Assessing the differential impact of online censorship on offline protest and repression

Governments worldwide suppress access to information, often during times of contentious politics, such as elections or social unrest. In the digital age, many governments have developed sophisticated forms of online censorship including filtering, blocking or throttling access to the internet. Political scientists started to examine offline consequences of online censorship, notably whether and to what extent censorship affects protest behavior and repression. To date, however, evidence largely relies on small-scale case studies that remain limited in their possibility to offer generalizable answers. In this paper, we systematically assess how different forms of online censorship affect offline repression and protest behavior. Theoretically, we expect to see an increase in violent repression offline when governments implement full blocking of internet access, whereas we expect internet filtering to be accompanied by legal forms of repression. We further expect that citizens respond to both filtering and blocking of internet access with increased protest in the short run, yet, protest is less likely to endure during events of full internet blocking. Empirically, we use large-scale Virtual Network Provider (VPN) data for African countries from 2017 to 2021 to infer periods of temporary filtering and blocking of access to the internet. We then employ duration models to assess how sudden filtering or blocking events affect state repression and protest behavior in the days and weeks following the event. Overall, this paper offers a novel way to measure different forms of online censorship and a systematic assessment of their consequences for offline repression and protest.