The Drivers of Polarity in Sentiments on Social Media: an Exploratory Study on the 2021 Canadian Federal Election

The drivers of polarity in sentiments on social media - an exploratory study on the 2021 Canadian Federal Election Hiba Mohammad Noor Master of Science in Management, 2022 Master of Science in Management, Ryerson University Social media is used by the public, voters, and politicians to share their political opinions leading to online political discourse. The opinions shared by voters on social media have different sentiments associated with them depending on voter needs and priorities. Understanding the factors that drive these sentiments can help policymakers and other political stakeholders to understand voter needs and expectations and develop policies that align with those needs. This research focuses on identifying the factors (keywords) that drive these sentiments. This research also investigates the relationship between these keywords and the number of retweets and hashtags. Sentiment analysis was performed on 779,169 tweets related to the 2021 Canadian Federal election followed by text clustering and keywords analysis. The topics and keywords that drive the sentiments were identified. Chi-Square test was used to investigate the relationship between these keywords, hashtags, and the number of retweets. The results suggest that some keywords are common in opposite sentiment types (positive and negative) which shows polarization in Twitter and some keywords are unique to a sentiment type which shows that these keywords drive that specific sentiment. The results also suggest that there is no significant relationship between the keywords and the number of hashtags but has a significant relationship between the keywords and the number of retweets for extremely negative tweets only.