Research Methodology
Research Methodology tests how to pose an answerable question, choose a design, control bias, and report transparently. The exam rewards understanding of validity over memorized design names.
Core rules
- A good question specifies population, intervention or exposure, comparator, and outcome. A hypothesis must be falsifiable.
- Designs: randomized controlled trial for causal inference; cohort for incidence; case–control for rare outcomes; cross-sectional for prevalence; systematic review for synthesis.
- Bias includes selection, information, confounding, recall, and publication. Randomization and allocation concealment address confounding and selection; blinding reduces measurement bias.
- Validity is internal when causal inference is sound within the study and external when results generalize. Reliability is consistency, not accuracy.
- Ethics require IRB approval, informed consent, confidentiality, and reporting of adverse events.
Worked example
A study finds coffee drinkers have less heart disease. Reasoning: this is observational; confounding by smoking, income, and health status is plausible. A randomized trial would be needed for causation, but blinding coffee is difficult. Adjusting for confounders helps but cannot remove unmeasured confounding. Report association, not proven causation.
Traps
Calling a cross-sectional study causal; confusing reliability with validity; ignoring intention-to-treat; changing outcomes after seeing data.