Pathways to vaccine hesitancy: Understanding personalized social media feeds and vaccine narratives
Researchers
Wendy Chun, Saemi Nadine Jung, Matt Canute, Brandon Lee, Pranjali Mann
Key takeaways
- Vaccine hesitancy is shaped through repeated algorithmic exposure rather than isolated misinformation encounters.
- Social media algorithms connect public health content with broader emotionally charged or polarizing narratives.
Overview
This project examines how algorithms shape exposure to vaccine-related content on social media feeds. By using a research persona method and custom mobile auditing tools developed by the Digital Democracies Institute (DDI), researchers trace how users are gradually shown emotionally charged posts and narratives even without explicitly seeking vaccine information.
How do algorithms influence the type of vaccine-related content that specific populations see on Facebook and Instagram? It is difficult to study social media while also protecting the privacy of real users.
By using custom personas (creating social media accounts for fictitious characters), researchers can audit social media feeds without monitoring or interacting with real users, providing an ethical way to learn about how social media algorithms influence public health discourse.
Personalized social media feeds direct some users toward emotionally charged and identity-relevant health content, even when they are not explicitly searching for vaccine information. Public health information is also linked to other types of emotionally charged posts or narratives — such as parenting, immigration, housing, and social conflict — shaping how trust and uncertainty are experienced online.
Want to better understand your own personalized social media feed?
This open source phone app allows anyone to be more aware of their online ecosystem and consumption over time.
The app allows people to anonymously record and upload a phone session while they scroll online. It then parses the content of your video recording and how much time people spend looking at certain types of content.
The app builds up a catalogue of the influence across different topics over time to help researchers understand why we get fed certain stories. Using this app can help test the bias in algorithmic recommendations across various topics.
Learn more