Policy Wonk

Prebunking videos shown as YouTube ads

Headline result
+5%
recognition of manipulation techniques in YouTube field test
RCT
Implemented by
University of Cambridge and University of Bristol, with Google Jigsaw
Type
NGO / nonprofit
Where
United States, United Kingdom
Region
Multi-region
When
2021–2022
Evidence
Randomized controlled trial

The challenge

Fact-checks cannot keep up with the volume of misleading content on large platforms.

The case

Researchers from the Universities of Cambridge and Bristol, working with Google's Jigsaw unit, made five short animated videos that each explain one manipulation technique: emotionally manipulative language, incoherence, false dichotomies, scapegoating and ad hominem attacks. Six preregistered randomized studies with 6,464 people found the videos improved people's ability to spot these techniques and to judge the trustworthiness of content. In a 2022 field test, two videos ran as YouTube ads; a random subset of viewers and a control group then answered a headline question within 24 hours (n = 22,632). Viewers were about 5% better at recognising the technique. The results were published in Science Advances in August 2022.

Why it worked

Teaching the technique rather than rebutting each false claim scales cheaply through ad slots; the real-world effect was small per person and fell just short of the authors' preset target, so reach and repeat exposure matter.

Sources