Echo Chambers: Online Segregation, Mechanisms, and Consequences

Introduction

The internet and social media dramatically transformed the news landscape. The abundance of online news options, recommendations of social media algorithms, and changes in news production may have segregated individuals into echo chambers, environments where people mostly consume like-minded news. Such echo chambers could exacerbate polarization, distort perceptions of reality, and threaten democracy.

Research Gap

However, a gap remains between these grave concerns and the limited evidence on echo chambers. A critical challenge is that research is typically conducted at the outlet level. Focusing on outlets rather than individual articles may underestimate segregation as individuals today consume specific articles from many outlets.

Project Overview

This project will analyze the demand and supply of news at the article level to provide new insights regarding the origins, extent, and implications of online echo chambers. The project consists of three parts.

Part One: Dataset Creation

1. I will create a new dataset of the slant (political leaning) of millions of articles using high-resolution data, expert ratings, and advances in large language models.
2. I will use this dataset to provide the first estimate of online segregation based on the slant of articles.

Part Two: Audience Analysis

1. Instead of fixing the set of articles and analyzing consumer behavior, I will fix the audience and analyze how outlets tailor articles to their consumers.
2. I will investigate whether this tailoring increases segregation.
3. I will complement the descriptive estimates of how outlets distribute news with a causal analysis of whether the internet and social media affect the news that outlets produce.

Part Three: Willingness to Pay

1. I will elicit individuals’ willingness to pay for various articles.
2. I will causally estimate how the articles people typically avoid affect their attitudes when they are consumed.
3. I will use these estimates to decompose the relative importance of two theories for how news polarizes attitudes:
   • Differences in preference for like-minded news
   • Heterogeneity in the effects of news.