2027 Public Health PhD Research: Writing a Dissertation Limitations Section Without Hiding Assumptions
Audited and Verified:
Our Editorial and Data Verification Process
Research.com uses government datasets, official education statistics and primary institutional information to support its career and degree guides. Our review process checks factual accuracy, occupational definitions, source dates and the requirements associated with the credentials discussed.
How This Article Was Fact-Checked
Our editorial team verifies salary estimates, employment projections, education requirements and program information against original data providers and official organizations. Salary figures are checked against the relevant occupational definitions and reporting periods. Information about licensure, accreditation and required in-person training is reviewed separately.
Fact-checked and verified on:
Co-Founder and Chief Data Scientist
A public health dissertation can report careful analysis and still overstate what its data show. The limitations section is where a PhD researcher defines the boundary between a defensible finding and an unsupported claim - especially when population data, community participation, or restricted records shape the evidence. This guide explains what to disclose, how to separate limitations from assumptions, and how to describe their likely effect on conclusions. Use it to revise a draft before your advisor or committee must ask which conditions your interpretation depends on.
Key Things You Should Know
- For each consequential limitation, identify its source, the finding it affects, the possible direction of bias when knowable, and any analysis that tests its impact.
- State assumptions explicitly: a model requirement, a claim about unmeasured factors, and a decision to exclude records require different kinds of justification.
- Restrict conclusions to the population, setting, period, and outcome actually supported by the study; transparency makes a bounded finding more credible, not less.
What belongs in a public health PhD dissertation limitations section?
A limitations section explains conditions that could change the interpretation, credibility, or applicability of the dissertation's findings. It is not an inventory of everything the researcher would have preferred to do. Prioritize issues consequential enough that a reader might reasonably reach a different conclusion if the issue were resolved.
For each major issue, name the affected claim, explain the mechanism, describe what you did to investigate or reduce the problem, and state what remains uncertain. A dissertation using administrative records, for example, might explain that an outcome appears in the records only when a person encounters a participating service. That limitation affects conclusions about population prevalence differently from conclusions about recorded service use.
Organize limitations around the dissertation's research questions rather than repeating a generic paragraph for every chapter. In a manuscript-based dissertation, distinguish constraints shared across studies - such as a common dataset - from limitations specific to one analysis. If restricted data access or funding shaped the design, say so without implying that another degree or a larger budget would necessarily have solved the methodological problem. Students weighing those practical constraints before starting doctoral work can consult a guide to the affordability of online public health PhD study.
How should researchers distinguish study limitations from research assumptions?
An assumption is a condition the design, analysis, or interpretation relies on. A limitation is a constraint on what the study can establish. An assumption may create a limitation when it is doubtful or cannot be checked with the available data.
Consider an observational study estimating an exposure-outcome association. Adjusting for measured confounders does not establish that all relevant confounding was measured. The assumption that no important unmeasured confounding remains supports a causal interpretation; the inability to verify it limits that interpretation. By contrast, incomplete outcome recording is a data limitation, and any assumption that the missing records resemble the observed ones needs separate justification.
A useful drafting test is to ask: What must be true for this sentence in the conclusion to hold? Then identify what the study can check. Model diagnostics may reveal a poor fit, but they cannot prove that an unobserved confounder does not exist. Researchers strengthening their statistical preparation can use an online biostatistics master's flexibility guide to compare training options; no credential substitutes for making an analysis's assumptions explicit.
How can public health researchers disclose limitations without weakening their findings?
Match the strength of each conclusion to the evidence. Rather than declaring that selection bias "may be present" and moving on, explain who could have been excluded, how exclusion might relate to the outcome, and whether its likely effect can be assessed. If the direction is unknown, say so; do not assert that a bias was negligible merely because the estimate remained statistically precise.
One useful pattern is finding → vulnerability → check → bounded interpretation. For a hypothetical study, a defensible passage might say: "The observed association concerns patients with recorded clinic visits. People who did not seek care are absent from these records, and their exposure and outcome patterns are unknown. Analyses using alternative visit definitions produced similar patterns, but they cannot establish that the association applies to all residents." The example claims no real results; it shows how a robustness check and a remaining limitation can coexist.
Avoid the opposite mistake of presenting every uncertainty as fatal. Explain which conclusions survive plausible alternatives and which require softer language. That distinction reflects doctoral research judgment, not simply admission to a research program; a resource comparing online doctorate admission routes in public health addresses a different decision from evaluating the strength of a dissertation finding.
Start with the design's ability to answer the stated question. An observational association may be informative without identifying a causal effect; a cross-sectional design generally cannot establish whether exposure preceded outcome. A program evaluation may describe changes after implementation while leaving other explanations for those changes unresolved.
Then examine limitations that could materially alter the answer: confounding, selection into the data, exposure or outcome misclassification, missingness, model specification, time-varying conditions, and uncertainty in estimates. Address survey weights or clustering when the sampling design requires them. Do not list every possible bias indiscriminately; explain why each named threat is plausible in this study and what evidence bears on it.
For example, if neighborhood exposure is assigned from an area average, acknowledge that individual exposure may differ and explain what that means for the proposed mechanism. If an alternative exposure definition is feasible, report its purpose and result elsewhere in the dissertation, then interpret it here. Readers deciding whether they need additional methods coursework before doctoral research might compare options using an online MPH tuition comparison guide, while checking that any program actually offers the methods depth their project requires.
Sampling determines who could enter the study; participation and data completeness determine who contributed to the analysis. State the source population, eligibility rules, recruitment or record-capture process, and any exclusions after enrollment. These details matter more than labeling a sample "small" or "nonrepresentative" without explanation.
In a community survey, people reached through a service organization may differ from those who do not use its services. In a health-system study, excluding patients without continuous records may preferentially retain people who use care more often. Such selection can affect an association as well as the population to which it applies. Compare included and excluded groups on available, relevant characteristics, but acknowledge that similar measured characteristics cannot rule out differences in unmeasured ones.
Sample size raises a separate question: whether the study can estimate the effect or describe a subgroup with useful precision. Report uncertainty rather than treating a nonsignificant result as evidence of no difference. Researchers earlier in their education can compare public health degree choices for undergraduate students if they need foundational training in population definitions and sampling before attempting independent research.
In quantitative research, connect each threat to the estimate or comparison it could affect. Discuss uncertainty in measurements, missing observations, model dependence, and plausible alternative explanations. Distinguish a sensitivity analysis that probes an assumption from one that actually resolves it.
In qualitative research, address how participants were recruited, whose perspectives may be absent, how the researcher's position and relationships could shape interviews or interpretation, and whether the setting allowed participants to speak freely. A small purposive sample is not automatically a defect if it fits the research aim; the limitation concerns the range of perspectives captured and the claims made from them.
For mixed-methods work, also explain whether the datasets describe the same population and period and what remains unresolved when findings disagree. Do not use interviews merely to "confirm" a numerical association if participants were asked a different question. A student considering a master's program as preparation for this work can review online public health master's no-letter application paths, but should evaluate the program's qualitative and quantitative training separately from its application requirements.
Generalizability concerns whether results apply beyond the people, places, and period studied. Begin by naming the target population and comparing it with the study population. Then examine features that could modify the result: age distribution, service access, policy environment, exposure levels, implementation conditions, or the way an outcome is measured.
For a qualitative study, transferability is often the more useful term: provide enough detail about participants and context for readers to judge relevance elsewhere, without claiming statistical representativeness. For an intervention study, distinguish whether a result could recur under similar delivery conditions from whether a different health department has the staff, infrastructure, and community relationships to deliver it.
Do not equate a large dataset with broad applicability. A large dataset from one system may omit everyone outside that system. Likewise, accelerated coursework is a scheduling feature, not evidence that research conducted within a course applies widely; students comparing public health BS programs with shorter academic terms should assess research preparation on its own merits.
Describe data constraints that affect the analysis without exposing participants or violating agreements. Relevant examples include suppressed small cells, unavailable variables, limited follow-up, imperfect linkage, restrictions on geographic detail, and inability to share row-level data or code containing sensitive information. Explain what a reader can and cannot reproduce, and whether an approved alternative - such as a data dictionary or analysis code without restricted data - is available.
State how consent, recruitment, institutional review, community permissions, and data-use terms shaped the study when those conditions bear on interpretation. Do not present an institutional review board determination as proof that data are complete or a study is unbiased. Conversely, do not imply that every use of public or deidentified data has the same review obligations; verify the determination and governing agreements for the actual project.
Community-engaged research may have limits on what can be quoted or released even when publication is otherwise permitted. Discuss those limits respectfully, separating legitimate governance from gaps in documentation. Prospective doctoral students evaluating whether a restricted-data project is feasible may also compare financial aid for epidemiology doctorates; funding and permission to access data are separate questions.
Committee expectations vary by institution and dissertation format, but a strong section usually demonstrates that the candidate understands the design better than a skeptical reader does. Advisors may look for consistency between the methods, results, limitations, and final claims. A limitation introduced only after a committee member identifies it can signal that the underlying assumption was not examined early enough.
Before submitting a chapter, work through the following review sequence with your advisor. It turns the limitations section into a check on the dissertation's argument rather than a last-minute disclaimer.
- Underline each conclusion that implies causation, population-wide prevalence, effectiveness, or applicability beyond the study setting.
- For each underlined claim, identify the design feature or assumption needed to support it and the most consequential challenge to that support.
- Check whether the methods or results provide a relevant diagnostic, comparison, sensitivity analysis, or explanation; revise the claim if they do not.
- Ask whether a reader could mistake an unavailable analysis for an analysis that found no problem.
Ask your advisor where the program expects cross-cutting limitations in a manuscript-based dissertation and how much chapter-level repetition is appropriate. Also confirm submission and defense milestones in your own handbook. Guides to affordable public health doctorates by study duration can inform an earlier program-planning decision, but cannot establish a particular committee's standards or revision timeline.
Which doctoral research standards support credible public health dissertation reporting?
Use reporting guidelines as checks on what a reader needs to evaluate, not as substitutes for methodological judgment or your university's dissertation rules. STROBE supports reporting of observational studies; CONSORT addresses randomized trials; COREQ and SRQR help researchers report qualitative work. If you use routinely collected health data, consider whether RECORD addresses relevant details. Confirm the appropriate guideline and version for each study design.
These standards prompt disclosure of matters such as participant selection, missing data, analytic choices, and study limitations. They do not certify that assumptions hold or require you to claim a bias has been eliminated. Pair the appropriate checklist with your protocol, approved amendments, analysis documentation, data-governance requirements, and committee feedback. A transparent account should let another qualified researcher understand what was done and where reasonable interpretations could differ.
If you still need foundational preparation, compare actual courses and supervised research opportunities rather than assuming any inexpensive credential teaches advanced dissertation methods. A guide to low-cost public health master's degrees can help frame that earlier education decision; the reporting standard for a PhD dissertation remains tied to its design and evidence.
Other Things You Should Know About Public Health
Follow your program's format and your advisor's guidance. Use enough space to explain consequential threats to interpretation; length alone does not indicate rigor.
No. Explain design constraints where readers first need them, then interpret their effect on findings in the discussion. A final chapter can synthesize limitations shared across studies.
Not by itself. Specify which unresolved question, population, measure, or design a future study should address and why resolving it would change a public health decision.
Tell your advisor or graduate office promptly and follow your institution's correction or deposit procedures. If the finding has also been published, ask the journal what disclosure or correction is appropriate.
- The prevention and handling of the missing data - PMC - NIH https://pmc.ncbi.nlm.nih.gov/articles/PMC3668100/
- Guidance for Industry: Patient-Reported Outcome Measures | fda.gov https://www.fda.gov/media/77832/download
- Sensitivity Analysis - NCBI - NIH https://www.ncbi.nlm.nih.gov/books/NBK126178/
- Dealing with Missing Data | va.gov https://www.herc.research.va.gov/include/page.asp
- Missing data: Issues, concepts, methods | sciencedirect.com https://www.sciencedirect.com/science/article/pii/S1073874624000082
- Missing Data and Multiple Imputation | Columbia University Mailman School of Public Health https://www.publichealth.columbia.edu/research/population-health-methods/missing-data-and-multiple-imputation
- Research Guides: Organizing Your Social Sciences Research Paper: Limitations of the Study | usc.edu https://libguides.usc.edu/writingguide/limitations
- Self-reported versus health administrative data | CMAJ Open https://www.cmajopen.ca/content/5/3/E729
