Social Media Listening for Improved Vaccine Acceptance

"...can definitely inform and help stakeholders who are involved in the vaccination process to design effective communication strategies that can be targeted in improving...vaccine confidence..." - Mayya Kelova

Organised by the Sabin Vaccine Institute, the Bay Area Global Health Alliance, and the Alliance for Advancing Health Online, this dialogue featured global health and communications experts sharing their experience using social media listening to drive programming decisions for improved vaccine uptake, specifically focused on minority and vulnerable populations. They discussed how they have used data from social listening to improve efforts to drive vaccine confidence and access.

Speakers included:

• Angus Thomson, PhD, United Nations Children's Fund (UNICEF)'s Vaccine Demand Observatory - Thomson began by clarifying that social listening, which is predicated on the basis of data equity, is not just social media listening. Perhaps half the world is not online but is still having conversations about vaccines. Today's event is focused on social media, but social listening is, more broadly, the mechanisms you put in place to understand not only the flow of misinformation but also the information needs of people, their concerns, and the knowledge gaps they have in relation to vaccines. Thomson talked about the need to view social listening data through a behavioural and social science lens to understand what might be driving people's decisions. The key is to "extract the signal from the noise". This information can guide intervention development. Thomson asserted that such approaches should be integrated into programming over the long term. He described the work the Observatory is doing with social listening for vaccination, noting that listening must be coupled with action - to inform what you do in the field. The Observatory, which is a partnership between UNICEF, the Public Good Projects (PGP), and the Yale Institute for Global Health (YIGH), has been in five countries in the past three weeks to feed its database of misinformation alerts. It is essential, per Thomson, to produce content (when responding to misinformation) that captures people's attention and that is eminently shareable; otherwise, you are wasting your time. What works in one country/context may not work in another, he stressed.
• Mayya Kelova, Minority Rights Group's Diversity: Impact on Vaccine Equality (DIVE) project - Kelova discussed one aspect of DIVE in Sri Lanka, which was to monitor social media to track and understand COVID-19 vaccine uptake and access specifically across three ethnic groups (the majority Sinhalese community, as well as the minority Tamil and Muslim groups). The goal was to understand the differences in sentiment by ethnicity. Kelova explained that Sri Lanka has a history of achieving high and equitable immunisation coverage across the island (e.g., with polio vaccines), but the pandemic brought its own issues that affected Sri Lanka's socio-political landscape. For example, members of the Muslim community were blamed for being mass spreaders of the virus; in general, health-based misinformation was rampant. DIVE gathered posts on Facebook and Twitter with many comments/replies and used the names and languages of the social media users who engaged in these conversations to identify their ethnicity. DIVE then coded over a thousand comments and tweets to assess each of them for confidence toward vaccines, issues related to vaccine access, and distrust of authorities expressed by each of these users. They found that low- to no-confidence levels were much more prevalent among Muslim and Tamil users. Of those who displayed low levels of confidence, their main concern was around vaccine safety (especially among Muslim users). Finally, there was selective preference for or confidence in certain brands of COVID-19 vaccines (e.g., with the Chinese-manufactured vaccine having a low rate of approval). She said Verité research (see below) is disseminating an online campaign to debunk COVID-19-related health-based misconceptions in order to promote vaccine uptake.
• Jia Hu, MD, 19 to Zero - Hu detailed how 19 to Zero use natural language processing (NLP) to analyse vaccine attitudes in Canada. Before the COVID-19 vaccine was even available, this group was gathering data through traditional surveys (reaching a sample of 1,500 Canadians every week for about 70 weeks). They reached out to certain segments that may have less confidence in the vaccine - e.g., indigenous populations - in this on-the-ground listening. To bolster these efforts, the organisation felt that it was important to look at what was happening online. Instead of qualitative approaches that involve going online and following influential accounts, 19 to Zero used artificial intelligence (AI)-based approaches, including NLP, to try to analyse what was happening in the North American social media discourse. These approaches enabled them to understand some predominant themes. They also did a social network analysis, which allowed them to organise different clusters of conversations. Ultimately, they were able to use these data to shift some of the influential political leaders into the pro-vaccine camp.
• Mahoshadi Peiris, Verité Research - Peiris responded to a question from the moderator about how offline strategies such as surveys and focus groups can be used in combination with social media listening to ensure the findings are relevant to all of one's audiences. She did so by discussing work done with survey partner Vanguard to conduct an island-wide survey in Sri Lanka among 2,400 respondents. They focused on achieving a sample that was representative of the three ethnic groups mentioned by Kelova. Peiris shared the survey findings, which gave Verité a sound understanding of on-the-ground realities and helped them when designing their online campaign. They were able to work to promote vaccine confidence among the small pockets of individuals who remained unvaccinated against COVID-19.

Some actionable recommendations for those designing messaging interventions based on social listening in the context of vaccination include:

• Take into consideration the figures in the community who are leaders with a certain kind of authority (including healthcare workers), and partner with them to reach out to communities.
• Make sure you are reaching out to the community through the platforms that are accessible to them (including, for example, radio and/or TV) and in languages they understand. These considerations are especially relevant when communicating with members of minority communities, who are more vulnerable to myths and conspiracy theories when content is not in their language.

The experts also shared tools and resources designed to help others implement and use social listening data in their own work, whatever their context. One example may be found at Related Summaries, below.