Research Methodology
Research methodology tests whether you can turn a clinical uncertainty into a valid, ethical study and interpret its limitations. The exam often asks for the design, sampling method, main bias or ethical safeguard.
Core rules
- A good question is focused: population, intervention or exposure, comparison, outcome and time frame.
- Cross-sectional designs measure prevalence; cohort studies measure incidence; randomised trials test causality best but may lack generalisability.
- Sampling must be defined; convenience samples threaten external validity.
- Bias is systematic error: selection, information, recall, observer and publication bias.
- Informed consent, confidentiality, ethics approval and data protection are mandatory.
- Report effect estimates with confidence intervals and adjust for confounders in the analysis plan.
- Pilot the questionnaire to improve reliability.
Worked example
A trainee wants to know the prevalence of undiagnosed hypertension in a primary-care centre. The reasoning: a cross-sectional study with a random or consecutive sample of eligible adults, standardised blood pressure measurement, repeated readings and a clear denominator. Convenience sampling of people attending for other reasons would overestimate or underestimate prevalence. Predefine the analysis. Report confidence intervals and response rate.
Traps
Calling a questionnaire study a cohort; confusing confounding with bias; changing the outcome after analysis; failing to report refusals and missing data; and using a statistical test as a substitute for design.