Public Preferences for Pandemic Response Strategies: A Discrete Choice Experiment on Vaccination, Restrictions, and Risk Communication in a Hypothetical Hantavirus Outbreak

Participants are presented with a hypothetical future hantavirus outbreak and complete a discrete choice experiment (DCE) consisting of eight randomized choice tasks. In each task, respondents choose their preferred public health response among three alternatives: (1) a vaccination strategy, (2) a restriction-based strategy, or (3) no new public health measures. The vaccination option varies in vaccine efficacy, risk of serious side effects, regulatory approval status, and recommending authority. The restriction option varies in the type, duration, and economic/social burden of non-pharmaceutical interventions. All alternatives are accompanied by projected numbers of hospitalizations and deaths per 100,000 population over the following six months. The order of the choice tasks is randomized across participants. The intervention is embedded within a broader online survey that also collects information on COVID-19 experiences, vaccine attitudes, trust in institutions, risk literacy, and sociodemographic characteristics.

This study employs a randomized, within-subject discrete choice experiment (DCE) embedded in an online survey administered to adult respondents in Italy and the United States. After completing background questions on demographics, COVID-19 experiences, vaccine attitudes, institutional trust, and risk literacy, participants are presented with a standardized hypothetical scenario describing the emergence of a future hantavirus epidemic.

Each respondent completes eight randomized choice tasks. In every task, participants choose one of three alternative public health response strategies: (1) a vaccination program, (2) a restriction-based strategy, or (3) no new public health measures. The attribute levels of the vaccination and restriction alternatives are varied according to an efficient fractional factorial experimental design.

The vaccination alternative is defined by four attributes: vaccine efficacy (60% or 90%), risk of serious side effects (1 in 10,000, 1 in 100,000, or 1 in 1,000,000), regulatory approval status (accelerated review pending, emergency authorization, or full approval), and recommending authority (national health authority or international health organization). The restriction alternative varies by restriction type (e.g., mandatory masking, workplace or school closures, local lockdowns, quarantine of close contacts), duration (1, 3, or 6 months), and economic/social burden (none, low, moderate, or high). The no-new-measures alternative remains constant across all tasks and represents continuation of voluntary recommendations without additional vaccination programs, mandates, or restrictions.

For each alternative, projected hospitalizations and deaths per 100,000 population over the following six months are presented and vary across choice tasks. The order of the eight tasks is randomized across respondents to minimize order and fatigue effects. Respondents select their preferred option in each task, generating repeated observations for panel analysis. Preference heterogeneity and trade-offs among policy attributes are estimated using mixed logit models, with respondent-level characteristics, including COVID-19 experiences, vaccine attitudes, trust in institutions, political orientation, and risk literacy, used as moderators of stated preferences. Cross-country comparisons between the Italian and U.S. samples are conducted to assess differences in public preferences for future pandemic response strategies.

2026-07-02

2026-07-31

The primary outcome is the respondent's stated choice in each discrete choice experiment (DCE) task. Each participant completes eight choice tasks and, in each task, selects one of three alternative public health response strategies:

Vaccination strategy
Restriction-based strategy
No new public health measures

The outcome is recorded as a categorical choice variable at the respondent–choice task level (up to eight observations per participant). These repeated choices are used to estimate the marginal utility (preference weights) of the experimental attributes—including vaccine efficacy, side-effect risk, regulatory approval status, recommending authority, restriction type, restriction duration, economic/social burden, and projected hospitalizations and deaths—using mixed logit (random-parameters logit) models.

The primary outcome is the respondent's choice in each of the eight discrete choice experiment (DCE) tasks. For every task, respondents select one of three alternatives: (1) a vaccination strategy, (2) a restriction-based strategy, or (3) no new public health measures. This choice is recorded as a categorical variable at the respondent-task level.

The outcome itself is not constructed from multiple variables. However, the analysis estimates latent preference weights (marginal utilities) for each policy attribute by modeling the observed choices using mixed logit (random-parameters logit) models. These preference weights are inferred from repeated choices across the eight tasks rather than directly observed. Respondent-level variables, including COVID-19 experiences, vaccine attitudes, trust in institutions, political orientation, and risk literacy, are used as moderators in interaction analyses to explain heterogeneity in preferences.

This study employs a randomized, within-subject discrete choice experiment (DCE) embedded within a cross-sectional online survey administered to adult participants residing in Italy and the United States. The objective is to elicit stated preferences for alternative public health response strategies during a hypothetical future hantavirus outbreak.
Each participant completes eight discrete choice tasks. In every task, respondents are presented with three mutually exclusive public health strategies and are asked to select the option they would most prefer. The three alternatives consist of: (1) a vaccination strategy, (2) a restriction-based strategy, and (3) a no-new-measures strategy. The "no new measures" alternative serves as a constant reference option across all tasks, while the vaccination and restriction alternatives vary according to experimentally assigned attribute levels.
The vaccination strategy is characterized by vaccine efficacy, the probability of serious side effects, regulatory approval status, and the authority recommending vaccination. The restriction strategy varies according to the type of restriction implemented, its duration, and the associated economic and social burden. In every choice task, each alternative is also accompanied by projected numbers of hospitalizations and deaths per 100,000 population over the following six months. Attribute combinations are generated using an efficient fractional factorial design to maximize statistical efficiency while avoiding implausible or dominant profiles.

The order of the eight choice tasks is randomized across respondents to reduce potential order and fatigue effects. All participants receive the same overall survey structure, but the randomized presentation of the experimental tasks ensures variation in exposure to the choice scenarios.
The DCE is embedded within a broader questionnaire collecting information on demographic characteristics, health status, previous experiences during the COVID-19 pandemic, vaccine attitudes, trust in institutions and information sources, political orientation, and risk literacy. These variables are measured observationally and are not experimentally manipulated.
Each respondent contributes up to eight observations, one for each completed choice task, allowing estimation of preferences for alternative policy attributes and investigation of heterogeneity in preferences across individuals and between the Italian and U.S. samples.

This study uses a randomized, within-subject discrete choice experiment (DCE) to investigate public preferences for alternative policy responses to a hypothetical future hantavirus epidemic. The experiment is embedded within a single-session online survey administered to adult respondents recruited from Italy and the United States through the Qualtrics survey platform. Country-specific quota sampling is used to obtain samples that broadly reflect the demographic composition of the adult populations.
After providing informed consent, respondents complete a structured questionnaire collecting information on sociodemographic characteristics, health status, COVID-19 experiences, COVID-19 vaccination history, attitudes toward vaccines, trust in institutions and information sources, political orientation, and risk literacy. Participants are then introduced to a standardized hypothetical hantavirus outbreak scenario and instructed that they will evaluate alternative public health response strategies.
Each respondent completes eight discrete choice tasks. In every task, participants choose their preferred option among three alternatives:
Vaccination strategy
Restriction-based strategy
No new public health measures
The vaccination alternative is defined by four experimentally manipulated attributes: vaccine efficacy (60% or 90%), probability of serious side effects (1 in 10,000; 1 in 100,000; or 1 in 1,000,000), regulatory approval status (accelerated review, emergency authorization, or full approval), and recommending authority (national health authority or international health organization).

The restriction alternative varies according to restriction type (mandatory masking, workplace and school closures, local lockdowns, quarantine of close contacts, or related combinations), restriction duration (1, 3, or 6 months), and the associated economic and social burden (none, low, moderate, or high).
The no-new-measures alternative remains constant across all tasks and represents continuation of voluntary recommendations without additional vaccination campaigns, mandates, or restrictions.
For every alternative in each choice task, projected hospitalizations and projected deaths per 100,000 population over the following six months are presented. These health outcome levels vary systematically across alternatives and choice tasks as part of the experimental design.
The combinations of attribute levels are generated using an efficient fractional factorial design that minimizes correlation among attributes while maximizing statistical efficiency for estimating marginal utilities. Implausible or dominant profiles are avoided during the experimental design process. The order of the eight choice tasks is randomized independently for each respondent to reduce potential order, learning, and fatigue effects.
The unit of analysis is the respondent-choice task observation, with each participant contributing up to eight observations. The primary outcome is the selected alternative in each choice task. Preference parameters are estimated using mixed logit (random-parameters logit) models that account for repeated choices within respondents and allow for individual-level preference heterogeneity. Conditional logit models are estimated as robustness checks, while latent class logit models are used as exploratory analyses to identify discrete preference segments.
Pre-specified interaction analyses examine whether preferences vary according to respondents' trust in public institutions, previous COVID-19 experiences, vaccine attitudes, political orientation, risk literacy, chronic illness status, and prior vaccination history. Additional analyses compare preference structures between the Italian and U.S. samples and estimate respondents' willingness to trade off health outcomes against the duration and burden of public health restrictions. Respondents who fail predefined eligibility or data-quality criteria, including incomplete surveys, attention-check failures, duplicate participation, or implausibly short completion times, are excluded from the primary analyses in accordance with the preregistered analysis plan.

Randomization was implemented automatically through the Qualtrics survey platform. Qualtrics randomized the order of presentation of the eight discrete choice experiment (DCE) tasks for each respondent using its built-in randomization functionality. The allocation of attribute combinations to choice tasks was predetermined by the experimental design, while the sequence in which the tasks were presented was randomized independently for every participant.

Individual respondent. Randomization occurs at the individual level through the Qualtrics platform, with each participant receiving a randomized order of the discrete choice experiment (DCE) tasks.

No

5,000 individual respondents (2,500 in Italy and 2,500 in the United States).

5,000 individual respondents (2,500 in Italy and 2,500 in the United States).

1250 by arm in each country