benefits (trust/mistrust of vaccine benefit), concerns about side effects (worries over unforeseen future effects), concerns about pharmaceutical companies pursuing economic interest over health (concerns about commercial profiteering), and individual preference for natural immunity (preference for natural immunity). The scale was initially validated by the authors (Martin & Petrie, 2017) and the original findings have been replicated in contexts such as in the UK (Wood et al., 2019), Spain (Paredes et al., 2021), Turkey (Yildiz et al., 2021), and many more (Godasi et al., 2021; Huza, 2020; Yildiz et al., 2021).

To the best of our knowledge, the VAX scale has been translated and adapted into a few languages other than the original, Romanian (Huza, 2020), Turkish (Yildiz et al., 2021), Hebrew (Shacham et al., 2021), Spanish (Paredes et al., 2021), and Telugu (Godasi et al., 2021) and Italian (Bruno et al., 2022). The translation by Bruno and colleagues (2022) closely resembles our but the present paper and Bruno and colleagues’ (2022) work differ on several aspects. First of all, our investigation was conducted independently and there was no contact between research groups. Second, our investigation adheres to open science practices to a greater extent. Our data and materials are openly available on the relative OSF repository (see Data Accessibility Statement), and our translation of the scale is openly available in the appendix. Finally, our investigation addresses two research questions unaddressed by Bruno et al. (2022). Specifically, we investigate if the scale can be used to predict responders support to spread containment policies which might be beneficial in implementing strategy to foster the general population’s health. Additionally, we investigate if the functioning of the scale is the same across to key populations: people who received a vaccination within the last year, and a more vaccination-averse population of people who have not received any vaccination in the previous year. This final step is of great importance as reaching the vaccination-averse population can have great benefits in terms of supporting the general population’s health.

In its original form, the scale cannot be used to assess vaccination related attitudes of non-English speakers, making the scale useless for a vast part of the global population. Adapting the scale and validating its psychometric properties would be beneficial for assessing vaccination resistance globally and tailoring surveys and campaigns at the local level helping to eradicate vaccination hesitancy. Moreover, assessing this kind of hesitancy on a general level instead of at a vaccine-specific level would be very insightful as negative attitudes toward multiple vaccines generally co-occur (Prislin et al., 1998). Therefore, the validation of the translation might be a useful regardless the specific disease tackled by the campaign or the intervention. Crucially, the present investigation aims at going beyond a mere translation exercise and it explores the link between anti-vaccination attitudes and the endorsement of policies aimed at containing the spreading of diseases. Therefore, our work differs from existing research (Bruno et al., 2022) in one major way besides being conducted independently. We think that assessing whether the scale can predict down-streaming real-life consequences will contribute to fostering vaccination efforts.

In the present work, we aim to validate our translation of the VAX Scale (Martin & Petrie, 2017) in Italian (Appendix 1) and at explore its link with spread-containment policies. This research effort is warranted by the need to reach the widest population possible, including non-English speakers, and the need for an assessment tool that might be used in the next investigations of the anti-vaccination attitudes in this population, both related to the anti-COVID-19 vaccination and the more generic upcoming ones. Crucially, reducing vaccination hesitancy is instrumental in reaching WHO’s vaccination goals. Moreover, exploring the scale capabilities is an essential research effort bearing research questions that have not been investigated so far. Indeed, no study has yet tested whether scores on the VAX scales predict endorsement of public health policies.

To do so, we took advantage of an existing survey run in the context of the recent pandemic. The survey was developed to investigate the incidence of COVID-19 on the responders and their social surroundings (i.e., close peers, family, etc.), and the role played by media consumption (i.e., from newspapers, TV, social media, etc.). More information can be found in the published article (Biella et al., 2023). In the survey, we employed our Italian translation of the VAX scale and added some items to check the scale’s validity (see Materials and Procedure). The ad-hoc analysis of those items and the analysis of the data produced by the survey will provide us with a comprehensive assessment of the psychometric properties of our version of the VAX scale and test its criterion-related validity.

Additionally, we will explore the down-streaming consequences of anti-vaccination attitudes. We will investigate whether having a more anti-vaccination attitude is related to lowered endorsement of safety measures and policies aimed at reducing the spread of the virus. Such an additional step is potentially useful as it explicitly addresses the question at whether the VAX scale is related to real-life decisions that might have an impact on the general population’s health status.

To summarize, we identify the three following goals we want to achieve with our investigation:

1. Assess psychometric properties (i.e., factor structure, measurement invariance, etc.) of the Italian version of the VAX scale.

2. Assess the scale’s criterion-related validity by showing that the scores can predict past behaviors related to vaccination following the approach used in the original validation (Martin & Petrie, 2017).

3. Explore the link between anti-vaccination attitudes measured via the scale and the endorsement of real-life public health policies.

Based on our aims, we outlined the following analytical strategy to assess the psychometric properties of the VAX scale. The first step consists of running a confirmatory factor analysis to test whether the expected theoretical structure of the scale holds. At this stage, we expect that the structure provided by the authors in the original paper (Martin & Petrie, 2017) will fit the new data adequately. Moreover, the individual factors’ reliability will be estimated by computing ω coefficients (Flora, 2020) and Cronbach’s alphas. In a second step, we will assess the scale criterion-related validity assessing whether the measure can predict and discriminate among participants exhibiting different vaccination behaviors in the past (e.g., having ever rejected a physician-recommended vaccination). Then, we will run a multi-group confirmatory factor analysis comparing our current data with existing data collected employing the original (English) version of the scale (Huynh & Senger, 2021) to test whether the translated version exhibits measurement invariance (Hirschfeld & Brachel, 2014; Meade et al., 2008). If all those steps are successful, we can conclude that our adaptation of the scale is suitable for its deployment in applied and theoretical research and test whether anti-vaccination attitude is related to different levels of infection-containment policies endorsement.

Once confident in the reliability of the measure, we explore its capabilities by testing whether the measure is related to the endorsement of safety measures and spread-containment policies in the context of the recent pandemic. This step will help us achieve our third research goal (see previous paragraph). Specifically, we test whether anti-vaccination attitudes are linked with diminished endorsement for the use of face masks, decreased support of lockdowns, disagreement with closing non-essential economic activities, and diminished endorsement of the creation of a vaccination passport. These policies were the first to be suggested by Italian authorities, the context in which our investigation unfolds, and received substantial coverage in the mainstream media. Moreover, these policies were subsequently adopted by other countries to fight against the diffusion of the disease. If this step is successful, we can assume that the anti-vaccination attitude measured using the VAX scale can be linked to the endorsement of real-life policies, proving that the scale is a suitable tool for managing the general population’s attitudes related to health issues.

An analytical power analysis for a multi-group confirmatory factor analysis is not available. Simulation-based studies recommend a sample size of at least 400 participants per group (French & Finch, 2006; Koziol & Bovaird, 2018; Meade et al., 2008; Meade & Bauer, 2007). Such a recommendation is very general and it should be considered as a starting point (Luong & Flake, 2023). Therefore, we focused on the original validation paper to determine the minimum sample size required. In the validation paper (Martin & Petrie, 2017), the main study enrolled 409 participants. Given our reliance on multi-group confirmatory factor analysis, we aimed to collect such a sample size in one group and to obtain the second set of observations at a later stage (see Measurement Invariance paragraph). We recruited a sample of N=479 Italian speakers, 342 females, 135 males, and 2 undisclosed, aged between 19-83 years old (M=40.78, SD=14.23). We posted the link to the survey on several Facebook groups and LinkedIn pages devoted to users’ discussions about vaccination. The names of the groups were indicative of the pro- or anti-vaccination stance. The selection of groups was deliberate and attempted to collect participants covering the whole attitude spectrum. Only participants who provided their informed consent were redirected to the survey while no data were collected from participants who did not provide their consent. Although non-representative, our sample was diverse in terms of geographical and socio-demographic characteristics (Table 1 ).

All procedures followed the APA ethical standards and were approved by the Ethics Committee of the University of Pisa (No. 0036344/2020). Survey data were collected by disseminating a link over social media such as Facebook and LinkedIn. The link remained available for a period of 5 weeks between April 19^th, 2021, and May 24^th, 2021. On the first page of the survey, information about the research was displayed, and participants were made aware of the anonymity of the data collection and provided their informed consent.

The first questions were related to generic demographics such as gender, age, education, occupation, region of residency, whether the respondent has kids and, if yes, how many. Then, 3 dichotomous items related to previous vaccination behaviors were presented. Following the original article by Martin and Petrie (2017) and a replication study (Wood et al., 2019), we recorded whether the participant received a generic non-COVID19-related vaccination within the previous year (“Have you received the flu vaccine last year?”), its history of vaccination rejections (“Have you ever rejected a vaccination suggested by your physician?”), and, if the participant has children, we recorded the participant’s vaccination behavior related to its child/children (“My child/children received a vaccination within the last year”). Then, the Italian translation of the VAX scale (Martin & Petrie, 2017) was presented. Participants were asked to rate their agreement with all 12 items on a 6-degree Likert scale with options ranging from “totally disagree” to “totally agree” as in the original scale. The original scale was translated from English to Italian by a native Italian speaker. In a second step, another Italian speaker who was never exposed to the original version of the scale provided feedback on the meaning and wording of the items. At the end of the survey, 4 binary items (Appendix 2) recorded whether the participants endorsed COVID19-related safety measures and spread containment policies (i.e mask usage, lockdowns, the closing of non-essential economic activities, and the creation of a vaccination passport).

It is worth noting that all previous vaccination behavior and support for spread-containment policies were treated as individual items and were not combined a vaccination behavior or policy support unitary factors. This was a deliberate decision as we are interested in a fine-grained investigation of previous vaccination individual behaviors and policy-specific support.

All analyses were performed using R version 4.2.2 (R Core Team, 2021). The packages used were lavaan (Rosseel, 2012), MVN (Korkmaz et al., 2014), and equaltestMI (Jiang & Mai, 2021). All models were fitted using Diagonally Weighted Least Squares (DWLS) as the items were treated as ordinal variables. Significance tests were two-tailed unless specified otherwise and the significance level was α=.05 across all analysis.

The first step of the validation checked the translation’s factorial structure and reliability. Specifically, we ran a confirmatory factor analysis relying on the original 4 factors structure (Figure 1 ) and using diagonally weighted least squares (DWLS) as in the original article (Martin & Petrie, 2017). The results confirmed that this structure fitted the data properly (Hu & Bentler, 1999). Specifically, the comparative fit index and the Tucker-Lewis index were satisfactory, CFI=1.00, TLI=0.99, as the root mean square error approximation, RMSEA=.045, and the standardized root mean square residual was excellent, SRMR=.030. Factor loadings and descriptive statistics for each item are available in Table 2.

In a second step, we compared fitted model against another one creating a single factor for the whole scale. The single factor model reached an adequate fit too, CFI=.97, TLI=0.97. However, the single factor model (AIC=18028.12, BIC=18128.24) was inferior to the main model (AIC=16392.50, BIC=16517.65) according to both AIC and BIC. Moreover, a likelihood-ratio test confirmed that the 4-factors structure fitted the data better, χ²(6)=2284.7, p<.001, than the single factor alternative. Additionally, we assessed the scale and the single factors’ reliability by computing ω coefficients and Cronbach’s alphas. This analysis confirmed that both the scale and its factors are reliable, both indexes were excellent (Trust/mistrust of vaccine benefit ω=.944, CI_95%[.928, .957], α=.944, Worries over unforeseen future effects ω=.878, CI_95%[.854, .898], α=.874, Concerns about commercial profiteering ω=.923, CI_95%[.904, .939], α=.921, Preference for natural immunity ω=.943, CI_95%[.928, .956], α=.943, and full scale α=.941). The same was true if we computed both indices considering a single factor structure (ω=.941, CI95%[.931, .951], α=.941). Additionally, the four factors correlated significantly with each other, ps<.0001, as in the original paper (Table 3 ).

To assess whether the Italian translation maintained its criterion-related validity, we followed a procedure in line with both the original validation paper (Martin & Petrie, 2017) and the replication study (Wood et al., 2019). Specifically, we computed a single score for each participant (greater score indicates greater anti vaccination attitude) by averaging the scale items, and we tested whether the scale is linked with previous vaccination behaviors in a series of logistic regressions on these variables. Indeed, a greater score on the VAX scale (greater anti vaccination attitude) was related to a lower likelihood of having received a generic vaccine in the previous year, β=-0.619, b=-0.488, b_CI95%[-0.713, -0.281], z=-4.44, p<.0001, to a greater likelihood of having rejected a physician-recommended vaccination in the past, β=0.957, b=0.754, b_CI95%[0.540, 0.979], z=6.76, p<.0001, and a lower likelihood of having vaccinated their child/children, β=-0.562, b=-0.443, b_CI95%[-0.696, -0.207], z=-3.56, p=.0004 (Table 4 ). As the regressions investigated previous behavior and are correlational in nature, no causal claim can be supported. However, the pattern of results is coherent with theoretically driven expectations in which increased anti-vaccination attitude is linked to decreased likelihood of receiving a generic vaccine in the previous year, to increased likelihood of rejecting a physician-recommended vaccination, and to decreased likelihood of vaccination their child/children.

To corroborate this finding, we replicated the same analysis as in Martin & Petrie (2017) in which the binary variable is used as independent variable in a series of Welch-adjusted t-tests having the VAX scale score as dependent variable. Indeed, this analysis confirmed the desired property of the scale.

To investigate measurement invariance, we used multi-group confirmatory factor analysis. Specifically, we compared our data with data (N=351) from a previous investigation in which the original English version of the VAX scale was used (Huynh & Senger, 2021). To test measurement invariance, we followed a classical approach (Hirschfeld & Brachel, 2014; Putnick & Bornstein, 2016). We computed one model for the configural, metric, scalar, and strict level of measurement invariance constraining the crucial parameter to be equal across groups and compared the fit of each model with the fit of the previous one. To overcome the chi squared test’s limitation of being sensitive to sample size, we relied on the comparative fit index (McDonald, 1989) for model comparison and used cutoffs of ΔCFI=-.002 to declare misfit (Meade et al., 2008).

The results of this analysis revealed that the Italian translation and the original version exhibit all levels of measurement invariance except the strict invariance (Table 5 ). Specifically, the scalar invariance model fitted the data well, CFI=0.999, TLI=0.999, RMSEA=.052, SRMR=.029 (Figure 2 ), and the decrease in fit in comparison with the previous model was negligible, ΔCFI<-.0007. However, the decrease in fit in the model related to strict invariance exceeded our cutoffs, ΔCFI=-.0038.

To further strengthen the capability of the measure to identify sub-population particularly prone to reject vaccination, we replicated our investigation of measurement invariance by grouping our participants into the group of those who received a vaccination within the last year (N=103) and those who did not (N=376). For this analysis, we relied on the Italian sample alone as the grouping variable is missing in the English-speaking sample. We selected such a grouping variable as we consider it the best suited to identify a sub-population of interest and as it was the most balanced one. This analysis revealed the measure exhibit scalar invariance as the indices exceeded our cutoffs, ΔCFI=-.003. This pattern replicates our investigation of measurement invariance comparing the English- and Italian-speaking samples.

Given these results, we can conclude that the original version of the VAX scale and its Italian translation share the same factorial structure and exhibit measurement invariance up to the scalar invariance level. Moreover, the scale seems to be suited to function in the same way across a vaccination-averse population and a population more willing to accept vaccination.

To explore whether the anti-vaccination attitude is related to the endorsement of safety measures and spread-containment policies, we ran a separate logistic regression with anti-vaccination attitude measured via the VAX scale as the independent variable for each one of the final four items. Specifically, the endorsement of the policy/safety measure was coded as a 1 while being against it was coded as a 0. This way negative coefficients will represent a decreased probability of endorsement as the attitude becomes more anti vaccination. Results suggested that greater anti-vaccination attitude leads to lower endorsement of each policy/safety measure (Figure 3 ).

In particular, greater scores on the VAX scale were significantly related to decreased likelihood of endorsing the use of face masks, β=-2.03, b=-1.60, b_CI95%[-1.94, -1.30], z=-9.80, p<.0001, to a decreased likelihood of being in favor of lockdowns, β=-1.55, b=-1.22, b_CI95%[-1.46, -1.00], z=-10.46, p<.0001, to a lower agreement that closing non-essential economic activities is needed, β=-1.08, b=-0.85, b_CI95%[-1.06, -0.66], z=-8.34, p<.0001, and to a decreased likelihood of endorsing the creation of a vaccination passport (i.e., EU Digital Covid Certificate), β=-1.64, b=-1.29, b_CI95%[-1.54, -1.06], z=-10.70, p<.0001. All these relationships remained significant even when age, gender, and occupation were introduced as covariates (Appendix 3) showing the robustness of the initial finding.

Our data collection, together with an existing administration of the original version of the scale in its English form, provided evidence that the Italian translation of the VAX scale exhibits great psychometric properties. Combining the existing data collection and our own data collection (involving our adaptation of the scale) was a required step for our investigation. Together, they show that our version of the scale exhibits the same structure as the original version. The present research provided evidence that the Italian version of this scale is sufficiently reliable and shows measurement invariance when compared with the original version (Martin & Petrie, 2017). Specifically, our version of the measure was capable of discriminating between responders who engage in vaccination behaviors within the last year from those who did not, between responders who rejected a vaccination suggested by their physician and those who complied with its recommendation, and, among respondents with children, between those who had their children vaccinated and those who did not. This pattern of results mimics the one used by the original authors of the scale to call for satisfactory criterion-related validity. Moreover, this finding was not an artifact of the analytical strategy as an alternative one provided the same pattern of results and demonstrated that the Italian version of the VAX scale has substantial criterion-related validity. Additionally, the present research demonstrated that our adaptation of the scale reached a high level of measurement invariance when compared with the original version (Huynh & Senger, 2021). Indeed, the analysis suggested that the translation exhibits scalar invariance with the original translation. This finding is obtained by relying on a formal test consistent with the state of the art (Hirschfeld & Brachel, 2014; Meade et al., 2008), and suggests that our translation can be used in further research and applied projects. Similarly, our analysis revealed that our translation exhibits scalar invariance even across a vaccination-averse population and a population more accepting of vaccination. This result is particularly valuable as it indicates that the scale can effectively employed in populations that should be reached by vaccination efforts.

The present research goes beyond the validation of our version of the measure as our investigation explores the link between anti-vaccination attitudes and the endorsement of spread-containment policies. We provided new insights suggesting that the attitude toward vaccination is related to the endorsement of policies and safety measures aimed at containing the spreading of a communicable disease such as COVID-19. Our data shows that anti-vaccination attitude is linked to refusing to endorse safety measures (i.e. wearing face masks) and policies (i.e., lockdowns, non-essential economic activities, and the creation of a vaccination passport). This additional step increases our confidence in the fact that our version of the scale can produce useful insights into managing the general population’s attitudes toward vaccination. Linking anti-vaccination attitudes to the endorsement of real-life policies suggests that campaigns that successfully address anti-vaccination attitudes measured using our version of the scale can potentially affect the endorsement of policies crucial for the containment of infectious disease.

In short, our findings support our claim that the Italian version of the VAX scale is a reliable tool ready to be deployed. Having a tool capable of capturing anti-vaccination attitudes, which successfully predicts vaccination behavior and policy endorsement, can help in testing interventions aimed at promoting vaccination and reducing vaccination hesitancy in the general population. Moreover, such a tool might help identify sub-populations that are particularly prone to avoid vaccination and/or particularly in favor of such a practice. This means that vaccination campaigns’ efforts can be better directed to those responders who still have to be persuaded and away to those who are already convinced. Addressing vaccination hesitancy, tailoring campaigns aimed at increasing vaccination rates, and research investigating reasons behind vaccination behaviors can all benefit from this tool. Moreover, as the VAX scale is not vaccine-specific and focuses on individual perception instead of parental decision, this tool will be extremely valuable for a wide range of domains and complement existing measures addressing vaccine-specific resistances and parental decisions shading on the vaccination hesitancy issue. Crucially, the tool validated in this project is promising in probing the potential endorsement of various policies against the spreading of infectious diseases, allowing researchers to investigate the psychological dynamics underlying such endorsement, and providing policymakers with useful insights for designing, monitoring, and assessing the effectiveness of life-saving interventions.

As any research project, our investigation is not free from limitation. For example, we had to rely on self-report measures related to vaccination behaviors. Further research could lift such a limitation by relying on objective data such as medical record on vaccination status. Another limitation is related to the lack of a-priori hypotheses. Indeed, we explored the predictive power of the scale instead of confirming a-priori (ideally, preregistered) hypotheses. Our findings align with the literature on the link between attitudes and behavior, and are theoretically sound (Szollosi & Donkin, 2021), but remain exploratory in nature. Further research could replicate our findings while pre-registering the hypothesis before collecting the data. Similarly, further research should probe the measurement discriminant validity, something not provided by our investigation.

One final limitation that is worth addressing is related to the design of our investigation. As we relied on a cross-sectional design, further research is needed to investigate temporal dynamics of the attitude toward vaccination and its relationship with vaccination behavior. Such future research will require a longitudinal design in which responders are probed at regular time intervals.

In conclusion, it is also important to consider the psychological impact of diseases like COVID-19. The risk of damage (physical and/or mental), increases exponentially when the consequences of a disease could be significant with potential long-term effects, especially in chronic clinical populations (Conversano & Di Giuseppe, 2021; Orrù et al., 2020). The psychological aspects of the COVID-19 pandemic, which has had a significant psycho-social impact on the entire global population across all age groups, should not be overlooked. This includes patients and healthcare professionals dealing with challenging working and communication conditions (Dell’Osso et al., 2011; Iasevoli et al., 2012). Understanding the long-term effects of the disease would enable health and psychological researchers to intervene with precise and innovative techniques to different clinical populations (Mazza et al., 2020; Orru et al., 2023). This further emphasizes the importance of tools like the VAX scale in understanding and addressing the complex interplay of factors influencing vaccination behaviors and disease response. Indeed, as the scale we validated can serve as a proxy for the endorsement of containment policies which, if properly implemented, can reduce the diffusion of viruses, the scale can be used to assess the general population support for policies which will be instrumental in limiting diseases and their aversive outcome. If the spreading is contained, fewer people will experience the disease’s aversive reactions. Therefore, the scale can serve as a useful tool in the hands of policymakers to prevent or at least reduce harmful consequences of future pandemics.

Data, materials, and analyses are available via the Open Science Framework (https://osf.io/pzajk/) with one exception: data on the English speaking sample were provided by Professor Ho Phi Huynh, Texas A&M University San Antonio, therefore, we cannot share them. For those data please, contact Professor Huynh directly. The authors would like to thank Professor Huynh for sharing the data needed to test measurement invariance, and the editor and two anonymous reviewers for the constructive feedback.

The first author received funding from the Publication Fund of the University of Basel for Open Access which covered cost associated to the publication.

The authors declare no conflict of interest.

Marco Biella and Graziella Orrù conceptualized the research and developed the survey. Marco Biella conducted the data analysis under the supervision of Graziella Orrù, and drafted the initial version of the manuscript. Graziella Orrù, Angelo Gemignani, Ciro Conversano, and Mario Miniati revised the initial draft before the submission.

Items of the Italian Version of the Vaccination Attitudes Examination (VAX) Scale

1. Mi sento sicuro/a dopo aver ricevuto un vaccino.*

2. Posso fidarmi dei vaccini per fermare la diffusione di pericolose malattie infettive.*

3. Mi sento protetto/a dopo aver ricevuto il vaccino.*

4. Sebbene la maggior parte dei vaccini sembrano sicuri, potrebbero causare problemi che non abbiamo ancora scoperto.

5. I vaccini possono causare problemi imprevisti nei bambini.

6. Mi preoccupo che i vaccini possano avere degli effetti sconosciuti in futuro.

7. I vaccini fruttano molti soldi alle case farmaceutiche ma non fanno molto per le persone.

8. Le autorità promuovono le vaccinazioni per i loro interessi economici e non per la salute delle persone.

9. Le campagne vaccinali sono una grande truffa.

10. L’immunità naturale dura più a lungo di quella dei vaccini.

11. L’esposizione naturale a virus e germi garantisce la protezione maggiore.

12. Essere esposti alle malattie in modo naturale è più sicuro per il sistema immunitario rispetto all’essere esposti mediante vaccino.

*Items 1, 2, and 3 are reverse coded.

Policy/safety measures endorsement items.

1. What do you think about the lockdown?

   • I am in favor. People should stay at home as much as possible to limit the spreading of the virus

   • I am against. People should have the chance to go out freely despite the spreading of the virus

2. What do you think about the usage of face masks?

   • I am in favor. Everybody should use them

   • I am against. The use of face masks should not be imposed to anyone

3. What do you think about the creation of a vaccination passport?

   • I am in favor. People should be allowed to travel only if they are not at risk of favoring the spread of the virus

   • I am against. People should be allowed to travel regardless the spreading of the virus

4. What do you think about the closings of non-essential economic activities?

   • I am in favor. All non-essential economic activities should remain closed until the situation does get better

   • I am against. All non-essential economic activities should remain open regardless the spreading of the virus