key: cord-0756199-6i4ifg9p
authors: Goodwin, R.; Ben-Ezra, M.; Takahashi, M.; Nguyen Luu, L. A.; Borsfay, K.; Kovacs, M.; Hou, W. K.; Hamama-Raz, Y.; Levin, Y.
title: Psychological factors underpinning vaccine willingness in Israel, Japan and Hungary
date: 2021-05-25
journal: nan
DOI: 10.1101/2021.05.24.21257465
sha: 0cd0d9c6918593d1c989cabbbdb1f72f1c2ca967
doc_id: 756199
cord_uid: 6i4ifg9p

The rapid international spread of the SARS-CoV-2 virus 19 led to unprecedented attempts to develop and administer an effective vaccine. However, there is evidence of considerable vaccine hesitancy in some countries and sub-populations. We investigated willingness to vaccinate in three nations with historically different levels of vaccine willingness and attitudes to the COVID-19 vaccine rollout: Israel, Japan and Hungary. Employing an ecological-systems approach we analysed associations between demographic factors and health status, individual cognitions, normative pressures, trust in government, belief in COVID-19 myths and willingness to be vaccinated, using data from three nationally representative samples (Israel, N=1011 (Jan 2021); Japan, N= 997 (Feb 2021); Hungary, N=1131 (Apr 2021)). In Israel 74% indicated a willingness to vaccinate, but only 51% in Japan and 31% in Hungary. Results from multigroup regression analyses indicated greater vaccine willingness amongst those who perceived benefits to vaccination, anticipated regret if not vaccinated and trusted the government. Multi-group latent class analysis of ten COVID-19 (mis)beliefs identified three classes of myths, with concerns about the alteration of DNA (Israel), allergies (Hungary) and catching COVID-19 from the vaccine (Japan) specific to vaccine willingness for each culture. Rather than focusing primarily on disease threats, intervention campaigns should focus on increasing trust and addressing culturally specific myths while emphasising the individual and social group benefits of vaccination.

for behaviours that afford personal control (e.g. wearing gloves) but not for acts that threaten autonomy (which may include vaccination) 31 , while others suggest that unlike regular conspiracy accounts paranoia-like beliefs are associated with adherence to safety guidelines 42 . In our analyses we focus on common vaccinerelated fears propagated across societies (e.g., about the association between vaccination and autism 43 ), drawing on ten common myths identified by health advisory bodies (CDC, NHS, WHO) at the start of our first data collection. We place particular emphasis on vaccine side-effects, the most prominent concern in January 2021 44 . We test the structuring of these myths in each country via multi-group latent class analysis (LCA), and associate each with willingness to vaccinate. This paper seeks to address four objectives. First, we compare rates of vaccine willingness in national surveys conducted in Israel, Japan and Hungary (objective 1). Second, we seek to understand the demographic and health risk factors associated with this willingness by considering associations between willingness and age, sex, education, employment and health status (objective 2). Third, we examine relative weight of each of the variables in our three-layer ecological model by conducting a multigroup path analysis in each culture (objective 3). Finally, (objective 4), we conduct a sensitivity analysis examining specific cluster structure of these ten different myths, and the impact of these on willingness to vaccinate in each country. Table 1 presents differences in the distribution of background and study variables between the three samples.

Willingness to vaccinate was higher in Israel (74.1%) than in Japan (51.1%), or Hungary (31%) (χ² (2) =397.86 , P=.001) (Figure 2) Objective 2: Willingness to vaccinate by demographics.

In Israel, willingness to vaccinate was greater amongst those who were: (1) older (M ages 41.20 (agree) vs. 38 .38 (disagree); t= -4.95 P<.001), (2) male (78% men vs. 71% women, χ 2 (1) = 6.19, P=.01), (3) more educated (χ2 (2) = 12.90 P=.002) (4) employed (χ2 (2) = 6.03 P=.05) (employed were more likely to vaccinate . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ;  https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint (76%), compared to those were not employed (73%) or lost their job during the pandemic (66%)). There was no difference in willingness to vaccinate between those who belonged to a risk group (73.5% with a condition vs. 76.2% of those without (χ2 (1) = 0.70 P=.41) or between secular vs. non-secular respondents in vaccine willingness (χ2 (1) = 2.35 P=.13).

In Japan, willingness to vaccinate was greater amongst those who were: (1) older (M ages 47.23 (agree) vs. 44.23 (disagree); t= -3.23 .001), (2) male (57% men vs. 45% women, χ 2 (1) = 13.20, p<.001), (3) more educated (57% University educated vs. 45% not University educated, In Israel ( Table 2 ) results showed that, with the exception of sex (where women were less willing to vaccinate, adjusted odds ratio (aOR) = .58, P=.018) there was no significant association between demographic variables and vaccine intention (aORs = .97-.99). Higher subjective rated health was associated with reluctance to vaccinate (aOR = .66, P=.048). Participants who had been formally diagnosed with COVID-19 as well as family member who had been infected were not significantly associated with willingness to vaccinate. There were positive associations between willingness to vaccinate and the cognitive factors of . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint benefits of vaccine (aOR = 2.11, P<.001), anticipated regret (aOR = 1.72, P < .001), while barriers to vaccination were associated with reluctance to vaccinate (aOR = .72 P < .001). Perceived severity and perceived likelihood were not associated with willingness to vaccinate, but subjective norms (aOR = 1.38, P = .008) was associated with more willingness to vaccinate. Trust in government was positively associated with willingness to vaccinate (aOR = 1.38, P = .002), false beliefs about COVID-19 significantly associated with unwillingness to vaccinate (aOR = .11, P < .001).

In Japan, participants who had a family member infected and ill with COVID-19 were more willing to vaccinate (aOR = 1.30 p = .042), but neither participant's own diagnosis with COVID-19 nor their demographic variables or health status was associated with vaccine intention. There were positive associations between willingness to vaccinate and the cognitive factors of the benefits of vaccination (aOR = 1.65, P = .004) and anticipated regret if not vaccinated (aOR = 1.74, P < .001) while barriers to vaccination were associated with reluctance to vaccinate (aOR = .20 P < .001). Subjective norms were associated with willingness to vaccinate (aOR = 1.45 p = .037). as was trust in government (aOR =1.98, P < .001), but not false beliefs (aOR=.66 P=.49).

Finally, in Hungary, findings indicated that, with the exception of sex (where men were more willing to vaccinate (aOR) = 1.99, P < .001)) there was no significant association between demographic or health status variables and vaccine intention (aORs = .99-1.00). Neither the participant's nor their family's formal diagnosis with COVID were related to willingness to vaccinate. There were positive associations between willingness to vaccinate and the cognitive factors of benefits of vaccine (aOR = 1.92, P < .001) and anticipated regret (aOR = 1.37, P < .001) but neither barriers to vaccination, perceived severity or perceived likelihood of COVID-19, nor subjective norms were associated with willingness to vaccinate. Trust in government was positively associated with willingness to vaccinate (aOR =2.20, P <.001) but false beliefs were not associated with such willingness.

A multi-group Latent Cluster Analysis tested one to four class solutions for the three samples, examining whether the solution demonstrated the same class pattern was obtained across samples 45 . As shown in Table 3 , decrease in BIC was greatest for a three-class solution, providing very strong evidence of best fit . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 

Respondents assigned to Class 1 (High False Beliefs, n = 125, 12.3%) exhibited high probability for all items (60% -81.30%) with the exception of the beliefs that the flu vaccine protects against COVID-19 (9.90%) and the vaccine causes autism (moderate probability, 54.60%). Those assigned to Class 2, a combination of the vaccine may change DNA + overall Low False Beliefs (n = 336, 33.2%) showed low probabilities for all false beliefs (8.90-36.30%) except for a medium-high probability that the vaccine alters DNA (62.30%) and that people can(not) benefit from the vaccine even if they had COVID (51.90%). The third class, Low False Beliefs (n = 550 54.4%) showed low probabilities for each belief (ranging from 0-25%).

Separate Logistic Regression showed Classes 1 and 2 were associated with greater unwillingness to vaccinate (vs. the low false belief reference group, class 3). (χ²(1) = 215.97 p < .001). High False Beliefs was most closely associated with unwillingness to vaccinate (b = -3.36 SE = .26 Wald = 170.86 p < .001 odds ratio = .035). Vaccine may change DNA + Low False Beliefs was also associated with unwillingness to vaccinate compared to the reference group (b = -1.14 SE = .18 Wald = 40.50, p < .001 odds ratio = .320).

Those assigned to Class 1 High False Beliefs (n = 136, 12.0%) exhibited high probability of all items (59.80% -100%) except for the belief that the flu vaccine protects against COVID-19 (low probability 10.80%) and that reactions to the vaccine are mild (moderate probability, 52.0%). Participants assigned to Class 2, High belief in Allergies + overall moderate-Low False Beliefs (n = 422, 37.3%) exhibited low probabilities for all false beliefs (6-31.50%) except for the belief that the vaccine causes allergies (77.40%) and the moderately-held beliefs the vaccine weakens the immune system (49.30%), causes autism (51.70%) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Across the world there is evidence of continued vaccine unease, vaccine resistance identified as a top ten threat to global health even before COVID-19 47 . Countries however have fared differently in the availability of vaccines and the trust of their populations towards a growing range of possible vaccines, with a nested set of factors influencing uptake.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 25, 2021. ;

In Israel, a well-established community-based health service, strong logistic capacity, large-scale public health campaigns and the early purchase of a large number of vaccines helped the country achieve a rapid and comprehensive roll-out of the Pfizer-BioNtec vaccine 48, 49 . Unsurprisingly therefore almost threequarters of our sample in this country demonstrated willingness to be vaccinated. In contrast, and despite greater personal exposure to the pandemic and perceived severity of threat, our Hungarian and Japanese respondents were less willing to vaccinate. In our sample only just over half (51%) of Japanese respondents indicated such willingness, higher than the 45% reported by the Imperial College COVID-19 tracker in January 2021 3 but lower than the 62% indicated by a cross-sectional survey also conducted that month 17 . In Hungary, the politicisation of the vaccine roll-out 7 , and disputes over the use of vaccines not approved by the EU 12 , contributed to high national rates of scepticism about efficacy 50 with only 31% of our national sample expressing willingness to vaccinate, and a further 21% uncertain.

Across our three country samples willingness to vaccinate was greater amongst older respondents, the age group most at risk of mortality/ morbidity from the SARS-CoV-2 virus, consonant with previous research 32, 36 . In further bivariate analyses in Japan and Israel the more educated and the employed were also more likely to vaccinate 36 . Rejection of covid-19 myths was associated with increased years schooling in Hungary (r = -.24 P=.001) or being a graduate in Israel (t (722) = -4.79 P=.001), suggesting that the more educated challenge some of the myths that contribute to vaccine hesitancy or rejection. This is consistent with crosscultural evidence suggesting the more numerate are less suspect to COVID-19 misinformation 32 . In Japan and Hungary those with an underlying health risk were more willing to vaccinate, in line with previous surveys on influenza vaccination 50 . In competitive regression which included all three levels of factors in our ecological model, however, demographic factors, individual disease vulnerability, and personal or family experience with COVID-19 had only a small association with vaccine willingness. Instead, in each culture, those who appreciated the benefits of the vaccine, those more likely to regret vaccinating, and respondents who trusted their governments and health authorities were more likely to indicate vaccine willingness. There were, however, only small and culturally variable associations between perceived likelihood or severity of infection and vaccine intention, indicating only a weak association between perceived threat and vaccination. This may be because while infection likelihood and severity are closely associated with viral threat, benefits and regrets may be more proximally associated with actual vaccine behaviour. Subjective pressure to vaccinate was . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 1 1 significantly higher in Israel and Japan compared to Hungary and associated with vaccine uptake in just these first two countries. This suggests that the importance of friends, families and others for vaccine willingness may be particularly significant where important others are prepared to be vaccinated.

Engagement and communication were further, additional predictors of vaccine willingness. False information about the virus, most likely to emerge from social media, has been shown to be negatively associated with COVID-19 health protective behaviour 38 , including vaccine willingness 20 where vaccine uptake, and choice of vaccine, has been particularly politically contentious. Polling data collected co-temporaneous to our Hungarian study showed political divisions between those willing to vaccinate with a Hungarian government approved vaccine (including Russian and Chinese vaccines at that time not approved by the European Medicines Agency(EMA)) and those on the political left more likely to endorse the EMA recommended Pfizer or Moderna vaccines 51 .

In our samples, misbeliefs about the vaccine correlated with each other, supporting the idea of a 'monological belief system' 32, 52 . Nevertheless, our latent class analysis allowed us to classify the structure of beliefs in each culture and their association with vaccine willingness, permitting a more nuanced understanding of how different groups may understand potential vaccine risks. In Israel and Hungary beliefs were similarly structured in classes, with one group of respondents (approximately 12%) demonstrating high rates of false beliefs and another (approximately half the sample) scoring low on false beliefs. In both countries high levels of false belief were associated with unwillingness to vaccinate, in accord with our regression analyses. However, in Israel the belief that COVID-19 can alter your DNA (held by 62% of those in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint the second category) was distinctive as a predictor of vaccine (un)willingness. Contrastingly, in Hungary it was the association between allergies and the vaccine (held by 77% of those in class 2) that distinguished a grouping of respondents reluctant to take the vaccine. In Japan, classes identified were generally more blended and less distinguishable, and subsequently less clearly relate to vaccine willingness. Despite this, all members of the largest latent class in Japan (43% of respondents) indicated their belief that the vaccine can give you . Notably, 61.5% of Japanese respondents who believed the vaccine can give you COVID in Japan were unwilling to vaccinate, compared to 48.6% Israel and 14.0% in Hungary, suggesting the particular significance of this misbelief in Japan for vaccine intentions.

Limitations. Our studies benefitted by including national samples from three very different cultures, at different stages in their vaccination programmes and with different histories of vaccine uptake. We go beyond most literature on vaccine uptake and hesitancy by examining vaccine willingness in an expanded model incorporating different 'levels' of influences: demographic, cognitive and societal, and by examining in more detail the structure of misbeliefs about the vaccine, by culture. However, we recognise a number of limitations to our survey. First, samples were cross-sectional, and were therefore not able to assess predictors of vaccine willingness over time. Data was first collected in early January, at the start of the first major vaccine roll-out, meaning that we did not include later misbeliefs that emerged in subsequent months and which often focused on the association between vaccination and government/'big Tec' control and monitoring. This may be particularly importance with the arrival of new variants of concern that have challenged potential vaccine efficacy. In addition, emergent concerns over vaccine safety (such as worries about blood clotting following the AstraZenica vaccine 53 ) may serve to directly inhibit uptake and perpetuate further new misbeliefs and distrust. Second, because of the speed of the evolving vaccination situation in both countries (the rapid programme vaccination in Israel, the introduction of vaccination in Japan) our survey companies expedited data collection within a short time period. Although widely used, and important particularly for the collection of time-sensitive data during a vaccination campaign, we recognise that the quota sampling we employed has important limitations in ascertaining accurate response rates 20, 54 . We distributed a large number of invitations at the potential expense of response rate and could not readily estimate non-response rate from those who saw the survey invitation. Our samples, while representative of the demographic characteristics of their wider . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint populations, were not genuine probability samples, and particular caution is needed when making crosscultural comparisons. In Hungary in particular the percentage of willingness to vaccine may have been underestimated by the omission of those already vaccinated. Third, we did not obtain information on occupation, despite significant variations by profession in exposure to COVID-19, which may have further impacted on vaccine willingness 55 , and we did not assess income, although those with lower income are less likely to indicate vaccine willingness 17 . Research in the UK suggest there is also likely to be variation by ethnicity in the misbeliefs that might mitigate against vaccination 44 . Fourth, we assessed only three countries, and future work should expand the testing of nested models across settings. Such expanded analyses can better assess the cultural values that play a further part in health protective behaviours during COVID-19 56 , with individualism helping encourage conspiracy beliefs 57 . Finally, we measure only behavioural intentions rather than actual vaccination behaviour. Although the link between the two has been well established 58 we recognise that attitudes towards any vaccination are likely to vary as populations acquire further experience with the virus and the vaccination rollout. Future longitudinal research could profitably examine the impact of cognitive and broader cultural factors on consistency of vaccine commitment and resilience to new countermessages, as well as behaviours post-vaccination, particularly in those critical weeks before full immunity is attained.

Implications for vaccine drives. Despite these limitations we believe our findings have important implications as other nations strive to accelerate their vaccination drives. Encouraging vaccine uptake requires a planned approach involving consistent, credible communications that emphasise vaccine benefits, confronts potential barriers, and warns of potential later regret if the vaccine is not accepted. Vaccine campaigns may need not focus on disease threat: instead, these initiatives would better focus on the effectiveness of the vaccine, confront misinformation, and seek to emphasise the trustworthiness of key actors, such as national health services. Those vaccinated should be encouraged to inform close others, in order to emphasise the normative nature of this activity. Public health agencies need to reach people beyond remote media campaigns and be present where individuals shop and work 59 ; doctors have been widely reported to be important in addressing myths 39 with pharmacists in Hungary significant for encouraging influenza vaccination in that . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; 1 4 country 50 . Social media is of course likely to have an important role, with vaccine resistance highest amongst who obtain their information from such media 20 , although source of the media is important (e.g. obtaining information from the WHO was associated with lower susceptibility to COVID-19 myths 32 ; those who use Instagram or What'sApp for information on the pandemic are more likely to believe that the vaccine negatively impacts the immune system than those using traditional media 44 ). Our analyses suggested that, despite some similarities in belief structure, there were distinctive beliefs in each culture important for understanding vaccine willingness. In Israel concerns need to be addressed about the possibility of DNA being altered by the (mRNA) vaccine employed; for Hungary vaccine campaigns need to be particularly wary of concern about allergies while in Japan such interventions need to be alerted to the belief that the vaccine can give you COVID-19. Sensitivity to these particular beliefs in one culture may be particularly important where views on vaccination reflect political divisions: our Israeli data was collected shortly before a highly contested election where vaccination roll-out was a major campaign topic, in Hungary the use of vaccines yet to be approved by European Medicines Agency caused controversy 60 , while the Japanese data was gathered only a few months before a controversial Olympic Games 61 , with vaccination a national priority in the run-up to this event. Finally, within country, group factors may be also particularly important. Younger respondents need to be encouraged in particular to vaccinate; this may require addressing in particular COVID-19 myths most prevalent in the social media consumed by younger audiences 40 . Two weeks after our data collection (1st Feb 2021), actual vaccination uptake in Israel amongst those aged over 60 varied significantly, with 66.1% uptake in the ultra-Orthodox community and 60% in the Arab population, compared to 84.9% in the wider populace 20, 61 . Uptake was also greater in settlements with higher socio-economic status, despite the greater morbidity from COVID-19 amongst poorer communities 62 . To address these variations in uptake the specific concerns of religious and other social groups need to be considered, with community leaders actively engaged through culturally appropriate conversations in order to allay fears, address specific myths and thus help further facilitate a successful vaccine campaign 63 .

Participants and procedure. Data were from nationally representative samples of adult populations collected in Israel (N = 1011), Japan (N = 997) and Hungary (N=1131), using large panel survey companies in each country (iPanel for Israel, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Because we were keen to obtain samples either early in vaccination campaigns (Israel, Hungary) or prior to its commencement (Japan) time sensitivity meant requesting the survey companies to obtain samples of 1000 respondents in each country. This resulted in sample sizes comparable to those gathered in several major international studies of vaccine willingness 3, 64 . For each sample participants were contacted as part of cloud panels administered by the survey company and asked to participate via email. They were then reimbursed by the companies for their participation. In each country inclusion criteria required participants be the approved minimal age set by ethical requirements (Israel, Hungary -18, Japan -20), and to successfully pass validation checks (both specific items and timing of responses) to ensure participant attention and accuracy. All respondents were fluent in the relevant national language (Hebrew, Japanese, Hungarian). 

Demographics Participants in each country indicated their age and sex. Because demographic information procedures vary across countries we obtained slight variations in each country, in common with other such cross-cultural comparisons 20 , while retaining the core ecosystems model variables in each country for comparative analysis. In Japan and Israel respondents identified whether they completed only high school prior to University or were currently a student/had graduated; in Hungary students indicated their level of schooling (up to secondary school (N=44 (3.9%); secondary school (N=647 (57.3%), college degree (N=276, 24.4%), masters' degree (N=147, 13.0%) or doctorate (N=16, 1.4%). In Israel respondents indicated if were employed (71.7%), unemployed (15.8%) or lost employment due to the pandemic (12.5%); in Japan only whether or not they were employed. Table 1 provides further descriptive statistics for each country.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint Health conditions. Respondents indicated their risk group membership using the US CDC risk group memberships (e.g. hypertension, diabetes). Participants also indicated whether they had been formally diagnosed with COVID-19 (yes, no), whether someone from their social circle had been thus diagnosed (yes, no), and their self-rated health (from bad (1) to excellent (4)).

Vaccine acceptance. Participants were asked Would you be willing to accept a vaccine approved safe and effective by the government? (strongly disagree to strongly agree)). Because we questioned respondents at the start of actual vaccine campaigns, rather than about a hypothetical vaccine, and were cautious about both the translation of Table 1 .

The first six sets of questions were drawn from the HBM and TPM and employed questions previously used to assess swine flu vaccination 18, 19 .

Here we asked three questions on likelihood of infection e.g. Getting COVID-19 is currently a possibility for me, which previously showed acceptable scale reliability for assessing vaccination intentions 19 Perceived severity of infection. We use a further 3 items drawing on the HBM and TPM e.g. I will be very sick if I get . These also showed acceptable scale reliability for assessing vaccination intentions 19 Perceived benefits of vaccination. These three items included Vaccination is a good idea because I feel will be less worried about catching COVID- 19 19 Barriers to vaccination indicate two daily impediments to vaccination e.g. The side-effects of COVID-19 will interfere with my usual activities 19 Anticipated regret is a single item drawn from an extension of TPB 19 and was previously a significant predictor of intention to be vaccinated for 2009 H1N1(If I did not have a COVID vaccination, I would later wish I had)

Subjective norms used items taken from the TPB with previous good reliability 19 . These indicate influence of others on willingness to be vaccinated, and was assessed by five items (e.g. My friends would approve of me having the COVID-19 vaccination)

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint Trust in government extended a similar item used in a previous global study of vaccine acceptance during COVID -19 36 and included three items to assessing generalised trust, trust related to COVID-19 and trust with respect to vaccination (e.g. I trust Government regarding vaccination) False beliefs were a set of ten true or false items taken from the myth-busting websites of the WHO, CDC and the UK NHS. Four of these were reverse coded. These included The COVID-19 vaccine can give you covid-19 (from the NHS website), and receiving an mRNA vaccine will alter your DNA (CDC)

We utilized layered multigroup logistic regression analysis using MPlus 8.1 66 to test the ecological model. Data were analysed using maximum likelihood estimation with robust standard errors (MLR) to handle non-normal distributions.

We used multigroup analyses 67 to test if paths from the covariates, demographic factors and health status, individual cognitions, normative pressures, trust in government, belief in COVID-19 myths and willingness to be vaccinated varied by culture. We included participant information on their own (or social network's) positive COVID-19 diagnoses.

Missing data due to nonresponse was minimal, ranging from .9% to 4%.

To analyse the structure of misbeliefs about the vaccine we employed Multi-Group Latent Class Analysis (MLCA) to identify subgroups within the three samples (Israel, Japan and Hungary). We use binary indicators of false beliefs. We specified models with one to four classes and compared the models to determine the optimal number of classes. Models with lower values on the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and sample-size adjusted Bayesian Information Criterion (aBIC) were prioritized. Entropy values and average latent class probabilities indicated classification accuracy, with preference for models with entropy values and probabilities of correct class assignment closer to 1.00 68 . Once optimal number of classes was determined we computed sample percentages assigned to each class and conditional probabilities by class, with labels for latent classes based on patterns of probabilities for endorsing each false belief.

The datasets generated and/or analysed during the current study are available in the Open Science Framework (OSF) repository, and can be accessed here: https://osf.io/dm587/.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

The authors report no competing interests.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint  F  i  g  1  :  E  c  o  l  o  g  i  c  a  l  m  o  d  e  l  o  f  f  a  c  t  o  r  s  c  o  n  t  r  i  b  u  t  i  n  g  t  o  v  a  c  c  i  n  e  w  i  l  l  i  n  g  n  e  s  s . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. ; https://doi.org/10.1101/2021.05.24.21257465 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 25, 2021. 1 3 1 S  u  p  p  l  e  m  e  n  t  a  r  y  m  a  t  e  r  i  a  l  :  I  t  e  m  s  i  n  c  l  u  d  e  d  i  n  t  h  e  q  u  e  s  t  i  o  n  n  a  i  r  e  .   S  c  a  l  e  ,  R  e  s  p  o  n  s  e  o  p  t  i  o  n 3  3   I  f  e  e  l  u  n  d  e  r  p  r  e  s  s  u  r  e  t  o  h  a  v  e  a  C  O  V  I  D  -1  9  v  a  c  c  i  n  a  t  i  o  n   P  e  o  p  l  e  w  h  o  a  r  e  i  m  p  o  r  t  a  n  t  t  o  m  e  i  n  f  l  u  e  n  c  e  m  y  d  e  c  i  s  i  o  n  t  o  h  a  v  e  t  h  e  C  O  V  I  D  -1  9  v  a  c  c  i  n  a  t  i  o . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 25, 2021. ;

Mathematical assessment of the roles of vaccination and non-pharmaceutical interventions on COVID-19 dynamics: a multigroup modeling approach

Covid-19 Behaviour Tracker

Mapping global trends in vaccine confidence and investigating barriers to vaccine uptake: a large-scale retrospective temporal modelling study

Expectations and Anxieties About the COVID-19 Vaccine in Japan

Hungary first in EU to approve Russian vaccine

Now is the opposition really anti-vaccination? Telex

What do we know about China's covid-19 vaccines?

Covid-19: EU looks to speed up vaccine rollout BMJ

The New York Times. Hungary pays big for a Chinese vaccine

Willingness to Use an Approved COVID-19 Vaccine: Cross-national Evidence on Levels and Individual-level Predictors

I Took the Trouble to Make Inquiries, So I Refuse to Accept Your Instructions": Religious Authority and Vaccine Hesitancy Among Ultra Orthodox Jewish Mothers in Israel

Timeliness and completeness of routine childhood vaccinations in young children residing in a district with recurrent vaccine-preventable disease outbreaks

Vaccine hesitancy: the next challenge in the fight against COVID 19

Acceptance of a COVID-19 Vaccine in Japan during the COVID-19 Pandemic

COVID-19 vaccination intention in the UK: results from the COVID-19 vaccination acceptability study (CoVAccS), a nationally representative cross-sectional survey

Using a theoretical framework to determine adults' intention to vaccinate against pandemic swine flu in priority groups in the UK. Public Health

Psychological characteristics associated with COVID-19 vaccine hesitancy and resistance in Ireland and the United Kingdom

Barriers of influenza vaccination intention and behavior-a systematic review of influenza vaccine hesitancy

Factors associated with uptake of vaccination against pandemic influenza: a systematic review

A multidisciplinary research agenda for understanding vaccine-related decisions

The Ecology of Human Development: Experiments by Nature and Design

Meta-analysis of the relationship between risk perception and health behavior: The example of vaccination

Perception of the 2020 SARS-CoV-2 pandemic among medical professionals in Germany: results from a nationwide online survey

The relationship between cultural tightness-looseness and COVID-19 cases and deaths: a global analysis

Cultural orientation, power, belief in conspiracy theories, and intentions to reduce the spread of COVID 19

Predicting intention to receive a seasonal influenza vaccination using Protection Motivation Theory

Officials gird for a war on vaccine misinformation

Hungarians receive Chinese Covid vaccine as Orban breaks with EU strategy

100 days to Tokyo: pessimism and fear remain in Japan as Games loom. The Guardian

Israeli Ministry of Education. Segmentation of the number of vaccinators by age and locality

Cultural competence in covid-19 vaccine rollout

Willingness to Receive COVID-19 Vaccination in Japan. Vaccines, 9

Response Style and Cross-Cultural Comparisons of Rating Scales among East Asian and North American Students

Statistical Analysis with Latent Variables: User's Guide Version 8

Principles and Practice of Structural Equation Modeling

Deciding on the number of classes in latent class