Echo Chambers After the Ban: Political Information on X in Brazil

class: center, middle, inverse, title-slide

.title[
# Echo Chambers After the Ban: Political Information on X in Brazil 
]
.author[
### Tiago Ventura , Christopher Barrie, Margaret E. Roberts, Christopher Schwarz, Joshua A. Tucker 
]
.date[
### APSA, 09/11/2025 
]

---

layout: true
<div class="my-footer">Tiago Ventura (Georgetown University) &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp &nbsp APSA 2025</div>

---
class:middle

## Motivation

Democratic governments often exert power over the governance of social media platforms. In recent years, these conflicts have sometimes escalated to full-scale block.

- Block of Russian website VKontakte in Ukraine; 
- India and Nepal have blocked access to TikTok;
- US bill to sell TikTok;

Despite this increasing trend, we know little about the effects of these - often temporary -  blocks on the information environment in democracies and how voters react to them:

- Information control in authoritarian contexts .midgrey[(Roberts, 2018;  Boxell and Steinert-Threlkeld, 2022; Pan and Siegel, 2020; Lutscher, 2023)]

--
   -  .midgrey[Golovchenko (2022)]: looks at the block of VKontakte in Ukraine in 2017, and finds no partisan effects on compliance with the block

---
class: middle
## Research Questions

- **RQ1:** How do partisan dynamics shape compliance with a state-imposed platform bans to social media platforms?

- **RQ2:** What are the consequences for information environments in a polarized democracy?

---
class: no-footer
background-image: url("motivation_tw.jpg")
background-size: cover
background-position: center

---
class: middle

# Design and Methods

- **Use X/Twitter data from Brazil from months before and after the Ban**

- **Pre-Ban Sharing Data:** 90 days of data, starting before August 30, via the X Decahose API. 
     - Use this data to estimate the ideology of media organizations and politically engaged users in Brazil 
--

- **Ideology Estimation:** News sharing scores .midgrey[(Eady et. al 2024)]
    - Columns: Political news organizations ~ 242 domains
    - Rows: Politically engaged users who share news often (>5 unique)~ 9061

- **Users Timeline:** Scrappe the timelines for these politically engaged users (~ 7,000 users)

---
class: middle, center, inverse

# Validation

---
class: middle, center
## Top-20 Domains: Ideology Score

---
class: middle, center
## Validating with survey data

---

class: middle, center, inverse
# Results

---
class: middle, center
## The Effects of the Ban on Posting Frequency

<img src="output_9k/fig_tweets_sharing.png" width="100%" />
---
class: middle

## Event Study Estimates: User posts

---
class:center, middle

## Event Study Estimates: News Sharing

<img src="output_9k/event_study_domains_ideology_poisson.png" width="100%" />
]

---
class: middle, center

## Tweets Posted Conditional on Users' Ideological Position

---
class: middle, center
## Concentration of Tweet Activity and News Sharing

<img src="output_9k/fig_cdf_user.png" width="100%" />
---
class: middle, center
## Concentration of Tweet Engagement
<img src="output_9k/cdf_engagement.png" width="100%" />

---
class: middle, center
## Changes Informational Environment

<img src="output_9k/fig_domain_ideology.png" width="100%" />
---
class:middle
## Conclusion

The ban substantially reduced use of the platform. **But** compliance with the ban was strongly driven by partisan dynamics

--
- Right-leaning users were the primary group to remain active
   
- After the ban: 
       
    - the information environment on X was still more conservative, 
    - right-leaning users more active, 
    - tweeting more frequently, and receiving a higher share of the engagement

- Bans in polarized society can backfire, increase fragmentation and create platform echo-chambers

- Even short bans can have long-lasting effects on information environments.

- Future research: multi-platform effects, user migration, separate effects on political elites.

--
---
class: middle, inverse, center

# Thank you!