coding science fair project ideas
Updated
A good coding science fair project tests one question with one variable: does a more specific prompt give a game with fewer bugs, or does a smaller target slow your clicks? Kids can build the experiment app with Launch, the AI coding platform for kids ages 5 and up, then measure, repeat trials and chart the results.
Start with a question you can measure
A science fair project needs a question, one thing you change (the independent variable), one thing you measure (the dependent variable) and everything else kept the same. Coding is a great fit because a program can change one thing exactly, time things to the millisecond and count every click. "Make a cool game" is not a project. "Does a smaller target make players miss more often?" is.
There are two ways to frame a coding project. A scientific-method project asks a question and runs an experiment, like the reaction-time ideas below. An engineering project starts from a problem, sets criteria, builds a solution, tests it and improves it, like an app that solves a real problem at school. Big fairs make room for both: the 2026 Regeneron International Science and Engineering Fair (ISEF) lists Software Design and Behavioral and Social Sciences among its 22 categories. Check your own fair's rules for which kinds of projects it takes.
Two habits make any coding project stronger. Run every condition several times (at least five trials each), because one run can be luck. And when an AI helps build something, run the same prompt more than once: AI output can change from run to run, so a single result proves little.
Ideas for grades K–2
1. Does a more detailed description give a game closer to my plan? Draw the game first and list five things it must have. Describe it to the AI in one short sentence, then in five detailed sentences, and count how many of the five things each version got right. Variable: how detailed the description is. Measure: things on the list the game got right.
2. Do bigger buttons get clicked faster? Make a page with one button that moves to a new spot after each click, then try a small, medium and large size. Time 10 clicks at each size. Variable: button size. Measure: seconds for 10 clicks.
3. Does the game get easier when the falling objects are slower? Make a catching game with three speeds and count catches in 30 seconds at each speed. Variable: falling speed. Measure: catches in 30 seconds. Younger kids can record tallies on paper while a grown-up times.
Ideas for grades 3–5
4. Does a more specific prompt produce a game with fewer bugs? Write a vague prompt and a specific prompt for the same game. Build each five times and count what doesn't work as asked (a score that never goes up, a character that walks off screen). Variable: prompt detail. Measure: bugs per build. 5. Does the AI build the same game twice? Use one prompt five times and list what changes between builds: colors, controls, rules. This is a question about the AI itself, with no people to test.
6. Which button color do classmates click first? Show two or more colored buttons in random positions and record which one each tester clicks first. Variable: button color. Measure: share of first clicks. Because it tests other people, read the rules section below before you start.
7. Does a bigger target mean fewer misses? Build a click-the-target game where the target jumps around, and try three target sizes. Count hits and misses in 30 seconds. Variable: target size. Measure: misses per round. Testing it on yourself over several days keeps it simple.
Ideas for grades 6–8
8. Is reaction time faster to a sound or a color change? Build a reaction-time app that waits a random time, then flashes the screen or plays a beep and records how many milliseconds it takes to press a key. Variable: type of signal. Measure: average reaction time over 20 trials. 9. Does my reaction time change with time of day? Use the same app on yourself morning, afternoon and evening for a week.
10. Do coin flips get closer to 50% the more you flip? Build a coin-flip simulator and run it 10, 100, 1,000 and 10,000 times. Variable: number of flips. Measure: how far the share of heads is from 50%. This is the law of large numbers, and the computer does the flipping. 11. Are people better at remembering colors or numbers? Build a memory game that shows a sequence and asks for it back. Variable: colors vs numbers. Measure: longest sequence remembered.
12. Does practice improve typing speed? Build a typing test that measures words per minute and test yourself each day for two weeks. 13. Engineering project: build an app that solves a school problem, such as a recycling tracker where every class logs what it collects. Set criteria first (works on a Chromebook, a class can log in under a minute), test it with users, fix what fails and record each version.
Testing on people: check the rules first
Ideas 6 and 11 test other people, and that changes the rules. Under the ISEF International Rules, student research with human participants must be reviewed and approved by an Institutional Review Board (IRB) before any recruitment or data collection begins. Surveys, questionnaires and tests count, and so does having anyone other than the student researchers use an invention, product or app. For minors, a survey must be attached to the written parental permission form.
Not every school fair follows the ISEF rules, but many science fairs do, so ask your teacher or fair coordinator early. The simplest way around the paperwork is to choose a question about the code, the AI or yourself, like ideas 4, 5, 10 or 12.
What to put on the board
Use the usual layout: question, hypothesis, variables (changed, measured, kept the same), materials, procedure, results, conclusion. For a coding project, add three things judges look for: a screenshot of the app, a short note on how the code measures the result (for example, "the app records the time between the flash and the key press"), and a table of every trial, not just the averages.
Show your data with a bar chart or a line chart, and say what went wrong as well as what worked. If an AI helped build the app, say so on the board and show the prompts you used. For an engineering project, show your criteria, each version and what you changed after each test. If the fair allows it, bring a laptop or Chromebook so judges can try the app.
Building the experiment app with Launch
Most of these projects need a small app: a reaction-time tester, a click-the-target game, a coin-flip simulator. On Launch, the AI coding platform for kids ages 5 and up, a kid describes the app they need and builds it with real code (HTML, CSS, JavaScript and React) alongside an AI set up for kids. On each build, the AI first says what it is about to make or change and explains one coding idea in words that fit the kid's age, which gives kids something to write on the board about how their app works.
Launch runs in a web browser on a laptop, Chromebook, desktop or iPad, with nothing to install, and needs no coding experience. Kids only chat with the coding AI, never with other kids, and the teacher or parent who manages the account can see and open their projects. Teachers can start free with no credit card and manage students from the teacher dashboard; families can choose Build for free and upgrade to the Launch Family Plan for $20/month.
questions parents and teachers ask
- What is a good coding science fair project for middle school?
- Pick one testable question with one variable. Good options: is reaction time faster to a sound or a color change, does a coin-flip simulator get closer to 50% with more flips, or does a more specific prompt produce a game with fewer bugs. Run at least five trials per condition and chart the results.
- Can an AI science fair project be a real experiment?
- Yes, if it has a variable and a measurement. For example, write a vague and a specific prompt for the same game, build each several times, and count the bugs in each build. Because AI output can change from run to run, repeat each condition and report every trial.
- Do I need IRB approval to test my app on classmates?
- Under the ISEF International Rules, yes: having anyone other than the student researchers use an app counts as human participant research and needs IRB approval before you start. Not every school fair follows ISEF, so ask your fair. Or test the code, the AI or yourself instead.
- Can my child build the experiment app with no coding experience?
- Yes. On Launch, kids ages 5 and up describe the app they need, such as a reaction-time tester, and build it with real code alongside an AI set up for kids. It runs in a web browser with nothing to install and is free to start, with no credit card.
try it with launch
Launch is the AI coding platform where kids ages 5 and up describe what they want to make and build it with real code, alongside an AI set up for kids.
sources
- Society for Science — Regeneron ISEF 2026 categories and subcategories (includes Software Design; Behavioral and Social Sciences) (checked )
- Society for Science — ISEF International Rules: Human Participants (IRB approval before recruitment or data collection; surveys; parental permission) (checked )
- Society for Science — ISEF Rules FAQ (use of an invention, product or app by others is a human participant study) (checked )