A new study suggests that public health messages which reveal how many people from an opposing political group have been vaccinated can influence individuals’ plans to get a COVID-19 jab. The work, led by Brandyn F. Churchill of American University with Xiaoxue Sherry Gao and Rong Rong from the University of Massachusetts Amherst, studied how information shocks tied to partisan identity affect health decisions. The researchers note the findings emerge from a real experiment and are documented in the study titled “Partisan differences in healthcare decision-making: Evidence from a vaccine experiment.”
Researchers began with a baseline survey in May 2024, recruiting 200 adults split evenly between Democrats and Republicans. Participants answered questions about their recent experiences with COVID-19, including infection and healthcare use, and whether they had been vaccinated. The two groups showed similar levels of infection and healthcare use, but vaccination behaviour differed: about 46 per cent of Democrats reported receiving a recent vaccine, versus 28 per cent among Republicans.
In August 2024, the team ran a larger main experiment with 579 participants—288 Democrats and 291 Republicans. Respondents were asked to estimate how the earlier group had answered the health questions. Half were told to guess their own party’s responses, the other half to guess the opposing party’s responses. To deter guesswork driven by bias, participants could win a monetary bonus by placing digital bets on how a randomly selected person from the first survey had answered the questions.
The results revealed broad misperceptions about the disease and its treatment. Across the political spectrum, participants overestimated how frequently people contract COVID-19 by around 20 percentage points and overestimated how often medical intervention is required by about 22 percentage points. They also underestimated the prevalence of vaccine side effects and the share of people who got vaccinated to protect others.
Interestingly, participants were not more accurate about the behaviours of their own party than of the other. Both groups tended to overestimate how many Republicans were vaccinated against the virus.
After guesses were settled, a computer programme revealed the actual statistics from the initial survey for participants to compare with their estimates. The researchers then asked whether respondents would consider receiving a COVID-19 vaccine within the next twelve months.
Learning the true statistics about infection rates, disease severity or side effects did not alter intentions to be vaccinated. Nor did learning the actual vaccination rate of one’s own party produce any change in plans.
However, knowledge of the opposing party’s vaccination rate did shift intentions. Participants who discovered they had underestimated how many in the opposing group were vaccinated became more likely to say they would be vaccinated themselves, while those who had overestimated the opposing rate became less likely.
To verify the robustness of this pattern, the researchers conducted additional checks. When they reanalysed the data as if the information shown had been different, the effect disappeared, suggesting it was specifically driven by the opposing party’s vaccination rate. There was no link between the information shock and any actual past vaccination behaviour, reinforcing that the observed shift related to stated future intentions rather than concrete actions.
The team interprets the finding within a social pressure framework: people are more inclined to adopt a health measure when they observe others doing so. Learning that a group they view as generally opposed is participating in vaccination can heighten social pressure to follow suit, and the reverse can occur if the opposing group appears less engaged than expected.
The study raises concerns about public health messaging. Campaigns that emphasise low vaccination rates may, because people tend to overestimate peers’ uptake, act as negative information shocks. In such cases, the message could dampen social pressure to vaccinate rather than boost it, at least in terms of reported intentions.
Limitations of the research include a relatively small sample size and the follow-up response rate, which meant the researchers could not reliably track how intentions translated into actual vaccination four months later. The authors acknowledge that the precise cognitive mechanisms behind why information about the opposing group proves more persuasive remain unclear and warrant further study.
Authors: Brandyn F. Churchill, Xiaoxue Sherry Gao and Rong Rong.
