Beliefs, Behavior, and Trusted Social Media Sources

How to cite

Bowles, Jeremy; Liu, Shelley; Larreguy, Horacio. 2020. "Beliefs, Behavior, and Trusted Social Media Sources." AEA RCT Registry. AEARCTR-0005761. https://doi.org/10.1257/rct.5761-1.0

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

Jeremy Bowles* , Shelley Liu , Horacio Larreguy

* Primary investigator

Registry ID
AEARCTR-0005761
DOI
https://doi.org/10.1257/rct.5761-1.0
First registered
Last updated
Status
On going
Trial dates
2020-04-10 – 2020-08-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/5761

Abstract

We examine how information from trusted social media sources can shape beliefs and behaviors when misinformation and mistrust is widespread. In the context of the COVID-19 outbreak in Zimbabwe, we partner with a highly reputable civil society organization to randomize the dissemination of misinformation-targeting messages about the crisis to over 27,000 newsletter subscribers through WhatsApp. We randomize the timing of the message dissemination to evaluate how exposure to these targeted messages affect citizens’ beliefs about how to deal with the crisis and their associated behavior. We also explore how these effects may vary by gender and rurality.

Intervention

In the context of the COVID-19 outbreak in Zimbabwe, we partner with a highly reputable civil society organization to randomize the timing of the dissemination of misinformation-targeting messages about the crisis to over 27,000 newsletter subscribers through WhatsApp.

Experimental design

We randomize lists to different treatment conditions relating to the timing of their exposure to different messages from the CSO.

Design details

Randomization method
Computer-based randomization.
Randomization unit
Given the CSO’s logistical constraints, the unit of randomization, within a given week, is the WhatsApp broadcast list (n = 133). Aggregating across weeks, the unit of treatment assignment becomes the list-week.
Clusters
133
Observations
Expected responses of around 500 per wave, over at least three waves, for a minimum total of 1500 observations.
Arms
Each week we evenly split lists into treatment and control. There are 62 treatment units and 63 control units.
Primary outcomes
We measure effects on beliefs using a standardized z-score index of responses to factual questions that relate to the message sent in a given week. We measure behavior using a list experiment, since directly asking about preventative measures would not only be likely to induce substantial social desirability bias, but may also force respondents to admit to illegal behavior.

Cited by

No citation data available for this registration yet.

Keywords