A topic is a claim about evidence you can actually obtain.
Interest is the wrong first filter. A dissertation topic is, before anything else, a statement about a body of evidence you will be permitted to collect or obtain, from people or records that exist, inside a window of time you control. Candidates who choose in the opposite order, subject first and evidence afterward, tend to meet the constraint in their second year, when changing it costs a committee meeting and most of a term.
The first draft of a topic is therefore best written as two sentences rather than as a title. One names the thing to be explained, compared or improved. The other names where the data for it lives, who controls access to it, and what standing you have to ask. If the second sentence cannot be written today, there is no topic yet, however well the first sentence reads.
Five tests, run before anyone becomes attached.
These are not tests of quality. A topic can pass all five and still be dull, and dullness is survivable; failing any one of them usually is not. Run them in the order given, since the later tests cost nothing once an earlier one has closed the question.
- Access. Name the participants, records or dataset, name the person who controls them, and state what would have to be true for that person to agree.
- Bounds. State the population, the setting and the timeframe in one sentence. A subject that will not fit into one sentence has been described rather than narrowed.
- Literature. There should be enough prior work to argue from and a visible gap to argue into. A subject with almost nothing published behind it is a career, not a dissertation.
- Method. Choose a design you could execute without a research team, using software your program teaches, and defend under questioning. Ambition in method is where timelines quietly die.
- Fit. The topic must sit inside somebody's expertise on your faculty, because a chair who cannot supervise it will not agree to chair it, and should not.
The scope error, and the four axes that correct it.
Almost every first topic is too large, which is a good sign rather than a bad one; broad thinking narrows more easily than narrow thinking expands. The correction is mechanical rather than inspired. Take the working topic and cut it along one axis at a time, checking after each cut whether the study has become executable by one person.
Two cuts are usually enough and three almost always are. A subject that still looks unmanageable after three is telling you the design rather than the scope is the difficulty, which is a conversation to have with a methodologist before the proposal is drafted.
- Population: one professional group, cohort or role rather than a whole sector
- Setting: one system, district or organization rather than a state or a region
- Timeframe: a defined window with a start and an end rather than a trend over years
- Construct: one outcome measured properly rather than four measured thinly
Where the review board quietly rewrites a topic.
A topic is also a proposal to an institutional review board, and the board's categories deserve consulting while the topic is still soft. Work involving minors, patients, people who cannot consent for themselves, or identifiable health and education records travels a longer road than work using publicly available data or consenting adults answering anonymous questions. None of that makes a topic impossible; all of it changes the calendar.
The commonest surprise is the candidate's own workplace. Recruiting people you supervise raises a question about voluntary participation that must be answered inside the protocol rather than waved at, and gathering data at a site you do not own requires that site's written permission before a board will open the file. Both consume weeks that a topic chosen in ignorance never budgeted for, and the documents themselves are unforgiving, as the anatomy of a consent form sets out.
The one-paragraph test, which settles it.
Before committing, write four sentences in a single paragraph: the problem, the purpose, the question or questions, and the method that answers them. Read it aloud. Each sentence should follow from the one before it, and the method sentence should obviously answer the question sentence rather than merely sit beside it. Misalignment is the most frequent reason a proposal comes back, and it is visible in a paragraph long before it is visible across three chapters.
Then hand the paragraph to somebody outside your field and ask them to say back what you intend to do. If they cannot, the topic is not yet clear enough to defend, and clarity is cheap at this stage and expensive at every later one. Reading the gap in the literature well enough to name it is its own skill, and a literature review that holds covers it. Where the paragraph is sound but the chapters after it stall, that is the ordinary work of the capstone and dissertation section, taken one chapter and one revision at a time.
Questions to the desk.
How narrow is too narrow for a dissertation?
A topic is too narrow when there is no literature to situate it in and no reader who would care about the answer. Those are the only two limits that matter. Narrowness itself is rarely the problem in doctoral work; committees return far more proposals for being unbounded than for being small, and a modest question answered well is a defensible study.
Can I change the topic after the proposal is approved?
Yes, though the cost rises steeply with time. A change before data collection means redrafting chapters and returning to the review board with an amendment. A change afterward usually means starting the empirical work again. Where the shift is small, in setting or sample rather than in question, an amendment often covers it, so raise the possibility with your chair as soon as it occurs to you.
Should the topic come from my own workplace?
It often should, because you have context, access and a genuine problem to solve. Two conditions apply. You need written site permission from someone with authority to give it, and you need a recruitment plan that does not lean on your position over potential participants. Both are solvable, and both should be settled before the topic is treated as final.
What if my chair and a committee member want different directions?
Ask for the disagreement to be made explicit in one meeting rather than resolved in separate emails. Usually the two directions differ in method rather than in question, and a study can be designed to satisfy both. Where they genuinely conflict, the chair's direction governs, since the chair carries the study to defense. Write down whatever is agreed and circulate it the same day.
Does a dissertation topic have to be original?
It has to be a contribution, which is a lower bar than novelty. Applying an established instrument to a population nobody has studied, replicating a finding in a different setting, or testing a familiar intervention under conditions that were previously assumed all count. What will not pass is a study whose answer is already known and unchallenged, since nothing follows from confirming it again.
