Stating Limitations and Contribution
Limitations and contribution are two halves of one claim: how far your finding travels. This page sets the two halves side by side in five examined works and gives you a test for whether yours agree with each other.
Where the five works put their limitations
Start with the observation that is easiest to miss because it is unanimous: every one of the five examined works states limitations. Not one of them skips the admission. Whatever else varies across a doctoral action research thesis, a doctoral mixed methods thesis, a portfolio thesis, a journal paper and a conference paper, this does not.
What varies is where the admission sits and how much room it gets. Four different placements appear across the five documents, and the placement changes how hard the admission is to find.
PLACEMENT OF THE LIMITATIONS STATEMENT IN FIVE EXAMINED WORKS
| Work | Design | Where the limitations sit | What that placement does |
|---|---|---|---|
| Exemplar A | Doctoral action research thesis | A section of the conclusions chapter | Findable from the contents page; reads as part of the closing argument rather than an afterthought |
| Exemplar B | Doctoral mixed methods thesis | A section of the conclusions chapter | Same placement, used here to bound the population the findings are claimed for |
| Exemplar C | Doctoral portfolio thesis | An explicit subsection inside each empirical paper's conclusion — not a separate chapter | Repeated once per paper because each paper must stand alone; a reader of any single paper cannot miss it |
| Exemplar D | Journal paper | In the conclusions prose, without its own heading | Present and specific, but unlabelled — a reader skimming headings can pass over it |
| Exemplar E | Conference paper | A single sentence embedded in the conclusion | The least visible of the four patterns, and the least complete |
Placements observed in these five documents only. There is no standard position, and none of these is prescribed by the supplied teaching source.
The trade-off is worth naming. A heading makes the admission auditable. Prose placement leaves it technically present and practically deniable, because nobody finds it without reading the whole conclusion.
What each work actually admitted
Placement is the easy half. The content of the admission is what decides whether the limitations section is doing work or performing humility.
THE CONTENT OF THE ADMISSION, WORK BY WORK
| Work | Limitations stated | What the admission bounds |
|---|---|---|
| Exemplar B | Low response rate; small dataset; findings said to hold for the 71 companies studied, but not necessarily for the whole industry | The population, explicitly. The claim is tied to a counted set of cases and withheld from the industry by name |
| Exemplar C | Single industry; single geography; small sample; self-selected volunteers; low response rate | Setting, sampling method and scale — five separate bounds, itemised in each empirical paper |
| Exemplar D | Only two case projects, both from the same organisation; the threshold-definition method might be an oversimplification; results may be affected by the single organisational context | Case count, organisational context, and one named methodological joint the authors knew was weak |
| Exemplar E | One limitation only: that the correlation can be considered to exist although its quantified level was not examined, as that was outside the research scope | The strength of a single claim. Nothing about the corpus, the coding frame or the attribution rule |
| Exemplar A | The structural record shows a limitations section in the conclusions chapter but does not itemise its contents | Not recoverable from the record available here — see the note below |
Every count and figure above is a finding of the single study that reported it. None is a standard, a benchmark or a norm.
On Exemplar A the honest answer is that the itemised content is not available. What the record does show are two method-level admissions made in the methodology rather than in the conclusions: every instrument except one standardised published inventory was researcher-built and was expressly stated to be non-standardised, and higher-order statistics were deliberately refused on two stated grounds — the sample was too small, and they would have added little to rich qualitative data. Both are treated in Rigour, Triangulation and Validity in Practice.
The widening-circles problem
The most instructive failure in the set is in the portfolio thesis. Its contribution is stated as a series of widening circles.
- Useful to the organisations studied
- Useful to the same profession elsewhere
- Useful to other industries
Each step outward is a substantially larger claim about generalisation than the one before it, and the generalisation is asserted rather than demonstrated. No evidence from a second profession or a second industry is presented, because none was collected.
It sits roughly a page after the limitations subsection of the same paper — the subsection that names a single industry, a single geography, a small sample, self-selected volunteers and a low response rate. In that work the survey strand distributed 150 questionnaires and had 31 returned, described as just over 20 per cent, and the interview strand ran eight interviews with volunteers described as an ideal sampling profile. Both figures are findings of that one study.
Three ways to bound a contribution claim
The same set of five works supplies three working counter-models. Each one makes a real contribution claim, and each one survives its own limitations because it is bounded before it is made.
Bounded contribution claims from the examined works
Name the population you actually studied
The mixed methods thesis states that its findings hold for the 71 companies studied, but not necessarily for the whole industry. The bound travels with the claim in the same sentence, so a reader cannot lift the finding out of it. That number is a finding of that study.
Claim the benefit you tested, not the one you would like
The journal paper claims its adapted risk technique gives a project team fewer risks to concentrate on — an efficiency gain. It does not claim the technique finds better risks, because it did not test that. Narrowing the kind of benefit is as effective as narrowing the population.
Convert the weakness into the next study
The conference paper frames its contribution as a stepping stone towards a future quantified study. The thing it could not do becomes the thing it recommends someone do. That is a modest claim that is also a useful one.
The journal paper is worth a second look because it does something harder. Its authors state that there is no systematic procedure and no scientific protocol for setting the critical thresholds the method depends on, select thresholds anyway because the analysis cannot proceed without them, and then list the threshold-definition method as a limitation. Flagging a known weak joint three times, in the method, in the analysis and in the conclusions, is more convincing than not using the method at all. The method itself is set out in Risk Scoring with RFMEA.
The limitations that were not stated
The conference paper states exactly one limitation, embedded in the conclusion: that the correlation it reports can be considered to exist although its quantified level was not examined, as that was outside the research scope. That is a fair admission about the strength of its central claim.
Two more consequential decisions are not admitted anywhere.
- The corpus came from one journal. Every research article in a single outlet across 35 years — 2,015 articles, a whole population of that outlet rather than a sample, a finding of that study. The single-source decision is defended in the method with four separate justifications including a cited methodological recommendation. It is defended; it is never listed as a limitation.
- Country attribution is by the nationality of the first author only. That one coding rule drives a country table and a collaboration matrix, and any co-author from anywhere else is invisible to both. It is not acknowledged at all.
The general rule falls straight out of that. A decision you felt the need to defend at length in the methodology is precisely the decision that belongs in the limitations. Defence and admission are not alternatives — a defended decision is a decision a reader might reasonably have made differently, which is the definition of a limitation.
A test for whether your two statements agree
The supplied teaching source prescribes nothing here. The following is synthesis, built from what the five works did well and badly, and it is deliberately mechanical — the failure it catches is one that reads perfectly well sentence by sentence.
The scope-match test
List every noun in your contribution claim
Populations, industries, geographies, organisational types, roles, project types, time periods and methods. Write them down as a flat list. Include the ones carried by words like "generally", "across sectors" and "in practice", which name a population without appearing to.
List every bound in your limitations
Sample size and how it was obtained, setting, sector, geography, response rate, instrument status, analysis restrictions, and any parameter you selected rather than measured.
Match the two lists item by item
Each noun from step 1 must correspond to something your design actually covered, and must not correspond to anything step 2 excludes. A noun that appears in step 1 and is excluded in step 2 is a contradiction, not a nuance.
Fix each mismatch with one of three moves
Demonstrate it — add evidence from the setting you are claiming. Bound it — claim only what you covered and name the rest as further research. Or reframe it — claim that the method, instrument or framework transfers, which is a different and often more defensible claim than saying the findings transfer.
Read the two passages aloud, back to back
Put the limitations paragraph immediately before the contribution paragraph with nothing between them. If the contribution only works for a reader who has forgotten what they just read, the contribution is the part that is wrong.
Which move to make when they disagree
Where to put yours
Placement follows format. The pattern below is drawn from what these five documents did, matched to the document type you are writing; it is not a rule any of them states.
PLACEMENT BY DOCUMENT TYPE — DRAWN FROM OBSERVED PRACTICE
| If you are writing | Observed pattern | Practical advice |
|---|---|---|
| A monograph thesis | A named section of the conclusions chapter, as in both doctoral theses in the set | Give it a heading that appears in the contents, and keep contribution and limitations as adjacent sections so the reader meets them together |
| A portfolio or multi-paper thesis | A subsection inside each paper's conclusion, repeated per paper | Accept the repetition. Each paper stands alone, so each needs its own bounds — see The Portfolio Thesis Format |
| A journal paper | Inside the conclusions prose, specific but unlabelled | Keep the specificity, add the label. There is no cost to a sentence that begins "This study is limited in three ways" |
| A short conference paper | One embedded sentence | One sentence is the floor, not the target. Name every decision that changes what the finding means |
Observed in five documents. The supplied teaching source specifies no placement.
What to carry forward
- All five examined works state limitations. Placement varied across four patterns; only the placement with a heading is reliably findable.
- The best admissions in the set bound something specific — a counted population, a case count, a named methodological joint — rather than gesturing at modesty.
- A contribution stated as widening circles is asserted generalisation. If you cannot show evidence from the second and third circles, claim the first and name the others as further research.
- A decision you defended at length in the methodology belongs in the limitations as well. Defence is not a substitute for admission.
- Run the scope-match test: every population, setting and method named in the contribution must be covered by the design and must not be excluded by the limitations.
- Three legitimate moves when they disagree — demonstrate, bound, or reframe the claim as a transferable method rather than a transferable finding.
Frequently asked questions
How many limitations should I state?
The supplied teaching source gives no figure and none of the five examined works states a target. What the set shows is a range from one to five stated limitations, and that the work stating only one had omitted two decisions that materially shaped its results. Count is the wrong measure; the test is whether every decision that changes what your finding means has been named.
Does a strong limitations section weaken my contribution?
It bounds the contribution, which is different. In the examined set the most credible contribution claims were the most tightly bounded ones — findings held to a counted population, or a benefit claimed as efficiency rather than accuracy. An unbounded claim is not stronger, it is simply easier to reject.
Where should the limitations go?
Four placements appear across the five works: a section of the conclusions chapter, a subsection in each paper's conclusion, unlabelled conclusions prose, and a single embedded sentence. None is prescribed. The practical difference is findability, so give it a heading that reaches the contents page and put it next to your contribution claim.
My contribution feels small. Is that a problem?
One of the examined works frames its contribution as a stepping stone towards a future quantified study, converting the thing it could not do into the thing it recommends. A small, accurate claim with a clear research agenda attached is more usable than a large claim the evidence cannot carry.
Do I need to list a decision I already justified in the methodology?
Yes. A decision that required justification is a decision a reasonable reader might have made differently, which is exactly what a limitation is. In the examined set, the two most consequential unstated limitations in one work were both decisions its methodology had defended at length.
Can I claim my findings apply to other industries?
Only if you collected evidence from them, or you reframe the claim. Saying your method, instrument or framework could be applied elsewhere is a defensible claim about transferability; saying your findings hold elsewhere is a claim about generalisation that requires data you do not have if you studied one setting.
References and source attribution
- Naoum, S. G. 2013, Dissertation Research & Writing for Construction Students, 3rd ed., Routledge.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford.
- Five examined and published research works in project and R&D management — a doctoral action research thesis, a doctoral mixed methods thesis, a doctoral portfolio thesis, a journal paper and a conference paper — read structurally from front matter, contents, methodology chapters, conclusions and abstracts. All counts and figures reported here are findings of the individual study concerned.
- The supplied teaching source: consolidated weekly teaching notes and slide material, which contains no guidance on stating limitations or contribution.
Suggested questions for Ask KEVOS
- Run the scope-match test on my draft limitations and contribution paragraphs.
- My contribution claims my findings apply across the sector but I surveyed one organisation. How do I rewrite it?
- Turn the design decisions I defended in my methodology into a limitations list.
- Draft a bounded contribution claim for a study with 40 survey responses from a single industry.
- What is the difference between claiming a finding transfers and claiming a method transfers?