I am trying to use proc power to determine sample size for a direct
mail test. I want to know how many people I need send package B (as
opposed to package A) in order to have a confidence level of .05 and
power of .80. For package B to be successful it will need to increase
the response rate from an estimated .03 which is what we expect from
package A to .035. I know I will be sending 100,000 people package A
as a control and want to use proc power to determine how many people
should be sent package B.