Back transformation of p-values

#1
Hi everybody,

I am currently working with some raw data which are not normally distributed (ND). I have now transformed the data with squareroot and they are ND. I then analysed my ND data with a 2-way ANOVA and got some p-values under 0.05. Do I have to back-transform my p-values or should I use the p-values from the ND data on my raw data?

Thank you very much

Kimmie
 
#2
No, I think you can use the p-values obtained by the ND data as they are. When you will present your analysis, you will have to specify which transformation you made before the ANOVA, and if you will add other data they will have to be transformed in the same way before the analysis, but p-values don't have to be "back-transformed".
 

Karabiner

TS Contributor
#3
I am currently working with some raw data which are not normally distributed (ND).
No ANOVA procedure requires normally distributed data. Please read your references carefully .

With Kind regards

K.