[R] p-values with pvclust

[Greg Snow]

I don't know much about pvclust itself, but you might consider the technique in this paper:

     Buja, A., Cook, D. Hofmann, H., Lawrence, M. Lee, E.-K., Swayne,
     D.F and Wickham, H. (2009) Statistical Inference for exploratory
     data analysis and model diagnostics Phil. Trans. R. Soc. A 2009
     367, 4361-4383 doi: 10.1098/rsta.2009.0120

You just need to understand your null hypothesis enough to be able to simulate data from the null (possibly permuting original data, or generating data from a distribution similar to your data but without clusters).

The vis.test function in the TeachingDemos package helps with an implementation of the test (you will still need to write some of your own code though).

Hope this helps,

-- 
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 syrvn
> Sent: Tuesday, August 10, 2010 6:45 AM
> To: r-help at r-project.org
> Subject: [R] p-values with pvclust
> 
> 
> Hi,
> 
> 
> if you look at the first image (Image1)  you see that there are 2 main
> clusters 7 and 8
> I wanted to use pvclust to calculate a p-value whether these clusters
> are
> due to chance
> or statistically significant. Unfortunately pvclust does not provide a
> p-value for the first
> brunch (7 and 8).
> 
> So I added a row to my matrix which is very different to the rest of
> the
> data to create an additional
> brunch. See image here
> http://r.789695.n4.nabble.com/file/n2319732/Image2.png . I finally got
> my
> p-value for the brunch (7 and 8) which is 98 (9).
> 
> I was happy to see that the p-value was significant until I realised
> that if
> I add an additional brunch which
> is not that different from the rest but still cluster in a sperate
> cluster
> (see image here  http://r.789695.n4.nabble.com/file/n2319732/Image3.png
> )
> the p-value is changing and not significant any longer (84).
> 
> 
> 
> I was wondering why this happens because I thought that for each brunch
> the
> p-value is calculated independently?
> 
> Does anybody know how to get a correct p-value for the first brunch (7
> and
> 8) maybe without adding an additional brunch?
> 
> Best regards
> syrvn
> --
> View this message in context: http://r.789695.n4.nabble.com/p-values-
> with-pvclust-tp2319732p2319732.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> 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.