Every good investigation begins the same way — with a sharp question that has a measurable answer. Here's how to spot one, why the fuzzy ones fail, and how to build one of your own.
Start sorting"Does temperature change how fast sugar dissolves?" is a great science question. "How does sugar work?" is not — and the difference is the entire secret.
The strong one is testable: you can change one thing, measure what happens, and let the result decide. The vague one points everywhere and nowhere — there's nothing single to measure, so an experiment can't settle it. Get the question right and the whole investigation falls into place: you'll know exactly what to set up, what to write down, and how you'll know when you have your answer.
So before you touch a single beaker, this is the move that matters most. A scientist spends real time sharpening the question, because a fuzzy question leads to a fuzzy experiment leads to a "results" page that doesn't actually prove anything. Sharpen the question first, and everything after it gets easier.
A research question is just the single, sharp thing your experiment is going to answer. It almost never starts that way, though. It starts as a big, blurry wondering — the kind of thought you have staring out a bus window. That's a wonderful place to begin. It's just not where you can stop.
Think of a kitchen funnel. You pour a whole jug of "I wonder…" in at the wide top — and a single clear stream comes out the narrow spout. Your job is the squeezing: taking something huge like "How does sugar work?" and narrowing it, again and again, until what drips out is one thing you could actually go and check.
Same starting curiosity — but now there's a spout. One thing in, one clear question out. That narrowing is the skill.
Here's the gentle truth: the blurry wondering and the sharp question are about the same thing. You're not throwing your curiosity away when you narrow it — you're giving it a doorway. "How does sugar work?" is still alive inside "Does temperature change how fast sugar dissolves?" You've just picked one small slice of the big mystery that you can actually pull on. Answer enough small slices and, bit by bit, you really do learn how sugar works.
Read the question on top of the stack, then drop it into a bin — tap a bin, or drag the card left or right. A strong question is one you could answer with a measurement. A weak one is too vague, too big, or just an opinion. You'll get told why after every drop, so even a wrong guess teaches you something.
Notice the test you're running in your head: could I put a number on the answer? If yes, it's strong. If you'd just be sharing an opinion, it's vague.
If a few caught you out, you're in good company — the tricky ones are tricky on purpose. "Do bigger dogs eat more?" feels casual, but you can weigh the food, so it's strong. "Is the ocean beautiful?" sounds science-y next to the word ocean, but "beautiful" is an opinion, so it's weak. The wrapping fools you; the measurement never lies.
Once you know the four classic traps, weak questions stop being mysterious — you can spot exactly what's wrong and, better still, fix it. Tap a wobbly question below to see which trap it fell into and how to rescue it.
See the pattern? Every fix does the same thing — it trades a fuzzy word for something you could point a ruler, a clock, or a counter at.
Keep these four in your back pocket and you've basically got a checklist for your own questions. Too broad? Pick one slice. An opinion? Swap the judgement word for a measurement. Can't be measured? Find an observable stand-in. Already answered? Find the genuinely unknown edge. Run any wondering through those four and the weak ones reveal themselves fast.
Weak questions all fail in different ways, but strong ones are weirdly similar — they're built from the same three pieces. Once you can see the pieces, you can build a strong question on purpose instead of by luck. Tap each part to light it up in our sugar question.
Those three parts even have official names, and they're worth knowing because your teachers and your lab reports will use them. The thing you change on purpose is the independent variable. The thing you watch and put a number on is the dependent variable — it "depends" on what you changed. And the things you deliberately freeze so they can't mess up your results are the controlled variables. Change one, measure one, control the rest. That's the whole machine.
Take the winning question — does temperature change how fast sugar dissolves? — and you can actually run it. Two identical cups, same sugar, same water. The only difference is temperature. Press play and watch the sugar vanish. The cup that empties first, and the stopwatch ticking beside it, are your answer.
Warmer water dissolves sugar faster — and the stopwatch turns that "faster" into a number you can compare. That number is what makes the question answerable.
This is the moment a question proves itself. Slide the hot cup down toward 20°C and the two cups race almost neck-and-neck — the gap between the percentages nearly disappears. Slide it up to 90°C and the hot cup runs away. That growing-and-shrinking gap is data: real numbers that change when you change the temperature. A strong question is one that hands you numbers like these. A vague one leaves you with nothing to put on the graph.
And notice what makes it a fair test: both cups got the exact same sugar, the same water, the same starting moment. The only difference was the one thing you chose to change. That's why you can trust the answer — because temperature was the only suspect in the room. A vague question like "How does sugar work?" has no cups to fill and no stopwatch to read. There's nothing to point the experiment at. That's the difference you can feel.
Here's the whole skill in one machine. Start with a hopelessly vague question, then make three choices: one thing to change, one thing to measure, and the things you'll keep equal. Watch the fuzzy question grow up into something you could actually run.
The colours match the three parts of every strong question. Change the dropdowns and watch the sentence — and the highlights — rebuild themselves.
Notice you never had to know the answer in advance. You don't write "Does temperature make sugar dissolve faster — yes, it does." You write the question and let the experiment fill in the answer. A research question is a promise to go and find out, not a sneaky way of stating what you already think. If you catch yourself already certain, that's a sign to make the question genuinely open again.
These are the ideas that trip up almost everyone when they first start writing research questions. Each one feels right, which is exactly why it's worth un-believing.
"Bigger questions are better questions."
TruthIt feels grand to ask "How does the universe work?" — but a giant question has nothing single you can test, so it can't be answered by any one experiment. The best research questions are small and sharp, small enough to fit on a lab bench. Big discoveries are built from stacks of tiny, answerable questions, not from one enormous unanswerable one.
"A yes-or-no question is too simple to be real science."
Truth"Does adding salt change the temperature water boils at?" answers with a yes or a no — and it's a perfectly real experiment. Most science questions have this exact shape: does changing X affect Y? What makes a question serious isn't whether the answer is long. It's whether you can measure your way to the answer at all.
"You should already know what the answer will be."
TruthIf you already knew for sure, there'd be nothing left to find out — that's not a question, it's a fact you're checking. A good research question is genuinely open. Having a hunch about the answer is great (that hunch is called a hypothesis), but the experiment, not your hunch, gets the final word.
"Anything about feelings or opinions just can't be science."
TruthYou truly can't measure "the best song." But you can measure an observable stand-in: how many times people tap their foot, how long they keep listening, how many pick one tune over another. Science can't settle taste, but it's brilliant at turning a fuzzy feeling-question into a measurable one sitting right next to it.
Spot the thread running through all four? Good science isn't about asking the most impressive-sounding question. It's about asking the most answerable one — and then being honest enough to let the measurement decide.
"Change one thing, measure one thing, keep the rest equal" isn't only for science fairs. The grown-up version of a strong research question is quietly steering medicine, technology, sport, and even the apps on your phone.
Before a new medicine reaches anyone, doctors ask "does this pill lower a fever faster than the old one?" — then give it to one group and not another, keeping everything else the same. A measurable question keeps people safe.
App makers run tiny experiments constantly: "does a green button get more taps than a blue one?" Same app, one thing changed, taps counted. That's a research question with millions of test subjects who never noticed.
"Does a longer warm-up change how high players can jump?" A coach changes the warm-up, measures the jump in centimetres, and keeps the players and shoes the same. Better questions, better training.
Climate and ocean scientists ask measurable questions — "does this water hold less oxygen as it warms?" — and compare careful measurements over time. Huge problems, broken into testable slices.
And it's about to matter for you directly. In your IB science investigations, the very first thing you write — and one of the first things that gets marked — is your research question. A sharp, measurable one practically plans the experiment for you: it tells you what to set up, what to record, and what your graph's axes should be. A vague one leaves you staring at a blank lab bench wondering what you're even supposed to do. Getting this right isn't extra credit; it's the foundation the whole investigation stands on.
Last stop: prove the shape works no matter what you pour into it. Pick one thing to change and one thing to measure. The moment both are things you could put a number on, your question is testable — even if the pairing is a little silly.
See the trick? You never have to know the answer in advance. You just have to make sure there is one you could go and measure.
Try the silliest combination you can — "does the music's tempo change how fast a plant grows?" Even that is a real testable question, because both parts are measurable. It might give a boring answer (probably "no difference"), but a boring answer is still an answer. That's the magic: once the shape is right, the question works whether the idea is brilliant or bonkers.
Trade "How does sugar work?" for one small, sharp slice you can actually look at.
Aim at something you can put a number on — a time, a height, a temperature, a count.
Change one thing, measure the result, keep the rest equal. Now the answer is findable.