Measurement, Statistics, Causal Inference, and Research Design
Medical Disclaimer
This content is for educational purposes only. Always consult a qualified medical professional or certified trainer before making decisions about your health, exercise, or fitness regimen.
How to know whether an intervention truly works
Evidence hierarchy is necessary but not sufficient
A randomized trial is powerful, but not all questions are randomized, and not all randomized studies are well designed. Researchers must understand bias, confounding, measurement error, missingness, and external validity.
Causal thinking
Association does not imply causation; the central challenge is identifying the counterfactual outcome that would have occurred without the intervention.
Core methodological terms
Study designs
| Design | Strength | Weakness |
|---|---|---|
| Cross-sectional | Fast and inexpensive | Cannot establish temporal sequence |
| Cohort | Tracks outcomes over time | Confounding and attrition |
| Randomized controlled trial | Strong causal inference | Costly and sometimes less generalizable |
| Mendelian randomization | Uses genetic instruments | Assumptions may fail |
Which design is generally strongest for causal inference when feasible?
Randomization helps balance both measured and unmeasured confounders on average.
Correct answer: Randomized controlled trial
What is a confounder?
Confounding can create a misleading association if not controlled.
Correct answer: a variable linked to both exposure and outcome
Reliability vs validity
Reliability
- Consistency
- Precision
- Can be high even if the measure is wrong
Validity
- Accuracy
- Construct relevance
- Depends on correct interpretation
Why is effect size important?
Effect size helps interpret whether a statistically significant finding is meaningful.
Correct answer: It describes practical magnitude
Define external validity.
It is the extent to which findings apply beyond the study sample and setting.
Correct answer: generalizability