Here is the fact that changes how you should prepare for every science fair interview you will ever do: the person scoring your project is almost never an expert in it.
At a county or regional fair, judges are volunteers. A mechanical engineer judges a microbiology project. A high school chemistry teacher judges a machine learning project. A retired physician judges an environmental study about watershed nitrates. Even at ISEF, where judges are assigned by category, "Biomedical and Health Sciences" is wide enough that a cardiologist is scoring a project on gut bacteria.
They will not follow your specifics. They will not admit that they did not follow your specifics. And they will score you anyway, on what they understood and how they felt about it.
Students who understand this win fairs with good projects. Students who do not lose fairs with great ones.
Who is actually holding the clipboard
Walk through who volunteers to judge a science fair on a Saturday:
- Working scientists and engineers from local companies and universities, usually recruited through a colleague. Smart, curious, and typically one or two fields away from yours.
- Teachers, often from the host school district. Broad science literacy, not deep research background.
- Graduate students, who are the closest thing to a specialist you will meet, and who are still usually in a different subfield.
- Retired professionals, who have been out of the lab for a decade and judge on fundamentals.
Fairs try to match categories. They cannot match subfields. There are not enough judges for that, and categories like "Environmental Science" or "Computational Biology" contain dozens of specialties that do not read each other's papers.
So the realistic judge profile is this: a smart, technically literate adult who knows roughly what a p-value is, has never heard of your specific method, and has 12 minutes to form an opinion before walking to the next table.
What a non-expert actually scores
The published rubrics say judges score creativity, methodology, data, and communication. Read how ISEF is actually judged for the line-by-line breakdown. But rubrics are filled in by humans after the conversation ends, and humans fill them in from memory and impression.
What a non-expert judge can actually assess in 12 minutes:
Number four is not on any rubric, and it moves every other score. A judge who felt understood by you, who followed your story and had a moment of "oh, that is clever," scores the methodology line higher even though they could not evaluate the methodology. That is human, not corrupt, and it is the same dynamic that decides grant reviews and job interviews.
A judge who got lost at minute two, nodded politely for ten more minutes, and walked away unsure what you did scores you in the middle of every line. Middle of every line does not place.
The interview is a pitch, not a defense
Most students prepare for the interview as if it were a thesis defense: anticipate the hardest technical question, be ready to justify every parameter. That preparation is not wasted, but it is the second half of the job, and most students never do the first half.
The first half is the pitch. Your project is the product. The judge is the customer. You have about 60 seconds to make them care, and if you spend those 60 seconds on methodology, you have already lost them.
The students who win are not always the ones with the deepest science. They are the ones who can communicate, every year, at every level. A clear student with a solid project beats a mumbling student with a brilliant one, because the judge could only score what reached them.
The 60-second opening
When a judge walks up, do not start with your method. Do not start with your title. Start with a person.
"My grandfather was diagnosed with Alzheimer's two years ago, and the test that confirmed it cost $4,000 and needed a spinal tap. I wanted to know whether a blood test could do part of that job. So I took 566 patients from a public Alzheimer's dataset and asked whether three proteins in their blood tracked with their diagnosis."
That is 60 seconds. Notice what it contains: a human stake, a problem a non-expert instantly understands, a question stated in plain words, and a hint at scale ("566 patients") that signals seriousness. Notice what it does not contain: the names of the proteins, the regression model, the correction for multiple comparisons. All of that exists, and the judge may ask for it. But not yet.
Compare the opening most students give:
"My project is titled 'Plasma Endothelial Biomarkers and Cognitive Status in the ADNI Cohort.' I used multivariable linear regression adjusted for age, sex, APOE epsilon-4 and education to test whether VCAM-1, ICAM-1 and VEGF..."
The judge has stopped listening at "endothelial." They will nod for eleven more minutes and give you a 7 out of 10 on everything.
Explain it like they are smart but not in your field
The test for every sentence in your first two minutes: could a smart 14-year-old follow this? Not a dumb one. A sharp, curious kid who has never taken your class. If the answer is no, the sentence needs to be simpler.
Rules that work:
- No jargon for the first two minutes. If a term is truly unavoidable, define it in six words and keep moving. "VCAM-1, a protein that shows up when blood vessels are inflamed."
- Translate every method into what it does, not what it is called. Not "I used elastic-net logistic regression with nested cross-validation." Instead: "I trained a model to tell sick from healthy, and I tested it on patients it had never seen, so it could not cheat."
- Use one analogy, and make it physical. Judges remember analogies. "Amyloid is the rust, these proteins are the smoke; I wanted to know whether the smoke tells you anything the rust does not."
Tell the story in four beats
Every winning pitch has the same skeleton. Say these four sentences, in order, before you touch a single graph:
Four sentences. Then stop and let the judge steer. Everything else in your project is detail they may or may not ask for. Students who front-load detail are not being thorough; they are hiding the story under it.
Make the result feel big
Your result has a technical form and a human form. The judge scores the human form.
- Not "accuracy improved by 12%." Instead: "That is three more patients out of every 25 caught early enough to treat."
- Not "a statistically significant difference of 0.55 standard deviations." Instead: "Patients with Alzheimer's had clearly higher levels, and the difference held up even after I accounted for the usual explanations."
- Not "the model achieved an AUC of 0.83." Instead: "If you handed the model two patients, one sick and one healthy, it would pick the sick one correctly 83 times out of 100."
The technical number should be on your board, and you should be ready to defend it. But what the judge writes on the sheet after they walk away is the version they can picture.
Read the judge and adjust
The interview is a conversation, and the best students run it like one.
- If they nod and lean in, go one level deeper. They are following and want more.
- If they go quiet or glance at the board, you have lost them. Pull back up a level and reconnect to the story.
- Ask "does that make sense so far?" exactly once, around the two-minute mark. It gives a lost judge a graceful way back in, and a judge who feels understood by you scores you higher.
- If a judge is clearly an expert, you will know within one question. Shift gears, give them the technical version, and enjoy it. This happens perhaps one interview in five.
Depth comes second, and only on request
None of this means the science does not matter. It matters enormously, but it is scored in the second half of the interview, when the questions start. That is when you show you can answer the four questions that decide the ISEF interview: what the biggest weakness is, why this method and not another, what you would do differently, what the next experiment is.
Know all of it cold. Then let the judge ask for it. Volunteering specifics to someone who did not want them is not rigor; it is blabbering, and judges experience it as a student who cannot read the room.
Say what you did yourself
Judges always ask, and they ask more pointedly when a project looks mentored. Be specific and honest: "My mentor suggested the dataset and reviewed my analysis plan. I wrote all the code, ran every model, and the composite score idea was mine after I saw the single-marker results." Confidence here is what makes the rest of your pitch believable. Hedging here makes a judge wonder what else is borrowed.
Rehearse on someone who knows nothing
Pitch to a parent, a sibling, a friend who has never heard of your topic. Then ask them to explain your project back to you in two sentences.
If they can, you are ready. If they cannot, your opening is too technical, and no amount of methodology practice will fix a pitch that loses the judge at minute one.
The YRI fellows who placed, like Avyay Gupta, who won first at his regional fair as a ninth grader, and Aditya Singla, who won Best Oral Presentation at an IEEE conference in Tokyo, did not win because their science was beyond the room. They won because they made the room care, and then proved the depth was there when asked.
Free download: The Science Fair Project Kit
Seven printable worksheets that take a project from first idea to judge-ready: hypothesis builder, experiment design, data log, board layout, and the ISEF scoring rubric with interview prep.
Frequently Asked Questions
Are science fair judges experts in my category? They are matched to broad categories, not subfields. A judge in "Biomedical Sciences" might be a cardiologist scoring a neuroscience project. Assume a smart, technically literate non-expert and you will be right most of the time.
Should I dumb down my project for judges? No. Simplify the explanation, not the project. Keep every technical detail on the board and ready for questions; just do not lead with it.
What if the judge is an expert? You will know within one question. Switch to the technical version and go as deep as they want. Preparing the simple pitch first costs you nothing in that case.
How long should my opening be? About 60 seconds, four beats: the problem, what you tried, what you found, why it matters. Then stop and let the judge ask.
Does this apply to JSHS and ISEF too? Yes. JSHS round one is a written paper reviewed by scientists, but the round-two talk is scored by a mixed panel. ISEF interviews are 25 of 100 points and the judges rotate across your category. See how to win JSHS and how to win the ISEF interview.
Related guides
- How science fair judging works, round by round
- How ISEF is judged: the 100-point rubric
- How to win the ISEF interview
- How to present your research at a science fair
- How to win a science fair