Programming the statistical procedures from SAS

How to explain the Normality Tests using KS,etc?

Accepted Solution Solved
Reply
Frequent Contributor
Posts: 99
Accepted Solution

How to explain the Normality Tests using KS,etc?

Dear All,

Thanks for explain the follwoing table for me,whether this field check for normality with the answer to its normal distribution  yes/no?

thanks.

Further, what if the three tests have contradiction?

Tests for Normality

Test

Statictice

P Value

Kolmogorov-Smirnov

D

0.064727

Pr > D

>0.15

Cramer-von Mises

W-Sq

0.21097

Pr > W-Sq

0.2014

Anderson-Darling

A-Sq

1.28515

Pr > A-Sq

0.1916


Accepted Solutions
Solution
‎03-26-2012 04:12 AM
Grand Advisor
Posts: 9,451

Re: How to explain the Normality Tests using KS,etc?

From your P value. Should refuse  H0 : distribution is Normal  , if alpha is 0.05 .

W is usually used when size of sample is small.

A is ususally used when size of sample is large.

I recommend to use A.

Maybe some expert can give you a whole explaination.

Ksharp

View solution in original post


All Replies
Solution
‎03-26-2012 04:12 AM
Grand Advisor
Posts: 9,451

Re: How to explain the Normality Tests using KS,etc?

From your P value. Should refuse  H0 : distribution is Normal  , if alpha is 0.05 .

W is usually used when size of sample is small.

A is ususally used when size of sample is large.

I recommend to use A.

Maybe some expert can give you a whole explaination.

Ksharp

Frequent Contributor
Posts: 99

Re: How to explain the Normality Tests using KS,etc?

Ksharp, Thank you very much.

Respected Advisor
Posts: 4,606

Re: How to explain the Normality Tests using KS,etc?

bbb_NG, the way to interpret the results you posted is that about one out of five (20%) samples of the size you analysed would have a distribution that looks further away from the normal than yours, just by chance. As Ksharp mentioned, we usually reject the hypothesis of normality when there is little chance, less than one in 20, that our sample could arise from the sampling of a normal population. Thus, we must conclude that the hypothesis of normality cannot be rejected for your data.

The tests do not contradict each other. They simply look at different aspects of the data distribution and compare it to what is expected from the normal distribution. By analogy, one could say that an animal is a bird because it has feathers and no teeth while someone else would require that it flies and lays eggs. They look at different aspects.

Hope this helps.

PG

PG
Frequent Contributor
Posts: 99

Re: How to explain the Normality Tests using KS,etc?

PG,Thanks, it do helps.

N/A
Posts: 1

Re: How to explain the Normality Tests using KS,etc?

BBB,

For your reference.

To others,

No need to read this post.for bbb only,thanks.

☑ This topic is SOLVED.

Need further help from the community? Please ask a new question.

Discussion stats
  • 5 replies
  • 2314 views
  • 3 likes
  • 4 in conversation