An introduction to observational and experimental design

On watching this video, students should be able to:

1. Distinguish between an observational and experimental study.
2. Explain the main reason why a randomised experiment can if done well, be used as a test of causality and why from an observational study we can often only describe associations, discussing in particular:
   • The concept of random assignment of the exposure or treatment as the mechanism that controls for unwanted sources of variation between exposure groups/levels and allows causal inference
   • Why with observation alone there is no random assignment of exposure, so confounding by other factors should always be considered as an explanation for associations found in an observational study.
3. Explain what is meant by stratification and how this is used to control for confounding in observational studies.