Willingness to Tweet (replication)

Last registered on November 04, 2022

View Trial History

Pre-Trial

Trial Information

General Information

Title

Willingness to Tweet (replication)

RCT ID

AEARCTR-0010269

Initial registration date

October 19, 2022

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

November 04, 2022, 1:40 PM EDT

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

Locations

Country

United States of America

Region

Illinois

Primary Investigator

Name

Aakaash Rao

Affiliation

Harvard University

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Georgy Egorov

PI Affiliation

Kellogg School of Management

Contact Investigator

PI Name

Leonardo Bursztyn

PI Affiliation

University of Chicago

Contact Investigator

PI Name

Ingar Haaland

PI Affiliation

NHH Norwegian School of Economics

Contact Investigator

PI Name

Christopher Roth

PI Affiliation

University of Cologne

Contact Investigator

Additional Trial Information

Status

Completed

Start date

2022-10-19

End date

2022-11-02

Keywords

Behavior, Crime, Violence, & Conflict

Additional Keywords

Twitter, social norms, social image

JEL code(s)

Secondary IDs

Prior work

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

Abstract

We replicate our results from AEARCTR-0008432.

External Link(s)

Registration Citation

Citation

Bursztyn, Leonardo et al. 2022. "Willingness to Tweet (replication)." AEA RCT Registry. November 04. https://doi.org/10.1257/rct.10269-1.0

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

We replicate our intervention from AEARCTR-0008432, varying the existence of a "social cover" associated with making a Tweet about a socially sensitive topic.

Intervention Start Date

2022-10-19

Intervention End Date

2022-11-02

Primary Outcomes

Primary Outcomes (end points)

Whether or not the participant chooses to authorize the Twitter post

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

(1) An indicator for whether the respondent indicates that the Tweet is misleading
(2) Conditional on (1), an indicator for whether the respondent indicates any concern that the Tweet is misleading about the timing of when they read the article

Secondary Outcomes (explanation)

Experimental Design

We vary the perceived social pressure associated with making a Tweet about a socially sensitive topic.

Experimental Design Details

Participants are shown an article written by Princeton criminologist Patrick Sharkey discussing the likely effect of defunding the police on violent crime and decide whether or not to (privately) join a campaign to oppose defunding the police. They are then told they will have the opportunity to schedule a post to be made on their Twitter account encouraging their followers to sign a petition to oppose defunding the police. The experiment manipulates only one word: whether the post indicates that the participant saw the article before joining the campaign, or after joining the campaign. (They are told that this post will be made public if and when we have reached survey respondents in all US counties. However, given that we are collecting fewer observations than the number of US counties, this will not happen, and thus the posts will never be made public.) We also collect various demographic characteristics of participants.

Randomization Method

Randomization through Qualtrics

Randomization Unit

Individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

Sample size: planned number of observations

We will ideally recruit 1000 Democrats. However, recruitment is difficult in this setting, and we may be unable to recruit 1000; in our previous experiment we are replicating, we were only able to recruit 523 respondents. We will keep data collection open for two weeks (ending November 2), or until we reach 1000 respondents, whichever comes first.

Sample size (or number of clusters) by treatment arms

Half of the sample in the "Rationale" condition; half in the "No Rationale" condition.

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

With 506 respondents, we will have 80% power to detect an 12 percentage point effect (the effect size estimated in the first experiment we replicate) at the 5% level. With 1000 respondents, we will have 80% power to detect a 8.6 percentage point effect at the 5% level.

Supporting Documents and Materials

IRB