answersLogoWhite

0

Your confidence level is 19/20=95%
1% is your margin of error
'either direction' means it's two-sided

Sample size, n, should be larger than (zalpha/2)2p(1-p)/m2

where

zalpha/2 is the upper critical value of the normal distribution. For your level of confidnce (95%) alpha=0.05. So alpha/2=0.025, making z=1.96

p is your 'best guess' about the true value of the proportion. You can estimate with a pilot study, or set to 50% which will maximise the sample size possibly much larger than required.

m is your margin of error (1%)

So choose a sample size of n>=(1.96)2*(0.5)2/0.012=9604

However, this will result in a test with only 50% power (i.e. probability of not getting a false negative is 50%). Typically power is set to around 80%. To increase power the sample size would have to be even larger.

User Avatar

Wiki User

17y ago

What else can I help you with?