WhatsApp increases users’ exposure to false rumors but has limited effects on political attitudes: Evidence from a deactivation experiment.

.title[
# <span style="font-size: 70%;"> WhatsApp increases users’ exposure to false rumors but has limited effects on political attitudes:</span> <span style="font-size: 70%;">Evidence from a deactivation experiment.</span>
]
.author[
### <span style="font-size: 60%;">Tiago Ventura, Rajeshwari Majumdar, Jonathan Nagler, and Joshua A. Tucker </span>
]
.date[
### <span style="font-size: 60%;"> Social Media & Democracy Conference <br> 03/10/2022 </span>
]

---

<div class="my-footer"><span>Social Media & Democracy, 09/03/2023</span></div>

---

## Social Media: The misinformation turn

---
## An Emerging Field

---
class:middle

## What we know:

- Externally-valid causal claims are hard, often small in size, and context-dependent.

- Most of the evidence come from a few countries (advanced democracies).

- And from a few platforms (mostly Twitter, and to some extent Facebook).

---
class:middle

## For years, WhatsApp has been the main social media applicattion in many Global South democracies.

---
class:middle

.panelset.sideways[

#### .fade[WhatsApp is a worldwide popular end-to-end encrypted messaging app allowing both direct and ingroup communications with close to 2 billion around the world]

]

- .red[66% of the Brazilian eligible voters] has a social media account, and .red[65% have an WhatsApp account] ([Datafolha 2018](http://media.folha.uol.com.br/datafolha/2018/10/27/44cc2204230d2fd45e18b039ee8c07a6.pdf)) – 150 million people in 2022.

- Most used app for all purposes: .red[talk to family & friends, do business & pay bills, consume news, talk politics, among others] ([Reuters Institute, 2021](https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2021)).

- Used by .red[48% of population to read news], despite having no “news feed” ([Reuters Institute, 2021](https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2021))

- Across the world, .red[Brazilians are the most concerned about misinformation]. And among the most active social media users

- And .red[WhatsApp is the main social media application] where users report to see more misinformation.

]
]

---

---
class: middle, center, inverse

### Even though WhatsApp has received a huge attention from the media and policy makers, no study has so far identified causal effects of WhatsApp on political outcomes.

---
class:middle, inverse, center

## Field Experiment: An WhatsApp Multi-Media Deactivation during the Brazilian Election

---
class:middle

## Motivation

Deactivation experiments have been recently used to causally identify the effects of social media on users' behavior and attitudes .red[(Asimovic et al 2021 and Alcott et al 2020)].

#### Deactivating WhatsApp users is **theoretically and empirically unfeasible**.

- Costs of deactivating on human subjects are too high.

- Findings from subjects enrolled in a complete deactivation were unlikely to be feasible (issues on external validity)

---
class: middle

## Our Design

- Disable their automatic download of media on WhatsApp 
- Do not purposefully click and watch any media on WhatsApp

- **Start:** Three weeks before general election in Brazil
- **End:** three days after the voting day

---
## Overview of the Experiment

---
.center[
<img src="output/recruitment.png" width="55%" />
]

---
.center[
<img src="output/filters.png" width="70%" />
]

---
.center[
<img src="output/deactivation.png" width="55%" />
]

---
<br>
.center[
<img src="output/iphone_deact.jpeg" width="30%" />
]

---
.center[
<img src="output/compliance.png" width="55%" />
]

---
<br>
.center[
<img src="output/iphone_compliance.jpeg" width="30%" />
]

---
## Post-Treatment

---
class:middle

## Hypotheses

H1: Deactivated users will report lower levels of previous exposure to FALSE rumors.

H2a: Deactivated  will display a higher ability to accurately identify FALSE rumors.

H2b Deactivated users will display lower levels of mainstream news knowledge.

H3: Deactivated will display lower levels of outgroup political polarization.

H4: Deactivated will display higher levels of the aggregated index of subjective well-being.

---
class: center, middle, inverse

# Results

---

### Effects on WhatsApp usage

---
### Effects on Exposure to Misinformation Rumors

---
### False > True reduction in exposure

---
### Effects on Belief Accuracy

---
### Effects on Polarization & Subjective Well-Being

---
class: middle

## Discussion

---
class:middle, inverse

#  BUT...

---

class:middle

### It would be a mistake to conclude that WhatsApp plays no role in politics.

- Scope conditions: focus on direct effects of WhatsApp.

- Do not rule out the use of WhatsApp as a mobilization and organizational tool by malicious actors.

- Largers effect might exist exactly in hard to reach populations.

- Small effect size vs Power

- Electoral context + Motivated Reasoning

---
class: middle

## Next steps.

### Replication, Replication, Replication

**Multi-Country Study**

- Contextual variation

- Partial vs full deactivation

- In and out of election periods

---

---
class:middle, center

## Thank you!

---
class: middle, center, inverse

## Additional Analyses

---
### Exposure and Partisanship

---
### Belief Accuracy Conditional on WhatsApp usage for Politics

---
### Polarization Outcomes

---
class:middle

## Examples of Misinformation Outcomes

### False Item:

_Only complete votes are counted by the electoral justice. If the voter only votes for the President, and votes blank for all the other races, the vote is considered a partial vote, and will be annulled_

### True Item:

_After consecutive reductions, the current oil price in Brazil is below the global average value_