Social Coordination in Response to Disaster: Experimental Evidence from the COVID-19 Pandemic
How to cite
Diep-Nguyen, Ha; Yang, Jun. 2020. "Social Coordination in Response to Disaster: Experimental Evidence from the COVID-19 Pandemic." AEA RCT Registry. AEARCTR-0005716. https://doi.org/10.1257/rct.5716-1.1
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We study the cooperative economic behavior of donations to provide public goods in response to disastrous events. We partnered with a charitable organization that aims to purchase and distribute surgical masks to smaller hospitals and facilities in areas hard-hit by COVID-19. We launched a GoFundMe campaign to raise funds for its operations in April 2020. We randomly rotated different versions of this campaign's GoFundMe webpage that vary in profile pictures, campaign descriptions, and operation updates. Our analysis has three components. First, we construct a measure of each version's trustworthiness based on the ratings provided by the subjects on Amazon MTurk. We then analyze the connection between the campaign page features and the level of trustworthiness. Second, we study how perceived trustworthiness affects contributions to the campaign. Finally, we extend our measure of trustworthiness to all campaigns that are related to PPE provision in the GoFundMe platform for generalization.
Intervention
We have two interventions in this project.
First, we partner with a charitable organization that aims to procure and distribute surgical masks to smaller hospitals and facilities in areas hard-hit by COVID-19 in launching a GoFundMe campaign. We randomly rotate different versions of the campaign's GoFundMe webpage that vary in profile pictures, campaign descriptions, and operation updates.
Second, we conduct a survey on Amazon mTurk that randomly displays different versions of the campaign's GoFundMe page and asks the participants to evaluate the displayed versions' credibility.
Experimental design
In the first intervention, we work with a charitable organization that procures and distributes surgical masks to smaller hospitals and facilities in areas hard-hit by COVID-19 in launching a fund-raising campaign on GoFundMe. GoFundMe allows for changes and updates to the campaign page. Specifically, we create 16 versions of the campaign page:
2 variations in the gender of the profile picture: Male and Female
2 variations in the ethnicity of the profile picture: White and Non-white
2 variations in campaign description: Long with more details and Short with fewer details
2 variations in campaign updates on how the organization has used the funds to procure and distribute masks: Show updates and No updates
We then randomly rotate the campaign page every 3 hours from 9 AM to 9 PM for 4 shifts (and return to the default page from 9 PM to 9 AM). We have the rotation up for 30 days from April 16 to May 15, 2020.
In the second intervention, we run a survey on Amazon mTurk to gauge the donors' perception of the campaign's credibility. In the survey, we first show the subjects a campaign page randomly chosen from the 16 versions we use in the first intervention. We then ask the subjects to rate the credibility of the displayed campaign version and the amount they would donate to the campaign. To incentivize truthful telling, we offer a raffle in which if they win, the indicated amount will be deducted from the prize money and donated to the campaign on their behalf.
Our analysis has three components.
First, we construct a measure of each version's trustworthiness based on the evaluation by subjects on Amazon MTurk. We then analyze the connection between campaign page features and the level of trustworthiness.
Second, we study how trustworthiness affects contributions to the campaign.
Finally, we extend our measure of trustworthiness to all campaigns that are related to PPE provision in the GoFundMe platform for generalization.
Design details
Cited by
No citation data available for this registration yet.