ACADEMIC WRITING GUIDE

How to Write the Limitations of a Study (With Examples)

By Acadelyra Editorial TeamPublished

A practical dissertation guide to identifying meaningful limitations, explaining their impact, and writing conclusions that match your evidence.

A limitations section should not read like an apology for imperfect research. It should tell readers where the evidence is strongest, where uncertainty remains, and what they should avoid concluding. Saying “the sample was small” is rarely enough. A reader needs to know why that matters for this particular research question.

This guide gives you a step-by-step writing method, a comparison table, and three fully explained examples for quantitative, qualitative and mixed-methods studies. The sample paragraphs are hypothetical teaching models—not results from published studies. Adapt only the points that your actual design, data and research record support.

Quick answer: what are the limitations of a study?

Research limitations are features of a study’s design, sample, measures, data or execution that restrict the confidence, precision, transferability or scope of the conclusions. Some are known before data collection; others become apparent during the research. A good limitations section explains their likely consequences, the steps actually taken to address them and what the evidence can still support.

Useful pattern: Limitation → Why it matters → What was done (if anything) → What conclusion remains justified.

For example, a one-time survey can identify an association between study habits and academic engagement. If the variables were measured at the same point in time, the study cannot on that evidence alone establish that study habits caused engagement to change. That is a specific, meaningful design limitation—not an admission that the study has no value.

Limitations, scope, delimitations and assumptions: the difference

These concepts overlap in a dissertation, but each serves a different purpose. Distinguish an intentional boundary from the consequence that boundary has for interpretation.

ConceptQuestion it answersIllustrative example
ScopeWhat does this study investigate?Academic engagement among undergraduates in one institution.
DelimitationWhich boundary did the researcher deliberately choose?Focus on second-year students because the question concerns their shared course experience.
LimitationWhat restricts the interpretation of the evidence?Voluntary single-site respondents may differ from nonrespondents.
AssumptionWhat premise is used without being directly tested?Participants understand the questionnaire as intended.

A single decision can create both a delimitation and a limitation. Choosing to study one university is an intentional boundary; caution about applying findings to other universities may follow from it. Do not label a well-justified qualitative purposive sample as automatically defective because it is not statistically representative: depth and transferability, rather than population-level estimation, may be the objective.

For a fuller explanation of intentional research boundaries, see Acadelyra’s guide on scope and delimitations of a study.

Where should you write limitations in a dissertation?

Follow your department’s dissertation handbook. Some programmes expect anticipated limitations in the introduction or proposal; others require a limitations subsection in the methodology or discussion chapter. Completed research often discusses limitations near the interpretation of findings, where readers can see exactly which claims need qualification.

The timing matters. In a proposal, describe what may limit the planned study: “A one-site voluntary sample may limit generalisability.” After conducting the study, report what actually occurred: “The survey was completed by volunteers from one institution; students who did not respond may have differed in ways relevant to engagement.” Never write as though a mitigation was performed if it was only planned.

What kinds of research limitations should you consider?

Choose the limitations that materially affect your question and evidence. The following examples are prompts for investigation, not a checklist to copy mechanically.

Sampling and nonresponse

Who could enter the study, who actually participated, and who is missing? Volunteers may differ from nonparticipants. In probability-based surveys, an inadequate response pattern can threaten representativeness; in purposive interviews, explain which perspectives may be missing rather than demanding statistical representativeness.

Research design and time order

Cross-sectional observational data may describe associations without establishing which event came first. A pre-post study without a comparison group may be affected by unrelated events over time.

Measurement and self-report

Recall errors, social-desirability responses, ambiguous survey wording or an instrument not adequately evaluated for a population can affect measurement. Say which variable is at risk and how the likely problem could alter interpretation.

Missing, inaccessible or inconsistent data

Incomplete records or unavailable groups may distort the analysed sample. Distinguish a small quantity of missing information from systematic missingness linked to the outcome.

Confounding and alternative explanations

A relationship between two variables may be influenced by a third factor. If important factors were unmeasured, say which alternative explanation remains possible rather than declaring a causal effect.

Researcher influence and context in qualitative work

Interview questions, the researcher’s professional role and participants’ concerns about disclosure can shape accounts. Describe reflexive practices and the specific settings needed for readers to assess transferability.

Integration in mixed-methods work

Surveys and interviews do not automatically represent the same people or time period. If interview volunteers differ from the survey population, their explanations may illuminate patterns without explaining every case.

See our guide to choosing a research methodology for the design decisions that create different evidence strengths and boundaries.

How to write a strong limitations section: six steps

Step 1. Start from the claims your study is trying to make

Write your research question and the two or three main conclusions you hope the evidence will address. For every claim, ask: What would have to be true for a reader to accept this inference? A study reporting how participants experienced a policy faces different interpretive risks from one estimating the policy’s causal effect.

Step 2. Identify the limitations that could change interpretation

Review your recruitment, measurements, data availability, design and analysis. Prioritise material constraints rather than filling the section with routine phrases about limited time or money. Ask whether the issue could change an estimate, leave a group unheard, weaken confidence, or narrow where findings apply.

When evaluating sample-related limitations, our article on choosing a sampling method and sample size helps distinguish sampling choices from their implications.

Step 3. Explain the mechanism, not just the label

Compare “The study used self-reports” with “Students who wish to appear diligent may overstate weekly study hours, potentially inflating the association between reported study habits and reported engagement.” The stronger sentence makes the measurement problem concrete. If you cannot know the direction or magnitude of bias, say it is uncertain rather than inventing an estimate.

Step 4. State what you actually did to reduce the concern

Name measures you completed, such as neutral recruitment language, piloting a questionnaire, reporting missing-data counts, using a documented interview procedure, or keeping a reflexive journal. Explain what each action could help with and what uncertainty remained. Do not say an instrument was validated or data were triangulated unless the research record supports it.

For measurement questions, Acadelyra’s research questionnaire design guide explains why wording, pilot testing and measurement choices matter.

Step 5. Recalibrate the conclusion without cancelling the study

Make a balanced statement: “These results describe associations in the participating students and should not be read as evidence that changing study hours would cause engagement to improve.” State what readers can still learn. A limitation restricts an inference; it does not erase credible evidence that answers the narrower question.

Step 6. Recommend follow-up research that addresses the limitation

Specific future work is more useful than “more research is needed.” Propose, where justified, recruiting from additional settings, following participants across time, using independent measurements, including a missing group, or gathering contrasting perspectives. Do not promise that the proposed next study would certainly resolve the problem.

The same careful interpretation is central to our guide on writing a research discussion chapter.

Three complete limitations-of-the-study examples

Each passage below describes an invented study scenario with explicitly assumed procedures. They show how a completed dissertation section might be written. Replace every sample detail and claimed mitigation with what actually happened in your project.

Example 1: Quantitative survey — study planning and engagement

Hypothetical design. A cross-sectional questionnaire was completed by 180 volunteer second-year undergraduates at one university. The study measured self-reported study-planning behaviour and academic engagement. Assume the questionnaire was piloted for clarity, but no objective measures of study time or engagement were collected.

Weak version. “The sample was small and time was limited. The results may not apply to everyone.” This does not identify the relevant inference or explain any mechanism.

Stronger limitations section (illustrative).

This study has three limitations relevant to the interpretation of its findings. First, participants were volunteers recruited from one university, so the responses may not represent students who did not participate or second-year undergraduates elsewhere. The findings are therefore most directly applicable to the students surveyed, rather than to all university students. Second, study-planning behaviour and academic engagement were measured through self-report. Students may have inaccurately recalled their routines or answered in socially desirable ways, and these errors could influence the observed association. The questionnaire was piloted for clarity, but the study did not independently verify reported behaviour. Third, because both variables were measured at one time point, their temporal order cannot be established. Any association should not be interpreted as evidence that study planning causes increased engagement. Future research could follow a more diverse group of students over time and, where appropriate, combine self-report with independently recorded indicators.

Why this example works. It connects recruitment to generalisability, self-report to measurement uncertainty, and cross-sectional design to causal inference. The pilot is described as a limited safeguard, not a guarantee of validity. It also identifies what the study can still report: an association among respondents.

Example 2: Qualitative interviews — hybrid-work experiences

Hypothetical design. Fourteen workers from two small businesses were interviewed about their experiences of hybrid work. Recruitment invitations were distributed through managers; participation was voluntary. Assume the researcher removed identifying details from transcripts and kept reflexive notes, but did not interview employees who declined.

Weak version. “This study had only fourteen participants, so the data cannot be trusted.” That treats a qualitative study as if it were designed to estimate a population percentage.

Stronger limitations section (illustrative).

The findings reflect the experiences of fourteen volunteers from two small businesses and should be interpreted in that organisational context. They were not intended to provide a statistically representative estimate of all employees’ views. Because recruitment invitations were distributed through managers, some employees may have doubted that critical accounts would remain private or may have chosen not to participate. Their perspectives could therefore be underrepresented. Removing identifying details helped protect participants’ confidentiality, but it cannot establish how employees who declined would have described hybrid work. The researcher’s prior interest in flexible work might also have shaped interview prompts or interpretation; reflexive notes were kept to examine this influence, although they cannot remove it entirely. The study offers detailed accounts that readers can evaluate for relevance to similar workplaces. Research involving differently sized organisations and alternative recruitment channels could help reveal perspectives not captured here.

Why this example works. The key concerns are recruitment, self-censorship, context and researcher influence—not the interview number by itself. It also correctly discusses transferability and safeguards without asserting that reflexivity eliminates bias.

Example 3: Mixed methods — missed outpatient appointments

Hypothetical design. A study analysed twelve months of appointment records from two clinics and interviewed a subset of adult patients about barriers to attendance. Assume the researchers checked a sample of records for coding consistency. Interview recruitment relied on patients reachable through contact details held by the clinics, including some patients with missed appointments.

Weak version. “The study used two methods, which removed most limitations.” Using multiple methods does not automatically remove selection or measurement problems.

Stronger limitations section (illustrative).

The quantitative analysis was based on routine appointment records from two clinics, and the recorded reasons or status of missed visits may not capture every scheduling change or patient circumstance. Checking a sample of records supported consistent coding, but inaccuracies in the underlying administrative data may remain. The qualitative interviews added context about transport, cost and scheduling experiences, yet recruitment depended on having usable patient contact details. Patients without current contact information, including some who missed appointments, may have been less likely to contribute accounts. The interview themes therefore cannot be assumed to explain the circumstances of every patient represented in the records. In addition, the observational design cannot establish that any single reported barrier caused nonattendance. The combined evidence can identify documented patterns and plausible explanations in the two clinics; it does not justify causal or nationwide claims. Further studies could test data quality across additional clinics and use ethically appropriate recruitment methods that better reach underrepresented patients.

Why this example works. It identifies distinct weaknesses in the records and interviews, then explains the problem created when those strands are combined. The paragraph does not confuse plausible patient explanations with proven causes. The assumed quality check is presented realistically, without claiming to fix all record errors.

A weak-to-strong revision you can copy as a method

Weak draft. “The research had a small sample, a short time frame, and limited funding.”

First improvement: identify the actual consequence. “Because the study surveyed volunteers from one university, respondents may differ from students who did not participate. The results should not be generalised automatically to other institutions.”

Second improvement: add the evidence-specific limit and an honest response. “Academic engagement was measured by self-report rather than observation or institutional records. Although questionnaire wording was piloted for clarity, the responses may still be influenced by recall or social-desirability bias.”

Final improvement: match the conclusion to the design. “Because the data were cross-sectional, the observed relationship should be interpreted as an association rather than proof that one behaviour caused the other.”

The final version is longer because it performs genuine analytical work: it identifies what went wrong or remains uncertain, describes the implication, and protects the reader from an overclaim. It is not longer merely for its own sake.

Limitations writing template

Use this as a planning aid, not a paragraph to submit unchanged. Replace the prompts only with constraints and actions supported by your own methods and results.

This study was subject to [specific limitation], arising from [design, recruitment or data feature]. This matters because [explain which measure, group or inference is affected and how]. To address the concern, [state only a safeguard actually performed, if any]. Nevertheless, [identify the remaining uncertainty]. The findings therefore support [appropriately narrow conclusion] but should not be used to claim [unsupported generalisation or causal interpretation]. Future research could [specific feasible improvement tied to the limitation].

Common mistakes to avoid

Listing generic problems without implications. “Time was limited” is meaningful only if it prevented a necessary comparison, follow-up, measure or recruitment step. Name that consequence.

Confusing chosen boundaries with weaknesses. A study designed around one organisation is not automatically flawed; explain the rationale as a delimitation and any transferability concern as a limitation.

Calling all qualitative samples too small. A purposive interview study is assessed differently from a population survey. Discuss the adequacy and range of accounts in relation to its purpose.

Inventing mitigations. Do not claim to have piloted, validated, randomised, triangulated or checked reliability unless you actually did.

Claiming a definite direction of bias without support. Bias may overstate or understate an estimate. Describe the mechanism and direction only if justified; otherwise acknowledge uncertainty.

Turning limitations into an apology or a guarantee. Neither “the results are worthless” nor “these limitations did not affect the findings” is supported without further evidence.

Using limitations as a substitute for fixing serious design problems. If a critical measure is missing or the research cannot answer the stated question, improve the design or narrow the research question rather than relying on a limitations paragraph.

Frequently asked questions

How long should a limitations section be?

There is no universal word count. Follow your institution’s guidance. A short, specific section addressing the important constraints is more useful than a long catalogue of minor inconveniences. The content should be proportional to the complexity of the study and the risks to interpreting its findings.

Can I write limitations before collecting data?

You can anticipate likely design constraints in a proposal, using appropriately conditional language. Revisit the section after data collection to describe what actually happened and remove predictions that did not materialise.

Should I mention a small sample?

Only explain it in relation to the research aim and design. A small quantitative sample may produce imprecise estimates or weak subgroup comparisons. A focused qualitative sample may be defensible for in-depth exploration, while still limiting which perspectives are represented.

Do I have to suggest future research?

Not always as a separate section, but a specific and proportionate suggestion can show how remaining uncertainty might be addressed. Do not add a generic recommendation for a bigger sample unless a bigger or different sample would answer something important.

Can limitations invalidate a study?

They can seriously undermine particular conclusions if the missing evidence is essential. Identifying limitations does not automatically repair that problem. When a design cannot answer its stated research question, revise the question, analysis or study design with your supervisor rather than writing around the problem.

Final checklist before you submit

  • Have I selected limitations that actually matter to my research question?

  • Does each limitation identify its source in the design, sample or data?

  • Have I explained how it could affect interpretation, precision or applicability?

  • Have I avoided inventing bias directions, statistics or safeguards?

  • Have I distinguished planned mitigations from actions actually completed?

  • Is my language suitable for a proposal or a completed dissertation?

  • Have I treated qualitative transferability differently from statistical generalisability?

  • Have I explained any weaknesses in combining methods, if applicable?

  • Does the discussion still say what the study legitimately contributes?

  • Do my proposed next steps address specific uncertainties?

  • Have I followed my supervisor’s and institution’s reporting requirements?

Final takeaway

Strong limitations writing is an exercise in sound judgement. State the constraint, show why it matters, disclose what you did, and narrow the conclusion only as much as the evidence requires. Readers should finish understanding both the value of your research and the uncertainty that remains.

When writing the final chapter, use Acadelyra’s guide to research conclusions that reflect the evidence to keep the contribution and its boundaries in balance.

If your department’s brief has unusual requirements, work from those instructions first. Acadelyra offers academic-writing guidance and support with understanding requirements, structuring material and improving clarity; examples in this article are models to learn from, not ready-to-submit findings.

Research guidance and further reading

These sources informed the distinctions and reporting principles in this guide; they do not substantiate the hypothetical study details above.