Crucially, we show when foreign propaganda appears most consequential: where institutional trust is low and integration weaker. Effects are strongest among cohorts partly schooled under Soviet rule, and we find no comparable pattern in Estonia or Lithuania.

Our design exploits two disruptions to Western vaccines that Russian state-backed media used to promote Sputnik V and attack Western institutions. After each, traffic shifted toward pro-Kremlin outlets and vaccination fell more where the ethnic Russian share was larger.

Now that I have more time, a bit more on the paper... We study when Russian state-sponsored misinformation/propaganda changes not just beliefs, but consequential behavior. We focus on Russian-speaking minorities in Latvia during the COVID vaccine rollout.

Our contribution: social media does not have a single effect on polarization. Its political consequences depend on who is speaking, whose messages are amplified, and how diverse users’ networks are—helping reconcile seemingly conflicting findings in the social media literature.

What matters is not simply access to social media, but the political exposure it creates. We find that gains from polarizing rhetoric are concentrated where online networks amplify one political voice—and weaken where networks expose voters to more diverse views.

So nice to see this with @jonnekamphorst.bsky.social in press!