[R-sig-Geo] Fwd: R-sig-Geo Digest, Vol 153, Issue 4

[Roger Bivand]

On Wed, 4 May 2016, John Lewis wrote:

> Hi,
> I think there is a more fundamental problem that you should consider. The
> use of p values and set significant levels are seriously being questioned
> as good statistical modelling practises. You might want to look at the
> recent policy statement by the American Statistical Association on this
> topic. There is even a strong movement to convince editors not to review
> papers which base the results on the above methods or at least advise the
> use of different procedures.
> If you are interested in reading this policy statement I would be happy to
> send a pdf copy.

I believe that this link (to a manuscript covering the background and 
including the statement) is open:

http://amstat.tandfonline.com/doi/pdf/10.1080/00031305.2016.1154108

Roger

> Cheers,
> John Lewis
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Tue, 3 May 2016 23:11:29 +0000 (UTC)
> From: "Thiago V. dos Santos" <thi_veloso at yahoo.com.br>
> To: R-sig-geo Mailing List <r-sig-geo at r-project.org>
> Subject: [R-sig-Geo] Mask a map using statistical significance
> Message-ID:
>        <1350390896.5790739.1462317089174.JavaMail.yahoo at mail.yahoo.com>
> Content-Type: text/plain; charset=UTF-8
>
> Dear all,
>
> In climate studies, it is a common practice to perform some statistical
> test between two fields (maps), and then plot the resulting map using a
> significance mask. This masking is usually done by adding some kind of
> pattern (shading, crosshatching etc) on top of the actual color palette.
>
> Examples can be seen here in this image https://www3.epa.gov/climatechange
> /images/science/GlobalPrecipMap-large.png and in the left images of this
> panel:
> http://www.nature.com/nclimate/journal/vaop/ncurrent
> /images_article/nclimate2996-f3.jpg
> In my case, I ran a statistical test for detecting trend on a time-series
> raster and I now have one raster with the trend for rainfall (in degree C
> per year) and one with the p-values associated to the test.
>
> My data looks roughly like this:
>
> require(raster)
>
> ## scratch raster objects and fill them with some random values
> r.trend <- raster(nrows=50, ncols=50, xmn=-58, xmx=-48, ymn=-33, ymx=-22)
> r.trend[] <- round(runif(50 * 50, min=0, max=3), digits=2)
>
> r.pvalue <- raster(nrows=50, ncols=50, xmn=-58, xmx=-48, ymn=-33, ymx=-22)
> r.pvalue[] <- round(runif(50 * 50, min=0.0001, max=0.1), digits=5)
>
>
> What I would like to do is to plot r.trend, and on top of it plot r.pvalue
> (as a pattern) where r.pvalues < 0.01 (corresponding to a significance
> level of 99%).
>
> Has anyone here ever tried to do a similar plot and can point me to any
> direction?
>
> I usually use rasterVis and ggplot2 to plot maps, but I would be open to
> try some other package and even other SIG software other than R.
>
> Many thanks in advance, -- Thiago V. dos Santos
>
> PhD student
> Land and Atmospheric Science
> University of Minnesota
>
>
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
> ------------------------------
>
> End of R-sig-Geo Digest, Vol 153, Issue 4
> *****************************************
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 91 00
e-mail: Roger.Bivand at nhh.no
http://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
http://depsy.org/person/434412