01Three candidates from one hospital
The first was a national comparison built on a dataset she would have to apply for, held by an organisation she had never dealt with. The second was interviews with clinicians across three employers, none of whom she worked for. The third was a records question inside her own unit, on a problem her manager had already described out loud as worth studying.
She listed them in that order because that is the order she liked them in, which is how most topic lists get written and why most of them need rebuilding. The national one felt like a real contribution. The interviews felt manageable. The unit one felt small, close to home and slightly disappointing, and it was the only one of the three that could actually be finished inside the time she had.
02The same questions, asked of each
Ask these before you become attached to anything, and ask them of a person who has actually done it at that organisation rather than of your own optimism. The answers take a week to gather and they save entire terms.
- Who physically holds the data, and is that a person, a department or an institution
- Who has to sign before you may see it, and does that person currently know you exist
- How long does that signature usually take there, according to somebody who has obtained one
- What happens to the project if the answer is no, and how long does finding out take
- Does a second approval sit behind the first, and does it start only once the first has finished
- Would the question still be worth asking if the sample were half the size you imagined
03The letter you obtain before you commit
Get a short letter of support from whoever controls the site or the data, and get it before the topic is defended rather than after. One paragraph is enough: that the organisation is aware of the proposed work, that the named person supports it, and that access to the relevant records or staff would be permitted subject to review. If nobody will write that paragraph, the topic is not yet a topic.
Two things happen when you obtain it early. A hopeful assumption becomes a dated document, which is a different kind of object entirely at a proposal meeting. And you have already had the conversation the review board will later want evidence of, which means the packet is stronger and the timeline you present is one you can defend.
Ask for the role as well as the name on it. Access resting on one supportive manager ends the week that manager takes another post, and posts change during a doctoral timeline far more often than a plan allows for. A letter that says the nurse manager of a named unit supports the work survives a reorganisation. A letter naming only an individual has to be sought again from a successor who agreed to nothing.
04Narrow by population, not by curiosity
When a topic is too big, the instinct is to make the question vaguer, which makes it worse. Shrink the population and the timeframe instead and leave the question exactly as sharp as it was. One unit rather than a system, one year rather than five, one cohort rather than a profession. The question survives intact and the approval chain shortens along with the collection, because fewer organisations means fewer signatures.
The doctoral detours worth fearing are almost never caused by a difficult question. They are caused by a topic whose access was assumed, discovered a year in, at the point where a dataset application is refused or a second organisation declines to participate. Interest can be added to a small accessible question. Access cannot be added to a large inaccessible one.
Where a topic has to be tested against a real timeline before it is defended, that assessment happens at the start of capstone and dissertation work rather than at chapter three.
FAQQuestions to control
Should I pick a topic my chair is interested in?
It helps considerably, and it is not the first filter. A chair who finds the question interesting reads drafts faster and defends your choices at a committee. A chair enthusiastic about a topic you cannot get access to will still be enthusiastic while you lose a year. Run the access questions first, then choose among the survivors with your chair's interests in mind.
Is using my own workplace a conflict of interest?
It can be, and it is usually manageable once declared. The concerns are recruiting people you supervise, using influence to obtain consent, and access to records your role gives you for other purposes. Declare the relationship in the submission, describe the safeguards, and let the board judge. Boards handle workplace research constantly; what they object to is discovering the relationship themselves.
How narrow is too narrow?
When the finding could not interest anyone who does not work in that building. That is a genuine limit and it sits much further down than most students think. A single unit, a single cohort or a single year is normal doctoral scope, provided the question connects to something the literature is already arguing about. The connection is what makes a small study meaningful.
What if my topic changes after the proposal?
Small changes are ordinary and expected. Changes that alter who the participants are, what they will do, or where the data comes from are a different matter, because approvals attach to those specifics. Any of them means going back to whoever approved the original, and the honest way to plan is to treat an approved design as a commitment rather than a starting position.
