The meaning of climate change in American politics: an embedding regression analysis of U.S. politicians on Twitter

Abstract

This study employs word embedding regression to examine the meaning of climate change in U.S. congressional discourse on Twitter from 2015–2022. By analyzing semantic variations across partisanship, district CO2 emissions, climate vulnerability, and constituent education levels, it offers novel insights into political communication on this issue. Results reveal less partisan polarization around climate change compared to contentious issues like abortion, guns, and immigration, with polarization decreasing in recent years. Notably, congressional representatives show greater polarization on climate change meaning in districts with fewer per capita emissions or higher shares of college graduates. This computational approach allows for systematic investigation of semantic meaning across social groups and time, contributing to our knowledge of how politicians frame climate change. The findings challenge common assumptions about partisan divides on climate issues and highlight the complex interplay between local factors and political discourse, offering implications for climate communication strategies and policy development.