Sorry for the misunderstanding! Yeah, it looks kind of hard to distinguish. Maybe someone has a clever method. The only thing I can immediately think of is an RCT on follow-up to different bits of media coverage, but I expect this would be super-messy to run and might not produce great data.
Sorry for the misunderstanding! Yeah, it looks kind of hard to distinguish. Maybe someone has a clever method. The only thing I can immediately think of is an RCT on follow-up to different bits of media coverage, but I expect this would be super-messy to run and might not produce great data.