Abstract
To better understand racial identity claims animating Canadian politics, this paper examines Twitter user activity and bio descriptions from two episodes in the lengthy history of #TrudeauMustGo. The first sample is drawn from 643,669 Tweets gathered between August 29 and October 22, 2019 using the Twitter Search API during the Canadian election campaign. The second sample is drawn from 249,046 Tweets gathered via the commercial data provider Meltwater between June 1, 2024 and March 1, 2025. The results of inductive qualitative and mixed methods research are presented through the lens of a machine-learning (ML) model based on 3.7 million labels applied to political Twitter user bios over five years. The core ML model is a binary: "Promoting Trump" versus "Not Promoting Trump" first developed using Canadian election Twitter data in 2019.
Supplementary weblinks
Title
DiscoverText Explained
Description
A multilingual text analytics and machine learning platform. This collection has our 30-second explainer video in 4 languages as well as other introductory lessons, use-case testimonials, background theory, and a very important argument about when to use, or not use, spreadsheets in Twitter research.
Actions
View