dear s-plus-users,
i have a lme-model (longitudinal data, 2-4 observations per subject
(CODE) with:
IN.THER = time in psychotherapy in days (continuous);
SCSW.010 = outcome (continuous);
THER = type of psychotherapy (3-level-factor);
Number of Observations: 1138;
Number of Groups: 400.
model.lme <- lme (data = data.gr,
random = ~IN.THER | CODE, fixed = SCSW.010 ~ IN.THER * THER)
resulting in the follwing fixed effects:
> fixef(model.lme)
(Intercept) IN.THER THER1 THER2
0.9847708 -0.0007369537 0.05211695 -0.0108868
IN.THERTHER1 IN.THERTHER2
-0.000227367 -0.00005134491
now i would like to plot the mean fixed effects which should result in
three pretty much overlapping "slowly" descending lines (one for each
factor level; time in pt on the x-axis and the mean fixed-effect values
on the y-axis), i.e.:
for 1st type of psychotherapy: 0.9740448 + 0.05211695 + IN.THER *
-0.0007369537 + IN.THER * -0.000227367
for 2nd type of psychotherapy: 0.9740448 + (-0.0108868) + IN.THER *
-0.0007369537 + IN.THER * -0.00005134491
for 3rd type of psychotherapy: 0.9740448 + IN.THER * -0.0007369537
i checked all lme-functions i could think of (predict, augPred, fitted,
etc.) but cannot find a suitable solution. any help would be highly
appreciated.
thanks a lot
bernd
puschner.vcf
Description: Visitenkarte für Bernd Puschner
|