stastistical test on normalized data

#1
I have to perform a stastistical test to asses if two grug show the same or different effect on my prepartion.
I tested on the same preparation only one drug and I have to use normalized data due to high variability in the control condictions.
What type can I use?
A non paramatric test right?
And how can do it on R software?
Thanks in advance

Simone
 
#2
I think I am reading you correctly. You have two treatment and one control datasets. You have somehow 'normalized' the two treatment groups by the control group and are left with the two 'normalized' treatment datasets. You now want to perform a statistical test to determine if there is a difference in the outcome between the two 'normalized' treatment groups.

If this is the case, and you feel that the data do not meet the assumptions of parametric tests for this kind of data (like the t-test, which assumes the data within each group are normally distributed), then you may use a variety of non-parametric tests. One popular non-parametric test for this kind of data is the Wilcoxon rank sum test. This test is used to test for a difference in the median response between groups.

In R, you can implement the Wilcoxon rank sum test with the command:

wilcox.test()

I suggest that you read the help file before using the test. This can be done with the following command:

?wilcox.test

or

help(wilcox.test)

-Matt
 
#3
Thanks Matt!!
I don’t have the normalized data using the same control for the both drugs because I could not test them on the same preparation. Is it the same??
I used the Shapiro Wilk test and the data are normal distributed

Thanks again

simone
 
#4
You, as the scientist will have to assume that the difference between each control was negligible. Then your approach is valid. The Shapiro-Wilk test is generally considered a good test for normality. Since you have concluded that your data are normally distributed, why not use a t-test?

R function:

ttest()

-Matt