barbara costa
2012-Sep-14 09:19 UTC
[R] linear mixed-effects models with two random variables?
Dear R users, Does anyone knows how to run a glmm with one fixed factor and 2 random numeric variables (indices)? Is there any way to force in the model a separate interaction of those random variables with the fixed one? I hope you can help me. #eg. Reserve <- rep(c("In","Out"), 100) fReserve <- factor(Reserve) DivBoulders <- rep (c(1.23,2.4,1.26,1.78,1.97,1.35,1.23,2.4,1.26,1.78), 20) Roughness <- rep(c(3.45,2.56,1.32,5.67,3.73,3.57,2.66,1.52,7.67,2.73),20) Biomass <- rep(c(8,5.3,3.5,12,25.4,10.1,9.8,2.4,5.6,5.3),20) myData <- data.frame (fReserve ,DivBoulders ,Roughness ,Biomass ) #glm glm1 <- glm (Biomass ~ fReserve * Roughness + fReserve * DivBoulders , family= Gamma, data= myData) #glmm: library (nlme) lme1 <- lme (Biomass ~ fReserve , random= ~1 + fReserve | Roughness + DivBoulders , data= myData ) # random intercept and slope #Error in getGroups.data.frame(dataMix, groups) : # Invalid formula for groups # if I only use one random variable I have: #Error in chol.default((value + t(value))/2) : #the leading minor of order 2 is not positive definite lme2 <- lme( Biomass ~ fReserve , random = ~1 | Roughness + DivBoulders ,data=myData) #random intercept #Error in getGroups.data.frame(dataMix, groups) : # Invalid formula for groups # if I only use one random variable my result is fine! lme3 <- lme (Biomass ~ fReserve , random= ~ Roughness + DivBoulders | fReserve , data= myData ) # from help (lme) summary (lme3) # I have a result. Is the model correct for what I want? # BUT my fixed effect (reserve) does not have a p-value due to zero degrees of freedom. However in glm1 it has. Can you help me please? Thanks a lot in advance, Barbara [[alternative HTML version deleted]]
Bert Gunter
2012-Sep-14 16:11 UTC
[R] linear mixed-effects models with two random variables?
Post on R-sig-mixed-models rather than here for better responses. -- Bert On Fri, Sep 14, 2012 at 2:19 AM, barbara costa <rbarbarahc at gmail.com> wrote:> Dear R users, > Does anyone knows how to run a glmm with one fixed factor and 2 random > numeric variables (indices)? Is there any way to force in the model a > separate interaction of those random variables with the fixed one? > I hope you can help me. > > #eg. > Reserve <- rep(c("In","Out"), 100) > fReserve <- factor(Reserve) > DivBoulders <- rep (c(1.23,2.4,1.26,1.78,1.97,1.35,1.23,2.4,1.26,1.78), 20) > Roughness <- rep(c(3.45,2.56,1.32,5.67,3.73,3.57,2.66,1.52,7.67,2.73),20) > Biomass <- rep(c(8,5.3,3.5,12,25.4,10.1,9.8,2.4,5.6,5.3),20) > > myData <- data.frame (fReserve ,DivBoulders ,Roughness ,Biomass ) > > #glm > glm1 <- glm (Biomass ~ fReserve * Roughness + fReserve * DivBoulders , > family= Gamma, data= myData) > > #glmm: > library (nlme) > > lme1 <- lme (Biomass ~ fReserve , random= ~1 + fReserve | Roughness + > DivBoulders , data= myData ) # random intercept and slope > #Error in getGroups.data.frame(dataMix, groups) : > # Invalid formula for groups > # if I only use one random variable I have: > #Error in chol.default((value + t(value))/2) : > #the leading minor of order 2 is not positive definite > > > lme2 <- lme( Biomass ~ fReserve , random = ~1 | Roughness + > DivBoulders ,data=myData) #random > intercept > #Error in getGroups.data.frame(dataMix, groups) : > # Invalid formula for groups > # if I only use one random variable my result is fine! > > > lme3 <- lme (Biomass ~ fReserve , random= ~ Roughness + DivBoulders > | fReserve , data= myData ) # from help (lme) > summary (lme3) > # I have a result. Is the model correct for what I want? > # BUT my fixed effect (reserve) does not have a p-value due to zero degrees > of freedom. However in glm1 it has. > > Can you help me please? > > Thanks a lot in advance, > Barbara > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list > stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.-- Bert Gunter Genentech Nonclinical Biostatistics Internal Contact Info: Phone: 467-7374 Website: pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm