[R-sig-eco] hierarchical partitioning in generalized linear mixed models

Carsten Dormann carsten.dormann at ufz.de
Fri Oct 29 08:12:46 CEST 2010

Dear Alessandro,

at the risk of inciting furious comments, I like to express my doubts 
about the usefulness of hierarchical partitioning as a way to quantify 
the contribution of separate variables. My concern has two facets: one 
is that the method is ad hoc, proposed rather tentatively by Chevan & 
Sutherland, more as an option one could consider for simple cases than 
as the silver bullet it is currently often sold. For example, it is 
restricted to linear combinations and hence doesn't accommodate 
polynomial terms for a variable.
Secondly, my experience is that it doesn't allow me to really 
differentiate between joined effects caused by the collinearity of 
variables and those caused by interactions. The first I would regard as 
"problematic" (in the sense that variance inflation combined with model 
selection can cause serious bias) while the second I find ecologically 
highly relevant and interesting.

(BTW, no, I don't know of a package to carry hier.part out on mixed 
models; it would require a leaps-like implementation for mixed models, 
which I also haven't seen. )



On 28.10.10 20:05, Alessandro Gherlenda wrote:
> Dear colleagues,
> I would to use hierarchical partitioning methods to evaluate variance
> contributions of different predictors as proposed by Chevan and
> Sutherland (1991) and Mac Nally (2000&  2002), within a generalized
> linear mixed modelling framework. The standard package in R for
> hierarchical partioning is hier.part.  However, the hier.part  package
> only works within a generalized linear modelling framework.  I was
> wondering wether there is any function out there that might perform
> hierarchical partioning for mixed models.
> thanks in advance
> Best wishes,
> Alessandro
> Chevan, A.&  Sutherland, M. (1991) Hierarchical partitioning. American
> Statistician, 45, 90–96.
> Mac Nally, R. (2000) Regression and model-building in conservation
> biology, biogeography and ecology: the distinction between–and
> reconciliation of–‘predictive’and ‘explanatory’models. Biodiversity
> and Conservation, 9, 655–671.
> Mac Nally, R. (2002) Multiple regression and inference in ecology and
> conservation biology: further comments on identifying important
> predictor variables. Biodiversity and Conservation, 11, 1397–1401.
> --
> Alessandro Gherlenda
> Ecological Consultant-Bologna Italy
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology

Dr. Carsten F. Dormann
Department of Computational Landscape Ecology
Helmholtz Centre for Environmental Research-UFZ	
(Department Landschaftsökologie)
(Helmholtz Zentrum für Umweltforschung - UFZ)
Permoserstr. 15	
04318 Leipzig				

Tel: ++49(0)341 2351946
Fax: ++49(0)341 2351939
Email: carsten.dormann at ufz.de
internet: http://www.ufz.de/index.php?de=4205

Registered Office/Sitz der Gesellschaft: Leipzig
Commercial Register Number/Registergericht: Amtsgericht Leipzig, 
Handelsregister Nr. B 4703
Chairman of the Supervisory Board/Vorsitzender des Aufsichtsrats: MinR 
Wilfried Kraus
Scientific Managing Director/Wissenschaftlicher Geschäftsführer: Prof. 
Dr. Georg Teutsch
Administrative Managing Director/Administrativer Geschäftsführer: Dr. 
Andreas Schmidt

More information about the R-sig-ecology mailing list