Non-normal Distribution

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

Non-normal Distribution
Message posted by kevin (via 161.142.15.152) on June 18, 2001 at 9:22 AM (ET)

What should I do if I want to compare more than 3 groups and 2 groups with non-normal data. I check for normality using kurtosis and skewness with the kurtosis and skewness values must be fall in the rank of +/- 2 times std error of kurtosis and +/- 2 times std error of skewness.

Please help me what kind of test that I can use ??

I've read before that we can assume the normality a set of data if we have large samples. How many samples actually can say as large samples ??

TQ !

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Non-normal Distribution
Message posted by JG (via 128.8.23.9) on June 18, 2001 at 11:13 AM (ET)

First, the best way to check for normality is to plot the data on special graph paper which can be constructed with the help of a Z-table or purchased.
The 'central limit theorem' tells us that the sampling distribution of sums of random variables approches normality as sample size increases. How fast this happens and how much normality one needs for certanin purposes are all separate issues that have to be solved on a case by case basis.
Also, remember that only populations can be have a normal distributions. In a sample it means nothing and small samples from normal or near normal populations can be very far from normal.

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.