Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 0 down vote favorite

There are two laboratories measuring identical samples. One of the lab's data is non-normal, therefore I am looking for a non-parametric test.

After comparing overall differences in the dependent variable (continuous) using Wilcoxon Mann-Whitney, I wish to test for differences between labs based on a third variable.

There are a number of third variables that I wish to test for, and some of them are categorical (2 levels, and 3 levels), and another is ordinal (by month).

Which test would be appropriate? I read about the van Elteren test, but the SAS output did not make sense.

Thanks in advance!

share|improve this question
2  
You said "One of the lab's data is non-normal, therefore I am looking for a non-parametric test." -- however, the usual assumptions nonparametric tests tend to make is that the distributions are the same (apart from location shift). If the shapes are all over the shop, nonparametric tests might be affected as much as ANOVA. Even the common bootstrapping-of-residuals (or a number of other resampling approaches) will be screwed up, because they rely on the ability to swap observations across groups. Resampling within groups might be used (one that resamples the rows of $[X|y]$ say). – Glen_b Nov 1 '12 at 21:42

1 Answer

up vote 2 down vote

The van Elteren test, a stratified Wilcoxon Rank Sum Test, is correct. It is available in PROC freq. There's an example on the SAS website support documents.

share|improve this answer
Could you add a link to the support docs? – naught101 Nov 1 '12 at 22:42
I'm confused as to interpreting my output. The PROC freq gives me a table showing lab on the row, and my dependent variable values in the column. However, the cells are mostly zeros. – mbee Nov 2 '12 at 1:25
This is the link that I referred to support.sas.com/kb/25/022.html – mbee Nov 2 '12 at 1:26

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.