Investors at a theme park would like to estimate the unknown proportion (p) of visitors who have visited the theme park in the past five years. They would like to be 90% confident of achieving an accuracy of at least 0.04. How large a sample of visitors should be selected?