[R-sig-eco] multiple regression (ONKELINX, Thierry)

[ONKELINX, Thierry]

Dear Alain,

Thank you for your comments.

> Interesting thought. Maybe correct. But there are a few 
> things to think
> about:
> 1. You have to assume that sampling was such that all species 
> out there have ended up in the data. Formulated 
> differently...you need to know the N_i (maximum number of 
> possible species) in the B(p_i,N_i).

I assume that one knowns all species that could occur in the region.
That is not the same as all samples species. 

> 2. Not sure how the variance structure would work out; 
> Poisson or NB versus Binomial. I guess it may work out ok.
> 3.  B(p_i,N_i) means N_i independent (!) trials, each (!) 
> with probability p_i of success. But what happens if these 
> trials (=species) are not independent?

Agreed that this might be problematic. But isn't this also a problem
with a Poisson model? Since you assume that the number of species only
depends on the covariantes and not on the presence of some species?

> 
> Anyway....it is not my problem....but you would need to think 
> about these things.

I was replying to a question. So it is is my problem neither ;-) Or are
least not at present, as I might get similar datasets in the future.

> 
> Alain
> 
> 

Best regards,

Thierry

Druk dit bericht a.u.b. niet onnodig af.
Please do not print this message unnecessarily.

Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer 
en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is
door een geldig ondertekend document. The views expressed in  this message 
and any annex are purely those of the writer and may not be regarded as stating 
an official position of INBO, as long as the message is not confirmed by a duly 
signed document.