Introduction
In political communication, "framing" refers to the strategic selection of specific aspects of a reality to make them more salient, thereby promoting a particular problem definition or moral evaluation. Recent research from the University of Rochester investigates the "framing gap" between traditional news media and social media users regarding three landmark 2023 U.S. Supreme Court decisions: affirmative action (Students for Fair Admissions v. Harvard), student loan forgiveness (Biden v. Nebraska), and abortion rights (Dobbs v. Jackson).
This study addresses a critical gap in our understanding of how divergent narratives can lead to disjointed public understanding, fuel political polarization, and ultimately sway policy-making. By comparing the linguistic frames used by journalists versus those used by the general public on Reddit, the authors highlight a significant divide in how legal controversies are consumed and discussed.
Methodology
The researchers employed a multi-stage computational pipeline to analyze a corpus of 4,784 news headlines (categorized by political leaning via AllSides) and 41,143 Reddit comments collected between December 2022 and August 2023.
Framing Detection
The team utilized a multi-label classification model based on the RoBERTa architecture. This model was trained to categorize text into two narrative frames (Episodic vs. Thematic) and 14 "Issue-Generic" frames, such as Economics, Morality, and Legality. The full taxonomy of these frames is provided below:
Stance Classification
To determine if a text was "for," "against," or "neutral" regarding a ruling, the researchers leveraged ChatGPT (gpt-3.5-turbo) in a zero-shot setting. They developed specific prompts to ensure the model treated news titles and Reddit comments appropriately for sentiment analysis.
Statistical Analysis
The study utilized log frequency ratios to compare frame usage and a Z-test (with Bonferroni correction) to ensure statistical significance. Additionally, K-means++ clustering was used to calculate "inertia," representing the level of disagreement or diversity in framing within a specific group.
Key Insights
Narrative Divergence: Episodic vs. Thematic
A primary finding is the stark contrast in narrative style. News media is predominantly episodic, focusing on the immediate "who, what, and where" of specific court rulings. Conversely, Reddit users engage in thematic framing, placing rulings within broader social contexts and discussing long-term societal implications.
The "Neutrality" Gap and Stance Polarization
There is a significant difference in stance between the two platforms. For example, in the context of affirmative action, news headlines were 72% neutral, while Reddit comments were only 6% neutral. Reddit users were overwhelmingly vocal and oppositional, with 65% taking an "against" stance compared to only 19% of news outlets.
Personal Impact vs. Legal Process
Reddit users frequently employed Quality of Life and Fairness and Equality frames, focusing on how rulings affect individuals' daily lives. News outlets leaned heavily on Legality, Constitutionality, and Jurisdiction, treating the issues as procedural or legal puzzles rather than human-centric events.
Results and Discussion
- Diversity of Thought: The "inertia" (disagreement) in framing was consistently higher on Reddit than in news media. On affirmative action, Reddit's inertia was 1.15, compared to 0.71 for news media.
- Consensus vs. Conflict: Student loan forgiveness showed the highest degree of consensus across platforms. In contrast, affirmative action and abortion showed the most extreme divergence in both stance and framing.
- Internal Media Polarization: While news outlets are more neutral than social media, they remain internally polarized. Right-leaning media strongly opposed affirmative action and abortion rulings, while Left-leaning media focused more on Health and Safety and Policy Prescription.
Limitations and Future Work
The authors acknowledge that classification performance was lower for Reddit comments (F1 ~0.53) than for news titles (F1 ~0.75), reflecting the complexity of informal, sarcastic social media text. Future research may expand to other platforms like X/Twitter or TikTok to investigate the "direction of influence"—whether social media framing eventually forces news outlets to shift their narratives, or vice versa.
Summary updated