Fair Tests

Two friends are arguing. Ravi says his paper plane flies furthest because it's his. Mia says hers wins because it's blue. How do you ever settle it? You don't shout louder — you hold a fair contest.

A fair contest is exactly what scientists call a fair test. It's how you find out what really makes a difference, instead of just guessing or hoping. Think of a race: it's only fair if everyone starts at the same line, at the same time, and runs the same distance. Change those and the race tells you nothing. A fair test works the very same way.

The three jobs in every fair test

Every fair test, from a paper plane to a rocket, comes down to three little jobs. Give them these kid-sized names and you'll never forget them:

What I change — the one thing you're testing. (Does a higher ramp help? Then ramp height is what you change.)
What I measure — the one thing you watch to see what happened. (How far the car rolls.)
What I keep the same — everything else, kept exactly, boringly the same. (Same car, same floor, same push.)

Get those three sorted before you start, and your test will actually answer your question. Miss one and it won't — no matter how carefully you do the rest.

Try it: the ramp-and-car test

Here's the classic fair test. You want to know: does a car roll further from a higher ramp? Slide the ramp up and down. Only one thing changes — the height — and you watch one thing: how far the car rolls (the dashed line). The car, the floor and the push stay exactly the same every single time.

Because nothing else was allowed to move, you can say it with confidence: the higher ramp is what sent the car further. That's the whole point of a fair test — you end up knowing, not just guessing.

Why only one change at a time?

Imagine you make the ramp higher and swap to a heavier car, both at once. The car rolls a different distance… but which change did it — the taller ramp, or the new car? You genuinely cannot tell. You changed two things, so the answer is jammed between them.

It's like a race where one runner starts early and has a shorter track. If they win, so what? You've learned nothing. Change just one thing and the answer is clean and clear: whatever happened, that one change caused it.

Nadia bakes biscuits and thinks they're a bit boring. So next time she adds more sugar and more butter. The new batch is delicious! Brilliant — but which change made them better: the sugar, or the butter? She has no idea, because she changed two things at once. To find out, she'd have to bake again changing only the sugar, then again changing only the butter. One change per batch, or the biscuits keep their secret. A fair test is just careful cooking.

Change only one thing — the thing you're testing. Change two and you won't know which one did it.
Keep everything else the same. Same equipment, same place, same push — the more you keep still, the more you can trust the answer.
One go isn't enough. Do it several times, because a single try might have been a fluke.

Do it again: repeats beat flukes

Suppose you roll the car once from the high ramp and it zooms miles. Was that the ramp… or did it just happen to catch a smooth patch of floor that one time? One go can be a fluke — a lucky or unlucky one-off that doesn't really tell you what usually happens.

So scientists do repeats: they run the same test several times and look at all the results together. If the high ramp wins again and again and again, that wasn't luck — it's real. If the results jump about all over the place, that's a warning that something sneaky is still changing when it shouldn't be. Repeating your test is how you become sure.

Back to Ravi and Mia and their paper planes. Instead of arguing, they hold a fair test. They fly each plane from the same spot, with the same gentle throw, down the same corridor — only the plane is different. Ravi's wins the first throw. "Told you!" But Mia says, "Once could be a fluke — throw again." So they throw five times each and measure every landing. Mia's plane wins four times out of five. The repeats settled it fairly, with no shouting — and Ravi even asks to copy her wing folds for next time.

Three fair tests you could do this week

The recipe is always the same — change one, measure one, keep the rest the same, repeat — but you can point it at almost any question:

Which paper plane flies furthest? Change the plane's shape. Measure how far it lands. Keep the same thrower, the same throw, the same spot — and throw each plane several times.
Does a car roll further from a higher ramp? Change the ramp height. Measure how far the car rolls. Keep the same car, floor and push — and roll from each height a few times.
Does a plant grow taller with more water? Change how much water each plant gets. Measure how tall it grows. Keep the same kind of seed, the same pot, the same soil, the same sunny windowsill — and grow more than one plant at each amount, in case a single seed is a dud.

Notice the plant one takes days, not seconds — but it's still the very same three jobs. Spot them in every experiment and you're thinking like a real scientist.