Prioritization preferences for COVID-19 vaccination are consistent across five countries

The study investigates how the public believes scarce COVID-19 vaccines should be allocated, asking who should receive priority access. In the context of limited initial vaccine supplies and varying national eligibility criteria, understanding public preferences is important for the perceived fairness and legitimacy of vaccination programs and for sustaining cooperation with public health measures. The authors fielded a cross-national survey before vaccine rollouts in Brazil, Germany, Italy, Poland, and the United States, using both a group ranking task and a forced-choice conjoint experiment to assess whether public preferences align with expert and policy guidelines and to identify potential gaps or non-utilitarian heuristics influencing prioritization judgments.

The paper situates vaccine allocation within two main ethical heuristics: utilitarianism (minimizing harm by prioritizing those at highest health risk and those essential to mitigating the pandemic, such as healthcare workers) and deservingness (prioritizing those perceived in need, lacking control over their circumstances, or meriting reward for past behavior). While these heuristics often converge on prioritizing the elderly, people with pre-existing conditions, and frontline workers, deservingness perceptions can also be shaped by in-group biases such as citizenship and partisanship. The authors also discuss popular but potentially non-utilitarian norms (e.g., women-and-children-first) that might influence public views. Prior studies across multiple countries using ranking, rating, and conjoint designs generally find strong support for prioritizing healthcare workers, the elderly, and high-risk individuals, with additional evidence that non-utilitarian factors (e.g., compliance with public health measures, healthy lifestyles, citizenship/ethnicity) can shape preferences. This study adds cross-national evidence with a consistent design, introduces novel attributes (elite status, citizenship, parenthood, early registration), and combines ranking with conjoint tasks to disentangle attribute effects.

Design: Cross-national online survey conducted September 8 to December 9, 2020, before public vaccine rollouts. Countries: Brazil, Germany, Italy, Poland, United States. Total N=4,366 (BR=799; DE=1,651; IT=1,002; PL=803; US=498). Respondents were recruited from Respondi AG access panels with quotas on age, gender, and education. Tasks: - Ranking task: Respondents ranked eight groups by vaccination priority (children, elderly, healthcare workers, people with medical preconditions, people with high-contact jobs, leading politicians, professional athletes, business leaders). Order randomized. - Forced-choice conjoint experiment: Each respondent evaluated four pairs of hypothetical profiles and chose who should be prioritized. Randomized attributes/levels: gender (female/male), age (27/42/61/76), has children (yes/no), occupation (nurse, physician, teacher [Germany only], cook, professor, unemployed), citizenship (citizen/non-citizen), pre-existing medical condition (yes/no), early registration (yes/no). Attribute order randomized across respondents; levels fully randomized. Covariates for heterogeneity analyses: age group, gender, parenthood, political ideology (left/moderate/right), institutional trust (scaled), self-reported pre-existing health condition (yes/no). Analytic strategy: - Ranking task: Compute average priority rank (1–8) for each group. - Conjoint task: Estimate marginal means and average marginal component effects (AMCE) for attributes; explore average marginal interaction component effects (AMICE) for subgroup heterogeneity using linear models with clustered SEs. - Attribute importance: Random forest variable importance (mean reduction in Gini) to assess predictive contribution of attributes; text analysis of an open-ended item on most important characteristics using a dictionary-based approach. The study also compares pooled (sample-size weighted) and country-specific results.

- Ranking task: Across countries, healthcare workers were consistently top-priority (average rank ≈ 2.3, the highest). In all countries except Germany, the top three were healthcare workers, people with medical preconditions, and the elderly. Elite groups (professional athletes, leading politicians, business leaders) clustered at the bottom; politicians were most often last. Children were typically mid-ranked (around 4th–5th). - Conjoint task (marginal means/AMCE): - Occupation: Nurses and physicians had about a 30 percentage point higher probability of selection than unemployed candidates and 10–20 pp higher than cooks and professors, holding other attributes constant. - Health status: A pre-existing medical condition increased selection probability by about 15 pp versus no condition. - Age: Older candidates (61 or 76) were about 10 pp more likely to be chosen than 27-year-olds in most countries. Germany differed, with middle ages (42, 61) favored over both 27 and 76. - Other attributes showed smaller but positive effects: having children, being a citizen, and early registration increased selection probabilities; women were slightly favored in some contexts. - Attribute importance: Random forest analyses highlighted occupation (especially medical professions) and health preconditions as the most predictive attributes of choices; age contributed modestly despite many respondents naming age as most important in open-ended responses, suggesting a gap between stated and revealed importance. - Preference heterogeneity: Minimal polarization. Effects were largely stable across respondent subgroups (sex, parenthood, health status). Modest ideological heterogeneity on citizenship: right-leaning respondents gave about a 10 pp advantage to citizens over non-citizens, compared to about 3 pp among left-leaning respondents, more salient in Italy, the US, and Germany. Overall, within-country coherence remained high.

The findings demonstrate that public preferences for vaccine prioritization are broadly aligned with expert and policy guidelines aimed at minimizing harm: prioritizing healthcare workers, individuals with medical preconditions, and older adults. This supports the view that the public recognizes the need to protect those at highest clinical risk and to maintain essential health system capacity, consistent with utilitarian and deservingness considerations. At the same time, the public also applies non-guideline heuristics, such as preferring citizens over non-citizens and rewarding early registration (first-come, first-served), and showing some preference for parents. These patterns underscore tensions between public perceptions of fairness and ethical principles or human rights-based guidelines that emphasize need and equity regardless of citizenship status. Importantly, the study finds little evidence of polarization within countries: preferences are remarkably consistent across demographic and ideological subgroups, with only modest differences on citizenship by ideology. This coherence suggests room for policymakers to communicate that prioritization policies largely reflect public preferences while remaining attentive to points of divergence that could affect perceptions of fairness and acceptance.

This cross-national study shows substantial consensus in public preferences for COVID-19 vaccine allocation across five countries: prioritize healthcare workers, people with medical preconditions, and the elderly. Results are consistent across tasks, countries, and subgroups, indicating limited polarization. However, non-utilitarian preferences—favoring citizens over non-citizens, early registrants, and to a lesser extent parents—diverge from many public guidelines, highlighting potential equity concerns. Policymakers can leverage the strong overlap with guideline principles to bolster legitimacy and acceptance, while addressing public perceptions that may conflict with rights-based approaches. Future research directions include: testing generalizability in additional contexts and over time; examining actual allocation decisions by medical personnel to assess behavioral discrimination; improving sampling beyond online access panels; and further investigating social desirability and the gap between stated and revealed attribute importance.

- External validity: Results from five countries may not generalize to other contexts. - Sampling: Online access panel samples, despite quota balancing on demographics, may differ on unobserved characteristics (e.g., knowledge of guidelines). - Social desirability bias: Respondents may align answers with perceived official guidance; conjoint design helps mitigate but cannot eliminate this. - Measurement scope: Findings concern preferences, not real-world allocation behavior; the study cannot infer actual discrimination by vaccinators or institutions. - Design constraints: Some attributes (e.g., teacher) only appeared in Germany; open-ended responses suggest potential discrepancies between stated and revealed importance (e.g., age).