key: cord-0887702-v66rqr6w authors: Duch, R.; Roope, L. S. J.; Violato, M.; Becerra, M. F.; Robinson, T.; Bonnefon, J.-F.; Friedman, J.; Loewen, P.; Mamidi, P.; Melegaro, A.; Blanco, M.; Vargas, J.; Seither, J.; Candio, P.; Cruz, A. G.; Hua, X.; Barnett, A.; Clarke, P. title: Who should be first in line for the COVID-19 vaccine? Surveys in 13 countries of the publics preferences for prioritisation date: 2021-02-02 journal: nan DOI: 10.1101/2021.01.31.21250866 sha: 34f20af0aabc3327fed284bdf0943c096067ceca doc_id: 887702 cord_uid: v66rqr6w How does the public want a COVID-19 vaccine to be allocated? We conducted a conjoint experiment asking 15,536 adults in 13 countries to evaluate 248,576 profiles of potential vaccine recipients that varied randomly on five attributes. Our sample includes diverse countries from all continents. The results suggest that in addition to giving priority to health workers and to those at high risk, the public favours giving priority to a broad range of key workers and to those on lower incomes. These preferences are similar across respondents of different education levels, incomes, and political ideologies, as well as across most surveyed countries. The public favoured COVID-19 vaccines being allocated solely via government programs, but were highly polarized in some developed countries on whether taking a vaccine should be mandatory. There is a consensus among the public on many aspects of COVID-19 vaccination which needs to be taken into account when developing and communicating roll-out strategies. How to allocate a COVID-19 vaccine is one of the most important decisions currently facing every government. There has been an unprecedented race to develop a vaccine. At the time of writing, 58 vaccine candidates are currently undergoing human trials (1) . Several vaccines have been shown to be safe and highly effective (2) and countries are starting to approve their use (3) . It has been recognised that, in many countries, public confidence in vaccination is fragile, and that policies for prioritizing vaccine allocation need to be seen as both equitable and evidence-based (4) . Ethical frameworks have been suggested for the allocation of scarce vaccine supplies between countries (5). The WHO has developed a values framework based on twelve objectives and six principles (Human Wellbeing, Equal Respect, Global Equity, National Equity, Reciprocity, Legitimacy). Importantly, the WHO does not provide any guidance on the order of importance of either the principles or the objectives. (6) Constraints on timely supply of vaccines mean that it is likely that it will not be possible to secure all of the objectives simultaneously. The WHO Strategic Advisory Group of Experts on Immunization (SAGE) has recently proposed a roadmap which prioritizes health workers and older adults. (7) At a national level, governments are having to rapidly develop guidelines to prioritize access to a vaccine. Based on a survey of governments' vaccine allocation policy plans, conducted in early December 2020, Table 1 indicates that there is considerable diversity across countries in the groups being prioritized. While prioritization of health workers and the clinically vulnerable appears universal, there is little consensus on which other groups to prioritize. The UK prioritization strategy is largely aged-based, starting with the oldest age categories followed by the clinically vulnerable (8) . No other criteria will be employed until after everyone over 50 and/or with underlying health conditions has been vaccinated. In contrast, an expert committee in France has recommended prioritizing workers who have contact with the general public 3 Occupation Age Transmission Vulnerability Essential infrastructure Health/social care Education/child care 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 February 2, 2021. and equitable. If this is not the case, for whatever reasons, governments risk the types of public resistance and polarization that occurred in some countries regarding the wearing of masks (21) . It also risks the creation of vaccine black markets that would threaten the safety and fairness of the vaccination campaign. To accomplish these goals governments should seek evidence of the public's opinions and preferences regarding the groups to be prioritized, public versus private distribution channels and mandatory requirements to be vaccinated. This information can aid in the design of better policies and the implementation of successful communication campaigns, both of which would help ensure a successful COVID-19 vaccination program (22) . To provide an evidence based understanding of public opinions on key aspects of vaccine allocation, we undertook a large scale online public opinion survey in 13 countries. Quota sampling was employed in order that the survey matched the demographic margins from the populations of each country. 1 The survey included a conjoint experiment to identify preferences for different vaccine prioritization schemes. Conjoint survey experiments are frequently employed to identify the importance individuals attribute to different features or characteristics of choices (23) . Examples include environmental migrants (24) , asylum seekers (25) and migration destinations (26) . (27) employed conjoint experiments that generated 40 million decisions to determine the ethical principles the public thinks should guide self-driving cars. 2 In the case of policy-oriented survey experiments, evidence suggests that the weights given to attribute characteristics in conjoint survey experiments map closely to the actual policy choices made by the population (29) . In our conjoint experiment each of the 15,536 subjects made eight binary choices over hy-pothetical vaccine recipients (a total of 124,288 pair-wise comparisons) that randomly varied on five attributes: occupation, age, transmission status (risk of contracting and transmitting the virus), risk of death from COVID-19, and income. 3 Table 1 suggests that these five attributes have played particularly important roles in the vaccine allocation policies being considered by our sample of countries. We estimated the importance of specific characteristics of vaccine allocation priorities with logistic regression (with standard errors clustered by participant). For each pair-wise choice we regressed the participant's binary decision on dichotomous variables representing the attribute values of the five vaccine allocation attribute variables. Table 6 in the Supplementary Materials presents the logistic regression results. Figure 1 reports the coefficients, along with their confidence intervals, from this logistic regression. We are interested in the relative effects of the attribute-levels in our models, and the logistic coefficients are sufficient to demonstrate the difference in relative magnitude and the direction of any causal effects within and between attributes. Readers should note that these coefficients should not be directly interpreted as the marginal effects on the probability of choosing a vaccine recipient. 4 The reference categories for the conjoint attributes are the neutral categories included as dots with coefficient zero in Figure 1 . The individual country conjoint results are organized by three regions in Figure 1 . There is no evidence in any of the 13 countries of respondents treating all potential vaccine recipients 3 Figure 4 in Supplementary Materials provides an example of the attributes and values that characterized the two vaccine candidates presented to respondents. Checking the proportion of times individual conjoint levels were shown to subjects confirms that they were adequately randomised (see Supplement Table 3) . 4 Identical models, using the same dichotomous outcome variable, were estimated using OLS regression with clustered standard errors. These results are reported in Supplementary Materials Table 7 . Supplementary Materials Figure 6 displays the estimated average marginal component-specific effects (AMCEs) generated from the OLS regressions (30) . The results are virtually identical to those presented in the main text. equally. With respect to each of the five priority attributes there is evidence that the global public favors some profile attributes over others. Moreover, the pattern of coefficient values across our sample of 13 countries is quite similar. The global public exhibits a surprising consensus on which population segments should have priority for a COVID-19 vaccine. Age matters. Respondents in virtually all countries favor vaccine candidates with age profiles greater than the young, 25 year-old, reference category. And there is evidence in a number of countries suggesting that the two oldest age categories (the 65 and 75 year old profiles) were favored over the younger 25 and 40 year old profiles. The one exception is China, where the older 65 and 75 year old profile attributes were less preferred than the younger, 25 and 40 year old, profile attributes. This distinct preference in China for younger vaccine recipients might be a function of the very young age composition of the China sample, which is heavily skewed towards younger participants. For a majority of the 13 countries in our study, the income attributes of the potential vaccine recipients affects allocation preferences. The reference category here is the lowest quintile of the income distribution. In most middle and low income countries (Brazil, Chile, Colombia, India and Uganda), respondents exhibit lower preferences for potential vaccine recipients in the average and high income categories. We see a similar, although somewhat less pronounced, pattern for the U.S., Canada, China, Australia, and European countries. It is worth pointing out that redistributive vaccine allocation preferences seem particularly salient in those countries with higher levels of income inequality. Vaccine allocation preferences related to occupation are virtually identical across all the sampled countries. The reference category here is "not working". In all of the countries, respondents accorded similar vaccine priority to the "not working" profile attribute and the "nonkey worker able to work at home" attribute. The "non-key workers unable to work at home" attribute had a significantly positive impact on profile selection (relative to the "not working" 9 . 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 February 2, 2021. ; https://doi.org/10.1101/2021.01.31.21250866 doi: medRxiv preprint reference category). All of the "key worker" occupational attributes had a positive effect on vaccine recipient selection in all sampled countries. This "key-worker" result is very much consistent with the recent vaccine allocation priority plans announced by the governments of our sampled countries (see Table 1 ). As Table 1 indicates, many of the governments of our sampled countries have prioritized COVID-19 vaccination for individuals who are at high risk of death from the virus and, to a lesser extent, those at high risk of contracting and transmitting the virus. Similarly, in Figure 1 , we see that vaccine allocation profiles with "high risk of COVID-19 death" and high risk of "COVID-19 contracting and transmitting" were significantly more likely to be selected by respondents in all 13 country samples. . 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) . 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 February 2, 2021. ; https://doi.org/10.1101/2021.01.31.21250866 doi: medRxiv preprint Segments of the population may display quite different views on vaccine allocation priorities. Figure 2 explores this possible heterogeneity by breaking out the main conjoint analysis by age, income and left-right political self-identification. Overall, the effects of the attributes were broadly similar across the different subgroups. As to which vaccine recipient attributes should be prioritized, there is a general consensus among those identifying with the left and the right, young and old, less and more highly educated, and richer and poorer citizens. In the Supplementary Materials (see Figure 7 ) we conducted additional sub-group analyses indicating that women and men, high and low educated, and those intending and not intending to be vaccinated all agree on the vaccine recipient attributes that should be prioritized. These are a diverse set of possible sources of heterogeneous treatment effects. This absence of heterogeneity here suggests that preferences regarding vaccine priorities are not affected by self-interest, political partisanship, vaccine hesitancy, or educational attainment. Instead, they reflect broad, societal consensuses on who should be vaccinated. . 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) . 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 February 2, 2021. Implementing COVID-19 vaccination programs will be challenging. The COVID-19 vaccines are a public good and the general public has expectations regarding how this public good should be provided; specifically, what vaccine allocation mechanisms are appropriate or acceptable. Figure 3 confirms that a very large majority of the public believes the government should assume the lead role in the distribution of COVID-19 vaccines. Moreover, this result is consistent across each of the 13 countries; at least two thirds of the population in each country said that distribution should only be available via government schemes. Nevertheless, in most countries there are still substantial percentages of the population who feel there should be at least some role for private distribution; this ranges from 13 percent in Canada to 28 percent in Chile. While a majority in all countries think COVID-19 vaccine distribution should be solely via government schemes, there is evidence that a large proportion would be willing to pay for a vaccine if it were available privately. We asked respondents "If a COVID-19 vaccine was also available for private purchase and you could receive it immediately [rather than wait six months] would you considering buying it." Panel B in Figure 3 suggests that roughly half of our global sample would be willing to purchase the COVID-19 vaccine on the private market, ranging from 18 percent in France to 79 percent in India and Uganda. The low and middle income countries 5 The exact wording of these questions is reported in the Supplementary Materials. are particularly enthusiastic about purchasing the vaccine on the private market (while at the same time strongly favoring government provision). There was no consensus on whether the COVID-19 vaccine should be mandatory, either globally or within national borders. We asked respondents to indicate on a scale from 0 (very much disagree) to 100 (very much agree) how much they agreed or disagreed with the statement " government should make the COVID-19 vaccine mandatory for everybody." As panel C in Figure 3 indicates, opinion overall is skewed toward making COVID-19 vaccination mandatory. But there is evidence of some polarization. In a number of countries there is substantial clustering of responses at both ends of the scale. About 24 percent of our global sample were strongly opposed to mandatory vaccination, while about 38 percent were strongly in favour. There was variation amongst countries. In France, there was a broad consensus of opinion strongly opposing mandatory vaccination (about 60 percent opposed mandatory vaccination). Opinion was highly polarized in the US and UK, with the majority of people either strongly opposed to mandatory vaccination, or strongly supportive. Opinion was also somewhat polarized, with little middle ground, in Australia, Brazil, Chile and Colombia -but with a much larger cluster supporting mandatory COVID-19 vaccination. In China, India, and Uganda, very few people were strongly opposed to mandatory vaccination and the majority were strongly supportive. . 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 February 2, 2021. vaccine. The profiles varied on a set of randomly assigned attributes. We found that the public would prioritize people for vaccination based on a broad range of factors. Not surprisingly, these include features directly related to contracting COVID-19 or developing severe symptoms, such as age, vulnerability and risk of transmission. Notably, however, the public would also prioritise according to what might be deemed more economic factors, including low income groups and quite a wide range of non-health related key occupations (e.g. teachers) and non-key workers that cannot work from home. While there is substantial variation in COVID-19 vaccine allocation policies in our sample of countries (see Table 1 ), public preferences for prioritization appear to be largely consistent across countries and larger regions. Moreover, when we estimate the conjoint model for sub-groups in the samples, we find that preferences are consistent across a range of respondent characteristics such as age and income ( Figure 2 ). Preferences for prioritization also do not vary by political leanings, suggesting that if governments implement policies in line with these public views they should not be politically polarizing. Conjoint methods have been widely used to understand preferences for other types of health care, including influenza vaccination, and have been shown to mimic real world decision making. However, it is important to note that our conjoint experiment is largely intended to capture preferences for prioritizing the access of others. Our analyses of heterogeneity indicate that in many instances respondents are willing to prioritize individuals that do not share their own characteristics. For example, those aged over 65 were found to have strong preferences for pri- . 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 February 2, 2021. ; https://doi.org/10.1101/2021.01.31.21250866 doi: medRxiv preprint oritising many types of essential workers. It is difficult to disentangle the degree to which these preferences are due to altruism, or to a desire to reduce the likelihood of being infected through interaction with unvaccinated individuals; or more broadly a desire to choose an optimal vaccination program that would help end the pandemic and bring life back to normal sooner. A limited number of studies have examined COVID-19 vaccine preferences at a national level. For example, a recent survey of the Belgian population (16) found a preference for vaccination of essential workers, those more likely to transmit the virus and those at high risk. Unlike our study, it did not find any preference for allocation to those aged over 60. An advantage of conducting comparative international studies of preferences and opinions is the ability to provide evidence on the degree to which public views are consistent across geographically, economically and culturally diverse countries. While there have been efforts to understand preferences for different methods of social distancing (31) and the likely uptake of a COVID-19 vaccine (32), we have not found other comparable international surveys that can help inform COVID-19 vaccine allocation. Our study also has important implications for vaccine prioritization policies. Importantly, in the countries in our sample, there appears to be much greater heterogeneity in national policies (see Table 1 ) than in public preferences. While current policies include some of the groups that the public think should be given priority (e.g. those at high risk of mortality), there are only a minority of countries that give priority to vaccinating groups of key workers (such as teachers). Further, our study indicates that the public feel that a broader set of economic factors should be taken into account in prioritization policies -these include low income groups and nonessential workers that cannot work from home. With regard to the former, while some countries have stressed the importance of ensuring good vaccine uptake among deprived groups, we are unaware of any countries that are explicitly targeting people according to levels of income or other markers of deprivation. (33) 18 . 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 February 2, 2021. ; Regarding the mode of allocation, there is a remarkably consistent preference for governmentonly allocation across all countries ( Figure 3 ). Nevertheless, the public will pay for a COVID-19 vaccine if it does become privately available. We saw in Figure 3 that in many countries a significant proportion indicated a willingness to purchase it on the private market in order to receive the vaccine faster. The highest proportion of those willing to purchase privately is in low and middle income countries such as Brazil, Chile, India and Uganda. Given the emergence of companies that are gearing to supply privately (17) , understanding implications for both coverage and affordability should be a priority. If COVID-19 vaccines are to be allocated privately, it will be essential to develop policies to ensure that private allocation does not jeopardize a country's ability to acquire the doses necessary for government-managed vaccination campaigns. Ideally, private access would complement public provision to maximize the health and economic gains of vaccination and minimise the potential for corruption (34) . Our survey finds that a very diverse set of allocation priorities are expressed by the general public. These encompass a concern for protecting the most vulnerable and reducing transmission while allowing life to return to normal and allowing productive sectors of the economy to open. These preferences could be used to help weight the multiple criteria on which to assess various vaccination strategies. These results should help inform the ongoing debate over how COVID-19 vaccination programs should be implemented. In particular, they identify opportunities for policymakers, who have often struggled during the COVID-19 pandemic to meet their obligations to protect public health and the economy while simultaneously respecting the public will. While approaches to lockdowns have been divisive and politicized, a positive message from this study is that, with the exception of mandatory vaccination, the public have generally consistent preferences regarding vaccination programs and these hold across political and geographic divides. This does not mean that government vaccination programs should fully accord with public preferences. Designing optimal vaccination programs is complex; there are many externalities to consider; and there is a clear role for expert input (36). Yet these programmatic choices incorporate important implicit value judgements; and the diversity of actual policies across countries emphasises the scope for experts to reach different conclusions. At a minimum, governments should take stock of public opinion. Governments, acting in the public interest, may enact vaccination programs that prioritize groups differently than would the general public. It is important though that governments recognize these differences between policy choices and public preferences and that these differences inform their efforts to gain public acceptance of their COVID-19 vaccination program. . 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 February 2, 2021. ; In all but Chile and Uganda, respondents were sampled by the sampling firm, Respondi. 6 The Respondi participants were compensated for completing the survey. For these twelve countries, the modal incentive was $3.50 for a median length of interview (LOI) of 25.53 minutes. In Chile and Uganda the respondents were sampled using Facebook Ad Manager (37). 7 Respondents in Chile received payments of $3.00 and in Uganda $2.25. The final sample is reasonably representative of the national population. We implemented a quota strategy that generated samples that roughly matched the populations on age, education, gender and region. Post-stratification weights were constructed to account for remaining imbalances, as explained below. Among the panelists invited to take our survey, the response rate (calculated as the fraction of complete responses over invited, eligible participants) was 21.3%, averaged across all countries. The final sample included an average of 1,195 respondents per country (15,536 respondents overall). Descriptive statistics are reported in Table 4 and Table 5 . Experimental Design A conjoint experiment was embedded at the beginning of the CAN-DOUR questionnaire on the theme of the COVID-19 vaccine. The experiment aimed to identify public preferences for which groups should be prioritised to receive limited available doses of COVID-19 vaccine. As an introduction to the overall survey, respondents were presented with a short definition of vaccines and how they work. Conjoint survey experiments are frequently employed to identify the importance individuals attribute to different features or characteristics 6 Further information on Respondi is available here: https://www.respondi.com/EN/. 7 The recruitment ad is available in the Online Appendix. . 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) of choices (23) . Examples include environmental migrants (24) , asylum seekers (25) and migration destinations (26) . Awad et al. (27) employed conjoint experiments that generate 40 million decisions to determine the ethical principles the public thinks should guide machine behavior. 8 We implemented a standard fully randomized paired profiles conjoint design (see Figure 4 for an example) in which each respondent was shown profiles of two different hypothetical vaccine recipients displayed side-by-side (30) . In the case of policy-oriented survey experiments, evidence suggests that the weights given to attribute characteristics in conjoint survey experiments map closely to the actual policy choices the population would make in real world decisions, such as referendums. (29) . In our conjoint experiment each of the 15,536 subjects made eight binary choices over hypothetical vaccine recipients (a total of 124,288 pair-wise comparisons) that randomly varied on five attributes that are being used, or have been proposed as being important criteria for vaccine allocation. Outcomes As Figure 4 indicates, respondents were shown two potential vaccine recipients (Person A and Person B). Respondents were first asked to chose which of the potential recipients should receive the COVID-19 vaccine immediately. The resulting choice outcome variable has a value of 1 for the preferred profile and 0 for the profile that was not selected. Figure 4 has the exact wording of the question. Table 3 presents the attributes and summarizes the distribution of the randomly assigned attribute values for the global sample (these are for the English versions of the survey that were administered in Australia, Canada, India, Uganda, the U.K., and the U.S.). The conjoint experiment is the first section of the survey. The experiment had a prelude introducing the vaccine allocation policy issue. Respondents were then informed of the 8 Other recent policy-related illustrations of conjoint experiments include (28) 25 . 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 February 2, 2021. ; profile attributes. Following this, respondents were asked a series of questions in which they were presented with a pair of profiles. In each question, they were asked to choose between the two profiles; and then to evaluate the extent to which their selected profile should be prioritised on a 7-point Likert scale. The wording of the items follows: Prelude: Across the world, COVID-19 has infected tens of millions, killed more than one million and resulted in loss of jobs, school closures and overall economic loss. A vaccination would prevent people from getting affected by the virus. It is like the flu vaccine: Some people who get the vaccine could still get COVID-19 because it is not 100% effective; the protection could last a few months or for years; and it could have side effects. Once a reliable COVID-19 vaccine is available, health officials will give some individuals priority over others. Some individuals will get the COVID-19 vaccine immediately and other individuals will have lower priority. We are interested in your opinion about these decisions. In the next screens you will be shown The survey instrument included a number of different themes in addition to the conjoint experiment. Many of the variables are described in Table 4 and Table 5 . A full version of the survey instrument is included in these Supplementary Materials. Here we present the wording of the questions presented in Figure 3 : • Talking about vaccines in general, in some countries vaccines are only available from the government either at low or no cost. In some countries vaccines are only available for private purchase. And in some countries vaccines are available from the government but citizens can pay privately to gain early access. Which of these three approaches do you think should be applied to the COVID-19 vaccine? Would you prefer -Vaccines only made available by government at low or no cost? -Vaccines are only available for private purchase? -Vaccines made available by government but citizens can pay privately to gain access? • Consider the following situation: a COVID-19 vaccine becomes available and is provided by government health agencies. For 80 out of 100 people the vaccine would provide 27 . 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. Captions to all relevant figures include a reference to a corresponding regression table. . 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. . 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 February 2, 2021. Figure 4 : Example profile pair. This figure displays an example of a pair of vaccine recipient profiles seen by the survey respondents. This example comes from the English version of the survey administered to respondents in the United Kingdom. Each respondent saw and evaluated eight separate pairs. The order of the attributes (rows) was fully randomized between respondents, but for each respondent, the order was kept constant across the five pairs they were shown. The specific attribute levels (values in the cells in the last two columns) were fully randomized between and within respondents. . 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 February 2, 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 this version posted February 2, 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 this version posted February 2, 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 this version posted February 2, 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 this version posted February 2, 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 this version posted February 2, 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 this version posted February 2, 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 this version posted February 2, 2021. Table 6 : Country Logistic Regression Results. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 1 . . 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 February 2, 2021. Table 7 : Country Linear (OLS) Regression Results. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 6 . . 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 February 2, 2021. ; https://doi.org/10.1101/2021.01.31.21250866 doi: medRxiv preprint Yes Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 8 : Logistic Regression Results for Age Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 2 . 43 . 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. Yes Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 9 : Linear (OLS) Regression Results for Age Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 10 . . 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. Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 10 : Logistic Regression Results for Income Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 2 . . 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. Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 11 : Linear (OLS) Regression Results for Income Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 10 . . 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. Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 12 : Logistic Regression Results for Ideology Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 2 . . 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 February 2, 2021. Figure 10 . . 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 February 2, 2021. Figure 7 . . 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 February 2, 2021. Figure 8 . . 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 February 2, 2021. Figure 7 . . 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. Yes Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 17 : Linear (OLS) Regression Results for Education Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 8 . . 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 February 2, 2021. Yes Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 18 : Logistic Regression Results for Gender Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 7 . . 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. Yes Yes Yes * * * p < 0.001; * * p < 0.01; * p < 0.05 Table 19 : Linear (OLS) Regression Results for Gender Categories. The dependent variable is the Forced Choice decision. These are the estimates used to construct the conjoint plots presented in Figure 8 . . 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 February 2, 2021. ; https://doi.org/10.1101/2021.01.31.21250866 doi: medRxiv preprint Coronavirus vaccine tracker WHO SAGE values framework for the allocation and prioritization of COVID-19 (World Health Organization WHO SAGE roadmap for prioritizing uses of Covid-19 vaccines in the context of limited supply Version 1.1 (World Health Organization Independent report priority groups for coronavirus (covid-19) vaccination: advice from the jcvi Cdc advisory panel takes first shot at prioritizing who gets the first shots of covid-19 vaccines Socio-demographic factors associated with self-protecting behavior during the covid-19 pandemic, Working Paper 27378 on Vaccination, Immunisation, Priority groups for coronavirus (covid-19) vaccination: advice from the jcvi