Scoping a research question so it survives the first year

The most common reason a thesis stalls in year two isn't a missing dataset or a slow advisor — it is a research question that was never scopable in the first place. Big, unfalsifiable, no obvious pivot if the data refuses to cooperate. This guide walks through the rubric we use across Research Goal's methodology cohorts to test a draft question before you've sunk a semester into it.

What "scopable" actually means#

Most first-year questions fail on the same three axes. Naming them makes the failure visible — and fixable — in an afternoon rather than at a committee meeting.

1. Tractability#

Can a competent peer in your shoes answer this question with the resources you actually have? Not in principle — with what is on your laptop and on your campus. If the honest answer is "only if I get a Nature paper's worth of data," the question is not yet tractable. Shrink the scope until a single Master's thesis could plausibly close it.

2. Falsifiability#

Write down what result would change your mind. If you cannot, the question is not yet a research question — it is an exploration. Explorations are fine, but they belong in pilot studies, not theses. A scopable question has a finding that is plausibly wrong.

3. Pivot distance#

How far is the nearest adjacent question you could pivot to if the data, the recruiting, or the field collapses under you? A robust question has a clear pivot one step away — same methods, same instrument, same broad area, slightly reframed. A fragile question has no pivot, which is why one supply-chain hiccup ends the project.

Three diagnostic questions#

Once a draft passes the three axes above, run it through these three diagnostics. We use them at the first cohort milestone and the second milestone four months in — the same question often passes one but not the other.

1. Could a competent peer answer this in your shoes?#

Hand the one-sentence question to a fellow graduate student in a neighbouring lab. Could they, with your data and your methods, produce a defensible answer in twelve months? If not, the question is leaning too hard on you specifically — usually because it is too narrow on the case and too broad on the claim.

2. What evidence would change your mind?#

Write the null result before you collect anything. One sentence: "I expect to find X; if I find ~X, I will conclude Y." If the ~X case leaves you with nothing to say, the question is asymmetric — and reviewers will notice it before you do.

3. How far is the nearest adjacent question?#

Name the pivot. "If recruiting collapses in Site A, I move to Site B and ask the same question." "If the registry refuses access, I reframe as a methods paper using simulated data." If you cannot name a single pivot in one sentence, you do not yet have a robust question — you have a single point of failure.

Common refinement traps#

Even with the rubric, the same handful of failure modes appear in nearly every cohort. They look like progress but they are not.

• Narrowing the case but not the claim — studying one hospital but still claiming "in healthcare systems"
• Adding clauses to fix scope — every "and also…" doubles the work without sharpening the question
• Confusing a method with a question — "I will run a random-effects model" is a method, not what you want to know
• Letting access drive the question — the dataset you can get is the constraint, not the question
• Polishing the question instead of testing it — twelve revisions before a single conversation with the field is a tell

Wrapping up#

Spend a disproportionate amount of time scoping. Two solid weeks of question-work in the first semester saves twelve solid months of redirection in the second year. The rubric — tractability, falsifiability, pivot distance — and the three diagnostics give you something better than a polished sentence: they give you a question that can actually be finished.