Graham M Smith
>
> This is probably a very broad question so I will be
> grateful to anyone who has the time to answer. I also
> apologise for its length
>
> A lot of ecological monitoring is undertaken by consultants
> with limited statistical abilities, even though it often
> involves a fairly sophisticated design. eg 10 transects
> with 5 sampling stations per transect, 10 samples collected
> from each Station, and samples collected 4 times a year for
> 5 years. Some of the transects might be impacted upon by
> development and you are trying to compare those with the
> ones not impacted upon. Or all transects might be impacted
> upon and it is a before and after type study.
>
> An unsophisticated approach to this, that I have often
> seen done, is to aggregate the results in a manner that
> allows an ANOVA to be done, usually Kruskal Wallis, or if a
> statistician is involved - to use a mixed effects model
> usually in SPlus.
>
> I have just started to read Lohr "Sampling: Design and
> Analysis" and discovered the facility in Stata to analyse
> survey sample data, which would seem to be an extremely
> useful tool for analysing environmental or ecological data.
> Yet I know of no one in the environmental/ecological field
> that uses Stata, or indeed programs like Sudaan. A quick
> poll of the professional statisticians I know (n=5)that
> work in ecology came up with a "never heard of Sudaan"
> "don't know anything about Stata" "can't see it doing
> anything that proper programming in S-Plus can't do"
>
> However, I am personally still interested in exploring this
> further. I am not a statistician, but I am interested in
> the design and analysis of cost effective and robust
> ecological/environmental monitoring programs. I have a
> reasonable understanding of statistics but it falls short
> of writing S-Plus programs and mixed effects modelling.
>
> Is anyone here using Stata in an ecological or
> environmental field (eg Pollution studies) and do the tools
> in Stata for analysing Survey sampling data offer something
> valuable to the analysis of ecological data. Not everyone
> has the skills or time to write an SPlus program - even if
> it can be done.
>
> Any way, I would be grateful if anyone can provide some
> comments on this, and improve my understanding of Stata and
> the use of its Survey Sampling routines.
There are Stata users with primary interests in environmental
applications. I am one. At the same time, such users appear
to make up a fairly small proportion of the Stata community.
I don't recollect anybody with a specific interest in
ecological monitoring. In contrast, S-Plus does have a
substantial presence in (its own definition of)
environmental statistics.
More specifically, I can't advise in detail on the usefulness of
Stata's -svy- commands for looking at ecological monitoring.
I would want to worry about their applicability to
situations with strong spatial and temporal element. However,
from what you say, these elements are often ignored in practice.
Stata is programmable; S-Plus is programmable.
Those who have programmed in both -- and that doesn't
include me -- might comment on the differences.
But if the issue is using Stata's -svy- commands
versus replicating them by programming in S-Plus,
then my guess is that the latter would be a very
great deal of work.
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
