On Tue, Oct 30, 2012 at 8:49 AM, Gabriel Go.Es. <[log in to unmask]> wrote:
>
> Thanks Guillaume and Donald,
>
> I'm questioning this because of an old entry on the SPM archives I'd found
> regarding a similar model;
> They say something like:
> " I want to compare GM between controls and patients, accounting for sex
> (categorical variable). Is it better that I do a full factorial design with
> a single factor (controls/patients) and I add as covariate a gender vector
> all made by zeroes and ones, or that I do a full factorial design with two
> factors (controls/patients and males/females)"
>
> to which Donald answer:
> "If you test the group differences in the first model, then you are testing
> the difference in controls males and patient males (e.g. the y-intercept if
> each group). You have assumed that the effect of being female is the same in
> both groups. I'll also suggest another model; add a single covariate of 1s
> and -1s to the model with patients and controls. This model will then test
> whether the groups - if they were balanced for males/females – are different
> from each other. This assumes the effect of being male/female is the same
> for both groups. This can be thought of as covariate-adjusted group means."
>
> Is not the same model I was trying to set?
The model that you proposed had your two groups (A and B) in the first
2 columns, then another column at the end of your model specifying
group A. This led to you having 2 columns for group A. This is where
the problem arose, not with having group columns and a gender column.
Removing the second column representing group A will solve the issue.
I did this, because I found no
> gender effect on my data, so I would like to thought that my model has a
> covariate-adjusted group means as Donald point in the last treat
This is correct.
>
> Kind Regards,
> Gabriel
>
>
>
>
>
>
>> Date: Tue, 30 Oct 2012 12:20:57 +0000
>> From: [log in to unmask]
>> To: [log in to unmask]
>> Subject: Re: [SPM] Error setting contrasts
>>
>> Dear Gabriel,
>>
>> when using a two-sample t-test, you are already modelling the group
>> effect (the first two columns of the design matrix) so you don't need to
>> add your 'group' covariate. When including it, you have an
>> over-specified model with redundancy (the sixth column is equal to the
>> first minus the second one, as highlighted with the grey boxes for
>> parameter estimability, which results in some contrasts to be
>> inestimable).
>>
>> Best regards,
>> Guillaume.
>
>>
>>
>> On 30/10/12 10:50, Gabriel Go.Es. wrote:
>> >
>> > Dear SPMers
>> >
>> >
>> > I'm having trouble when trying to set some constrasts for my statistical
>> > design (see atteched image)
>> > group was entered as a categorical variable with 1's and -1's and gender
>> > as 1 & 2's
>> >
>> > My problem is that, when I make the same design but without the group
>> > variable, everything works perfectly, but when I do introduce the group
>> > in the model
>> > I always get the following error:
>> >
>> > 1 -1 0 0 0 0 0 <- !invalid contrast
>> > Failed 'Contrast Manager'
>> > Error using ==> spm_run_con at 228
>> > Error in contrast specification
>> > In file "C:\Program Files\MATLAB\R2009b\work\spm8\config\spm_run_con.m"
>> > (v3993), function "spm_run_con" at line 228.
>> >
>> >
>> >
>> > The only valid contrast is possible when I do set the ANCOVA option to
>> > YES, is when thesting the columns of globals@ as:
>> > 0 0 0 0 0 1 -1
>> > but not sure what I'm testing on such contrasts.
>> >
>> > How can I test my group differences with my initial model? by accounting
>> > for the effect of all my variables. Is it real possible? or I do have to
>> > somehow change to fix my model.
>> >
>> >
>> > Many thanks in advanced,
>> > Gabriel
>>
>> --
>> Guillaume Flandin, PhD
>> Wellcome Trust Centre for Neuroimaging
>> University College London
>> 12 Queen Square
>> London WC1N 3BG
>>
|