[R-sig-ME] Lme doubt...

[Mike Dunbar]

Dear Achaz

Two suggestions:
If you include the extra lake level predictors and they are significant then you will reduce the magnitude of the lake random effect, and hence you may improve the estimation of the fish effect which is also at the lake level. I'm guessing that reducing the lake random effect won't affect the testing of the month and year factors. 
Secondly, if you didn't have any lake-level predictors that you were interested in a priori then it simply depends on whether you are interested in the magitude of these additional effects on the zooplankton, if you are then include them, if not then their effects will be mopped up into the random effect.

regards

Mike Dunbar



>>> Achaz von Hardenberg <fauna at pngp.it> 23/01/2008 12:22 >>>
Dear all,

I am analyzing data regarding the effects of introduced fish on the  
biodiversity of 12 alpine lakes. I have 6 lakes with fish and 6  
without and my dependent variables are various repeated density  
measurements of zooplankton taxa.
As we repeated the biodiversity measurements twice a year for two  
years (2006 and 2007) I am analyzing this data with  the lme function  
of the nlme package with lake identity as random grouping factor  
(random = ~1|as.factor(Cod)), and the presence/absence of fish as  
fixed effect. I included also the year and month of sampling as fixed  
effects.
The full model specification is the following:

zoopcrost.lme<-lme(Logcrostacei~as.factor(year)+as.factor(month) 
+as.factor(fish),method="ML", random=~1|as.factor(Cod),  
data=zooplankton, na.action=na.omit)

Now, I have a doubt....
The lakes differ among each other for some physical characteristics  
such as altitude, maximum depth etc....Obviously these parameters do  
not vary within each sampling as they are specific for each lake  
(they do not vary from one year to the next or from one month to the  
next) but they may have an influence on the densities of zooplankton  
taxa, regardless of the presence/absence of introduced fish. Should I  
nonetheless consider these variables among the fixed effects? Or  
should the fact that I impose the lake identity as grouping factor  
take care of the fact that there are individual differences among the  
lakes?

Sorry for possibly a naive question and thank you for your help!

Dr. Achaz von Hardenberg
------------------------------------------------------------------------ 
--------------------------------
Centro Studi Fauna Alpina - Alpine Wildlife Research Centre
Servizio Sanitario e della Ricerca Scientifica
Parco Nazionale Gran Paradiso, Degioz, 11, 11010-Valsavarenche (Ao),  
Italy
-----------------------------------------------------------------





	[[alternative HTML version deleted]]

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models


-- 
This message (and any attachments) is for the recipient ...{{dropped:6}}