Uncovering Media Bias in Eviction Reporting: A Comprehensive Analysis Utilising Sentiment Analysis Framework and Social Media Data

Abstract

This study investigates the prevalence of evictionsin South Africa and examines potential disparitiesbetween traditional media reporting and socialmedia discourse. Employing a sentiment analysisframework, we extend its application to comparethe reporting of evictions in newspaper articles(i.e. conventional media) and Twitter data (i.e.social media). Statistical machine-learningmethods are utilized to predict sentiment scoresfor both types of content, and a chi-square test isemployed to evaluate bias between news articlesand tweets. The test results reveal a significant biasin the sentiment distribution, suggesting that thedissimilarities observed between articles andtweets are not merely coincidental.