Public-data topic guide
Is your public-data research topic feasible?
Turn “can this topic be done?” into a feasibility memo Requested outputs / delivery goal: A feasibility memo, a gap list and a research-plan discussion draft.
A feasible public-data study is not simply one whose data can be downloaded. The research question, permitted uses, target population and design, core variables, effective sample, analytical methods, and added value must align before the topic is ready for a protocol.
Assess feasibility before formal statistical analysis, and allow conclusions of 'needs adjustment' or 'currently infeasible'. If the main exposure, outcome, temporal sequence, or authorization conditions are missing, a larger sample, more models, or a rewritten title cannot close the fundamental gap.
A concrete task with RigorMed
Turn “can this topic be done?” into a feasibility memo
These are suggested task inputs and outputs to request, not a new execution, a customer outcome or a guarantee of completion.
Materials to provide: The research question, lawful data sources, variable documentation, existing sample summaries and verified similar studies.
Requested outputs / delivery goal: A feasibility memo, a gap list and a research-plan discussion draft.
An example request
Separate established conditions, pending questions and blockers, linking each to a source. Check variables, design, sample and added value separately, then recommend retaining, narrowing or postponing the research plan.
What still needs review
A popular database, overall sample size or a single significant result does not establish feasibility. Do not invent ethics facts, event counts or statistical power.
Three possible feasibility conclusions
Feasibility is not a simple yes-or-no decision. A clear conclusion category makes the next action explicit.
| Conclusion | Typical conditions | Next step |
|---|---|---|
| Ready for a protocol | The source and permitted uses are clear, the design fits, core variables are verifiable, initial sample and method checks are supportive, and no unexplained duplication problem has been identified | Freeze the research question, variable definitions, and analysis plan before formal data preparation |
| Conditionally feasible | The main framework is sound, but cycles, variable mapping, sample size, missingness, similar studies, or governance still have gaps that can be investigated | List the gaps, responsible parties, and verification methods; reassess after resolving them |
| Currently infeasible | Data are unavailable or their use is not permitted, core variables are missing, the design cannot answer the question, the sample is clearly inadequate, or the intended conclusion exceeds what the data support | Stop this topic, revise the question, or choose a suitable data source |
'Conditionally feasible' is not automatic clearance. Each gap must concern a specific source, variable, sample, or methodological fact and have a documented verification result before the topic can become ready for a protocol.
Seven checks for a public-data study topic
Is the research question specific enough?
Specify the target population, main exposure or predictors, primary outcome, time window, and study design. A disease or database name alone is not a question that can be checked.
Are the data genuinely available and their use permitted?
Retain the official access point, version, access date, license or data-use agreement, application requirements, citation requirements, and redistribution restrictions. A public web page, a searchable catalog, and downloadable patient-level data represent different access states.
Does the design support the intended conclusion?
Check the sampling frame, included population, observation period, temporal order of exposure and outcome, and target population. Cross-sectional surveys, registries, and hospital data do not answer the same questions.
Can each core variable be verified?
The main exposure, outcome, and required covariates need evidence for official definitions, measurement methods, units, coding, eligible populations, and cycles. If a key concept is represented only by an incomplete proxy, specify measurement bias and interpretive limitations in advance.
Does the effective sample support the planned analysis?
Estimate the final sample using eligibility criteria and variable availability, not the database's total enrollment. For binary outcomes, examine events and non-events. Prediction studies need separate development and validation data; complex surveys require attention to design degrees of freedom.
Have similar studies been checked?
Search sources such as PubMed for the closest published studies and check registries such as ClinicalTrials.gov when relevant. Compare populations, data cycles, variable definitions, designs, models, and conclusions, rather than checking only for identical titles.
Can the analytical and governance boundaries be implemented?
Before statistics, specify the main model, missing-data strategy, sensitivity analyses, and reporting standards. Responsible parties must confirm ethics, exemptions, data-use conditions, and research registration requirements. Software can flag gaps but cannot make governance decisions for institutions or researchers.
Four gaps that wording or models cannot fix
First, the data cannot be obtained lawfully or the permitted uses do not cover the study. Second, the main exposure or outcome has no verifiable variable. Third, the design cannot establish the required temporal sequence or target population. Fourth, the effective sample or outcome events cannot support the planned analysis.
These are topic-level blockers. Revise the research question, narrow the conclusions, or change data sources instead of adding covariates, changing statistical methods, or presenting exploratory results as prespecified findings.
Six signals that do not establish feasibility on their own
- The database has a very large total sample;
- A key term appears on a variable-search page;
- Published papers have used the same database;
- An initial model produces a statistically significant result;
- No paper with an identical title was found;
- A sophisticated model can run on the available data.
Checklist before reaching a feasibility conclusion
Record each item as confirmed, pending, missing, or blocked, and retain its evidence source:
- Are the research question and intended conclusion explicit?
- Are the official access point, version, access requirements, and permitted uses confirmed?
- Do the target population, sampling frame, and temporal structure match the topic?
- Do the main exposure, outcome, and necessary covariates all have verifiable definitions?
- Is there evidence of variable comparability across cycles or sources?
- Have the final sample, events, non-events, and missingness been provisionally quantified?
- Do model parameters, survey design, and validation strategy match the sample?
- Has search evidence been retained for the closest publications and relevant registered studies?
- Does the topic add value beyond changing a variable, cycle, or model?
- Are responsibilities for ethics, authorization, registration, and reporting clear?
Passing feasibility means only that the topic is ready for the protocol stage. It does not imply that results will be statistically significant or that a paper will be published.
Frequently asked questions
Why can a topic be infeasible even when a public database is downloadable?
Downloadability shows only that an access point exists. The intended use must also be permitted, the design must support the question, core variables must be verifiable, the effective sample must be adequate, and analytical and interpretive boundaries must be implementable. Any missing core condition can block the topic.
If all the variables exist, can statistical analysis begin?
Not yet. Confirm variable eligibility, measurement timing, coding, missingness, comparability across cycles, and the alignment of design, sample, and weights. Starting formal statistics before freezing these conditions can turn data problems into problems of interpretation.
Does finding no identical paper establish novelty?
No. A missed result may reflect the search strategy, terminology, database coverage, or timing. Retain the strategy and compare the closest studies by population, variables, design, and methods. Novelty requires verifiable differences.
Can the topic change after passing feasibility assessment?
Possibly. Data cleaning, complete-sample counts, and ethics or authorization checks may reveal new limitations. Record the reason for any important change and reassess the question and analysis plan, rather than silently changing the primary outcome or hypothesis.
Verified sources
The official sources below were checked on 2026-09-07 / RigorMed: 2026-10-13. Access conditions and update dates are rechecked before publication.
Start with a specific task
Turn “can this topic be done?” into a feasibility memo
Bring legally usable materials and this page's example request to the workspace. Check the pre-execution material and credit notices before confirming, then review the resulting outputs.