#5 — statistical test for distributions comparison

State	Resolved
Area	Statistics and Mathematics
Issue type	Fast question
Severity	Low
Submitted by	Luke Skywalker
Submitted on	May 09, 2014
Responsible	Enza Colonna

Return to tracker

Hi,

I have two series of data relative to cell diameters. One set is form cell treated with a drug and the other from cells treated with placebo. I want to compare the two distributions to asses if there are significant differences, and I thought about using the Kruskal-Wallis test to obtain a p-value.

Do you think this is appropriate?

thanks.

Added by Enza Colonna on May 09, 2014 07:39 AM

Issue state: Unconfirmed → Resolved

Severity: Medium → Low

Responsible manager: (UNASSIGNED) → enza.colonna

Hi Luke,

thanks for your question.

Yes, the test you want to use is appropriate and especially valid for data not normally distributed, if you are interested in comparing median values of the diameters you can also use the Wilcoxon Mann-Whitney test.

However, if your data (diameters) is normally distributed, than you can apply the t-test, and thus we suggest you first check if your data is normally distributed, perhaps just making an histogram.

Hope this helps
Enza

No responses can be added.