Steve E. answered 03/29/16
Tutor
5
(15)
Patient and a good listener
For a margin of error of a 95% confidence interval, you can use the following formula:
n = ( p-hat*(1 - p-hat)*(1.96)^2 ) / MOE^2
where p-hat is the estimate of proportion success. Since you don't know what this is before taking your survey, a conservative estimate for the number in the sample is to use p-hat = 0.5. This proportion will produce the largest number of participants for your desired MOE.
So,
n = ( (0.5)*(0.5)*(1.96)^2 ) / (0.04)^2 = 600.25 => 601 participants
For the MOE for your voting survey, the proportion voting for Hillary is 422/813 = 0.52. So p-hat = 0.52.
p-hat +/- (1.96)*sqrt( ( (.52)*(.48) ) / 813 ) is your 95% confidence interval for your MOE
0.52 +/- 0.034, so MOE = +/- 0.034
(0.486 , 0.554) is the confidence interval for your proportion estimate. This tells you that if you were to do many repeated surveys like this one, the true population proportion would fall in this range 95% of the time.
Steve E.
05/17/16