Curating your Feed: Experimental Evidence from Instagram

How to cite

Kaeppel, Katharina; Widmer, Philine; Gauthier, Germain; Hodler, Roland. 2026. "Curating your Feed: Experimental Evidence from Instagram." AEA RCT Registry. AEARCTR-0019180. https://doi.org/10.1257/rct.19180-1.0

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

Katharina Kaeppel* , Philine Widmer , Germain Gauthier , Roland Hodler

* Primary investigator

Registry ID
AEARCTR-0019180
DOI
https://doi.org/10.1257/rct.19180-1.0
First registered
Last updated
Status
In development
Trial dates
2026-07-24 – 2027-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/19180

Abstract

We study how an intervention that teaches and encourages Instagram users to curate their own feed affects their content consumption, well-being, political attitudes, and gender norms. The study is a two-wave, individually randomized controlled trial with active Instagram users aged 18 to 35 from the United States (US). In the baseline wave, participants complete a survey and share data on their Instagram content; a randomly assigned treatment group is then shown how to use two built-in curation tools. Roughly six weeks later, an endline wave re-measures the same outcomes, measures further outcomes, and collects the same data on consumed content. We estimate intention-to-treat effects with baseline covariate adjustments and local average treatment effects using random assignment as an instrument for compliance.

Intervention

The treatment follows an encouragement design. Treated participants are shown how Instagram's built-in tools let them shape their feed, and are encouraged to use these tools during the baseline wave. The control group receives no such information and proceeds directly through the survey.

Experimental design

Participants are US residents aged 18 to 35 who are active Instagram users, recruited via Prolific and Cloud Research in the US. Participants provide informed consent before any data are collected. We have access to basic demographics (e.g., age, gender, partisanship).

After answering a short battery of baseline questions and sharing data on their Instagram use, participants are randomized into either the treatment or the control group using the platforms' built-in randomizers. The treatment group receives information on different tools to curate their feed, and is encouraged to use these tools. The control group receives no such information.

Outcome measures and data on Instagram usage is collected in a second wave approximately six weeks after the baseline.

Assignment is at the individual level, stratified (blocked) by the interaction of gender and partisanship, all measured at baseline prior to randomization.

Design details

Randomization method
Built-in platform randomizer on Qualtrics
Randomization unit
Individual
Clusters
1,000 to 5,000, with the final total determined by the budget.
Observations
1,000 to 5,000
Arms
~500 - 2,500 Treatment, ~500 - 2,500 Control (~50% - 50%)
Primary outcomes
We will organize outcomes into broad families to answer questions regarding the effect of our treatment. 1. Instagram user experience 2. Well-being and mental health 3. Political attitudes 4. Views on current political events 5. Gender norms Our descriptive analysis of Instagram use will involve content analyses of the Instagram data and is treated separately.

Cited by

No citation data available for this registration yet.

Keywords