between-subject design
if three-way interaction is signifcant , compute the simple interaction effect.
**********simple interaction effect******************** manova
total by gender (1,2) depart (1,2) grade(1,3) /design = gender*grade within depart(1) gender*grade within depart(2).
if the simple interaction effect is not signifcant ,
compute the simple effect.如果简单交互效应不显著，那么就计算
***************simple effect*********************** manova
total by gender (1,2) depart (1,2) grade(1,3) /design = grade within depart(1) gender within depart(1).
**************multiple comparisons***************** manova
total by gender (1,2) depart (1,2) grade(1,3) /contrast (grade) = special (1 1 1 1 -1 0 1 0 -1) /design = grade within depart(1).
manova
total by gender (1,2) depart (1,2) grade(1,3)
/contrast (grade) = special (1 1 1 1 -1 0 0 1 -1) /design = grade within depart(1).
if the simple interaction is significant ,compute the simple simple effect.简单交互效应显著，计算简单简单效应。
***********simple simple effect********************** manova
total by gender (1,2) depart (1,2) grade(1,3)
/design = grade within gender(1) within depart(2) grade within gender(2) within depart(2).
******************multiple comparisons******************* manova
total by gender (1,2) depart (1,2) grade(1,3) /contrast (grade) = special (1 1 1 1 -1 0 1 0 -1) /design = grade within gender(2) within depart(2).
manova
total by gender (1,2) depart (1,2) grade(1,3) /contrast (grade) = special (1 1 1 1 -1 0 0 1 -1) /design = grade within gender(2) within depart(2).
The within-subject design （组内设计）is similar to the between-subject design . You can change the
“design” into “wsdesign（组内）”. And the command contrast is not adap to multiple comparisons in within –subject design.
mixed-design
if three-way interation effect is
significant ,compute the simple interation effect.（如果三项交互效应显著，计算简单交互效应）
*************simple interaction effect************* manova
a1b1 a1b2 a1b3 a2b1 a2b2 a2b3 by freq(1,3) /wsfactors = a (2) b (3) /wsdesign = a*b
/design = mwithin（混合设计） freq(1) mwithin freq(2) mwithin freq(3).
if simple interaction effect is significant , compute the simple simple effect.
************simple simple effect ****************** manova
a1b1 a1b2 a1b3 a2b1 a2b2 a2b3 by freq(1,3) /wsfactors = a (2) b (3) /wsdesign = a within b(1) a within b(2) a within b(3) /design = mwithin freq(1).
manova
a1b1 a1b2 a1b3 a2b1 a2b2 a2b3 by freq(1,3) /wsfactors = a (2) b (3) /wsdesign = b within a(1) b within a(2)