Starting research

“I have read for a year. Why do I still not have a research question?”

Turning a gap area into a specific, finishable question.

You have read widely and thought hard, and somehow the research question is further away than when you started; every paper opens three more. That is not a failure of effort. Reading maps the territory. It does not, by itself, produce a question.

By Heidi Andersén, MD, PhD · Updated 25 July 2026 · 7 min read

“I’m looking now for over a year for a feasible research question. The more I work and read, the more I feel away from an actual research question.” — a doctoral student, a year in

The problemMore reading is not a question.

The trap is intuitive: if you do not have a question yet, read more. But reading widens the field faster than it narrows it. Each paper adds context, caveats, and adjacent problems. A gap area (“not much is known about X in setting Y”) is not the same as a research question (“does A change B among C?”). The area is where you are standing; the question is a specific, answerable thing you decide to do inside it. You get there by choosing and constraining, not by reading one more review.

Why it mattersAn open question has no finish line.

Without a specific question, there is nothing to design, no obvious data, and no way to know when you are done. That is exactly why a year can pass. A one-sentence question is the most useful thing you can produce early, because it converts an open-ended reading project into a study.

Common mistakesFour ways the question stays out of reach

Waiting for it to “emerge”. Questions are decided, not discovered. At some point you choose one and commit.
Confusing a topic with a question. “Burnout in nurses” is a topic. “Does X predict burnout among first-year nurses?” is a question.
Skipping feasibility. A perfect question you cannot answer with the data, time, or access you have is not usable. Finishable beats interesting.
Never pressure-testing it. A question that has not met your supervisor’s objections is untested. Better to hear them now than at the viva.

Stop reading. Start choosing.

The moveGap area to finishable question

Name the gap in one sentence.
Not a paragraph. “We don’t know whether [thing] affects [outcome] in [group].” If you cannot compress it to one sentence, the gap is still an area.
Turn it into a question with named parts.
Use a frame that forces specifics: PICO for clinical or quantitative work (population, intervention or exposure, comparison, outcome), or for qualitative work, whose experience, of what, in what setting.
Constrain until it is finishable.
Narrow the population, the setting, the time window, until you could actually collect or access the data within your project. Each cut makes it more doable and more defensible.
Check it is answerable and worth answering.
Can existing methods answer it? Would the answer matter either way it comes out? A question whose answer is already obvious, or unmeasurable, is not yet the one.
Pressure-test it with your supervisor.
Take the one-sentence question and invite the hardest objection. If it survives “hasn’t this been done?” and “can you actually get that data?”, you have a question, not an area.

Worked exampleFrom a year of reading to one line

Narrowing until it finishes

A year of reading left one broad line: moral distress in healthcare workers is under-studied. That is an area, too broad to answer and never finished. The gap sentence inside it: we do not yet know how newly qualified nurses experience moral distress in their first year. And the question inside that: how do first-year nurses in acute hospital wards experience and respond to moral distress?

Constrain that to two hospitals and the first twelve months, and a 20-interview study could finish it. It survives the two hardest questions too: it is novel (the first-year focus is thin in the literature) and the data is reachable (access agreed at both sites). Committed.

ChecklistRefine your own question

Four checks before you commit

Write the gap as a single sentence, then turn it into a question with named parts (PICO, or whose experience, of what, in what setting).
Constrain the population, setting, and time until you could actually collect or reach the data.
Check existing methods can answer it, and that the answer matters either way it comes out.
Pressure-test it against your supervisor’s two hardest objections before you commit.