Tweeting, or Posting, that is the question: Measuring the Platform Effect in Political Campaigns on Covid-19 in the UK

Social media has become an integral part of political campaigns nowadays. The activities of politicians and their interactions with the wider public are recorded as digital trace data every day; these data offer enormous opportunities for the study of political communication. Although a considerable amount of works have been conducted on political communication on social network sites, many of them focus on a single platform. Among the platforms, Twitter has been a particularly popular choice among researchers, possibly due to its comparatively open data policy, and this is especially the case after various other platforms severely restricted access to platform data via their Application Programming Interfaces (APIs) in recent years. However, previous works have identified potential sources of bias caused by platform selection, including the effect of platform affordances on users’ behavior, the across-platform discrepancy of user demographics, and the differences in types and scopes of data made available by the platform providers. Using the UK as a case, we investigate to what extent the selection of social media platforms affects the results of political communication on COVID-19. To address the question, we collected all the posts on the official Facebook and Twitter profile of major political parties in the UK in 2019. We then performed commonly adopted analysis in computational social science, including topic identification and sentiment analysis, on the posts from the two platforms separately to identify the discrepancies between the results. This paper has significant implications for the research that use social media data to understand political communication.