- Create an evidence/judgment/consequence table.
Read with a purpose
Find the central question and claim, and identify the paper’s role in your decision. Reconcile the abstract and conclusion with methods and tables. Distinguish educational reviews, primary studies and systematic evidence syntheses.
Assess validity using appropriate tools
Inspect selection, measurement, comparison, timing, missingness and analysis. Use an appraisal tool suited to the design instead of one generic score for everything. Reporting guidance is not identical to risk-of-bias assessment.
Examine estimates and interpretation
Compare collected and analyzed samples. Read effect estimates, uncertainty and model assumptions rather than significance alone. Attractive figures or a prestigious affiliation do not establish valid findings.
Check applicability and document status
Compare populations, settings and measurement with your own question. Check the publisher for corrections, withdrawal and version status. Clearly label preprints. Citation counts may reflect visibility or criticism, not methodological quality.
Keep an evidence record
Record question, design, key finding, limitations, appraisal and relevance to your argument. Separate quotation, summary and personal interpretation. Retain conflicting evidence and do not change eligibility criteria simply to favor preferred results.
Practical research checklist
- Reconcile claims, design and evidence.
- Use a design-appropriate tool.
- Record estimates and limitations.
- Check versions and publication status.
Worked case and implementation decisions
The following is a fictional teaching case. Do not use its numbers or wording as actual study findings.
A reputable journal or large sample does not finish appraisal of the fictional observational paper. Examine the question, selection, measurement, missingness, confounding and precision separately. Compare another paper by design and bias rather than only agreement in direction. CASP or a suitable specialist tool structures questions; connect each answer to evidence or absent information. An unexplained total score is not a methodological judgment.
| Stage | Teaching example | Verification question |
|---|---|---|
| Question | An assessable claim | Can the design answer it? |
| Sample | Origin and selection | Are generality limits explicit? |
| Measure | Definition and quality evidence | Is likely error assessed? |
| Analysis | Missingness, confounding and intervals | Are assumptions defensible? |
| Conclusion | Strengths, gaps and claim consequences | Is judgment tied to evidence? |
Exercise output: Create an evidence/judgment/consequence table.
Sources and further reading
Official sources for verification and further reading