But Now What: Can We Still Use Twitter/X Data for Political Research?

Throughout the recent decades, especially text and interaction data from various social media platforms (Facebook, Twitter/X) allowed researchers to offer better tests of core theories of political communication and behaviour and also develop new theories focused on representation, agenda setting, and political dialogue. This period was marked by extensive debates regarding how representative social media data are, which politicians use it and how, and whether we can benchmark these platforms to other campaign platforms, ultimately trying to understand opportunities and limitations of social media data for political science. Recently, following a change in ownership and subsequent loosening of moderation policies, one of the main platforms used in political research, Twitter/X, was flooded by extremist views formerly associated with the fringes of the political spectrum. This led to a substantial loss of revenue, with major companies stopping advertising on the platform. It is still unclear if this impacted how politicians use the platform and the interactions on it. Capitalizing on large and comparable data collection of Twitter/X activity of candidates for the 2014, 2019, and 2024 European Parliament (EP) elections, we use the same principles and benchmarks that were used when social media research was on the rise, and we ask if we can still learn anything from social media data for politics, and if so, how. We investigate the pattern of use (i.e., presence, number of posts, and number of followers), the content (issue coverage), and engagement with the communication of returning candidates (within-individual design). We also analyse overall changes across the three elections using a between-individual design, with a special focus on candidate characteristics (e.g., gender) and party characteristics. Substantively, we provide quantitative evidence corroborating anecdotal evidence suggesting that the content and user profile notably shifted towards the extreme right of the political spectrum. While these changes helped far-right politicians increase their notoriety, we show that mainstream politicians have been disincentivized from using the platform both for demand reasons (lack of broad audiences) and quality of interactions (with increased abuse).