R help - Jul 2006 - seeking robust test for equality of variances w/ observation weights

Hello R community,

I am looking for a robust test for equality of variances that can take
observation weights.
I realize I can do the F-test with weighted variances, but I've read that
this test is not very robust.

So I thought about maybe adding a "weights" argument to John Fox's
code for
the Levene Test (in the "car" library, "levene.test"),
substituting his "median" function for a " weighted.mean"
and also including
the observation weights in his "lm" run--
after all, Levene's original test used the mean, not the median.

I asked John about it and he doesn't know what the properties of this
weighted Levene test would be.
Does anyone have any thoughts or suggestions, or know of a robust weighted
hypothesis test for equality of variances?

Thank you in advance for any advice you can provide,

Alexis Diamond
adiamond@fas.harvard.edu

	[[alternative HTML version deleted]]

I am using the multcomp package for doing multiple comparisons. Since
the data I am handling is huge the number of comparisons are also large.
I am interested in:
1>  Breaking down my plots to get rid of the clutter that happens when
plotting the entire data set. How do I pass only part of the data to the
plot function ?
 
fungus.cirec<-simint(Fungus.yield~Habitat,
data=fungus,conf.level=0.95,type =c("Tukey"))
plot(fungus.cirec)  #This plots the entire data. I want to plot part of
the data only
 
2>I am also interested in getting rid of the field name associated with
each categorical variable. 
Here is what the part of the data looks like
 
Habitat Fungus.yield
Birch 20.83829053
Birch 22.9718181
Birch 22.28216829
Birch 24.23136797
Birch 22.32147961
Birch 20.30783598
Oak 27.24047258
Oak 29.7730014
Oak 30.12608508
Oak 25.76088669
Oak 30.14750974
Hornbeam 17.05307949
Hornbeam 15.32805111
Hornbeam 18.26920177
Hornbeam 21.30987049
Hornbeam 21.7173223
It plots labels as HabitatBirch-HabitatOak for example.  How do I get
rid of the field name Habitat in the plot? 
 
3> How do I tell the method to mark the significant comparisons? (i.e
those that do not intersect the zero line).
 
Thanks ../Murli

You can always bootstrap any robust spread measure (e.g. mad or higher
efficiency versions from robustbase or other packages).

-- Bert Gunter
Genentech Non-Clinical Statistics
South San Francisco, CA
 
"The business of the statistician is to catalyze the scientific learning
process."  - George E. P. Box
 
 
> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch 
> [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Alexis Diamond
> Sent: Friday, July 21, 2006 9:56 AM
> To: r-help at stat.math.ethz.ch
> Subject: [R] seeking robust test for equality of variances w/ 
> observationweights
> 
> Hello R community,
> 
> I am looking for a robust test for equality of variances that can take
> observation weights.
> I realize I can do the F-test with weighted variances, but 
> I've read that
> this test is not very robust.
> 
> So I thought about maybe adding a "weights" argument to John 
> Fox's code for
> the Levene Test (in the "car" library, "levene.test"),
> substituting his "median" function for a "
weighted.mean" and
> also including
> the observation weights in his "lm" run--
> after all, Levene's original test used the mean, not the median.
> 
> I asked John about it and he doesn't know what the properties of this
> weighted Levene test would be.
> Does anyone have any thoughts or suggestions, or know of a 
> robust weighted
> hypothesis test for equality of variances?
> 
> Thank you in advance for any advice you can provide,
> 
> Alexis Diamond
> adiamond at fas.harvard.edu
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>