Environmental Racism In The News: Analyzing Racial Biases in Minority Coverage

Abstract

Black, Indigenous, and people-of-color (BIPOC) communities are disproportionately exposed to polluting factories and environmental hazards. This form of systemic racism - Environmental Racism - is a concept that has gained traction in recent years, especially among grassroots activists and marginalized communities. This thesis aims to explore the portrayal of minority races afflicted by environmental racism in the media. A massive news dataset of over 3.3 million articles is scraped and labeled automatically using a selection of conventional machine learning models and neural networks to create a novel dataset of news articles covering environmental racism. Sentiment analysis techniques reveal that environmental racism articles are found to evoke stronger emotions of fear, disgust, and negativity compared to general news. Word embeddings trained on these datasets find evidence of racial biases, with news stories associating African Americans with high threat and low status stereotypes while Latinos are more often stereotyped as being low status. Analyses extended to compare articles published by media outlets with different political affiliations find differences in the strength of these biases by political affiliation.