A first shot would be a look at the introduction of digital media in particular places or times. Some people have done this with the internet by using e.g. distance to an ADSL exchange as an instrument.
Or, a randomized controlled trial. If it has an effect worth caring about, then it's worth running an experiment on. A real positive or negative effect would be a big deal for policy.
Or, a randomized controlled trial. If it has an effect worth caring about, then it's worth running an experiment on. A real positive or negative effect would be a big deal for policy.