Before You Settle on a Dissertation Topic, Test the Evidence
A Promising Title Can Hide an Unanswerable Question
In this illustrative September 2026 scenario, final-year student Rohan proposes a dissertation on whether university sustainability campaigns change student behaviour. He has found campaign posters and public strategy documents. He assumes the university will also share purchasing records and help him recruit students for interviews.
Rohan, his project and the institution are fictional. His proposal contains a useful decision to examine: does the evidence he can obtain support the question he wants to answer? Posters can show what a campaign says. They cannot, by themselves, establish whether students changed their behaviour because of it.
Before polishing the title, he needs to separate material already available from material he hopes someone will provide. That distinction can change the entire project.
Write Down What Would Count as an Answer
Start with a sentence describing the result your research could reasonably produce. For Rohan, “describe the messages used in a defined set of campaign materials” requires a different design from “measure the campaign's effect on purchases.” Neither becomes stronger simply by sounding more ambitious.
The University of Leeds Library's guidance on starting a research project, checked on 17 September 2026, asks students to consider whether their question is manageable, whether necessary resources are accessible, and whether they have the time and skills for the method. It also recommends discussing topic ideas with a supervisor early. These are useful planning prompts; your own department determines the requirements for your dissertation.
Try this working sentence: “I will examine this material, within these boundaries, using this approach, to answer this question.” An empty slot is a decision to resolve before committing to the project.
Separate Available Evidence From Assumed Access
For the fictional proposal, Rohan makes three short entries:
- Public campaign documents: He can open them now, but still needs a clear rule for which documents belong in the study.
- Purchasing records: Access, coverage and the meaning of each field are unconfirmed. He cannot yet treat these as the foundation of his analysis.
- Student interviews: Recruitment, the interview design and the department's applicable approval process need discussion before he approaches participants.
This record should describe the actual status of each source. “A staff member might help” is different from permission to use a defined dataset. Even confirmed access does not establish that the material contains the information your question needs.
Ask your supervisor which permissions and research approvals apply, and follow the institution's process before starting activities that require them. Public availability alone is not a substitute for checking the proposed use of material.
Test One Small Piece Before Expanding the Project
Rohan's next step is a limited feasibility exercise with material he is allowed to examine. He selects two fictional campaign documents and tries a draft coding sheet: document date, intended audience, requested action and evidence offered for the claim. He records where a category is unclear rather than forcing every passage into it.
The exercise reveals that one document promotes a reusable cup scheme while another reports campus-wide waste figures. Combining them as if they measure the same activity would weaken the analysis. He now has a concrete scope problem to discuss: should he restrict the study to communications about one campaign?
This trial is not a completed study or evidence that the campaign worked. Its purpose is to expose practical problems while the proposal is still easy to change. Keep the trial material, your notes and unresolved questions so the reasoning behind the final design remains visible.
For a numerical project, a comparable check might be opening an authorised sample and inspecting its variables, dates and missing values. For an archival project, it might be confirming that the relevant records exist and can be consulted. Choose a check that tests the weakest assumption in your own proposal.
Take Two Bounded Options to the Supervisor
Rohan prepares two options for discussion. One retains the behavioural question but depends on suitable evidence and an agreed design for interpreting it. The other asks how a specified campaign presents individual responsibility in its public materials. That second option changes what the dissertation could conclude, so it needs academic review rather than a quiet substitution halfway through the work.
For each option, write a short scope note: the question, included material, excluded material, intended analysis and unresolved dependency. Add a date by which a dependency must be resolved to leave enough time for the remaining work. Agree that date with the supervisor around the actual course timetable.
If access falls through, revisit the question and method together. Removing an unavailable dataset while keeping the original claim can leave a proposal that no longer has a way to answer itself.
Where Edupath Could Help
A proposed dissertation planning feature in Edupath's Learning Path could keep the research question beside its evidence requirements and unresolved access decisions. A student could attach a feasibility note, record a supervisor discussion and mark a dependency as confirmed only when there is a supporting response. These are proposed capabilities, not claims about features currently available.
Institution support could provide links to the relevant departmental handbook, research support contact and approval process. The academic decision would remain with the student and authorised university staff. A completed checklist would show preparation, not certify that a method is valid or a proposal approved.
For your next planning session, choose the least certain part of your topic. Inspect one permissible sample or identify the person who can resolve access. Bring the result to your supervisor alongside a question whose boundaries you can explain.