Hi Annie,
The formula for this is:
n = (z*/m)2*p*q
n= sample size
z*=z-critical value
m=margin of error
p=probability of success (person supports candidate)
q=probabiility of failure (person opposes candidate)
When you know nothing in advance about your population, you must assume p and q are 0.5. Now, in general, the z-critical value for 99% confidence intervals is 2.5760. You may want to memorize this as it comes up a lot in statistics classes. Anyway, that means:
z*= 2.5760
m= 0.002–divide your 0.2 by 100 to get the appropriate margin of error
p=0.5
q=0.5
n=(z*/m)2*p*q
n= (2.576/0.002)2*0.5*0.5
n=414736
This is very high because your candidate wanted an extremely low margin of error. This value wound up in the denominator, which means the sample size will increase substantially, I hope this helps.