Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
sample size and normal distribution
Message posted by Lee Bascom (via 205.188.201.209) on January 18, 2002 at 3:55 PM (ET)
What is the smallest sample size one can use and still have a normal distribution?
Thanks
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: sample size and normal distribution
Message posted by JG (via 64.10.126.132) on January 19, 2002 at 5:47 AM (ET)
It all depends on what the physical process is, if sampling is independent, etc. Usually, 30 is assumed adequate, 120 definitely good, and in some cases, even 5 or 6 is assumed to be reasonable.
In any case, we only approximate normality - adequate for some purposes but other purposes.
Re: sample size and normal distribution
Message posted by JG (via 64.10.124.110) on January 19, 2002 at 4:54 PM (ET)
when we average normals we get a normal and when we average anything else, we approach normality. if we start with close to normal we approach normality very quickly. thus 1 or 2 good if start with near normal,etc.
Re: sample size and normal distribution
Message posted by JG (via 64.10.124.92) on January 19, 2002 at 10:57 PM (ET)
Thus, to answer the sample size question, we need to know how close to normality is the population we are samplig, how independent are the observations, what the sample is used for, etc.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
