In: Statistics and Probability
Suppose you want have a coin that you think might not be exactly fair, that is the probability of a head might be slightly different from 0.5. You would like to produce a 95% confidence interval for p, the true probability of a head. You decide that you would like the margin of error for your interval to be plus or minus 0.001 (that is plus or minus 0.1%).
How many times, n, do you need to toss the coin to be sure your CI has this margin of error, or less?
The following information is provided,
Significance Level, α = 0.05, Margin of Error, E = 0.001
The provided estimate of proportion p is, p = 0.5
The critical value for significance level, α = 0.05 is 1.96.
The following formula is used to compute the minimum sample size
required to estimate the population proportion p within the
required margin of error:
n >= p*(1-p)*(zc/E)^2
n = 0.5*(1 - 0.5)*(1.96/0.001)^2
n = 960400
Therefore, the sample size needed to satisfy the condition n
>= 960400 and it must be an integer number, we conclude that the
minimum required sample size is n = 960400
Ans : Sample size, n = 960400