April 9, 2024

Improving vaccine messaging

Sarah Eichmeyer discusses ways to increase vaccine take-up rates.

Source: StudioRomantic

The COVID-19 pandemic highlighted the importance of vaccines, but it also underscored the reservations and low take-up rates among US citizens.

In a paper in the American Economic Journal: Economic Policy, authors Marcella Alsan and Sarah Eichmeyer tested several approaches to improving messages aimed at boosting vaccine demand. Their main finding was that messages delivered by laypersons were more effective than those delivered by persons perceived to be doctors. 

Eichmeyer says that video messages delivered by experts who were of the same race or were perceived as empathetic can be effective for some types of viewers, but for the most hesitant, ordinary citizens may be the best positioned to dispel myths about vaccines. 

She recently spoke with Tyler Smith about the design of her and Alsan’s experiment and what their results imply about vaccine messaging.

The edited highlights of that conversation are below, and the full interview can be heard using the podcast player.


Tyler Smith: Why did you want to study vaccine messaging, and why messaging to Black and White men without a college degree in particular?

Sarah Eichmeyer: In this study, my coauthor and I focused on vaccines because they are very effective at preventing many diseases, but they often have pretty low take-up rates. It's an area where increasing take-up can have really big positive effects on health. And also in the case of communicable diseases, for example the flu, increasing the take-up of vaccines can also have really big positive externalities on the community. We focused on men of low socioeconomic status because they're the ones with the lowest vaccination rates. Figuring out what kind of campaigns may increase vaccine take-up in this group is particularly important. Also, long standing systemic discrimination may contribute to racial health disparities, so we developed additional messaging options for Black men in particular.

Smith: What kind of vaccine messages were you interested in trying to get to Black and White men without a college degree?

Eichmeyer: We wanted to test the effectiveness of different information messages about the importance of getting the flu vaccination on people's take-up decisions. We did this by recording several information videos about the safety and the effectiveness of the flu vaccine, which followed current medical advice. Each video basically featured one person, either a Black or a White man between the age of 30 to 50, who would provide the information in a simple one-minute message. We recorded the videos with ten different individuals. And then we recruited almost 3,000 people online, mostly through a survey panel provider, and we randomly assigned each of them to one of the video messages. And afterwards, we surveyed our participants about their intentions to receive the flu shot. And a few weeks later, we surveyed them again and asked whether they got a flu shot in the meantime. 

Figuring out which design features make these types of messages more persuasive is really important for public policies. And in terms of the types of messages we tested, there is a huge space of potential messages we could have tested, so we had to narrow it down to something we could handle in a randomized experiment. We focused our messages on increasing people's trust in the advice that we provide in the message. And that's because we know from existing work that often it's not lack of information, per se, or financial constraints that may prevent people from getting a vaccine like the flu shot or the COVID-19 vaccine. Rather, it's issues around trust in the medical advice given. 

When it comes to the features of the sender, the first element we varied was that of the perceived medical expertise of the sender. So in our case, whether the information is given by a medical professional or by a layperson. The second feature about the sender we varied is that of racial concordance—so whether the race of the messenger matches that of the recipient or not. And the third variation we used was in what is said in the messaging specifically. We either added or omitted a part of the message where the messenger or the sender acknowledges past injustices committed by the medical community.

Smith: When you did this experiment, what did it reveal about the importance of expertise, racial concordance, and acknowledging past wrongs?

Eichmeyer: For Black respondents, we found that when it comes to both the intention to get vaccinated against flu and also the actual self-reported take-up of the vaccine, which we measured in a follow-up survey a few weeks after people saw the video, the concordant layperson message performed the best. We found about a 4 to 5 percentage point increase in the intention to get vaccinated, and also in the likelihood that someone reported having received the vaccine. Given that our intervention was this very light-touch, one-minute video, this effect size is actually substantial. On the other hand, we found that overall adding the acknowledgment statement had only very small, noisy, positive effects on the take-up rate and on intentions compared to omitting the acknowledgment statement. Similarly, the concordant expert arm also only had relatively small and noisy positive effects on vaccination intentions and self-reported take-up rates compared to the discordant expert message. When we looked at White respondents to whom we only assigned either the concordant expert message or the discordant expert message, we did not find any concordance effects at all among our White respondents. 

Finally, we found striking heterogeneity across respondents who are more or less vaccine hesitant. We found that the layperson message was particularly persuasive among the most vaccine-hesitant individuals. On the other hand, we found that among those who were the least vaccine hesitant—the people who did receive a flu vaccine in at least one of the previous two flu seasons—adding this empathetic acknowledgement message or providing the message by a concordant expert were more effective than the standard message by a discordant expert. In sum, while messages from concordant and empathetic experts may resonate most among individuals familiar with vaccination, our study suggests that peer figures like community health workers or citizen ambassadors could play an important role in communicating benefits and dispelling myths about vaccines among the least inclined to receive one.

While messages from concordant and empathetic experts may resonate most among individuals familiar with vaccination, our study suggests that peer figures like community health workers or citizen ambassadors could play an important role in communicating benefits and dispelling myths about vaccines among the least inclined to receive one.

Sarah Eichmeyer

Smith: Do you know why nonexpert messengers were more effective than the expert messengers?

Eichmeyer: In hindsight, it makes sense that these nonexperts were particularly persuasive among the people who are the most vaccine hesitant because their hesitancy may reflect low trust in the medical profession more generally. So community health workers or citizen ambassadors may end up being perceived as more trustworthy because they may be perceived as being outside of the medical establishment.

Smith: Do you think that if this were to be scaled up to a larger, real-world messaging campaign, it would put a dent in low take-up rates?

Eichmeyer: Our study relied on self-reported take-up rates and self-reports shortly after our experiment. So it's difficult to extrapolate to the scaled-up natural setting where these messages would be disseminated through other platforms. But I think disseminating such successful campaigns at scale could put a dent in these big disparities in vaccination rates across socioeconomic groups. These overall disparities are on the order of about 15 percent. So there's about a 15 percentage point difference in the flu vaccination rates among men of low socioeconomic status compared to their counterparts. If messages like these may increase the take-up rate of vaccines on the order of 2 to 5 percentage points, that would already close this gap substantially.

Smith: Beyond just increasing take-up rates, are there any lessons here for medical experts and public health institutions with messaging responsibilities? What should they take from your study?

Eichmeyer: I think there are three pieces that may have important implications for medical experts and public health institutions that may run these kinds of messaging campaigns. The first is that there is quite clearly no one-size-fits-all policy. As I mentioned before, we found this important heterogeneity in which the most persuasive message depended on the respondent’s baseline vaccine hesitancy. This distinction of whether the messages are aimed at vaccine-hesitant individuals or those more open to vaccines already seems really important. Second, it's often very difficult to forecast, ex ante, which message may be most persuasive. This was the case in our study, where we didn't necessarily predict that our layperson message would be so successful. It's really vital to diligently test the effectiveness of messaging campaigns. That is, I very much recommend randomizing the messages in the initial stage of the campaign and to diligently track the outcomes. And, third, I would suggest testing more messages. Public health institutions could follow approaches that start from a much larger message space and then narrow in on the most effective messages.

Experimental Evidence on the Effectiveness of Nonexperts for Improving Vaccine Demand” appears in the February 2024 issue of the American Economic Journal: Economic Policy. Music in the audio is by Podington Bear.