what would happen to the width of a confidence interval if the size of the population being sampled was doubled, from 25 million to 50 million?