Observational analysis identifies slight liberal bias in articles across controversial topics, suggesting influence of journal quality and author backgrounds.
Key Points
A slight liberal bias was observed, with more liberal articles published across various topics, highlighting an intriguing publishing trend.
The dataset included over 20,000 article titles and abstracts focused on 12 controversial topics, providing a broad context for analysis.
Analysis combined human and GPT-supported coding methods to assess bias, revealing that political bias may not be as influential as presumed.
Further exploration of author self-censorship and editorial gatekeeping is essential to fully understand their impact on controversial research publications.