[R] Extracting correlation in a nlme model
Kenneth Roy Cabrera Torres
krcabrer at une.net.co
Wed May 20 06:31:45 CEST 2009
Hi R users:
Is there a function to obtain the correlation within groups
from this very simple lme model?
> modeloMx1
Linear mixed-effects model fit by REML
Data: barrag
Log-restricted-likelihood: -70.92739
Fixed: fza_tension ~ 1
(Intercept)
90.86667
Random effects:
Formula: ~1 | molde
(Intercept) Residual
StdDev: 2.610052 2.412176
Number of Observations: 30
Number of Groups: 3
I want to obtain \rho = \sigma_b^2 / (\sigma_b^2 + \sigma^2)
I know that I obtain \sigma_b^2 and \sigma^2 with
> VarCorr(modeloMx1)
molde = pdLogChol(1)
Variance StdDev
(Intercept) 6.812374 2.610052
Residual 5.818593 2.412176
But, I want to know if I can obtain
\rho = 6.8123/(6.8123 + 5.8185) = 0.53934 straightforward.
Thank you for you help.
Kenneth
More information about the R-help
mailing list