How do behaviors of role models in the media affect offline behavior? 📺🎬🎥
Our field experiment in Nigeria 🇳🇬 is out today @ScienceAdvances w/ @reblitt @betsylevyp
Short thread on how we designed a study to answer this question 👇
advances.sciencemag.org/content/5/3/ea…
Our field experiment in Nigeria 🇳🇬 is out today @ScienceAdvances w/ @reblitt @betsylevyp
Short thread on how we designed a study to answer this question 👇
advances.sciencemag.org/content/5/3/ea…
Nollywood, Nigeria’s film industry, is 2nd largest film producer in world after Bollywood & reaches every corner of Africa & beyond
To study effects of media, we teamed up w/ Nollywood filmmakers & produced new feature-length film about corruption starring well-known actors
2/n
To study effects of media, we teamed up w/ Nollywood filmmakers & produced new feature-length film about corruption starring well-known actors
2/n
Seeing a film about corruption could shift behaviors on its own (& we hoped it would!)
To isolate effect of *behavioral modeling*, we made treatment & placebo versions of film. Treatment has 17 min of extra scenes showing actors reporting corruption
3/n
vimeo.com/322907329
To isolate effect of *behavioral modeling*, we made treatment & placebo versions of film. Treatment has 17 min of extra scenes showing actors reporting corruption
3/n
vimeo.com/322907329
We wanted to know the effect of watching media *as people actually consume it*
Nollywood films are sold in small shops, wheelbarrows, market stalls, etc. We worked with team of film distributors to run promotion with local vendors. If you bought a film, you got ours free
4/n
Nollywood films are sold in small shops, wheelbarrows, market stalls, etc. We worked with team of film distributors to run promotion with local vendors. If you bought a film, you got ours free
4/n
Film encouraged viewers to text in their own corruption report
To study how reporting in film affected offline behavior, we randomized which film *version* was distributed to communities. Each version had own text #, so we could match incoming reports to treatment condition
5/n
To study how reporting in film affected offline behavior, we randomized which film *version* was distributed to communities. Each version had own text #, so we could match incoming reports to treatment condition
5/n
We wanted to know whether behavioral modeling shifted perceptions of corruption-related social norms
During film promotion, we asked for film *takers’* phone # & entered them in lottery for a smartphone. Takers didn’t know there were diff versions: packaging looked the same
6/n
During film promotion, we asked for film *takers’* phone # & entered them in lottery for a smartphone. Takers didn’t know there were diff versions: packaging looked the same
6/n
What did we find⁉️
Reports came in from communities who received treatment version of film, but almost none from placebo communities
The treatment film shifted viewer perceptions of norms in their community: that corruption, and anger about corruption, were widespread
7/n
Reports came in from communities who received treatment version of film, but almost none from placebo communities
The treatment film shifted viewer perceptions of norms in their community: that corruption, and anger about corruption, were widespread
7/n
Media, survey instruments, data, and code that we used for study are available via @OSFramework osf.io/9a7h5
We also tested a nudge intervention in a stepped-wedge design. To read more about the full study, see waterofgoldstudy.com
8/n
We also tested a nudge intervention in a stepped-wedge design. To read more about the full study, see waterofgoldstudy.com
8/n
This project would not have been possible without the help of many 👇
Thanks especially to extraordinary research assistance from @RobinGomila
9/9
Thanks especially to extraordinary research assistance from @RobinGomila
https://twitter.com/reblitt/status/1105898081831272448
9/9
• • •
Missing some Tweet in this thread? You can try to
force a refresh
