Factors in partialRDA

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Factors in partialRDA

Andrew Halford
Hi Listers,

I am running a partial RDA analysis and I have an environmental dataset
which includes a non-ordered factor representing reef type.

I want to test for a significant linear trend in the environmental dataset
as part of the process of testing for spatial autocorrelation.

e.g. lineartest <- anova(rda(env,latlong))

I have coded this Factor variable up as a group of dummy variables equal to
the number of different reef types to run the test but I am unsure if I
need to drop one of the dummy variables before running the linear test?

I will revert back to a single Factor variable when I run the RDA as told
in the vegan documentation?

My reading tells me that the call to RDA in Vegan will automatically deal
with creating dummy variables from a Factor variable presented in the model
but this is not the case

cheers

Andy


Andrew Halford Ph.D
Research Scientist (Kimberley Marine Parks)
Dept. Parks and Wildlife
Western Australia

Ph: +61 8 9219 9795
Mobile: +61 (0) 468 419 473

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
Reply | Threaded
Open this post in threaded view
|

Re: Factors in partialRDA

Jari Oksanen
Andy,

That’s correct: if you have coded your variable as a single factor, R (and hence vegan) will automatically create dummy variables. If you have originally coded your variable as a set of dummy variables, and you have done that correctly, you will get the same result. If you do  not drop one of the dummy variables, R will do it for you and tell that some of your original variables were aliased. This will not influence the results, though.

In conclusion: everything goes, but it is best to use the R way and define your factor as a single factor.

Cheers, Jari Oksanen

> On 7 Feb 2017, at 10:48 am, Andrew Halford <[hidden email]> wrote:
>
> Hi Listers,
>
> I am running a partial RDA analysis and I have an environmental dataset
> which includes a non-ordered factor representing reef type.
>
> I want to test for a significant linear trend in the environmental dataset
> as part of the process of testing for spatial autocorrelation.
>
> e.g. lineartest <- anova(rda(env,latlong))
>
> I have coded this Factor variable up as a group of dummy variables equal to
> the number of different reef types to run the test but I am unsure if I
> need to drop one of the dummy variables before running the linear test?
>
> I will revert back to a single Factor variable when I run the RDA as told
> in the vegan documentation?
>
> My reading tells me that the call to RDA in Vegan will automatically deal
> with creating dummy variables from a Factor variable presented in the model
> but this is not the case
>
> cheers
>
> Andy
>
>
> Andrew Halford Ph.D
> Research Scientist (Kimberley Marine Parks)
> Dept. Parks and Wildlife
> Western Australia
>
> Ph: +61 8 9219 9795
> Mobile: +61 (0) 468 419 473
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-ecology mailing list
> [hidden email]
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology

_______________________________________________
R-sig-ecology mailing list
[hidden email]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology