Combating Fake News in Health: An Online Experiment

Last registered on January 23, 2023

View Trial History

Pre-Trial

Trial Information

General Information

Title

Combating Fake News in Health: An Online Experiment

RCT ID

AEARCTR-0010724

Initial registration date

January 17, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

January 23, 2023, 7:49 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Ghana

Region

Country

Kenya

Region

Country

Nigeria

Region

Country

Tanzania, United Republic of

Region

Country

Uganda

Region

Country

South Africa

Region

Primary Investigator

Name

Maximilian Guigas

Affiliation

Bernhard-Nocht-Institute for Tropical Medicine

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Jan Priebe

PI Affiliation

Bernhard-Nocht-Institute for Tropical Medicine

Contact Investigator

PI Name

Kerstin Unfried

PI Affiliation

Bernhard-Nocht-Institute for Tropical Medicine

Contact Investigator

Additional Trial Information

Status

In development

Start date

2023-02-01

End date

2023-06-30

Keywords

Behavior, Health

Additional Keywords

Fake-News, Sub-Saharan Africa, Vaccination, Social Media

JEL code(s)

I12, D83, D91

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

The dissemination of health-related fake news , especially on social media platforms, is a strong thread to the efficient provision of health care. Several media platforms have recently established tools to combat the spread of fake news. In this research project, we aim to assess the effectiveness of such tools, and elaborate on the underlying psychological mechanisms. Specifically, we will conduct an online survey experiment in six sub-Saharan African countries via Facebook to evaluate the impact of two kinds of such tools – a pre- and a debunking tool - on individuals’ sharing behavior. Moreover, our experimental set-up allows to shed light on the causal impact of confirmatory search behavior and article accuracy on the impact of these tools on people’s article sharing behavior. We consider two indicators of sharing behavior: (i) intentions (willingness to share) and (ii) action (clicking a Facebook sharing button). Apart from the impacts of those tools, we plan to analyze differences in sharing behavior regarding sociodemographic factors, personality and risk aversion, prior health and vaccination attitudes and own health and vaccination history.

External Link(s)

Registration Citation

Citation

Guigas, Maximilian, Jan Priebe and Kerstin Unfried. 2023. "Combating Fake News in Health: An Online Experiment." AEA RCT Registry. January 23. https://doi.org/10.1257/rct.10724-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

We use Prebunking and Debunking tools to test their effectiveness against the spread of fake news.

Intervention Start Date

2023-02-01

Intervention End Date

2023-02-18

Primary Outcomes

Primary Outcomes (end points)

I) Article sharing: Our main outcome of interest is an individual’s sharing behavior.
II) Attitudes toward vaccination: We measure attitudes and emotions towards vaccination before and after the interventions to capture potential belief updating.

Primary Outcomes (explanation)

For the measurement of article sharing we use two indicators: First, we ask respondents about their intention to share the article. Second, we provide respondents the opportunity to share the article on their Facebook page using a Facebook share button within the survey providers environment. Our second measure assesses whether participants click on the sharing button in the survey that leads to a pop-up window, where they have to reconfirm the sharing with Facebook. Moreover, we observe the actual spread of the article on Facebook to control for attrition of sharing numbers due to the reconfirmation necessary within the Facebook environment.

Secondary Outcomes

Secondary Outcomes (end points)

Secondary outcome indicators are as follows:
1. Individual level
a. Sociodemographic factors
b. Personality, risk aversion and wealth
c. Numeracy
d. Social media behavior and news preferences
e. Own health
i. Current health status
ii. Health history
iii. Vaccination history
f. Survey questions on vaccination knowledge
g. Health and vaccination attitudes

Secondary Outcomes (explanation)

Secondary outcomes include variables that we will use to investigate mechanisms through heterogeneous treatment effects. Secondary outcomes are related to three distinct features of the intervention: (a) respondent’s characteristics such as personality traits, education levels, and risk, (b) health and vaccination attitudes of the participants, vaccination knowledge and reading preferences in the article selection stage (c) social norms of individuals and reasons/motives behind the sharing decision.

Experimental Design

Our experiment uses a between subject design. It has three treatment arms following a 2 x 2 x 3 design.

Experimental Design Details

There are 3 types of interventions that are of interest in our study. Intervention 1 relates to exposing respondents to their most or least preferred article choice. While initially respondents state their preferences for a certain article (based on the displayed article title), they are randomized (50:50) into (not) receiving their preferred article. Intervention 2 relates to whether respondents receive an article that contains true or false information. Respondents are randomized (50:50) into receiving either type of an article. Intervention 3 relates to whether respondents are viewing articles featuring prebunking or debunking tools. In intervention 3 individuals are randomized into one of the three groups (control, prebunking, debunking) with equal probabilities.

Randomization Method

Randomization in our study happens on the individual level. The randomization into the respective treatment arms is done by a so-called randomization trigger provided by the survey platform provider UNIPARK.

Randomization Unit

Randomization in our study happens on the individual level.

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

N.A.

Sample size: planned number of observations

We aim for a sample size of at least 5000 participants, however it is difficult to estimate a precise sample size, since by now very few surveys in Sub-Saharan Africa, which acquired the participants via Facebook Ads were conducted.

Sample size (or number of clusters) by treatment arms

With respect to intervention 3 (prebunking, debunking, control), we assume that about 1/3 of the overall sample is in any given treatment group. With respect to sub-randomizations (intervention 1, intervention 2), the sample size in each treatment arm is about ½ of each main treatment arm.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

We provide an estimate of the MDE assuming an average sharing decisions of 10 percent in the control group (intervention 3). Assuming a power of 80% and a significance level of 0.05, we obtain an MDE of 9.6pp. We believe that the actual MDE is smaller in reality once we can control for important covariates such as age, gender, etc..

Supporting Documents and Materials

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

IRB