Evidence and evaluation
Research design and statistical analysis plans
Grant panels and journal reviewers judge a study by its design long before results exist. A clear estimand, a credible comparison and a pre-specified analysis plan often decide whether an application is funded.
Questions this answers
- Is our research question specified precisely enough to analyse?
- What comparison will answer it, and what are the main sources of bias?
- How large does the study need to be?
- What exactly will we analyse, and how, before we see the data?
Methods
The estimand framework, causal diagrams, target-trial specification, power and sample-size calculation (including by simulation), and statistical analysis plans to reporting-guideline standards.