In early 2021 the United States Capitol in Washington was stormed during a riot and violent attack. A similar storming occurred in Brazil in 2023. Although both attacks were instances in longer sequences of events, these have provided a testimony for many observers who had claimed that online actions, including the propagation of disinformation, have offline consequences. Soon after, a number of papers have been published about the relation between online disinformation and offline violence, among other related relations. Hitherto, the effects upon political protests have been unexplored. This paper thus evaluates such effects with a time series cross-sectional sample of 125 countries in a period between 2000 and 2019. The results are mixed. Based on Bayesian multi-level regression modeling, (i) there indeed is an effect between online disinformation and offline protests, but the effect is partially meditated by political polarization. The results are clearer in a sample of countries belonging to the European Economic Area. With this sample, (ii) offline protest counts increase from online disinformation disseminated by domestic governments, political parties, and politicians as well as by foreign governments. Furthermore, (iii) Internet shutdowns tend to decrease the counts, although, paradoxically, the absence of governmental online monitoring of social media tends to also decrease these. With these results, the paper contributes to the blossoming disinformation research by modeling the impact of disinformation upon offline phenomenon. The contribution is important due to the various policy measures planned or already enacted.

翻译：摘要：2021年初，华盛顿美国国会大厦在骚乱和暴力袭击中被冲击。2023年巴西也发生了类似冲击事件。尽管这两次袭击都是更长时间序列中的事件，但它们为许多观察者提供了佐证，这些人曾声称包括虚假信息传播在内的在线行为会产生线下后果。此后不久，大量论文围绕在线虚假信息与线下暴力等关系展开研究。然而，迄今为止，虚假信息对政治抗议的影响尚未得到探讨。因此，本文利用2000年至2019年间125个国家的时序截面样本评估了这种影响。研究结果复杂多样。基于贝叶斯多层次回归模型：(i)在线虚假信息与线下抗议之间确实存在关联，但这种关联部分受到政治极化的中介作用。在属于欧洲经济区的国家样本中，结果更为清晰。基于此样本，(ii)国内政府、政党、政客以及外国政府传播的在线虚假信息增加了线下抗议次数。此外，(iii)网络关闭倾向于减少抗议次数，但矛盾的是，政府缺乏对社交媒体的在线监控也倾向于减少这些抗议次数。基于这些结果，本文通过建模虚假信息对线下现象的影响，为蓬勃发展的虚假信息研究做出了贡献。鉴于已规划或已实施的各种政策措施，这一贡献具有重要意义。