ANOVA Design question

[Roy Money]

Hi Listers

I am trying to understand the implications of these two different
UNIANOVA specifications.
In the first I have a one way anova with covariate and specify the
interaction of the  factor A and the covariate B in the DESIGN statement.
In the second I  use a computed interaction term which I specify in the
opening covariate list as well as the DESIGN statement.

The second specification gives me results equivalent to using REGRESSION
rather than UNIANOVA and I can easily interpret the results.
The first specification gives me a very different result for the main
effect of A - not sure why and not sure how to interpret the meaning of
the result.


Title Interaction term as cross product in design.
UNIANOVA Y BY A WITH B
   /CONTRAST(A)=Simple
   /METHOD=SSTYPE(3)
   /INTERCEPT=INCLUDE
   /PRINT=PARAMETER
   /CRITERIA=ALPHA(.05)
   /DESIGN=A B A*B.

Title interaction term as covariate.
compute  AtimesB=A*B.
UNIANOVA Y BY A WITH B AtimesB
   /CONTRAST(A)=Simple
   /METHOD=SSTYPE(3)
   /INTERCEPT=INCLUDE
   /PRINT=PARAMETER
   /CRITERIA=ALPHA(.05)
   /DESIGN=A B AtimesB.

Thanks in advance.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

Roy Money wrote

Hi Listers

I am trying to understand the implications of these two different
UNIANOVA specifications.
In the first I have a one way anova with covariate and specify the
interaction of the  factor A and the covariate B in the DESIGN statement.
In the second I  use a computed interaction term which I specify in the
opening covariate list as well as the DESIGN statement.

The second specification gives me results equivalent to using REGRESSION
rather than UNIANOVA and I can easily interpret the results.
The first specification gives me a very different result for the main
effect of A - not sure why and not sure how to interpret the meaning of
the result.


Title Interaction term as cross product in design.
UNIANOVA Y BY A WITH B
   /CONTRAST(A)=Simple
   /METHOD=SSTYPE(3)
   /INTERCEPT=INCLUDE
   /PRINT=PARAMETER
   /CRITERIA=ALPHA(.05)
   /DESIGN=A B A*B.

Title interaction term as covariate.
compute  AtimesB=A*B.
UNIANOVA Y BY A WITH B AtimesB
   /CONTRAST(A)=Simple
   /METHOD=SSTYPE(3)
   /INTERCEPT=INCLUDE
   /PRINT=PARAMETER
   /CRITERIA=ALPHA(.05)
   /DESIGN=A B AtimesB.

Thanks in advance.

Please post your REGRESSION syntax too.