How social media creators shape mass politics

How to cite

Marshall, John; Chmel, Kirill; Kim, Eunji. 2024. "How social media creators shape mass politics." AEA RCT Registry. AEARCTR-0013994. https://doi.org/10.1257/rct.13994-1.3

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

John Marshall* , Kirill Chmel , Eunji Kim

* Primary investigator

Registry ID
AEARCTR-0013994
DOI
https://doi.org/10.1257/rct.13994-1.3
First registered
Last updated
Status
On going
Trial dates
2024-07-29 – 2025-06-30
Countries
Private
JEL codes
D72; D83; L82; L86
Registry record
http://www.socialscienceregistry.org/trials/13994

Abstract

Political apathy and skepticism toward traditional authorities are increasingly common, but social media creators (SMCs) capture the public's attention. Despite their prominence, whether these seemingly-frivolous yet engaging actors shape political attitudes and behaviors remains largely unknown. Our field experiment estimates the impact of encouraging Americans aged 18-45 to follow five progressive policy-minded SMCs on Instagram, TikTok, or YouTube for five months. Participants are randomly assigned to follow SMCs producing predominantly-apolitical and non-partisan or predominantly-political content; we cross-randomize recommendation-only or additional financial incentives to follow assigned SMCs. We then measure participants' political engagement, policy attitudes and priorities, progressive worldviews, voting behavior, non-electoral participation, and institutional and interpersonal trust. The comparison between treatment conditions, and against a pure control and several placebo groups, will evaluate the persuasive potential of sustained exposure to previously-unfollowed SMCs. Our findings will illuminate the role of SMCs as modern-day opinion leaders defining political communication in American democracy.

Experimental design

The treatment and incentive conditions will be cross-randomized to generate the distribution of participants. Treatments will be (dynamically) randomized within blocks of participants who are similar in terms of pre-treatment observables.

Design details

Randomization method
By computer.
Randomization unit
Individual.
Clusters
4550 individuals.
Observations
Up to 9100 observations - one midline and one endline survey for participating individuals who complete the midline and endline survey.
Arms
4,716 participants were assigned a treatment condition, of which 4,566 completed the baseline survey. The sample size by treatment arm is shown in the pre-analysis plan.
Primary outcomes
Media consumption and engagement; political knowledge and engagement; policy and political attitudes; progressive worldviews and political polarization; voting behavior; non-electoral political participation; trust.

Cited by

No citation data available for this registration yet.

Keywords