Recommendation Feeds and Teen Autonomy

ISEF Category: Behavioral and Social Sciences

Subcategory: Other  ·  Difficulty: Intermediate  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

Your phone can act like a tiny lab. Every swipe can nudge what you want next, and that can change how much control you feel over your choices. This project tests whether repeated recommendation feeds shift teen self-reported autonomy during the day.

What Is It?

Algorithmic fatigue is the mental drag that can build after you keep seeing more of the same from a recommendation feed. Think of it like hearing the same song on repeat. At first, it helps you predict what comes next. After a while, it can feel pushy, noisy, or boring. In this project, you ask whether that steady stream of suggestions changes how much control teens feel over their own choices.

Experience-sampling means collecting short answers in real time, not asking someone to remember the whole day later. A free app like PACO can send quick prompts during normal school days. That gives you a closer look at how feed exposure and autonomy move together across the day, instead of relying on one big end-of-day guess.

Why This Is a Good Topic

This is a strong science fair topic because you can measure it with real people, real behavior, and clear numbers. The question connects to screen habits, teen decision-making, and digital well-being, so it has a real-world angle students care about. You can learn how to design prompts, clean time-stamped data, and test whether a pattern holds across different days and participants.

Research Questions

• How does the number of recommendation-feed sessions in a day relate to adolescents' self-reported autonomy later that day?
• What is the effect of spending more consecutive time on a recommendation feed on autonomy ratings at the next prompt?
• Does feed diversity change the link between recommendation exposure and autonomy?
• To what extent do autonomy ratings differ between school days and weekend days after feed exposure?
• Which type of feed content, entertainment, shopping, news, or school-related, is associated with the lowest self-reported autonomy?
• Does self-reported fatigue mediate the relationship between feed exposure and autonomy?

Basic Materials

• A smartphone for each participant or a shared check-in device with notification access.
• Free PACO app installed and tested on each device.
• Parent and student consent forms approved by your school or district.
• A short autonomy survey with three to five clear Likert-scale items.
• Google Sheets for timestamps, response logs, and participant IDs.
• A simple recruitment script and participant instruction sheet.

Advanced Materials

• Institutional review board approval materials for adolescent research.
• A secure data store for de-identified participant IDs and time-stamped responses.
• REDCap or Qualtrics on an institutional account for larger-scale experience sampling.
• A validated autonomy scale with permission-free or institutionally approved use.
• R or JASP on a university computer for mixed-effects or repeated-measures analysis.
• Optional device screen-time logs or app-use summaries for cross-checking self-reports.

Software & Tools

• PACO: Sends short experience-sampling prompts and records responses on a free app platform.
• Google Sheets: Organizes timestamps, response logs, and participant-level summaries.
• JASP: Runs repeated-measures tests and other basic statistics without paid software.
• R: Fits mixed-effects models and checks whether results hold across participants and days.
• Jamovi: Offers a free point-and-click option for cleaning data and testing simple models.

Experiment Steps

1. Define exactly what counts as recommendation-feed exposure, such as session count, total scroll bursts, or self-reported feed check-ins.
2. Choose one autonomy scale and keep the wording identical across every prompt.
3. Build a sampling schedule that covers school time, after school, and evening time without overloading participants.
4. Plan controls for mood, sleep, and context so you can separate feed effects from daily life effects.
5. Decide how you will turn raw prompts into daily and participant-level scores before you start collecting data.
6. Select your main analysis method, such as a repeated-measures or mixed-effects model, before you run the study.

Common Pitfalls

• Letting participants define feed exposure differently, which makes one person's scroll time look like another person's app use.
• Sending prompts at random times without school-day coverage, which hides the pattern you care about.
• Mixing autonomy with mood or boredom in the same question, which blurs what you are actually measuring.
• Using memory-based reports of feed use days later, which misses the fast shifts experience sampling is meant to catch.
• Treating every prompt as independent, which inflates your sample size and can make weak patterns look real.

What Makes This Competitive

A stronger project would separate time spent on feeds from the type of feed and the timing of exposure. You could compare recommendation-heavy feeds with chronological feeds, or test lagged effects, where yesterday's feed use predicts today's autonomy. If you pair that with mixed-effects models and clear hypotheses, your project moves beyond a simple survey and into careful behavioral measurement. The strongest entries also show who the pattern applies to, and who it does not.

Project Variations

• Compare recommendation-heavy feeds with chronological feeds to see which one tracks lower autonomy.
• Test whether autonomy drops more after entertainment feeds than after school-related feeds.
• Add a nightly reflection survey to see whether the day's last feed exposure predicts bedtime autonomy.

Learn More

• PubMed: Search review articles on experience sampling, adolescent autonomy, and social media use.
• NIH Office of Behavioral and Social Sciences Research: Background on behavioral measurement and real-time assessment, found on the NIH website.
• PACO: A free experience-sampling app and project page, found by searching for the PACO app documentation.
• JASP: Free statistics software for repeated-measures tests, found on the JASP website.
• PubMed Central: Free full-text papers on ecological momentary assessment, found through the PubMed Central database.