Pawn or King : Does Political Candidates' Use of Twitter Tell us Anything About the Likelihood of Electoral Success for Political Parties?

Twitter is now an unavoidable communication platform for most politicians. Data generated by their online activities allow researchers to scrutinize in great detail the different politicians’ messages and interactions. While a lot of scholarly attention is given to the study of politicians’ activities on Twitter, little has been written on whether their inactivity can also be meaningful. This article examines the use of Twitter by 1745 candidates during the 2021 Canadian federal election campaign. It aims at exploring whether inactivity on Twitter is associated with low potential of winning. Riding-level data on each parties’ past results are used to measure the probability of winning of each candidate. Statistical models are then used to study the relationship between the chance of winning and the lack of intensity, or absence, of Twitter activity collected using the platform’s API. Results show indeed a clear correlation, which suggests that even politicians’ or parties’ silence on social media could be meaningful and reveals information on their own evaluation of their chance of winning. From a broader perspective, by focusing on social media inactivity, this research sheds new light on the inferential value and limitations of social media data for political science.