[R-sig-ME] A question about CLMM in R

Georgina Southon g.southon at sheffield.ac.uk
Thu Dec 3 18:01:05 CET 2015


Dear CLMM experts,

I have been using the ordinal package in R to model ordinal response data with CLMM, with the inclusion of categorical predictor variables and interaction terms. The model returns perfectly coherent results thankfully (a truncated example attached), however it does not provide an overall P value for each interaction term. Is this possible to obtain when modelling with clmm? 

Call:				
clmm2(location = preference ~ treatment + siteuse +treatment*siteuse,random+sample,Hess=TRUE
 

 random = sample, 
    data = plops, Hess = TRUE)		
Random effects:			
            Var  Std.Dev			
sample 1.660947 1.288777		
Location coefficients:			
                      Estimate Std. Error z value Pr(>|z|)  	
treatmentB             0.0736   1.4480     0.0508 0.95947987
treatmentC             0.8667   1.0871     0.7972 0.42531178
treatmentD             0.1548   1.1883     0.1303 0.89636709
treatmentE            -0.9512   1.3482    -0.7056 0.48046420
treatmentF             4.0151   1.2521     3.2066 0.00134313
treatmentG            -0.0489   1.0773    -0.0454 0.96376758
treatmentH             1.1422   1.2055     0.9474 0.34342605
treatmentI             0.2228   1.0840     0.2055 0.83716432
siteuse               -0.0004   0.0013    -0.3159 0.75211206




treatmentB:siteuse     0.0012   0.0019     0.6217 0.53413206
treatmentC:siteuse     0.0011   0.0015     0.7460 0.45565644
treatmentD:siteuse     0.0009   0.0016     0.5841 0.55913734
treatmentE:siteuse     0.0004   0.0016     0.2355 0.81378524
treatmentF:siteuse     0.0002   0.0016     0.1011 0.91943399
treatmentG:siteuse    -0.0006   0.0015    -0.4202 0.67432023
treatmentH:siteuse     0.0000   0.0018    -0.0278 0.97782476
treatmentI:siteuse    -0.0019   0.0016    -1.1369 0.25559690



Any insights would be most gratefully received.

Thank you,


Dr Georgina Southon
Post-doctoral Researcher
Department of Landscape
The University of Sheffield







	[[alternative HTML version deleted]]



More information about the R-sig-mixed-models mailing list