How to write a research question you can actually answer

A research question is a promise. It tells the reader what you will find out, and it tells you what you can stop reading about. Students often skip the step of writing one down carefully, because the topic feels obvious. The topic is never the question.

Start with a topic, then refuse to stop there

"Microplastics" is a topic. "Sleep and teenagers" is a topic. "Mycelium as a building material" is a topic. None of them can be investigated, because none of them say what you want to know, about whom, measured how.

Write your topic at the top of a page. Underneath it, write ten questions about it, as fast as you can, without judging them. Some will be enormous ("Are microplastics harmful?"). Some will be trivially factual ("How many microplastics are in Taipei tap water?"). Most will be somewhere in between. The point of the exercise is to see the range.

Narrow along four axes

A question becomes investigable when you narrow it along at least some of these:

1. Population or object. Not "teenagers" but "senior high school students in Taipei who commute more than 40 minutes."
2. Variable or phenomenon. Not "sleep" but "self-reported sleep duration on school nights."
3. Comparison. Not "is it bad" but "compared with students who commute under 20 minutes."
4. Context. A place, a period, a material, a species, a dataset.

A narrowed version of the sleep topic might read: Among senior high school students in Taipei, is commute length associated with self-reported sleep duration on school nights, and does the association differ between students who use public transport and those driven by family?

It is long. That is fine. A good question is often a sentence with several clauses, because it carries the design of the study inside it.

Test it against three criteria

Can you answer it with what you have? Access to people, a dataset, a laboratory, a material, a collection of texts. If the answer requires a resource you do not have, narrow again or change the method.

Would the answer change anything? Ask who would care about the result and what they would do differently. If no one, the question is a puzzle, not a project.

Does it have more than one possible answer? If you already know what you will find, you are writing an essay to confirm a belief, not conducting research. Research questions must be able to come out the other way.

Write it down, and keep rewriting it

Your question will change. Reading the literature will show you that part of it has been answered, or that the terms you used mean something different in the field. Early data will reveal that a variable is impossible to measure. Each time, rewrite the question and date the version. The log of revisions is itself evidence of thinking, and mentors find it more useful than any single draft.

A short checklist

• The question names a population or object, a phenomenon, and a context.
• It can be answered with resources you can reach in the time you have.
• Someone outside your project would want to know the answer.
• The answer is not already obvious.
• You have written down the current version with a date.