The evolution and persistence of echo chambers over time and across topics: German Covid-19 and Russo-Ukrainian War conversations on Twitter

The literature on echo chambers hosts a wealth of discussions on the absence or presence of cross-cutting exposure among homogeneous groups forming online. In terms of the former, echo chamber effects are likely to manifest when controversial, conflicting, and polarizing topics are discussed. For example, the handling of Covid-19 has seen the birth of strongly opposing camps that are either defending safety measures or prioritizing individual freedoms. Thereby, social media, such as Twitter, are prominent tools to vocalize opinions on Covid-19 policies. In Germany, these discussions were particularly salient during the 2021 federal election campaign with the pandemic becoming an electorally decisive issue. The paper revisits the German-language Covid-19 chatter and analyzes echo chamber effects among users commenting on the pandemic. In contrast to previous studies, however, it takes on a time-sensitive approach and a longitudinal perspective. Going beyond questions on the mere existence of echo chambers, we focus on their evolution and potential persistence over time. For this purpose, we constructed retweet-based networks based on 8,629,236 Covid-19-related (re-)tweets, which yielded eight clearly separable clusters. Those were further investigated for cross-community interactions and homophily. The latter was computed using topological properties of the network and an independent estimation of the political leanings of users. We find that the heyday of the political cycle, the German federal election, clearly serves as a stimulus for Covid-19 chatter with temporally stable echo chambers in two cluster. However, echo chambers are not only stable over time but also across topics. To that end, we estimated retweet-based networks for Twitter chatter on the Russo-Ukrainian War and compared the overlap in users with those in the Covid-19 bubbles. According to our findings, there is a remarkable overlap characterized by ideological alignment. We therefore conclude that echo chambers are not issue-specific but enabled, reinforced, and stabilized by crises that may be different in content but similarly polarizing in character.