[R] Hotelling T2 test extension for multigroup data

[Vickie S]

Hey Peter,

I agree that Hotelling's T test, although technically possible for two groups but rejecting the Ho in this case is not going to convey any meaning.

But I am sure there is a possibility. My assumption is based on using simulated reference set for each comparison.

-- Vickie 




----------------------------------------
> Subject: Re: [R] Hotelling T2 test extension for multigroup data

> From: pdalgd at gmail.com

> Date: Thu, 9 Feb 2012 14:08:22 +0100

> CC: r-help at r-project.org

> To: isvik at live.com

>

>

> On Feb 9, 2012, at 12:11 , Vickie S wrote:

>

> >

> > Hi all,

> > I've got the following matrix :

> >

> > mat <- matrix(rnorm(700), ncol=5, dimnames=list( paste("f", c(1:140), sep="_"), c("A", "B", "C", "D", "E")))

> >

> > I can see that currently most of the multivariate Hotelling T2 tests are limited for application on two groups/samples.

> >

> > I wud appreciate if someone can provide me a suggestion how to implement the same for multiple groups.

> >

>

> Please don't start up a new thread, cloning the same misconceptions as an earlier one! It doesn't help, you will only get people misunderstanding you in the same way...

>

> Apparently, you have rows and columns reversed relative to the common layout in multivariate analysis. Also, you have only one observation for each of your "groups" (species)?

>

> What makes you think that T^2 would work for you if you had only "A" and "B" species? I don't think anything will work unless you have either replication or some sort of additional assumptions on the covariance structure of your 140-dimensional response.

>

> --

> Peter Dalgaard, Professor

> Center for Statistics, Copenhagen Business School

> Solbjerg Plads 3, 2000 Frederiksberg, Denmark

> Phone: (+45)38153501

> Email: pd.mes at cbs.dk Priv: PDalgd at gmail.com

>