R help - Jun 2010 - k-sample Kolmogorov-Smirnov test?

Hello,

I am curious if anyone has had any success with finding a R version of a
k-sample Kolmogorov-Smirnov test. Most of the references that I have able to
find on this are fairly old and I am wondering if this type of analysis has
fallen out of favour. If so, how do people tend to compare distributions
when they have more than two? Is it reasonable to pursue an adjusted p-value
method. That is, could you compare say three distributions by performing
three two-sample K-S test's then apply a bonferroni correction?

Just curious what some peoples approaches are when they want to compare more
than two distributions.

Thanks in advance.

Sam

-- 
*****************************************************
Sam Albers
Geography Program
University of Northern British Columbia
3333 University Way
Prince George, British Columbia
Canada, V2N 4Z9
phone: 250 960-6777
*****************************************************

	[[alternative HTML version deleted]]

If you want to test the null that all datasets come from the same distribution,
then I would probably do this as a permutation test.  Find the largest K-S
distance between 2 groups, then randomly permute the data into a new set of k
groups (with same sample sizes) and find the largest K-S distance again, repeat
this a bunch of times (total around 2,000 is good, the replicate function works
nicely for this).  This gives the distribution of the largest K-S distance under
the null, check to see if the value for the original data is in the tail or not.



-- 
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at imail.org
801.408.8111

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of Sam Albers
> Sent: Tuesday, June 22, 2010 12:28 PM
> To: r-help at r-project.org
> Subject: Re: [R] k-sample Kolmogorov-Smirnov test?
> 
>     Hello,
> 
> I am curious if anyone has had any success with finding a R version of
> a
> k-sample Kolmogorov-Smirnov test. Most of the references that I have
> able to
> find on this are fairly old and I am wondering if this type of analysis
> has
> fallen out of favour. If so, how do people tend to compare
> distributions
> when they have more than two? Is it reasonable to pursue an adjusted p-
> value
> method. That is, could you compare say three distributions by
> performing
> three two-sample K-S test's then apply a bonferroni correction?
> 
> Just curious what some peoples approaches are when they want to compare
> more
> than two distributions.
> 
> Thanks in advance.
> 
> Sam
> 
> --
> *****************************************************
> Sam Albers
> Geography Program
> University of Northern British Columbia
> 3333 University Way
> Prince George, British Columbia
> Canada, V2N 4Z9
> phone: 250 960-6777
> *****************************************************
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-
> guide.html
> and provide commented, minimal, self-contained, reproducible code.