Ideological Media Markets, Cross-Pressured Voters, and Spatial Voting: A Computational and Experimental Approach

Abstract

Cross-pressured voters (CPVs) lack major parties that represent them on both the economic and cultural ideological dimensions. CPVs consist of two groups -- economically-leftist and culturally-conservative voters ("left-conservatives") and economically-rightist and culturally-progressive voters ("right-progressives"). Despite being ideological opposites, both groups usually support right-conservative parties. Past research has explained this puzzle by pointing to party messaging and differences in salience. My dissertation builds on these explanations by examining the role of media in shaping three "spatial voting mediators" -- salience, ideological positions, and especially perceived party positions -- to alter CPV voting behavior. Since CPVs have ideological reasons to support both left-progressive and right-conservative parties, media can play a key role in pushing them in either direction.

To this end, my dissertation performs two main tasks. First, I examine the supply side of media ideology to determine whether there are ideological gaps for CPVs -- that is, a relative paucity of left-conservative and right-progressive outlets. I develop a novel way to measure media ideology using embeddings and large language models (Paper 2), crawl an original dataset of one million randomly selected online media articles, and test for ideological gaps (Paper 3). Indeed, I find persistent gaps in the ideological spaces that CPVs inhabit -- the left-conservative and right-progressive quadrants. This means that CPVs must often select outlets that are incongruent on either the economic or cultural dimension, which highlights necessary extensions to the literature on selective exposure.

Second, I examine the effects of these media supply gaps on CPVs. I use an original survey with six experiments (including two factorials), three panel datasets, and a quasi-experiment to test how CPVs select political media outlets and content (Paper 4) and how this media exposure influences their spatial voting mediators and electoral behavior (Papers 5 and 6). My results indicate that both left-conservatives and right-progressives select more right-conservative media and that this media exposure increases their support for right-conservative parties, especially by altering their perceptions of party positions. Since CPVs are a large and growing share of many electorates, this research could be consequential for understanding media effects, shifting cleavages, and recent spikes in electoral volatility.