[R] A covariate correlated with other explanatory variables

Christoph Scherber Christoph.Scherber at agr.uni-goettingen.de
Fri Oct 17 11:51:56 CEST 2008


Dear all,

I have here a difficult problem: We are currently analysing data from a biodiversity experiment, 
where the following models are compared using AICc:

a) y ~ plant.diversity
b) y ~ plant.biomass
c) y ~ plant.diversity + plant.biomass

The problem now is that plant.diversity and plant.biomass are correlated at about r=0.6.

Plant.diversity is a variable that is maintained experimentally at fixed values, while plant.biomass 
is the result of plant.diversity (i.e., higher plant.diversity leads to higher productivity).

Is it allowed to compare these models at all? And is it allowed to have these two correlated terms 
in the same model (assuming Type I sums of squares)?

Or should we ignore all plant.biomass effects, because they are caused by plant.diversity?

I would highly appreciate your help on that problem.

Many thanks and best wishes
Christoph







-- 
Dr. rer.nat. Christoph Scherber
University of Goettingen
DNPW, Agroecology
Waldweg 26
D-37073 Goettingen
Germany

phone +49 (0)551 39 8807
fax   +49 (0)551 39 8806

Homepage http://www.gwdg.de/~cscherb1



More information about the R-help mailing list