R help - Oct 2009 - COVARIANCE MATRIX FOR RANDOM EFFECTS - nlme

R-Help,

I been using nlme to fit a model with 2 random effects. The correlation
matrix I get with the VarCorr command does not seem to have the correct
value for the correlation entry. E.g., below is a VarCorr matrix of random
effects from data that I am working on:

         Variance     StdDev     Corr 

b1       14.386191885 3.79291338 b1   

b3        0.002872538 0.05359606 0.109

Residual  0.052819504 0.22982494 

 

The Corr value 0.109 does not appear to be correct. SAS gives a Corr value
that is the square of the value got in R. I get nonsensical results when I
use the value given by R. Could someone check this out.

 

Charles O. Sabatia


	[[alternative HTML version deleted]]