A national exam is given yearly to fifth graders. At one school they found 3 to 5 times as many erasures as at other schools, and 89% of those erasures resulted in correct answers as compared to less than 70% for other schools. Has there been tampering? An investigator wants to determine the proportion of test answers that result in correct answers after eraser.
How large should the sample be to estimate within 2% having 99% confidence?