↗
Where gaps genuinely are
Examine the taxonomy, because this is the most useful thing a candidate can be told and the thing they most often get wrong. Gaps are not subject areas. They are narrower than that, and they come in about eight recognisable forms. A population studied elsewhere but not in this setting. A place studied, but not with this population. A measure, where the same concept has been measured badly or with an instrument nobody has checked locally. A mechanism, where the prevalence of a problem is well established and the explanation is not. A period, where everything is known about before a change and after it and nobody has looked during. An outcome nobody chose to measure because everyone measures the same one. A combination two bodies of work that have not been put together. And a method that works elsewhere and has not been applied here. Then the test: a gap you can state in one sentence, in those terms, is a gap. A subject you like is not.
◎
Has it already been done?
Examine the literature test, which candidates run in the wrong direction. The useful question is not whether there is writing on your topic, because there always is and its absence is not a frontier but a desert with no theory in it. The useful question is whether the specific study you have in mind has been done, and the way to find out is to search for it rather than for the topic. Type your proposed study as a phrase, not your subject, and read the first two pages. If several teams have already asked it and answered it, the claim is not available to you however open the subject seems. If work exists but measured the wrong thing, or in the wrong place, or without explaining a condition, you have found the gap. The second search, which catches more, is to look at the closing paragraphs of the closest work, where a limitation is usually named and is often exactly your study.
✦
Supersized topics
Examine the most common single fault, which is a topic that is a field wearing a question mark. The tells are recognisable. It contains words like impact, effects, challenges, perceptions and the role of, which are not operations. It names a population with a word like youth, students or staff and no cohort, so there is no way to count them. It covers a period nobody could cover, or every level of a system from household to state. It cannot be finished by one person in one academic year, and its answer would be a report rather than a study. The narrowing move is mechanical. Replace the population word with a countable group, replace the general outcome with one you could measure, replace the span with a period, and add the place. If what survives is still a field, you have not narrowed far enough. If what survives is too small to matter, widen the cohort rather than the question.
◌
The problem with a gap you cannot reach
Examine a gap that is real but unreachable, which is the failure that wastes the most time. A real gap is still the wrong gap if you cannot reach the evidence. So the second test after finding a gap is reachability, and the two have to pass together. Then the trap, which is staying loyal to a gap because you found it first and it seems like yours.
↗
Topic against research question
Examine the difference, because candidates often think choosing a topic means writing a question and end up with neither. A topic is a subject bounded enough to study: use of a course management system by final-year undergraduates at one Nigerian university. A research question is what you will ask about it: what factors are associated with how often those students use the platform. A study has both, plus a design and a claim, and the sequence matters because each stage narrows the last. Choosing the topic first and the question later is normal and often right, because the question is what your access and your design can actually answer. Choosing the question first and the topic never is the common failure, because a good question in a field you cannot reach is not a project. A useful discipline is to write the topic, then the question, then check that the topic could have supported a different question.
◎
The supervisor is a constraint too
Examine supervision as a selection criterion, because it is one and it is treated as an afterthought. A candidate who picks a topic nobody locally can supervise is stuck in a way that is expensive to undo, since changing supervisors late is harder than changing topics late. So before you commit, find out who here can supervise it, and ask them, in plain terms, whether the topic is one they would want to supervise. A second useful question is whether they have supervised anything similar, since it tells you what you will be held to. And a third, which is the one candidates skip, is who else in the department could read your work if that person is unavailable.
✦
The four readiness questions
Examine the questions, because a topic that passes all four is ready and a topic that fails one is not, whatever its merits. Can you state the study in one sentence to someone outside your field. Can you name the three papers closest to it, and say what each of them did not do. Can you say what you would do if your first finding came back negative, which is the question that tells a study from a fishing trip. And can you name who will let you near the data, and by when. Three out of four is a topic to develop; two out of four is a topic to replace. The test takes an hour, and running it in week one saves a term.
✓
Having a second and a third
Examine the reserve, which takes very little and is the one candidates refuse to make. Two further topics, at the same level of seriousness and with the same access checks done, turn any of those from a disaster into a delay. Then the practical detail that makes this work. The reserve has to be genuinely ready, not a list of ideas, so it needs the same four questions answered. And it should be adjacent enough that the reading you have already done transfers, which is why a second topic in the same field is worth more than an exciting one in another. Then the habit worth forming, which is to write your selection decisions down as you go, because the reasons are what you need when you have to choose again.