# Kurtosis and Skewness coefficents and ANOVA

#### Lolo

##### New Member
Hello,
A friend of mine told me that despite my variable is not ditributed normally, I am allowed to do an ANOVA if both Kurtosis and Skewness coefficents are between -3 and 3.
Do you confirm that?

E

#### elnaz

##### Guest
hello
not at all
when i see my data are not normal 1) i use cox-box transformation or 2) i use non parametric test
i advice you do like me
i dont do that at all
Best regards
Elnaz

#### Lolo

##### New Member
Thanks a lot.
I want to do a non parametric test with 3 factors. I have been said to do a Kruskall-Wallis test, but this is for 1 factor, isn´t it?

E

#### elnaz

##### Guest
Hello
is factor for you , population?
if your intent is that you have three population and you want to compare them .
yes ,Kruskall-Wallis test is true.

Last edited by a moderator:

#### Lolo

##### New Member
Hello Elnaz,
Thanks very much for your help.
Do you know if a non-parametric test with 3 factors (=2 studied factors + 1 controlled one) does exist?
Regards

#### Lolo

##### New Member
(Sorry Elnaz, I didn’t see your reply before sending my latest message!!)

This is the aim of my study:
I am studying the post harvest shelf-life of melon fruits, according to the melon variety and the duration of storage (in a cold chamber)
I want to:
- find the varieties that show the longer shelf life
- show the fruit quality loss all along the storage
- if possible : find the maximum shelf-life for each variety.

So I made a split-plot design trial with 2 blocks (referring to the position of the fruit)
I made several measurements referring on fruit quality (weight loss, external and internal appearance, firmness...), which are the variables of my analysis.

What I call "factor" is what I made vary during the experiment. It is to say: the melon variety (I tested 9 different varieties), the storage duration (1 day, 5 days, 10 days...) and the block (which is a factor with a random effect)

My datas are in the following table:

--------------------------------------------------------------------------
Storage Variety Block Weight loss Appearance (mark) ...
--------------------------------------------------------------------------1 day No1 No1
No2
1 day No2 No1
No2
1 day No3 No1
No2
1 day … No1
No2
5 days No1 No1
No2
5 days No2 No1
No2
5 days No3 No1
No2
5 days … No1
… No2
--------------------------------------------------------------------------

Since some variables (for example Appearance) doesn’t follow a normal distribution and given that I didn’t find any suitable transformation for them, I have to do a non-parametric test with 3 factors (= storage, variety, block)

Do you understand my problem?

Best regards

#### Lolo

##### New Member
The file!

Sorry the table was not readable! This is a file.