A good scientific question is one that can be answered through observation or experimentation, tested against evidence, and potentially proven wrong. That last part is what separates science from philosophy, opinion, or belief. If no possible result could disprove your idea, it’s not a scientific question. If you can design an experiment or collect data to find out, it is.
Whether you’re working on a school project or designing a real study, the difference between a weak question and a strong one comes down to a few concrete traits.
What Makes a Question Scientific
A scientific question has four core properties: it’s focused, it’s answerable through investigation, it addresses a gap in what we currently know, and it leads to a hypothesis you can test. That testability requirement is the most important filter. The philosopher Karl Popper called this “falsifiability,” meaning a true scientific claim must make predictions that could be disproven by experimental results. If your question doesn’t allow for that possibility, it belongs to a different category of inquiry.
Consider these examples of questions that fail the test:
- “Which animal is the best animal?” This is opinion-based. No experiment can determine “best” because the criteria are subjective.
- “Can ghosts move objects?” This refers to a supernatural phenomenon that can’t be measured or controlled in an experiment.
- “Whom should you vote for?” This relies on moral and social values, not measurable evidence.
A nonscientific question typically falls into one of those three buckets: it’s based on opinion, it involves moral or social values, or it refers to something that can’t be measured. A scientific question, by contrast, asks about phenomena you can observe, quantify, and repeat.
Narrow Focus Beats a Broad Topic
One of the most common mistakes is asking something too broad. “What causes cancer?” is technically scientific, but it’s so vast that no single study could address it. A stronger version narrows the scope: “Does daily exposure to a specific chemical increase tumor growth rates in lab mice?” That question identifies exactly what you’re changing, what you’re measuring, and in whom.
This is where variables come in. A well-built scientific question contains two key ingredients. The independent variable is the factor you’re manipulating or examining. The dependent variable is the outcome you’re measuring. In the example “Do high concentrations of vehicle exhaust increase asthma rates in children?”, vehicle exhaust concentration is the independent variable and asthma incidence is the dependent variable. If your question doesn’t have a clear relationship between a cause and an effect, it probably needs tightening.
The FINER Framework
Researchers use a checklist called FINER to evaluate whether a question is worth pursuing. Each letter represents a quality your question should have.
- Feasible: Can you actually carry out this study? Do you have the time, money, equipment, and access to enough participants? A question about deep-sea organisms is scientifically valid but not feasible if you don’t have a submarine. Feasibility also includes whether you can recruit enough subjects and measure outcomes reliably.
- Interesting: Does this question matter to anyone beyond you? A good question engages other researchers or presents a fresh angle on a known problem.
- Novel: Does it fill a gap in existing knowledge? Repeating a study that’s already been done dozens of times with the same design isn’t contributing much. Your question should push understanding forward, even incrementally.
- Ethical: Can you investigate this without causing unjustified harm? Some questions are scientifically testable but ethically off-limits. You can’t deliberately expose people to a toxin to see what happens. Ethical review boards weigh potential risks against potential benefits before any human research moves forward.
- Relevant: Will the answer be useful? The strongest questions produce results that change how people practice medicine, design technology, manage ecosystems, or understand the world.
You don’t need to use FINER formally for a class assignment, but running your question through these five filters quickly reveals whether it’s ready or needs work.
How to Structure a Question Using PICO
In health and clinical research, there’s a more specific formula called PICO that helps you build a question piece by piece. It stands for Population, Intervention, Comparison, and Outcome.
- Population: Who are you studying? This could be defined by age, disease, geography, or another characteristic.
- Intervention: What treatment, exposure, or change are you testing?
- Comparison: What are you comparing it against? This might be a placebo, standard treatment, or no treatment at all.
- Outcome: What result are you measuring?
A vague health question like “Does exercise help with depression?” becomes much stronger through the PICO lens: “In adults aged 30 to 50 with moderate depression (population), does 30 minutes of daily aerobic exercise (intervention) compared to no structured activity (comparison) reduce self-reported depression scores after 12 weeks (outcome)?” Every piece of the question now points toward a specific, measurable study.
Turning a Topic Into a Testable Question
Most people start with a topic, not a question. You’re curious about sleep, or plant growth, or screen time. The work is in converting that curiosity into something precise enough to investigate. Here’s a simple process that works at any level.
Start with your broad interest: “I’m curious about how light affects plants.” Then add specificity by identifying your variables. What kind of light? What kind of plant? What outcome are you measuring? You might land on: “Does the color of light (red vs. blue) affect the height of basil seedlings over 21 days?” That question is focused, testable, and measurable. You know exactly what to set up, what to change, and what to record.
A few quick checks can tell you if your question is ready. Can you imagine collecting data to answer it? Could the results go either way? Is it specific enough that two people would design roughly the same experiment from it? If the answer to all three is yes, you have a workable scientific question.
Mistakes That Weaken a Question
Beyond being too broad, several other pitfalls trip people up. Asking a yes/no question without a measurable dimension is one. “Does music affect mood?” technically has a testable core, but it’s so open-ended that the results won’t tell you much. Specifying the type of music, the population, and how you’ll measure mood transforms it into something useful.
Another common error is embedding an assumption into the question. “Why does sugar make children hyperactive?” presupposes a cause-and-effect relationship that hasn’t been established. A better version asks whether the relationship exists at all: “Does consuming 50 grams of sugar increase physical activity levels in children aged 6 to 10 over a two-hour period?”
Finally, some questions are simply not feasible given your resources. Feasibility research shows that practical constraints like cost, time, recruitment challenges, and available equipment are real barriers. A question you can’t afford to answer or don’t have the tools to measure isn’t a bad question in principle, but it’s not one you can act on right now. Scaling it down to something you can investigate with what you have is always better than designing a study you’ll never complete.