I want to calculate the sample size required to detect a 50% change in the probability of an binary outcome.

The base probability is .2. A 50% increase would go from .2 to .3. A 50% decrease would go from .2 to .1.

In this example, assuming a normal distribution, it should be easier to detect an increase compared to a decrease because there will be more observations in the former. Using power.prop.test in R confirms that the former requires a smaller sample size. However, using pwr command gives only a single sample size. How is that single sample size derived?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.