The limitations section of a public health thesis should name the specific constraints on your data, design and generalizability — not apologize generally. In Canadian public health research this usually means addressing self-reported survey bias, the population your data actually represents versus the population you are claiming to speak to, and any gap between administrative/surveillance data and the outcomes you are really measuring. State each limitation, its likely direction of effect, and how it bounds your conclusions.
What is the difference between a limitation and a delimitation in a public health thesis?
A limitation is a constraint outside your control that affects how far your findings can be trusted or generalized — a low survey response rate, a data source that undercounts a population, a cross-sectional design that cannot establish causation. A delimitation is a boundary you chose deliberately — studying one health region instead of the whole province, one age cohort instead of all ages, one time window instead of a longer trend. Public health examiners specifically check whether you have separated the two: presenting a deliberate scope decision as an apologetic limitation reads as if you did not control your own design.
Why does the limitations section matter more in public health than in some other fields?
Public health research is regularly used to justify policy and resource-allocation decisions, so examiners in this field hold the limitations section to a higher practical standard than a purely theoretical discipline might. A thesis that recommends a screening program, a public awareness campaign, or a resource shift based on findings that carry unstated limitations is read as methodologically incomplete even if the statistics themselves are correct. Framing your limitations section as “what a policymaker reading only my abstract needs to know before acting on this” is a useful discipline for deciding what belongs in it.
What are the most common methodological limitations in Canadian public health theses?
- Self-report bias in survey-based measures of health behaviour (smoking, alcohol use, physical activity, diet) — respondents systematically under- or over-report certain behaviours, and the direction of that bias is often predictable and should be stated, not just acknowledged.
- Cross-sectional design in studies using a single wave of a survey such as the Canadian Community Health Survey (CCHS) — a real association at one point in time cannot establish that exposure preceded outcome, and a public health thesis proposing an intervention based on cross-sectional association needs to say so explicitly.
- Healthy-user and healthy-volunteer bias in any study recruiting participants who opt in, which tends to skew toward more health-engaged respondents than the general population.
- Ecological fallacy when population-level or area-level data (postal-code-level income, health-region-level rates) is used to draw conclusions about individuals — a genuine and specific risk whenever administrative or geographic data substitutes for individual-level measurement.
- Surveillance and administrative-data lag, where a data source such as provincial hospitalization records or the Canadian Institute for Health Information’s (CIHI) databases reflects health-system contact rather than true incidence, undercounting conditions that go undiagnosed or untreated.
How do you write the limitations section without undermining your own thesis?
The structure that satisfies examiners without reading as self-defeating has three parts, repeated for each limitation:
- Name the limitation specifically. Not “the sample may not be representative” but “the sample overrepresents respondents with post-secondary education relative to the health region’s population, per the most recent regional census profile.”
- State the likely direction of effect. If a limitation would plausibly make your effect estimate larger or smaller than the true value, say which direction, and say so even when you are not certain of the magnitude.
- Bound your conclusion accordingly. Rewrite the claim so it holds even given the limitation — “findings apply to health-engaged CCHS respondents in this age band” rather than an unqualified population-wide claim.
A limitations section built this way reads as methodological control, not apology, and it is the version examiners are trained to look for.
What does a worked limitations paragraph actually look like?
A usable template that names the limitation, states its direction, and bounds the conclusion in three sentences:
“This study relied on self-reported physical activity from the Canadian Community Health Survey, which is known to overestimate moderate-to-vigorous activity relative to accelerometer-measured data in validation studies. If this bias operated similarly in the current sample, the true association between activity and the outcome measured here is likely weaker than reported. Findings should therefore be read as an upper-bound estimate of the activity-outcome relationship in this population, not a precise effect size.” Notice the structure does three jobs in three sentences: names the specific data source and known bias, states the direction the bias likely pushes the estimate, and rewrites the claim so it survives the limitation rather than ignoring it.
What limitations are specific to population-health and surveillance data?
When your thesis relies on surveillance or population-health data rather than data you collected yourself, the limitations shift toward measurement validity and coverage rather than sampling:
- Reportable-disease surveillance depends on testing and reporting behaviour, which itself varies by region, time period, and access to care — a rise in reported cases can reflect increased testing rather than increased incidence, and your thesis needs to address which explanation your design can and cannot distinguish.
- Postal-code-based linkage to socioeconomic indicators introduces the ecological-fallacy risk described above, plus a specific Canadian wrinkle: postal codes do not map cleanly onto census geography, and the conversion file used (Statistics Canada’s Postal Code Conversion File) has known, documented imprecision that should be named if your thesis relies on it.
- Indigenous health data specifically carries additional governance and interpretation limitations under OCAP® (ownership, control, access, and possession) principles — a thesis using any data describing First Nations communities without engaging OCAP® considerations is missing a limitation examiners in this field will expect to see addressed, and this sits alongside — not instead of — your standard REB approval obligations under TCPS 2.
For a broader look at what Statistics Canada makes available, and the access route (public tables versus microdata via a Research Data Centre) that determines how precisely you can describe your data source’s limitations, see how to get Statistics Canada data for your thesis.
How do limitations differ between quantitative and qualitative public health research?
Quantitative public health limitations centre on measurement validity, sampling frame, and confounding — the items above. Qualitative public health research (interview- or focus-group-based studies of health behaviour, access, or lived experience) has a distinct limitations vocabulary: transferability rather than generalizability, the researcher’s own positionality relative to the community studied, and saturation — whether your final sample size actually reached the point where new interviews stopped producing new themes. A mixed-methods public health thesis needs both limitation vocabularies stated separately, not blended into one generic paragraph. Saturation claims specifically connect back to your original sample-size reasoning — a limitations section that never revisits that earlier justification looks disconnected from the methods chapter.
What limitations come from using linked administrative or secondary data?
A growing share of Canadian public health theses use linked administrative data — combining, for example, hospital discharge records with vital statistics, or CIHI datasets with regional health-authority data — rather than data collected specifically for the study. This introduces its own limitation category, distinct from survey-based limitations:
- Purpose mismatch. Administrative data was collected for billing, operational, or vital-registration purposes, not for research, so variables you need may be coded coarsely, missing, or defined differently than a research instrument would define them.
- Linkage error. Probabilistic record linkage between datasets introduces both false matches and missed matches, and the direction of resulting bias depends on which records are more or less likely to link successfully — often not a random subset of the population.
- Privacy-driven data suppression. Small-cell suppression (a standard Statistics Canada and CIHI practice to prevent re-identification) can hide exactly the small, high-risk subgroups your thesis may most want to describe, and this needs to be named as a limitation rather than silently worked around.
- Access lag. Secondary datasets are often released one to three years after the reference period, meaning your “current” findings describe a period that may already be historical by the time your thesis is examined — worth stating plainly rather than implying currency the data does not have.
What do examiners actually check in this section?
Across Canadian public health programs, examiners consistently look for three things in the limitations section specifically:
- Whether every major methodological choice earlier in the thesis (data source, design, sample) has a corresponding limitation here — a methods chapter with no matching limitation is read as an oversight, not as evidence of a flawless design.
- Whether the discussion and conclusion chapters actually respect the stated limitations, rather than making claims the limitations section itself says the data cannot support.
- Whether recommendations for future research follow logically from the specific limitations named, rather than generic boilerplate (“future research should use a larger sample”).
Frequently asked questions
Should I discuss limitations in the methods chapter or save them all for the end?
Both, in different forms. Brief methodological limitations belong in the methods chapter, where you justify a specific design choice against its known constraint. The dedicated limitations section near the conclusion is where you assess the cumulative effect of all limitations together on your specific findings.
Is it acceptable to have a long limitations section in a public health thesis?
Yes, when every limitation is specific and tied to a real feature of your data or design. A long list of vague, generic limitations reads worse than a short list of precisely stated ones — length is not the signal examiners respond to.
How do I handle a limitation I only discovered after data collection was complete?
State it plainly and explain what you would do differently with hindsight. Examiners generally respond better to a candidly acknowledged late-discovered limitation than to one a reader has to infer was hidden.
Do limitations weaken my thesis grade?
A well-argued limitations section is evidence of methodological maturity and typically strengthens an examiner’s assessment. What weakens a thesis is a limitations section that is missing, generic, or contradicted by overreaching claims elsewhere in the document.
What is the difference between a limitation and a threat to validity?
They overlap but are not identical terms — “threat to validity” (internal, external, construct, statistical conclusion validity) is the more technical epidemiological framework some Canadian public health programs specifically require you to use instead of, or alongside, a general limitations discussion. Confirm which vocabulary your supervisor and program expect.
Should I mention funding sources or conflicts of interest as a limitation?
Yes, if any funder or partner organization had a role in study design, data access, or the decision to publish. This is normally a short standalone disclosure rather than folded into the methodological limitations list, but public health examiners specifically look for it given the field’s frequent industry- and government-funded research.
Can a limitation also be a strength, depending on framing?
Occasionally, but resist the temptation to spin every limitation as secretly a strength. A large administrative dataset with a specific bias is genuinely both large (a strength for statistical power) and biased (a real limitation) — state both honestly rather than using strength-language to minimize a real constraint.
If your public health thesis is itself a synthesis of existing evidence rather than primary data collection, the limitations vocabulary shifts again — see how a systematic-review thesis in an adjacent health field documents search comprehensiveness and study quality as its own specific limitation category.
Where Tesify fits
Tesify cannot decide which limitations apply to your specific data — that judgement has to come from you and your supervisor. What it speeds up is turning a rough list of methodological concerns into the three-part structure examiners expect (name it, state the direction, bound the conclusion), consistently across every limitation, so the section reads as deliberate methodological reasoning rather than a bolted-on afterthought written the night before submission.