Imagine you're evaluating an A/B test with both a test group and a control group.
How would you determine the necessary sample size for a reliable measurement?
Assuming the sample size is adequate and comparable between the two groups, should the power increase or decrease to detect very small differences?