spss code for decomposition of a significant three-way interaction into two-way interactions, preserving degrees of freedom

this is a "trick" to make spss resolve two-way interactions in the presence of a significant three-way interaction in an anova. a more general way that can also deal with continuous predictors is described in aiken & west:

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks: Sage.

and elaborated on in How to probe simple and complex slopes in a regression analysis with three (or more) predictors using SPSS in this here blog.

MANOVA dv BY iv1(x,y) iv2(a,b) iv3(o,p) /ERROR=WITHIN /DESIGN=iv1*iv2 WITHIN iv3(o) iv1*iv2 WITHIN iv3(p) /*[and so on...]*/.

where

dv	dependent var
iv1, iv2	independent variables
iv3	independent variable for whose levels you decompose the three-way into two-way interactions (iv1*iv2)
x	first level of iv1
y	last level of iv1
a	first level of iv2
b	last level of iv2
o	first level of iv3
p	last level of iv3

IMPORTANT: iv1 and iv2 need to be coded in consecutive integers corresponding to x,y,a,b.

you can specify any simple or higher-order effects at a level of another variable in the /DESIGN= command.