Framing of Culture War Issues in Congressional Campaign Websites

• Activity Number: 417 - SPEED: Methodological Developments in Social Statistics, Part 2
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 2:45 PM
• Sponsor: Social Statistics Section
• Author(s): Jack Wolf* and Christopher Chapp and My Khe Nguyen and Paul Roback and Jessica Whittenburg
• Companies: St. Olaf College and St. Olaf College and St. Olaf College and St. Olaf College and St. Olaf College
• Keywords: topic modeling; natural language processing; text modeling

Abstract

Research shows that politicians strategically frame issues to gain public support for their views and affect voters’ opinions on issues. Previous research done on issue framing has two primary limitations: first, the labor-intensive nature of manually coding documents, and second, the necessity of determining all possible frames of an issue. We use a unique dataset of 2008–2018 House of Representative candidates’ campaign website pages to examine framing of culture-war issues such as LGBT+ rights and abortion through topic modeling and dictionary based methods. Through the R package stm, we identify underlying topics in our data without pre-specifying what they may be. Additionally, we apply the Moral Foundations and Linguistic Inquiry and Word Count dictionaries to score pages based on predetermined categories. We use mixed-effect models to examine how party affiliation, incumbency status, congressional district characteristics, and time affect framing. Our results provide insight into how politicians frame issues and contribute to existing methods for topic modeling.