Hallo!

I have data of the following design: 
N subjects were measured at Baseline (visit 1) and at 3 following time
points (visit 2-4). There is or is not a treatment. Say, the results
of visits 1 to 4 are Y1, Y2, Y3, Y4.

T: Variable Treatment (yes/no)
V: Variable Visit (2, 3, 4 - without baseline)
id: Variable subject identification number
N: number of subjects

Most interesting is the question if there is a difference in treatment
between the results of visit 4 and baseline. (The other time points
are also of interest.)

I am thinking of different possibilities for the analysis (balanced
design):

1) ANOVA with differences to baseline without patient effect
Let YD be the differences to baseline (Y2-Y1, Y3-Y1, Y4-Y1).
The model will be:
YD ~ T + V + T*V + E
Var DF  Meaning
T   1   Treatment
V   2   Visit (2-4)
T*V 2   Interaction
E 3*N-1-5       Error

2) ANOVA with differences to baseline with patient effect
Let YD be the differences to baseline (Y2-Y1, Y3-Y1, Y4-Y1).
The model will be:
YD ~ T + V + T*V + id[T] + E
(E: Error)
Var DF  Meaning
T   1   Treatment
V   2   Visit (2-4)
T*V 2   Interaction
id[T]   N-2     subjects nested under T
E 3*N-1-(N-2)-5 Error (if size of both treatment groups equal)

3) ANCOVA without patient effect
Let Y be the values at visit 2-4 (Y2, Y3, Y4).
The model will be:
Y ~ T + V + T*V + Y1 + E
Var DF  Meaning
T   1   Treatment
V   2   Visit (2-4)
T*V 2   Interaction
Y1      1       baseline
T*Y1 1  Interaction
E 3*N-1-7 Error

4) ANCOVA with patient effect
Let Y be the values at visit 2-4 (Y2, Y3, Y4).
The model will be:
Y ~ T + V + T*V + Y1 + id[T] + E
(E: Error)
Var DF  Meaning
T   1   Treatment
V   2   Visit (2-4)
T*V 2   Interaction
Y1      1       baseline
id[T]   N-2     subjects nested under T
E 3*N-1-7       Error
witout intercept! (with intercept I get singularities)
(Interaction T*Y1 Interaction is not possible)

5) t-test with difference to baselne (visit 4 minus visit 1, Y4-Y1,
group by treatment T)

So I can think different possibilities to analyse the data. WHAT'S
BEST???

Hope somebody can help!

Karl
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to