CEU Electronic Theses and Dissertations, 2025
| Author | Baniamerian, Sara |
|---|---|
| Title | Computational Framing Of Abortion Rights: Comparing Reddit With Liberal And Conservative News Coverage Before And After Roe V. Wade'S Reversal |
| Summary | The Dobbs draft leak in 2022 marked a major turning point in U.S. abortion policy, overturning the constitutional right established by Roe v. Wade and sparking heated debate around reproductive rights and abortion access. This research investigates how the issue was framed across traditional media—represented by CNN and Fox News, reflecting two sides of the U.S. political spectrum—and how the public framed it through Reddit posts and comments. The study also explores whether and how emotional tone and framing shifted following the Dobbs leak, and whether political affiliation or platform contributed to increased polarization. To conduct this analysis, I used televised news transcripts and Reddit data from before and after the Dobbs draft leak. Transformer-based Natural Language Processing (NLP) models were applied for sentiment analysis, emotion classification, and embedding-driven topic modeling. Compared to traditional lexicon-based or qualitative methods, these models offer both scale and nuance—enabling the analysis of thousands of posts and transcripts while capturing subtle, context-dependent shifts in framing and emotional tone. The results show clear differences in sentiment and framing across sources. CNN maintained a more positive tone and emphasized healthcare access, reproductive rights, and electoral implications. Fox News presented abortion in relation to political unrest and social critique, while omitting rights-based framing. Reddit displayed more emotionally diverse and morally charged discussions, with frames including bodily autonomy, personal experience, and ideological division. Statistical analysis using Chi-Square and Jensen–Shannon Divergence confirmed that polarization between CNN and Fox News significantly increased after the leak. Traditional media emphasized legal framing, while Reddit users expressed stronger emotional tones. This thesis contributes by demonstrating how political ideology and platform shape discourse on abortion, especially during moments of legal and political change. It also illustrates the value of applying transformer-based NLP to capture complex patterns of framing, sentiment, and emotion across diverse platforms on a highly contested and politically sensitive issue. |
| Supervisor | Fodor, Eva; Kralj Novak, Petra |
| Department | Network Science MSc |
| Full text | https://www.etd.ceu.edu/2025/baniamerian_sara.pdf |
Visit the CEU Library.
© 2007-2025, Central European University