Almost every thesis states its limitations. Far fewer state their delimitations, and the difference costs marks at the defence. A limitation is a constraint you suffered — the sample was what it was, the instrument measures what it measures. A delimitation is a boundary you chose, and choosing a boundary is a claim about what your question requires. Examiners read the two very differently: a limitation invites sympathy, a delimitation invites a question, and unanswered boundary questions are the most common way a defence goes long.
This page separates the two properly, writes out six scope paragraphs by discipline, and gives you the template plus the questions committees ask about each kind of boundary. The paragraphs are shapes, not claims: no real study, [figure] where your own number belongs.
The distinction, made operational
| Delimitation | Limitation | |
|---|---|---|
| Origin | Your decision | A constraint on you |
| What it needs | A justification | An acknowledgement and an effect |
| Test | Could you have decided otherwise? | Could you have prevented it? |
| Typical wording | “This study is confined to…, because…” | “Because …, the findings cannot establish…” |
| Where it goes | Introduction, and restated in the methods chapter | Methods chapter and the discussion |
| Failure mode | Missing — so every boundary becomes a question | Generic — “small sample”, “time constraints” |
The test in row three settles almost every hard case. A single-province study is a delimitation if you decided one province was the right unit; it is a limitation if you wanted three and two refused. Same fact, different sentence, and an examiner can tell which one is true — so write the true one. Where a field has its own boundary conventions, follow those; the public health limitations guide works the distinction through for one discipline in depth.
The five axes a boundary can run along
- Population. Who is in, who is out, and what characteristic defines the edge.
- Setting or geography. One province, one network, one sector, one type of institution.
- Time. The observation window, and why it starts and stops where it does.
- Constructs. Which concepts the study addresses and which adjacent ones it deliberately does not.
- Method. What the chosen design can and cannot establish — association rather than causation, for instance.
A complete scope paragraph names a boundary on at least three of these. Fewer than three usually means the study has not been specified tightly enough to be finished in the time available.
Six written-out scope paragraphs
1. Nursing
This study is delimited to registered nurses on adult medical and surgical inpatient units, and excludes critical care, paediatrics, emergency and community settings, because the discharge-planning process under study is organised differently in each of those and combining them would make the exposure incoherent. It is delimited to one health network, chosen because the network operates a single electronic discharge record that makes the exposure measurable at all. The observation window covers
[figure]consecutive months within one fiscal year, so that no change in network policy falls inside it. The study addresses discharge preparedness and help-seeking; it does not address clinical outcomes, which would require record linkage beyond the approvals obtained. Because the design is observational, it establishes association and not causation.
What each sentence buys you. Sentence one removes “why not the ICU?” Sentence two turns a convenience constraint into a measurement argument, which is honest here because the single record genuinely is why the study is possible. Sentence three removes “did the policy change mid-study?” Sentence four removes “why not outcomes?” Sentence five removes the causal question before the discussion chapter has to.
2. Education
The study is delimited to secondary teachers of subjects carrying a provincially set examination, because the tension the study examines exists only where an external assessment competes with classroom assessment; teachers of subjects without such an examination are outside the scope rather than a comparison group. It is delimited to one school board. That boundary was not entirely chosen: approval was sought from three boards and granted by one, which makes it a limitation as well as a delimitation, and the study reports it as both. The study addresses enacted assessment practice and the interpretations behind it; it does not address student achievement, which the design cannot attribute.
The honest move. The second boundary is both a choice and a constraint, and the paragraph says so instead of dressing a refusal up as a design decision. Examiners rate that sentence highly precisely because most theses do the opposite.
3. Psychology
The study is delimited to community-dwelling adults aged 18 to 65; older adults and clinical populations are excluded because the instruments used were validated in general adult samples and their interpretation in those groups is not established. It is delimited to English-language administration, which excludes a substantial portion of the Canadian population and is a genuine boundary on generalisability rather than a technicality. The study addresses self-reported regulation and one behavioural index; it does not address physiological indices, which the available facilities cannot support. The design is cross-sectional and therefore establishes covariation at a single point rather than change.
Note the second sentence. In a Canadian thesis, a monolingual instrument is a boundary worth naming explicitly rather than burying. Saying it plainly is stronger than hoping the committee shares your assumption that it does not matter.
4. Civil engineering
The study is delimited to buried potable-water distribution segments and excludes transmission mains and wastewater infrastructure, whose failure mechanisms differ. It is delimited to one municipality’s asset register, and within it to segments with complete material, diameter and installation-year fields; incomplete records are excluded, and because incompleteness rises with age, the boundary skews the analysed population toward newer infrastructure. The observation window is
[figure]years, bounded at the start by the first year of consistent break recording and at the end by the most recent complete fiscal year. The study models technical inputs to renewal priority and does not model budget allocation or political decision processes.
Note the second sentence again. It states a delimitation and its consequence in the same breath, which is the highest form of this paragraph: a boundary whose effect on the results is named rather than left for the reader to infer.
5. Public policy
The study is delimited to federal programs with published evaluation reports, which excludes programs evaluated internally without publication and therefore cannot speak to whether publication itself affects responsiveness. It is delimited to evaluations published between
[figure]and[figure], a window chosen so that every case has at least three subsequent years of program documents against which change could be observed. The study measures documented change in program design and does not measure implementation or effect on beneficiaries.
The window justification. “Three subsequent years for every case” is a boundary derived from the measurement rather than from convenience, which is the strongest form of a time delimitation.
6. Computer science
The study is delimited to retrieval-augmented question answering over closed document collections, and excludes open-web retrieval, where the abstention problem does not arise in the same form. It is delimited to English-language corpora, since the evaluation resources required do not exist for the other languages of interest. The study evaluates abstention behaviour and answer accuracy; it does not evaluate latency, cost or user satisfaction, each of which would require a separate instrument. Results are reported for the model versions and configurations named in Chapter 3 and are not claimed to hold for later versions, which are not controllable by the researcher.
The last sentence is a modern necessity. Any thesis evaluating a commercial model needs a version boundary, because the artefact changes underneath you and an examiner reading a year later will be using a different one.
The template
- This study is delimited to [population]; [excluded group] is outside the scope because [reason grounded in the question, not in convenience].
- It is delimited to [setting or geography], chosen because [what that setting makes possible].
- The observation window is [period], bounded by [what fixes each end].
- The study addresses [constructs]; it does not address [the adjacent construct a reader will think of], because [reason].
- Because the design is [design], the study establishes [what it can] and not [what it cannot].
Sentence four is the one that stops defence questions, because it names the thing your reader was about to ask about. Write it by asking a colleague outside your area what they expected the study to cover and did not find.
Where the paragraph goes, and where it must match
The scope paragraph belongs in the introduction, as move six alongside significance — the full architecture is in the introduction chapter guide — and it is restated more technically in the methods chapter. Three places must agree word for word:
- the boundary in your introduction;
- the inclusion and exclusion criteria in your sampling section;
- the eligibility criteria in your ethics application and on your consent and recruitment materials.
A mismatch between those three is a real finding for an examiner, because it suggests the boundary was written after the data rather than before it. Read them side by side once, late, as a deliberate check.
Five things examiners ask about boundaries
- “Why this population and not the adjacent one?” Because the mechanism differs there — not because it was easier to reach.
- “Would your conclusion change one province over?” The honest answer is usually “it might, and here is the feature that would have to be the same”.
- “Is this a delimitation or a limitation?” Have the answer ready for every boundary. A boundary you present as a choice but that was really a refusal will be found out.
- “What did excluding that group do to your results?” Name the direction if you can. “Excluding incomplete records skews toward newer assets” is worth more than any hedge.
- “You say the design cannot establish causation. What would?” This is an invitation, not an attack. Answer it and you have written your future-research section.
Most boundary questions at a defence are predictable, and the full set of what to expect is in the defence questions guide. Boundaries that appear only in the discussion, having never been declared, are the ones that turn into long exchanges.
The relationship to your problem statement
Every boundary narrows the gap you claimed to be filling, which means your scope paragraph and your problem statement have to be consistent. If your problem statement asserts a national gap and your scope delimits to one network, one of them is wrong: either narrow the claim or widen the study. Doing this check while the study is still a proposal is the cheapest version of it; doing it after data collection means rewriting the introduction to match what you actually did, which is legitimate but demoralising.
Write the boundaries before the data
Tesify drafts the scope and delimitations section from your own design decisions, with the justification attached to each boundary, so the paragraph exists before a committee asks for it. Start your thesis with Tesify — 100% written by you.
Frequently asked questions
What is the difference between a limitation and a delimitation?
A delimitation is a boundary you chose and must justify; a limitation is a constraint you could not prevent and must acknowledge, along with its effect. The test is whether you could have decided otherwise.
Where does the scope section go?
In the introduction, usually alongside significance, and restated more technically in the methods chapter. Some programmes ask for a separate “Scope and Delimitations” heading.
How long should it be?
One or two paragraphs in the introduction. Five sentences that each name a boundary and its reason beat a page of hedging.
Can something be both a limitation and a delimitation?
Yes, and saying so is stronger than picking one. A single-site study that was also the only site that granted access is honestly reported as both.
Do qualitative studies have delimitations?
Yes. Which participants, which setting, which period and which aspects of the phenomenon are all boundaries, and they are as consequential as in quantitative work.
Is a small sample a delimitation?
Usually a limitation. It becomes a delimitation only if you deliberately chose depth over breadth, in which case say that and defend the choice.
Should I say my study is not generalisable?
Say precisely what it does and does not generalise to. A blanket “not generalisable” throws away the claim you are entitled to make about your accessible population.
Do delimitations belong in the abstract?
The most consequential one usually does, because it tells a reader in a sentence whether the study is about them.
What if a boundary changed during the study?
Report the change, when it happened and why, in the methods chapter. A boundary that moved silently is the version that causes trouble.
How do I know I have named enough boundaries?
Ask someone outside your area what they expected the study to cover. Anything they name that you did not address is a boundary you have not yet declared.