Where Limitations Are Placed, and How Long They Run
Six of seven examined works state limitations. Not one puts them where the teaching material says they go. The shortest is the best, the longest is the least disciplined, and one describes the limitations of an experiment that was never attempted.
Seven works, seven answers
The section skeleton these works share has no heading for limitations — see The Thesis Pro-Forma for what else it omits. What happens next is instructive precisely because the form does not decide it: four researchers created a heading, two wrote unheaded prose, and one wrote nothing at all.
PLACEMENT, LENGTH AND CONTENT ACROSS SEVEN EXAMINED WORKS
| Work | Heading? | In the contents? | Placement | Words | % of document | About the study actually run? |
|---|---|---|---|---|---|---|
| F | No — an unheaded bold line | No | Inside Conclusions and further research | 364 | 3.1% | Yes — six, each causal |
| G | None at all | — | — | 0 | 0% | No — none stated anywhere |
| H | Yes, 6.2 | Yes | Inside Conclusions and further research | 210 | 2.0% | No — describes an experiment never attempted |
| J | Yes, 6.2 | Yes | Inside Conclusions and further research | 151 | 1.4% | Partly — three real, one displaced |
| K | Yes, 6.1 | Yes | Inside Conclusions and further research | 295 | 2.8% | Yes — four, each causal, three converted into further research |
| L | No | No | Split: Introduction plus the last two paragraphs of the Conclusion | 369 + ~200 | ~6.3% | Partly — five real, two misplaced, one is a defence |
| M | Yes, 8.2 | Yes | Inside Conclusions and further research | 105 | 1.2% | Yes — three, all real, two causally linked |
Word counts and percentages are this library's, taken by counting the works' own text. They describe seven examined and passed documents; none of them is a target.
All of them are in the wrong chapter
Of the four works examined first, three state limitations and all three place them inside the Conclusions chapter; the three works examined afterwards add two more in the same place and one split between the Introduction and the Conclusion. The direction is unanimous across two independently produced groups. This is not a filing quibble. Placement determines what a limitation can still do. A limitation in the Discussion constrains the claims being made as they are made. A limitation in the Conclusions arrives as a closing concession, after the Discussion has already made its claims, and the reader has to go back and re-weigh arguments that have already landed.
The corpus shows exactly that happening, three times over:
- One work's Discussion asserts that the organisation studied is not equipped for the method in question — and its limitations then note that 35% of respondents had been employed there for under twelve months and "may not be able to provide accurate testimony" about that environment.
- One work's Conclusion claims superior performance on five project deliverables, and two paragraphs later concedes that the framework it credits "can be assumed with a high degree of confidence to not be the sole contributing factor".
- One work's Conclusions state that all four objectives were "successfully tackled", 400 words before the limitations state that "an accurate snapshot would require a much larger sample size to ensure accuracy".
In each case the limitation is correct, honestly stated, and arrives after the claim it should have shaped. If you write only one sentence differently as a result of this page, move your limitations forward — see Writing the Discussion Section for what a discussion is supposed to be doing when they arrive.
The limitations of an experiment that was never attempted
One work is a literature-only study. It collects data from nobody: no participants, no instrument, no sample. Its limitations section is a genuine sub-heading, appears in the contents page, and runs to 210 words — 2.0% of the document. It reads, in order:
- The design, restated.
- "…the sample size and sample profile are among the limitations of this study."
- What would be needed instead: a specific sample profile and a large sample size applying the same model to the same change — "however the results will still be somewhat skewed as it is nigh on impossible to have precisely the same organisations to run the same experiment on that the results would be entirely unbiased."
- "Time and financial resources are the other limitations of this study."
- What that would require: experiments in several organisations, on each model, "over the span of many years", plus "organisations who are willing to grant access to their sensitive, private workings and information".
The diagnostic is simple enough to apply to your own draft. Read each limitation and ask: does this describe the study in this document? If it describes a better study you might have run instead, it belongs in further research, not in limitations.
The work with no limitations, and the ten in its own pages
One work of the seven states no limitations of its own study anywhere. There is no heading, no paragraph and no sentence. The word "limitation" occurs twice in 14,081 words and both occurrences are general claims about method types, both immediately neutralised — "while there are limitations to the qualitative data that questionnaires can gather, they have numerous strengths", and "all methods individually are flawed, but these limitations can be mitigated through mixed-methods research". The document is one of the eight mapped at Eight Examined Theses Compared.
The second is the more consequential: mixed methods is offered as the mitigation of method-specific flaws, in a study whose second method is its own literature review. The only two mentions of limitation in the document are arguments that its design has none.
Ten limitations are recoverable from that work's own pages. This list is this library's audit, not the thesis's, and every item is visible from the document's own text: no recruitment method is stated, so 100 respondents cannot be located in any population; no response rate, because no denominator is given; a 98% male sample reported and never discussed; a second industry inside the sample and outside the title; self-reported safety behaviour used to rank occupational groups by safety, with no mention of social desirability; cross-sectional data used for a developmental claim about complacency; a qualitative strand that is the study's own literature review, so convergence between strands is not independent; a stated eligibility criterion that was not enforced, with the excluded group then reported as a finding; an instrument whose item count reconciles only under an undeclared counting convention; and an appendix printing 75 of 100 responses with no statement of what happened to the other 25.
Length is not the measure
Rank these seven on the quality of their limitations and the ranking inverts the ranking on structure, twice over.
Proper heading, weak content
- The 210-word section with a heading and a contents entry describes an experiment that was never attempted.
- The 151-word section with a heading contains three real limitations and one displaced one, and records an external shock affecting who answered a survey twenty-two pages after the response rate is computed.
- The 569 words split across two chapters, with no heading anywhere, include a defence of the study inside its own limitations — "However, as discussed and articulated below, the comparison between the two projects does illustrate the importance of the framework…"
No heading or few words, strong content
- The 364-word unheaded section — invisible from the contents page — is causal throughout: each item names a constraint, then names what the constraint did to the design. Thirteen weeks of research window; therefore one focus group; therefore an established group rather than a purpose-built sample; therefore four participants at the floor of a cited range.
- The 105-word section is the shortest in the library and every sentence does work: security sensitivity limits what can be discussed; the political nature of the projects compounds it, so participants hedge even when anonymous; and the population is many thousands, so this sample cannot be an accurate snapshot.
- The 295-word section states four limitations, each with a cause and a consequence, and converts three of the four directly into a named further-research question.
Having the heading is not the same as doing the work, and neither is having the words. That is the finding, and it holds in both directions across two independently examined groups of works.
What a limitation looks like when it works
Four moves recur in the sections that do their job. None of them is taught in the supplied teaching source; all four are these researchers' own. For the wider question of how limitations sit beside a claimed contribution, see Stating Limitations and Contribution.
Four moves worth copying
Constraint, then consequence
Name the constraint and then name what it did to the design. "Time was the biggest limitation" is a complaint; "time was the limitation, so only one focus group was run, so an established group was approached rather than a purpose-built sample, so four participants attended" is a limitation.
Convert it into further research
One work turns three of its four limitations directly into successor questions. A limitation that produces a research question is doing double duty and reads as design awareness rather than apology.
Name the limits of your own analysis
One work states that "fragmentation may also have occurred because of the generalisation and categorisation of the responses received" — a stated limitation of the researcher's own coding, and the only one in the library. It is right, and it is unverifiable for exactly the reason it identifies, because no coding procedure is reported.
Follow one constraint through the whole document
One work names a single mechanism three times in three roles: the reason its literature is thin, the reason it admitted general rather than sector-specific literature, and the reason its own participants would not say everything. One constraint traced through gap, design and limitation is the most coherent methodological argument in the corpus.
Writing yours
Before you submit
- Move your limitations into the Discussion, or at minimum reference them there before you make the claim they constrain.
- Give them a heading and make sure the heading reaches the contents page. Three works in this corpus have limitations a reader cannot find from the front of the document.
- Read each item and ask whether it describes the study in this document. If it describes a better study, move it to further research.
- Make every item causal: constraint, then the specific design consequence. One work in seven does this throughout and it is the clearest section of its kind in the corpus.
- Convert what you can into further research questions, and say which limitation generated which question.
- Include the limitations of your own analysis, not only of your sample — coding, categorisation, the judgement calls you made alone.
- Delete any sentence that defends the study. A defence inside a limitations section cancels the section.
What to carry forward
- Placement decides what a limitation can still do. Six examined works put theirs after the claims they should have shaped, and all six were passed in that state.
- Ask of every item: does this describe the study in this document? One work's entire limitations section describes an experiment it never attempted.
- The gap between what a document contains and what it admits is the thing to close. Ten limitations sat recoverable in one work's own pages and none was written down.
- Short and causal beats long and defensive. The best section in the corpus runs to 105 words; the longest contains a defence of the study inside itself.
- A limitation that generates a further-research question is doing double duty — and reads as design awareness rather than apology.
Frequently asked questions
Where should limitations actually go?
The supplied teaching source places them in the Discussion: "application of the findings to the research questions and highlights limitations of the study". No examined work in this library does that. Five of six put them in the Conclusions and one splits them between the Introduction and the Conclusion. The practical consequence is that the limitation arrives after the claim it should have constrained.
How long should a limitations section be?
This corpus gives a range of 0 to 569 words, or 0% to about 6.3% of a document, and no rule. Length correlates with nothing useful here: the best section is the shortest at 105 words and the longest is the least disciplined. Judge by whether each sentence names a constraint and its consequence for the design you actually ran.
What is the difference between limitations and delimitations?
Delimitations are boundaries you chose — the population, the period, the sector you excluded — and in these works they belong in the Research Boundaries section. Limitations are constraints you did not choose that bound what the study can claim. One work in this corpus puts a limitations section in the boundaries slot and mixes both, which is why the distinction is worth keeping.
Can a study have no limitations?
No study in this corpus has none; one document states none. Its only two mentions of the word are general arguments that its design has none, and ten limitations are recoverable from its own pages by anyone who reads it. An unstated limitation does not stop existing; it stops being yours to frame.
Should I list limitations of the study I wish I could have run?
Not under the limitations heading. One work's whole 210-word section describes the sample size and sample profile of a comparative experiment it never attempted, and names no limitation of the literature study it did conduct. That material belongs in further research, where it reads as a research agenda instead of a misfiled critique.
How do I write a limitation about my own analysis?
Name the judgement you made and what it may have cost. One work states that fragmentation may have occurred because of the generalisation and categorisation of the interview responses — the only stated limitation of a researcher's own coding in the library. Pair it with a reported coding procedure so a reader can judge the size of the loss; that work does not, which is why its own limitation cannot be checked.
References and source attribution
- Seven examined master's works in project management, supplied as student work, whose limitations sections were located, counted and read in full. Researchers, supervisors, institutions, jurisdictions, employers, communities and programmes scrubbed. Used as observed practice rather than as model answers.
- The consolidated extract of four of those works, which records limitations placement, heading status, word count and proportion of document for each, and the ten limitations recoverable from the pages of the work that states none.
- The consolidated extract of the three works examined afterwards, which extends the comparison to seven works and records the shortest and the longest limitations sections in the library.
- The supplied teaching source: weekly study notes, slide decks and assessment activities for a master's-level research methods subject in project management, whose report structure places limitations in the Discussion, whose proposal component list requires them in the research-design section, and which never distinguishes anticipated from encountered limitations. Author, institution and year not stated in the supplied files.
Suggested questions for Ask KEVOS
- Where should the limitations section sit in my thesis?
- How do I write a limitation that names its design consequence?
- What is the difference between a limitation and a delimitation?
- How do I turn my limitations into further research questions?
- How do I find the limitations my own draft does not admit?
- Should I discuss limitations of a study design I could not afford to run?