This website uses cookies.

By clicking the "Accept" button or continuing to browse our site, you agree to first-party and session-only cookies being stored on your device to enhance site navigation and analyze site performance and traffic. For more information on our use of cookies, please see our Privacy Policy.

  1. Home
  2. ASSA Annual Meeting
  3. 2024
  4. Program
  5. Media and Perceptions
« Back to Results

Media and Perceptions

Paper Session

Sunday, Jan. 7, 2024 8:00 AM - 10:00 AM (CST)

Convention Center, 221D
Hosted By: American Economic Association
  • Chair: Milena Djourelova, Cornell University

Toxic Content and User Engagement on Social Media: Evidence from a Field Experiment

George Beknazar-Yuzbashev
,
Columbia University
Rafael Jiménez-Durán
,
University of Chicago
Mateusz Stalinski
,
University of Chicago

Abstract

As much as forty percent of social media users have been harassed online, but there is scarce causal evidence of how toxic content impacts user engagement and whether it is contagious. In a pre-registered field experiment, we recruited participants to install a browser extension, and randomly assigned them to either a treatment group where the extension automatically hides toxic text content on Facebook, Twitter, and YouTube, or to a control group without hiding. As the first stage, 6.6% of the content displayed to users was classified as toxic by the extension relying on state-of-the-art toxicity detection tools, and duly hidden in the treatment group during a six-week long period. Lowering exposure to toxicity reduced content consumption on Facebook by 23% relative to the mean – beyond the mechanical effect of our intervention. We also report a 9.2% drop in ad consumption on Twitter (relative to the mean), where this metric is available. Additionally, the intervention reduced the average toxicity of content posted by users on Facebook and Twitter, evidence of toxicity being contagious. Taken together, our results suggest a trade-off faced by platforms: they can curb users’ toxicity at the expense of their content consumption.

Experience, Narratives, and Climate Change Beliefs: Evidence from Extreme Weather Events

Milena Djourelova
,
Cornell University
Ruben Durante
,
Pompeu Fabra University
Elliot Motte
,
Pompeu Fabra University
Eleonora Patacchini
,
Cornell University

Abstract

We study the media discourse and public opinion on climate change in the aftermath of extreme weather events. In both cable news and local media, and for both national-interest and local events, we find that left-leaning media consistently increase their coverage of climate change in the aftermath of natural disasters, while conservative media do not, despite equal disaster-related coverage. We then link the experience of disasters to concerns about climate change expressed in large-scale electoral surveys. We find a polarizing effect: disaster experience increases concerns about climate change and support for environmental policies among liberal respondents, but has the opposite effect on conservative respondents. Both effects are driven by areas where the ideology of the respondent is in the minority.

Officer-Involved: The Media Language of Police Killings

Bocar Ba
,
Duke University
Patrick Bayer
,
Duke University
Aurelie Ouss
,
University of Pennsylvania
Jonathan Moreno-Medina
,
University of Texas-San Antonio

Abstract

This paper studies the language used in television news broadcasts to describe police killings in the United States from 2013-19. We begin by documenting that the media is significantly more likely to use several language structures - e.g., passive voice, nominalization, intransitive verbs - that obfuscate responsibility for police killings compared to civilian homicides. We next use an online experiment to test whether these language differences matter. Participants are less likely to hold a police officer morally responsible for a killing and to demand penalties after reading a story that uses obfuscatory language. In the experiment, the language used in the story matters more when the decedent is not reported to be armed, prompting a final research question: is media obfuscation more common in high leverage circumstances, when the public might be more inclined to judge the police harshly? Returning to the news data, we find that news broadcasts are indeed especially likely to use obfuscatory language structures when the decedent was unarmed or when body camera video is available. Through this important case study, our paper highlights the importance of incorporating the semantic structure of language, in addition to the amount and slant of coverage, in analyses of how the media shapes perceptions.

Measuring, Tracking and Analyzing Inequality using Social Media

Carlo Schwarz
,
Bocconi University
Paul Bose
,
Bocconi University
Dirk Hovy
,
Bocconi University
Lorenzo Lupo
,
Bocconi University

Abstract

The MENTALISM project develops predictive models for the measurement of inequality based on social media data. We show that it is possible to create representative, high-frequency measures of inequality at the municipality level for an entire country (Italy) using social media data for the near universe of Italian Twitter users. Our deep-learning-based NLP model analyzes the text and language of social media content to provide real-time tracking of attitudes, concerns, opinions, and grievances surrounding a plethora of inequality dimensions (e.g., economic or gender inequality). The advantage of our approach is that it can create measures of inequality even in situations where survey or administrative data are not available at the high time-frequency or fine geographic granularity. Based on these measures, we provide a comprehensive analysis of the determinants and inequality shocks over the last decade.

Discussant(s)
Lisa George
,
CUNY
Elliott Ash
,
ETH Zurich
Rafael Jiménez-Durán
,
University of Chicago
Jonathan Moreno-Medina
,
University of Texas-San Antonio
JEL Classifications
  • D0 - General
  • L8 - Industry Studies: Services