In: Statistics and Probability
Write a report to your boss at C-Span, explaining to her what you are doing.
z value for 95% confidence is 1.96
Margin of error E = 3/100 = 0.03
Sample size =
= 1040 (Rounding to nearest integer)
If we do not know the proportion estimate, we assume p = 0.5 so as to get the conservative estimate of sample size. This is because the standard error (and margin error) is maximum at p = 0.5 for a given sample size.
Sample size =
= 1067 (Rounding to nearest integer)
For 99% confidence level, z = 2.576
Sample size =
= 1843 (Rounding to nearest integer)
We require larger sample size for 99% confidence level as compared with 95% confidence level. The polling company use 95% to get a smaller confidence interval length which will provide more precise estimate than 99% confidence interval.
If we know the proportion estimate, we need 1040 samples to estimate interval within 3 points with 95% confidence. If we do not know the proportion estimate, we need 1067 samples to estimate interval within 3 points with 95% confidence. If we need to increase the confidence level to 99%, we require a large sample size of 1843.