Power simulations with MANOVA

[JOHN ANTONAKIS]

Hi:

Does anyone have some code to share for doing a power
simulation with MANOVA.  I would like to specify code for a
two-way as well as a repeated measures model. I have some
basic code, extended from a posting by Jeremy Miles (below),
that I'd like to build on (I would like to add a second
factor, and then a repeated measure on all three dependent
variables).

Best,
J.

Matrix data variables = cue rowtype_ tf tr pa
/factor cue
/format = lower nodiagonal

begin data.
1 mean 2.5 2.0 1.5
1 n 100 100 100
2 mean 3.0 2.5 1.0
2 n 100 100 100
. sd 1.2 1.1 .90
. corr .70
. corr .20 .45
end data.

MANOVA tf tr pa by cue (1,2)
/method unique
/error within+residual
/matrix = in (*)
/power t (.05) F(.05)
/print signif (mult averf)
/noprint param (estim) .



___________________________________

Prof. John Antonakis
Faculty of Management and Economics
University of Lausanne
Internef #527
CH-1015 Lausanne-Dorigny
Switzerland

Tel: ++41 (0)21 692-3438
Fax: ++41 (0)21 692-3305

http://www.hec.unil.ch/jantonakis
___________________________________

[Maguin, Eugene]

John,

It doesn't look like you got any replies. I've done this. Here are some
examples I used. Gene Maguin

*************************************************************.
*  COND BY DV SIGMA(m)=.0808, R=.4, N = 125. REPEATED MEASURES.
*************************************************************.
matrix data variables = group rowtype_ dv1 dv2 dv3/
   factor = group/format = full.
begin data
. n           375  375  375
1 mean        0.00 0.21 0.42
1 n           125  125  125
2 mean        0.00 0.21 0.42
2 n           125  125  125
3 mean        0.00 0.00 0.00
3 n           125  125  125
. stddev      1.0  1.0  1.0
. corr        1.0  0.4  0.4
. corr        0.4  1.0  0.4
. corr        0.4  0.4  1.0
end data.

manova dv1 dv2 dv3 by group(1,3)/wsfactors dv(3)/method=unique/
   error=within+residual/matrix=in(*)/power t (.05) F (.05)/
   print signif(multiv averf efsize)/noprint param(estim).


*************************************************************.
*  COND BY DV SIGMA(m)=.0808, R=.4, N = 125. ANCOVA DESIGN.
*  COV1 AND COV2 ARE SAME. NEEDED TO MAKE SPSS WORK.
*************************************************************.
matrix data variables = group rowtype_ dv1 dv2 cov1 cov2/
   factor = group/format = full.
begin data
. n           375  375  375  375
1 mean        0.21 0.42 0.00 0.00
1 n           125  125  125  125
2 mean        0.21 0.42 0.00 0.00
2 n           125  125  125  125
3 mean        0.00 0.00 0.00 0.00
3 n           125  125  125  125
. stddev      1.0  1.0  1.0  1.0
. corr        1.0  0.4  0.4  0.4
. corr        0.4  1.0  0.4  0.4
. corr        0.4  0.4  1.0  1.0
. corr        0.4  0.4  1.0  1.0
end data.

manova dv1 dv2 by group(1,3) with cov1 cov2/wsfactors dv(2)/method=unique/
   error=within+residual/matrix=in(*)/power t (.05) F (.05)/
   print signif(multiv averf efsize) cellinfo(means corr)/
   noprint param(estim).

manova dv2 dv3 by group(1,3) with (cov1)/wsfactors dv(2)/method=unique/
   error=within+residual/matrix=in(*)/power t (.05) F (.05)/
   print signif(multiv averf efsize)/noprint param(estim).

***These are equivalent statements because (cov1) designates that cov1 is a
constant covariates. See point 4 under the MANOVA Variable List heading in
the Manova procedure documentation.