R help - Aug 2009 - Autocorrelation and t-tests

Hi,

I have two sets of data for a given set of (non-lattice) locations. I would
like to know whether the two are significantly different. This would be
simple enough if it wasn't for the fact that the data is spatially
autocorrelated. I have come across several possible solutions (including
Cliff & Ord which however appears to be for gridded data), or using gls.
However, they don't quite fit the bill (I think). Ideally it would simply be
a modified t-test which somehow 'takes care' of the spatial
autocorrelation,
and it would be perfect (perhaps asking too much!) if it were a
function/package in R. I could also take the differences between the two
sets and test whether they are significantly different to 0 or not. I get
the impression that this issue must have come up again and again, so I am
hoping someone might know of an appropriate solution! Thank you very much in
advance for any help!

Bernardo

	[[alternative HTML version deleted]]

Hi Bernardo,

I suggest you give a look at:
Dale MRT & Fortin MJ, 2009. Spatial Autocorrelation and Statistical Tests:
Some Solutions.
Journal of Agricultural, Biological and Environmental Statistics,
14(2):188-206.

Cheers

milton

On Tue, Aug 25, 2009 at 1:08 PM, B Garcia Carreras <
bernardo.garcia-carreras08@imperial.ac.uk> wrote:
> Hi,
>
> I have two sets of data for a given set of (non-lattice) locations. I would
> like to know whether the two are significantly different. This would be
> simple enough if it wasn't for the fact that the data is spatially
> autocorrelated. I have come across several possible solutions (including
> Cliff & Ord which however appears to be for gridded data), or using
gls.
> However, they don't quite fit the bill (I think). Ideally it would
simply
> be
> a modified t-test which somehow 'takes care' of the spatial
> autocorrelation,
> and it would be perfect (perhaps asking too much!) if it were a
> function/package in R. I could also take the differences between the two
> sets and test whether they are significantly different to 0 or not. I get
> the impression that this issue must have come up again and again, so I am
> hoping someone might know of an appropriate solution! Thank you very much
> in
> advance for any help!
>
> Bernardo
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
>
http://www.R-project.org/posting-guide.html<http://www.r-project.org/posting-guide.html>
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]