Anonymous vs Identified Brainstorming Project Ideas

ISEF Category: Behavioral and Social Sciences

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

The Hook

Two teams can write the same number of ideas and still end up with very different thinking. When names are hidden, people often feel freer to take risks. When names are shown, they may edit themselves more. Your project asks which setup leads to ideas that are farther apart in meaning.

What Is It?

This project studies how identity changes group idea generation. In one condition, people brainstorm with their names attached. In the other, they post ideas without names. You then compare the ideas to see which group produced more novelty.

Think of each idea as a point on a map of meaning. Ideas that mean almost the same thing sit close together. Ideas that are very different sit far apart. Semantic-distance embeddings turn text into numbers so you can measure those gaps instead of guessing by eye.

The key question is not just how many ideas a team writes. It is how varied those ideas are. A group can produce a long list that all points to the same answer, or a shorter list that spreads into new territory.

Why This Is a Good Topic

This is a strong science fair topic because you can test it with real students, clear controls, and measurable text data. It connects to teamwork, creativity, and how social pressure changes behavior in groups. You can learn experimental design, data cleaning, and simple text analysis without needing a professional lab.

Research Questions

• How does anonymity change the average semantic distance between ideas in a team?
• What is the effect of anonymous versus identified brainstorming on the number of unique idea clusters a team produces?
• Does the impact of anonymity on idea novelty change when the prompt asks for practical, social, or science-based ideas?
• To what extent does group size change the difference between anonymous and identified brainstorming?
• Which condition, anonymous or identified, produces more ideas that are far from the team’s first idea?
• To what extent do self-rated confidence scores relate to semantic novelty in each condition?

Basic Materials

• Laptop or Chromebook with internet access.
• Shared Google Doc or similar collaborative document.
• Three to six high-school volunteers per team.
• Spreadsheet for logging each idea and team condition.
• Simple participant instructions and consent script.
• Timer or stopwatch.
• Text prompt list for each brainstorming round.

Advanced Materials

• University-approved human subjects protocol.
• Secure survey platform such as Qualtrics or REDCap.
• Python environment with sentence-transformers, pandas, and scipy.
• R or Python notebook for mixed-effects modeling.
• Blinded human rating form for independent novelty scores.
• Secure storage for raw text and de-identified participant data.
• Access to a larger student sample for replication.

Software & Tools

• Google Docs: Runs the brainstorming sessions and records each idea in one shared place.
• Google Sheets: Organizes conditions, team size, and idea counts for quick review.
• Python: Cleans the text, calculates embeddings, and runs the distance analysis.
• Jupyter Notebook: Keeps your code, charts, and notes together while you analyze the data.
• Sentence-Transformers: Converts each idea into an embedding so you can compare semantic distance.

Experiment Steps

1. Define how you will score novelty, and decide what counts as one unique idea.
2. Choose whether you will compare different teams or reuse the same teams in two rounds.
3. Fix the prompt, group size, and session length so identity status is the main variable that changes.
4. Plan how you will convert each brainstorm into embeddings, pairwise distances, and one team score.
5. Pick your statistical test and a backup check for outliers, duplicates, and uneven team output.

Common Pitfalls

• Letting anonymous and identified groups see different prompts, which makes the identity effect impossible to separate from prompt difficulty.
• Counting repeated or near-duplicate ideas as new ideas, which inflates novelty scores.
• Using one noisy embedding score without cleaning filler text, which lets wording quirks drive the result.
• Comparing teams with different group sizes, which can make larger groups look more creative just because they write more.
• Forgetting to blind human raters if you add manual scoring, which can bias ratings toward the condition you expect.

What Makes This Competitive

A stronger project goes beyond raw idea counts and tests how far apart the ideas really are in meaning. You can compare multiple prompts, control for group size, and use a mixed-effects model so team differences do not swamp the main result. If you also compare embedding scores with blind human novelty ratings, you get a second lens on the same behavior. That makes the project feel like real social psychology, not just a class survey.

Project Variations

• Compare anonymous and identified brainstorming on practical, social, and science prompts to see whether topic type changes the effect.
• Test whether anonymity matters more in smaller teams than in larger teams.
• Measure novelty with semantic distance, then compare it with blind human ratings of idea originality.

Learn More

• NIH PubMed: Search review articles on brainstorming, group creativity, and anonymity.
• Google Scholar: Search for papers on social loafing, evaluation apprehension, and group idea generation.
• OpenStax Psychology 2e: Review core ideas in social psychology and group behavior in a free textbook.
• MIT OpenCourseWare: Find free course materials on social psychology and research methods.
• PLOS ONE: Search open-access studies on creativity, collaboration, and idea novelty.