0

I would like to consider my nested study design in the HOF function of the eHOF package. We sampled in points which where nested in sample blocks and those where nested in villages (random).

So far I used GLMM to describe the effect of habitat type(Habitat, explanatory, fixed) on species richness (SpR, response variable, fixed). Therefore I used lme (of nlme-package) where I could incorporate the nested study design in the function: 

model1<-lme(SpR~Habitat,random=~1|Village/Block,data=birds) # full model

Now I would like to describe the effect on SpR along a gradient. My HOF function looks like this: 

hof.blocks<-HOF(SpR,Indexc,M=max(SpR),bootstrap=100,
                 family=poisson)

Is there a way to incorporate the nested study design in HOF?

Also I was not able to obtain P-values, nor F- or R2-values. Did I overlook something?

Package: eHOF Version: 1.4 R version 3.0.2

Lula
  • 3
  • 3

1 Answers1

0

No, not at the moment. You could try to fit the nonlinear models by hand using nlme(), perhaps looking in the eHOF package code for hints on how to specify the formula etc, but that package uses a non-linear optimiser directly, and not nlme() so you'll need to do the translation work to get a model formula that nlme can use and work out other details.

Gavin Simpson
  • 170,508
  • 25
  • 396
  • 453
  • Thank you for this idea! But unfortunately this approach would be very time-consuming to do with all my different species to test. I was looking for other possibilities to incorporate the nested design. The author of the eHOF package ran afterwards a GLM with and without including a nested study design and compared the AIC for the models. Nevertheless, how to obtain a p- and the F-values trough the eHOF package? – Lula Sep 22 '14 at 11:14