🤝 If you are also working on the application of behavioural science in the public or private sector and want to exchange thoughts, please do connect – and if you are (ever) in the @harvard.edu‬ / Cambridge / Boston area, do reach out for lunch or coffee!

🏆 The Shorenstein Center is known for, among other things, hosting the HKS Misinformation Review and the Harvard Behavioral Insights Group. The Center also hosts a wide range of practitioners and researchers from the worlds of media and governance.

📜 After 7 years of fundamental research on the behavioural science of misinformation, it is time for me to focus more on applying my insights. I will be spending the Fall Semester reflecting on how I can best put my research insights into policy solutions, preparing for my post-academia career.

For an overview of the findings, you can find a more detailed thread here: bsky.app/profile/rako...

For a more detailed scientific summary of the findings, you can find the Bluesky thread here: bsky.app/profile/rako...

All materials, datasets, and scripts can be found on our OSF repository at: doi.org/10.17605/OSF...

I would also like to thank my brilliant co-authors:

Jon Roozenbeek
@jonsimons.bsky.social
@lewan.bsky.social
Vanessa Maturo
@bgoldberg.bsky.social
Rachel Xu
@profsanderlinden.bsky.social

(10/10)

See pp. 49–58 of the supplement for discussion of:

1) The memory model of inoculation & how it advances the field
2) Booster implementation
3) Practical implications for practitioners

Also check:

Capewell et al. (2024)
doi.org/10.1111/jasp...

Leder et al. (2024)
doi.org/10.1037/xge0...

(9/10)

Impact conclusion:

1) Inoculation works well, but effects decay fast!
2) We _need_ "booster shots" to be able to speak of an *effective* intervention

What constitutes a good booster?

- Testing/quizzing people (applying knowledge)
- Repeating the intervention (in full or shortened form)

(8/10)

The inoculation effects also seemed to work well across subgroups based on demographics, political views, and a range of psychological traits.

(7/10)

Using dominance analysis we tested a range of variables traditionally linked to inoculation effectiveness

We find similar results across text-based, gamified, and video-based inoc:

=> objective memory of what was learned in the intervention is by far the best predictor of inoc longevity

(6/10)

In each study we tested memory & motivation:

1) Memory (objective recall) = predictor of inoc longevity
2) Motivation helps with discernment & memory formation, but not with predicting longevity
3) Memory-boosters worked, motivation-boosters not

=> evidence for Memory-Motivation Model

(5/10)

The key insights are similar across the 5 studies:

1) If no booster, much effect decays within 1–2 weeks
2) With memory booster at ~10 days, decay slows down, with effects up to 1 month
3) Motivation does not seem to be linked to longevity, while memory was an excellent predictor

(4/10)

In five preregistered longitudinal experiments (US; N = 11,759), we explored the inoculation longevity questions in three inoculation paradigms:

1) Text-based (perceived scientific consensus)
2) Gamified (misinformation reliability ratings)
3) Video-based (manipulativeness discernment)

(3/10)

Inoculation has been studied for >60 years. 3 key questions remained unclear:

1) How long do inoc effects last
2) Why do inoc effects dissipate over time
3) How can we boost inoc effects

We defined & tested 3 models, centered around 2 key variables:

a) Motivation
b) Memory

(Model C won)

(2/10)

That's it for the Misinformation and Belief Science preconference at #SPSP2025. @rakoenmaertens.bsky.social and I are so proud of what this conference has become and that it will continue without us in future years. 21/