Hi, everyone, I have a question on using lme on a mixed effects model with nested error structure. After applying lme to the data, and put the outcome in, say TR.lme. I can extract the fixed effects by TR.lme$coef$fixed. However, when I use TR.lme$coef$random.effects, it does not give the variance components that I need, but a vector of values at each nested level. What I want are the estimated variance components that show up with summary(TR.lme), but I can not extract them from TR.lme. Could you please give any suggestion? I have tried ranef( ), but it seems not to be the right function. Thank you very much! Zonghui
Zonghui Hu <zhu at stat.tamu.edu> writes:> I have a question on using lme on a mixed effects model with nested > error structure. After applying lme to the data, and put the outcome > in, say TR.lme. I can extract the fixed effects by > TR.lme$coef$fixed. However, when I use TR.lme$coef$random.effects, it > does not give the variance components that I need, but a vector of > values at each nested level. What I want are the estimated variance > components that show up with summary(TR.lme), but I can not extract > them from TR.lme.It is better to use the extractor functions fixef (or fixed.effects) and ranef (of random.effects) to extract the estimated coefficients. You should not depend on a particular internal representation of the object returned from a model-fitting function. For example, objects returned by the version of lme in the lme4 package uses a different internal representation. As indicated in the output from help(package = "nlme") the function VarCorr returns the variance and correlation components. -- Douglas Bates bates at stat.wisc.edu Statistics Department 608/262-2598 University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/