Hi Tugan,
You're absolutely right. If you have only 3 points on your line
(I1,I2,I3) then the question "Is the linear trend significantly greater
than zero?" is identical to the question "Is I3 significantly greater than
I1?".
I don't think there is any advantage in doing a second level contrast
between (I3-I2) and (I2-I1). You will still effectively get (I3-I1).
The only real way to be more confident about your trend is to collect data
at more points on the line!
Hope this is clear
Tim
On Thu, 5 Jun 2003, tgn mft wrote:
> Hi,
>
> I was reading through the overview of GLM analysis and
> there is an issue about contrasts that confuses me. It
> is said that "[-1 1 0] shows where the response to
> level 2 is greater than that for level 1. [-1 0 1]
> shows the general linear increase across all three
> levels." But in my opinion, [-1 0 1] also shows (and
> maybe more likely to show) regions where response to
> level 3 is greater than that for level 1 (but level 2
> is completely ignored). It is impossible to tell
> whether it is (I3-I1) or [(I3-I2)+(I2-I1)]. If there
> is no response to levels 1 and 2 and there is positive
> response to level 3, then the above contrast will also
> show that nad it is not a linear increase in responses
> to tasks. So, how do you differentiate the two
> activation patterns?
>
> Maybe it'd be better to avoid such a misleading
> contrast and get two contrast images (I3-I2) and
> (I2-I1) separately and then enter them into a second
> level analysis to compare the difference of two
> contast images? Of course, here you have to do some
> masking so that any region where (I2-I1)=0 does not
> show up in the comparison.
>
> I'd be interested in hearing your thoughts on this
> confusing matter.
>
> Thanks,
>
> Tugan
>
>
> Univ. California, Irvine
> Center for Functional onco-imaging
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
> http://calendar.yahoo.com
>
--
-------------------------------------------------------------------------------
Tim Behrens
Centre for Functional MRI of the Brain
The John Radcliffe Hospital
Headley Way Oxford OX3 9DU
Oxford University
Work 01865 222782
Mobile 07980 884537
-------------------------------------------------------------------------------
|