R help - Jan 2008 - 95% confidence and prediction intervals for linear mixed models

Hi R-users,
>From the last week I've been working fitting a linear mixed model with
random intercept and fixed shape (model4) for a data set with 37 individuals
measured over time, using lme package. Results are at the end of this
message. The outcome is score and the covariate is age.

My question is: is possible (and how) to estimate both 95% confidence and
prediction intervals for the mixed model as in linear regression models?
I've tried this using
predict(modelc,interval="confidence",level=0.95) and
predict(modelc,interval="prediction",level=0.95) but the result is
always
the same given by fitted(model4).

Thanks in advance,

Jorge Velez


# Results for the lme model

Linear mixed-effects model fit by maximum likelihood
 Data: NPC.n
       AIC      BIC    logLik
  805.3955 822.3171 -396.6977

Random effects:
 Formula: ~1 + age | subject
 Structure: General positive-definite, Log-Cholesky parametrization
            StdDev     Corr
(Intercept) 13.1125118 (Intr)
age          0.7672007 -0.087
Residual     2.5729797

Fixed effects: total ~ age
                 Value Std.Error DF   t-value p-value
(Intercept) -10.658901  2.731439 86 -3.902303   2e-04
age           1.872654  0.178246 86 10.506009   0e+00
 Correlation:
    (Intr)
age -0.421

Standardized Within-Group Residuals:
        Min          Q1         Med          Q3         Max
-2.04501756 -0.42833252 -0.01521936  0.30176415  2.16862549

Number of Observations: 124
Number of Groups: 37

	[[alternative HTML version deleted]]