Re: How to reassemble $components in coleo(ade4)?

column.names is not an argument (it is col.names). Try:

 > data(coleo)
 > write.table(coleo$tab, file = "coleo.txt", sep = " ", 
row.names=coleo$species.names, col.names=coleo$moda.names)

Cheers.

Mark Krueger wrote:
> Dear Stéphane,
>
> thanks for your explanation. The question is how do I export the 
> complete data set as one matrix.
> The write.table() function as such did not work since there are 
> different argument length (110, 32, 9). Following your hint I used
>
> > write.table(coleo, file = "coleo.txt", sep = " ", 
> row.names=species.names, column.names=moda.names)
>
> but this failed as well due to "unused arguments".
>
> What did I miss?
>
> Thank you for your help!
>
> Best,
> Mark
>
>
>

interpreting Fisher's classification function coefficients

  
Dear colleagues,

I run discriminant analysis on a dataset with plants species presence as  variables (binary,
independent) and lakes # as cases. I also have a classification variable with 4 classes (representing
each PC of PCA of the same matrix, but transposed). 

I have some problems with interpreting Fisher's classification function coefficients. 
It is said that a case belongs to the class for which the score is highest. In all books you can find
something about this, one compares (e.g.) score= 0.1 with score= 2.5 (of which 2.5 is largest).

My question is: would an object belong to a class with score=0.1 or to one with the score= -75 ?. I
don´t know if I should take that state literary.

Any suggestion much appreciated.

Thanks

Dragos Zaharescu
@VigoUni.

      
	[[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
R-sig-ecology@...

pairwise comparisons in community structure analysis data

I have a question that I'm not sure has a right answer, but I would
appreciate any and all opinions, especially if you know of any citations to
back them up.

In the past, when dealing with univariate data, I have always been a
proponent of using Fisher's Protected LSD for pairwise comparisons because
it is the most powerful procedure, and the accuracy of its experimentwise
error rate is based on a sound argument that was subsequently proven to be
true with simulations in Carmer and Swanson (1973 ).  I included the full
reference below.

At the moment, I'm working with community structure data (multivariate
nonparametric) at multiple sites and there is an need to determine which
sites are different from which other sites.  My colleague has used a
Bonferroni correction  for this type of question the past, but I tend to
think that is most likely too conservative.  I'm interested to know if any
of you have dealt with a similar problem and/or if you know if anyone has
done any work on comparing pariwise comparisons procedures for multivariate
nonparametric data.

I've done a preliminary literature search with no success, but am still
looking.

Thanks for your help.

-Penelope

Carmer, S. G. and M. R. Swanson (1973). "An evaluation of ten pairwise
multiple comparison procedures by Monte Carlo methods." Journal of the

Re: pairwise comparisons in community structure

Non-metric Multidimensional Scaling (NMS or NMDS) is one procedure used to
analyze differences in sites regarding vegetation structure. It uses a
dissimilarity matrix to orient sites in species space. It is available as a
base package in R as isoMDS, or in the package Vegan as metaMDS. It is not a
pairwise analysis, as you are requesting, but it is useful for showing
community differences among multiple sites.
Jonathan White
UofL Biology
Louisville, KY

On Tue, Aug 11, 2009 at 6:00 AM, <r-sig-ecology-request@...>wrote:

> Send R-sig-ecology mailing list submissions to
>        r-sig-ecology@...
>
> To subscribe or unsubscribe via the World Wide Web, visit
>        https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
> or, via email, send a message with subject or body 'help' to
>        r-sig-ecology-request@...
>
> You can reach the person managing the list at
>        r-sig-ecology-owner@...
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of R-sig-ecology digest..."
>
>
> Today's Topics:
>
>   1. pairwise comparisons in community structure analysis      data

Re: pairwise comparisons in community structure analysis data

On Mon, 2009-08-10 at 16:43 -0400, Penelope_Pooler@... wrote:
> I have a question that I'm not sure has a right answer, but I would
> appreciate any and all opinions, especially if you know of any citations to
> back them up.
> 
> In the past, when dealing with univariate data, I have always been a
> proponent of using Fisher's Protected LSD for pairwise comparisons because
> it is the most powerful procedure, and the accuracy of its experimentwise
> error rate is based on a sound argument that was subsequently proven to be
> true with simulations in Carmer and Swanson (1973 ).  I included the full
> reference below.
> 
> At the moment, I'm working with community structure data (multivariate
> nonparametric) at multiple sites and there is an need to determine which
> sites are different from which other sites.  My colleague has used a
> Bonferroni correction  for this type of question the past, but I tend to
> think that is most likely too conservative.  I'm interested to know if any
> of you have dealt with a similar problem and/or if you know if anyone has
> done any work on comparing pariwise comparisons procedures for multivariate
> nonparametric data.

You don't say how you are comparing sites, i.e. to what is the
Bonferroni correction applied? The p-value from ...?

You are also a bit vague about the data structures/layout. If your data
represent a set of sites with multiple samples taken at each of these
sites, then you might want to look at adonis() and betadisper(), both
functions in the vegan package.

adonis() partitions the dissimilarities between sites on the basis of

Re: pairwise comparisons in community structure

Penelope,

Following up on Rodney's email, it may be useful to ask whether pairwise 
comparisons are necessary.  Depending on your question and what you want 
to gain from the research, pairwise comparisons may not be the best 
approach.  If you are interested in the general patterns and potential 
processes across your study system, NMS would be a useful approach to 
determine how sites differ and what measured "environmental" variables 
are correlated with those differences.  You may wish to also use 
Multi-response Permutation Procedures (MRPP), either independently of, 
or in connection with, the results from the NMS to test for group 
differences. 

Good luck,
Chris

Christopher D. Stallings, Ph.D.
Postdoctoral Associate
Florida State University Coastal & Marine Lab
St. Teresa, FL 32358, USA

phone: 850.697.4103
fax: 850.697.3822
web: http://www.marinelab.fsu.edu/faculty/stallings.aspx

Rodney White wrote:
> Non-metric Multidimensional Scaling (NMS or NMDS) is one procedure used to
> analyze differences in sites regarding vegetation structure. It uses a
> dissimilarity matrix to orient sites in species space. It is available as a
> base package in R as isoMDS, or in the package Vegan as metaMDS. It is not a

Re: pairwise comparisons in community structure

Penelope,

One of the nice things about the Bonferroni correction is that it is  
simple and straightforward to implement.  However, many do blanch at  
the rapid loss of power.  One simple alternative is the False  
Discovery rate (Benjamini and Hochberg 1995 J. R. Stat Soc. B) and the  
Sharpened FDR (Benjamini and Hochberg 2000 Journal of Educational and  
Behavioral Statistics, Verhoeven et al Oikos 2005).  The change in p- 
value is linear rather than multiplicative.  R actually implements the  
FDR in a few cases, and I have some code for the sharpened FDR if you  
need it.

-Jarrett

----------------------------------------
Jarrett Byrnes
Postdoctoral Associate, Santa Barbara Coastal LTER
Marine Science Institute
University of California Santa Barbara
Santa Barbara, CA 93106-6150
http://www.lifesci.ucsb.edu/eemb/labs/cardinale/people/byrnes/index.html
>>>
>>>
>>>
>>> I have a question that I'm not sure has a right answer, but I would
>>> appreciate any and all opinions, especially if you know of any  
>>> citations to
>>> back them up.
>>>
>>> In the past, when dealing with univariate data, I have always been a

Re: pairwise comparisons in community structure

Thanks for all of your great comments and questions.  I think I got what I
needed.  Two people (Elgin Perry and Jarrett Byrnes) recommended a paper
about controlling the FDR (False Discovery rate) by Benjamini and Hochberg
(1995) and Jarrett also mentioned newer paper by the same authors that
updates their work with another procedure called an adaptive FDR.  These
procedures will be ideal for what I am doing.

To answer some of your questions, I am using adonis(), because it is
analogous to what has been used in the past (anosim()) for this project,
but I feel it is better algorithm and has the added benefit of outputing
the results in a format similar to a univariate ANOVA table.  Gavin
mentioned the possibility of also checking out betadisper() in the same
vegan package and I will.

Unfortunately, for this particular research I can't use NMDS because I am
presenting the results to park managers and other non scientists so the
methods as well as the results and how I explain them have to be very
straightforward.  Otherwise, I think that would be a great solution.  As
for why I am doing pairwise comparisons, I agree that they are not always
the best approach, but in this particular case the park managers want
simple straightforwad answers as to whether or not site a differs from site
b differs from site c, etc and what species, characteristics contribute to
those differences.

Thank you again for your help and useful suggestions and comments.

-Penelope

==============================
Penelope S. Pooler

Re: pairwise comparisons in community structure

Penelope,

Check ?p.adjust for the adjustment methods for multiple comparisons. It is
in standard R (stats) so that it is immediately available for use. There are
several methods available, and in my installation those include both FDR and
Hochberg among others:

> p.adjust.methods
[1] "holm"       "hochberg"   "hommel"     "bonferroni" "BH"
[6] "BY"         "fdr"        "none"

Cheers, Jari Oksanen

On 11/08/09 21:03 PM, "Penelope_Pooler@..." <Penelope_Pooler@...>
wrote:

> Thanks for all of your great comments and questions.  I think I got what I
> needed.  Two people (Elgin Perry and Jarrett Byrnes) recommended a paper
> about controlling the FDR (False Discovery rate) by Benjamini and Hochberg
> (1995) and Jarrett also mentioned newer paper by the same authors that
> updates their work with another procedure called an adaptive FDR.  These
> procedures will be ideal for what I am doing.
> 
> To answer some of your questions, I am using adonis(), because it is
> analogous to what has been used in the past (anosim()) for this project,
> but I feel it is better algorithm and has the added benefit of outputing
> the results in a format similar to a univariate ANOVA table.  Gavin
> mentioned the possibility of also checking out betadisper() in the same
> vegan package and I will.
> 

Re: pairwise comparisons in community structure

Hi Penelope,

You're right that Bonferonni can be overly conservative for multiple
testing. Several other options exist, see ?p.adjust for some examples.

As far as I know, the same methods apply for multivariate data, see
Legendre & Legendre (1998) Numerical Ecology, Chapter 1, Box. 1.3 on
"Multiple testing" (p. 18). He recommends Holm's or Hochberg's
procedures.

Cheers

Etienne

> Date: Mon, 10 Aug 2009 16:43:27 -0400
> From: Penelope_Pooler@...
> Subject: [R-sig-eco] pairwise comparisons in community structure
>        analysis        data
> To: r-sig-ecology@...
>
> I have a question that I'm not sure has a right answer, but I would
> appreciate any and all opinions, especially if you know of any citations to
> back them up.
>
> In the past, when dealing with univariate data, I have always been a
> proponent of using Fisher's Protected LSD for pairwise comparisons because
> it is the most powerful procedure, and the accuracy of its experimentwise
> error rate is based on a sound argument that was subsequently proven to be
> true with simulations in Carmer and Swanson (1973 ).  I included the full
> reference below.