Help System (web edition)

If both multi-way and one-way analyses are requested in the same
ANOVA operation then the error mean square used in multi-way
univariate analysis is automatically transformed to one-way
analyses (e.g. multiple comparison of means), e.g.
    ANOVA IEADATA,END+2
    DEPENDENT=KNowldge
    GROUPING=Attitude,Sex
    MEANS=A,S,AS   METHOD=Tukey(s)
The families for multiple comparisons are defined by MEANS statement
using the initials of grouping variables.

If the design involves several sizes of experimental units then
several error terms are needed. Therefore, the user has the
possibility to define by the EMS statement which mean squares
form the error term while a specific main effect or interaction
is tested, e.g.
    EMS=A[AB],C[AC+ABC]
states that the interaction AB is used as the error term while
the main effects of A are tested. When the main effects of C are
tested then the error term is formed from interactions AC and ABC.
If one-way analyses are performed at the same time, these error
terms are automatically used in those analyses.

Further information:
   V = More on analysis of variance for multiple factors 
   C = More on analysis of covariance 
   A = More on ANOVA operations