Countering Deepfake Misinformation Among Low Digital Literacy Populations

How to cite

Ali, Ayesha; Qazi, Ihsan Ayyub; Raza, Agha Ali. 2023. "Countering Deepfake Misinformation Among Low Digital Literacy Populations." AEA RCT Registry. AEARCTR-0012005. https://doi.org/10.1257/rct.12005-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Ayesha Ali* , Ihsan Ayyub Qazi , Agha Ali Raza

* Primary investigator

Registry ID
AEARCTR-0012005
DOI
https://doi.org/10.1257/rct.12005-1.1
First registered
Last updated
Status
On going
Trial dates
2023-09-15 – 2024-02-01
Countries
Private
JEL codes
D83, L86, L82, O10
Registry record
http://www.socialscienceregistry.org/trials/12005

Abstract

The proliferation of misinformation can increase polarization, erode trust in public institutions, and pose significant risks to democracy. This problem is exacerbated in developing countries where social media growth has brought many online, many of whom lack the essential digital skills needed to competently navigate digital spaces. Through a randomized controlled trial in Lahore, Pakistan, we test whether interventions designed to educate social media users about misinformation (including deep- fake misinformation) can improve indicators of truth discernment, reduce emotional reactions to news items, and decrease polarization.

Experimental design

Our experiment has been designed to include two stages to ensure that subjects do not leave with inaccurate beliefs about political elites while ensuring a valid control group for both stages of the experiment. In stage-1 RCT, we will test the effect of priming participants about misinformation (including deepfake misinformation) and paying attention to accuracy. Respondents will be informed about the veracity of news stories after they have seen/heard the news stories and answered questions about them. Subsequently, the control group will also receive T1 to ensure that all participants have received the same exposure before stage-2. In stage-2 RCT, we will test the remaining three treatments.

Design details

Randomization method
Randomization using a random number generator on a computer
Randomization unit
Individual
Clusters
NA
Observations
2800 individuals
Arms
700 per arm
Primary outcomes
Accuracy rate

Cited by

No citation data available for this registration yet.

Keywords