Emotions and Social Media Content

How to cite

Noy, Shakked. 2026. "Emotions and Social Media Content." AEA RCT Registry. AEARCTR-0018673. https://doi.org/10.1257/rct.18673-1.0

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

Shakked Noy*

* Primary investigator

Registry ID
AEARCTR-0018673
DOI
https://doi.org/10.1257/rct.18673-1.0
First registered
Last updated
Status
In development
Trial dates
2026-05-19 – 2026-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/18673

Abstract

This study examines the effects of exposure to different kinds of short-form video content on viewers' contemporaneous emotions. We recruit participants who are paid to spend up to 30 minutes using an interface that allows them to scroll through a randomized feed of short videos scraped from social media platforms (TikTok, Instagram Reels, YouTube Shorts) presented alongside either (i) a multi-choice interface that allows them to indicate which emotions the video triggered or (ii) an interface that lets them record short voice messages explaining their reactions to the video. The goal of this exercise is to analyze how different types of social media content affect viewers' emotions. We are particularly interested in the effects of (i) content prominently displaying 'enviable' individuals likely to trigger unfavorable comparisons and (ii) political content on viewers' emotions.

Intervention

In this experiment, participants will be paid to scroll through a randomized feed of short videos and report the emotions they experience while watching each video. The "intervention" consists of within- and across-person randomization of videos into these feeds, which will allow us to estimate the effects of different types of videos on emotions.

Via a previous data collection, we have access to the TikTok, Instagram, and YouTube viewership histories of about 390 young Americans. We will sample videos for this experiment through the following procedure. (1) We randomly sample a subset of the creators who appear in this data (by randomly picking videos and adding their creators to our subsample). (2) We scrape videos posted by these creators in the week preceding the rollout of the experiment. (3) We scrape the metadata of these videos (transcripts, captions, thumbnails, etc.) and classify features of the videos using LLMs. Of particular interest is whether the video prominently features a creator whom viewers would consider 'enviable' on one or more dimensions (appearance, success, wealth, etc.), and whether the video is about politics. Politics will take primacy in our codings, so that a video classified as political cannot be classified as enviable. (4) From the videos scraped and classified this way, we will randomly sample a subset of videos, maintaining equal weight on all the sampled creators, and stratifying so that we end up with 33% each of politics, enviable, and other videos. We will sample a number of videos such that, given our expected sample size of participant-video pairs, each video in our data is seen by 5-10 participants. We will exclude advertisements and non-English-language videos.

Once we have this set of videos, participants in the experiment will see a feed of videos randomly sampled from this set, in a random order, stratified to be 33% politics, 33% enviable, and 33% other.

Experimental design

See description of Intervention above.

We will recruit American participants aged 18-30 on Prolific, screened on being regular users of TikTok, Instagram Reels, or YouTube Shorts, and pay them to watch a 30-minute video feed containing videos sampled and randomized using the methods described in the Intervention section.

Design details

Randomization method
Computer randomization of videos shown and order in which they are shown.
Randomization unit
Randomization is at the participant-by-video level, i.e. each participant will get their own randomly selected 30-minute video feed randomly chosen in a stratified way using the methods described in the Intervention section.
Clusters
We will recruit 200 participants.
Observations
We will recruit 200 participants. The number of participant-by-video observations depends on how long people end up taking on each video and therefore how many videos they get through in 30 minutes, but our rough projection, assuming 30 seconds per video, is 2*30*200 = 12,000 observations.
Arms
200
Primary outcomes
Our main outcomes for participants given the multi-choice emotions interface will be (1) an index of emotional positivity constructed by adding +1 for each positive emotion selected by the respondent and -1 for each negative emotion selected and then z-scoring within respondent, and (2) separate dummies for each specific emotion we elicit. Similarly, for participants for whom we collect voice recordings, our outcomes will be LLM-coded dummies for (1) the transcript of the recording mentioning any negative emotions and (2) dummies for specific negative emotions extracted from the transcript.

Cited by

No citation data available for this registration yet.

Keywords