Describe a sample
A sample share does not establish population prevalence, representativeness or causation.
Analysis
An analysis should make its denominator, units and comparison visible before presenting a conclusion.
In practice
Describe a group as a share of the actual sample. Avoid silently switching between the number invited, the number responding and the number included in the final analysis.
Explain uncertainty and the limits of generalising beyond those observations.
Study numbers
Calculate observed change
A before-and-after difference alone cannot attribute the change to an intervention.