Dear Veronica,
this sounds more complicated than most other flexible factorial
designs I have seen so far. Your contrasts are probably not estimable
because both group-by-sex and group-by-replication interaction columns
are not orthogonal to the subject columns. Things would be much easier
if you tried a full factorial design without an explicit subject factor.
Volkmar
Quoting Veronica Garcia Vazquez <[log in to unmask]>:
> Dear list,
>
> We have problems with creating T-contrasts in my design. I would
> appreciate any kind of help We receive...
>
> 1))) My conceptual design is the following one:
>
> |--Day 0: 9 images (1 image per
> subject)
> |--- VH (9 subjects) --- |
> | |--Day 2: 9 images (1 image per
> subject and same subjects as Day 0)
> |-- Female----|
> | | |--Day 0: 8 images (1 image per
> subject)
> | |--- CP (8 subjects) --- | |
> |--Day 2: 8 images (1 image per subject and same subjects as
> Day 0)
> |
> --|
> | |--Day 0: 6 images (1 image per
> subject)
> | |--- VH (6 subjects) --- |
> | | |--Day 2: 6 images (1 image per
> subject and same subjects as Day 0)
> |---- Male ---|
> | |--Day 0: 7 images (1 image per
> subject)
> |--- CP (7 subjects) --- |
> |--Day 2: 7 images (1 image per
> subject and same subjects as Day 0)
> 2))) In SPM5, we choose Flexible Factorial and 3 factors
> in the following order: subject, group (for levels VH and CP),
> replication (for levels Day 0 and Day 2) and sex (for levels female and
> male).
>
> Subject: Independent Yes and Variance Equal
> Group: Independent Yes and Variance UnEqual
> Replication: Independent No and Variance UnEqual
> Sex: Independent Yes and Variance UnEqual
>
> For example, for the two images (day 0 and day 2) of subject female VH,
> its conditions are: 1 1 1; 1 2 1; For example, for the two images
> (day 0 and day 2) of subject female CP, its conditions are: 2 1 1; 2
> 2 1; For example, for the two images (day 0 and day 2) of subject
> male VH, its conditions are: 1 1 2; 1 2 2; For example, for the
> two images (day 0 and day 2) of subject male CP, its conditions are:
> 2 1 2; 2 2 2; 3))) The main effects and interactions are:
> Main effect: subject (1)
> Interactions: group and replication (2 and 3); group and sex (2 and 4)
>
> 4))) We want to test main effect in sex (FEMALE>MALE)
>
> We tried with the t-contrasts: zeros(1,38) 1 1 -1 -1
> 1/17*ones(1,17) -1/13*ones(1,13) zeros(8)
> 1 1 -1 -1
>
> and they doesn´t work..., why?
>
> Regards
>
> Veronica
>
>
>
>
>
>
>
>
>
> my design is as follows:
> 2 groups (N1, N2), 2 conditions (C1, C2). i specified 3 factors
> (subject, group, condition) and got a design matrix comparable to
> darrens and matts one (see attached file). the first 26 columns are for
> the subjects, the last 4 columns are for the factor interactions
> (N1xC1, N1xC2, N2xC1, N2xC2).
>
> so i want to have the following contrasts, which i managed to specify
> in spm. my first question is:
>
> are the following contrast definitions correct?
>
>
>
>
> i read the discussion in the list about the 2x2 ANOVA with great
> interest, because i like to setup a similiar design. so i followed the
> procedure of darren and matt for the flexible factorial design setup.
> i designed 3 factors (subject (2xn), group (2 groups), condition (2
> conditions) /independence yes, yes, no / variance equal, unequal,
> unequal). specified the design by subject. for each subject of my group
> 1 the condition matrix is [1 1;1 2], for each subject of my group 2 [2
> 1;2 2]. There is a main effect of subject (factor number 1) and an
> interaction of group x condition (factor numbers 2 3).
> This produces a design very similiar to the ones of darren and matt. i
> got n columns for my n participants, each with 2 rows. Now the
> difference and my problem:
> i got only 3 columns of group x condition instead of my expected 4.
> these are column 1 (group1 x condition2), column 2 (group1 x
> condition2), and column 3 (group2 x condition2). the lacking column
> (group2 x condition1) seems to be mixed into column 2 (see attached
> file). does anybody know what my mistake is?
>
> thanks markus
|