Union officials want to estimate the percent, p, of workers at Chi Chi's who favor a strike. The union wants to have 90% confidence that its estimate for p is within 2.5% of the true population of workers who want a strike if no accurate estimate for p is available, how large a sample is needed?