Folio

How to write the limitations section of a research paper

Last updated · 4 min read

Short answer: The limitations section explains the weaknesses in a study's design, data or analysis that affect how its findings should be interpreted. It usually sits near the end of the discussion, covers the two to four limitations that matter most, and treats each one in four moves: name the limitation, explain how it could affect the results, say what you did to reduce it, and suggest how future research could address it. It should be specific to your study, not a generic list or an apology.

On this page
  1. What counts as a limitation
  2. Where the limitations go
  3. Step 1: Choose the limitations that matter most
  4. Step 2: Name each limitation precisely
  5. Step 3: Explain the likely effect on your findings
  6. Step 4: Say what you did about it
  7. Step 5: Connect it to future research
  8. Common limitations by research design
  9. Example sentences you can adapt
  10. What not to write
  11. Quick checklist

Every study has limitations. Reviewers and examiners know this, so a limitations section does not weaken a paper; a missing or evasive one does. Done well, it shows that you understand your design, helps readers judge how far your conclusions travel, and sets up your suggestions for future research.

What counts as a limitation

A limitation is a feature of your design, sample, measures, data or analysis that constrains what you can conclude. It is different from a delimitation, which is a boundary you chose on purpose (for example, studying only first-year students). Delimitations belong in your methods or introduction; limitations are the consequences you could not fully avoid.

A useful test: if the issue could plausibly change the size, direction or generalizability of your findings, it is worth discussing. If it could not, leave it out.

Where the limitations go

In most empirical papers, limitations form a subsection or a few paragraphs near the end of the discussion, before future research and the conclusion. Some journals ask for a separate "Strengths and limitations" heading, and theses often give limitations their own section in the final chapter. Check your target journal or your department's guidelines. Our guide to the discussion section shows how limitations fit into the larger argument.

Step 1: Choose the limitations that matter most

Start by listing every weakness you can think of, then keep the two to four that most affect interpretation. Prioritize problems with internal validity (could something other than your explanation account for the results?) and external validity (to whom and where do the findings apply?) over minor practical issues.

Step 2: Name each limitation precisely

Be concrete. "The sample was small" says little. "The sample of 84 participants gave limited power to detect small interaction effects" tells readers exactly what is at stake.

Step 3: Explain the likely effect on your findings

Say which way the limitation could bias the results, or which conclusions it constrains. Random measurement error in a predictor, for example, usually weakens (attenuates) an observed association rather than creating one, whereas a confounder can create or reverse one. If you do not know the direction, say what remains uncertain.

Step 4: Say what you did about it

Describe any steps that reduce the problem: a robustness check, a sensitivity analysis, controlling for a confounder, triangulating data sources, or a time-lagged design. This is not about defending yourself; it tells readers how much weight the limitation deserves.

Step 5: Connect it to future research

Close each limitation, or the section, with a specific suggestion: a longitudinal design, a more diverse sample, an objective measure, a replication in another context. Specific suggestions are far more useful than "more research is needed".

Common limitations by research design

DesignTypical limitationsWays to reduce or address them
Cross-sectional surveyNo causal inference; common-method bias; self-report errorTime-lagged measures; multiple data sources; procedural remedies; acknowledge correlational design
ExperimentArtificial setting; short-term outcomes; student or convenience samplesField replication; follow-up measures; pre-registration; power analysis
Longitudinal or panel studyAttrition; time-varying confounders; measurement changes across wavesAttrition analysis; fixed-effects models; sensitivity checks
Secondary or administrative dataVariables not designed for your question; missing data; limited measuresValidate proxies; report missing-data handling; robustness checks
Qualitative interviews or case studyTransferability; researcher influence; small, purposive samplesThick description; reflexivity statement; member checking; clear sampling logic
Systematic or literature reviewSearch and language restrictions; publication bias; heterogeneous studiesTransparent search strategy; bias assessment; sensitivity analyses

Example sentences you can adapt

Because all variables were measured by self-report at a single time point, the associations reported here may be partly inflated by common-method variance, and they do not establish causal direction. We reduced this risk by separating predictor and outcome measures within the survey, but longitudinal or multi-source designs are needed to test the proposed sequence.
The sample was drawn from two universities in one region, so the findings may not generalize to institutions with different admission policies or student populations. Replications in more diverse settings would show whether the pattern holds.
Interviews were conducted by a single researcher with prior experience in the sector, which may have shaped both the questions asked and the interpretation of responses. A reflexive journal and a second coder for 20% of transcripts were used to limit this influence.

What not to write

Quick checklist

If you are still planning the study, many limitations can be anticipated in the methods; see how to write a methodology section and how to write a research proposal.