[log in to unmask] wrote:
> Dear Will,
>
> I still have some basic questions. Please see below. Thanks!
>
>
>>> If I want to see if there is a interaction among tasks(A,B,C) *
>>> hands(uncross,crossed)* compatibility(compatible, incompatible)? what
>>> I should do? Thanks!
>> A 3-way interaction at the 2nd level !
>>
>> Perhaps the easiest way is, for each subject, create the two-way
>> interaction contrast - the interaction between hands and compatibility.
>> This will be 1 -1 -1 1 over the uncross,crossed,compatible,incompatible
>> columns of your first level design.
>
> Here is 1 -1 -1 1 the same as -1 1 1 -1?
Yes !
>> Then you have one con image per subject per task.
>>
>> At the second level create a one-way ANOVA with 3 levels (using full
>> factorial as before). You don't need subject effects in your 2nd level
>> design here because the data you are taking up are differences (ie.
>> subject effects already removed). This will create
>> a design matrix with 3 columns A, B, C.
>>
>> To test for an interaction among your two-way interactions ie. a
>> 3-way interaction, use the F-contrast [1 -1 0; 0 1 -1].
>
> This may be a silly question. Here, I am not exactly sure why the contrast is [1
> -1 0; 0 1 -1]. Could you suggest me reading something regarding how to specify
> contrast? I always have trouble to do this as I'm not good at math. Thanks!
>
This is a 'multi-dimensional' contrast ie. it has more than one row.
The first row ([1 -1 0;]) tests for a difference between the first and
second levels. The second row ([0 1 -1]) tests for a difference between
the second and third levels.
The overall test [1 -1 0; 0 1 -1] tests for a difference between the
first and second levels OR a difference between the 2nd and 3rd levels
OR any linear combination thereof.
Have a look at this book chapter for more info:
http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/pdfs/Ch8.pdf
This has been updated to chapter 9 in the more recent SPM book.
See also:
http://www.fil.ion.ucl.ac.uk/~wpenny/publications/spm-book/anova.pdf
Best,
Will.
> Best wishes,
> Chunhong
>
>
--
William D. Penny
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|