Hi Debbie,
Although you think that these are 3 ways to ask the same question it
seems to me that they have different answers so the questions are also
probably different ;-)
On Tue, Aug 17, 2010 at 1:55 PM, dtalmi <[log in to unmask]> wrote:
> hi vladimir,
> this is a really naive question i'm afraid. i will ask it in 3 different
> ways but they are all the same:
>
> *is there any implicit baseline or baseline correction in SPM, e.g. does SPM
> consider activation at early time points in the epoch a baseline?
If you use the epoching function (as opposed to defining trials at
conversion) then by default the negative part of the time axis will be
used as baseline. You can also do your own baseline correction and
define the baseline in any way you want.
> *when a parameter estimate in SPM equal zero for a specific condition, does
> this mean that activation in this condition was equal to the simple average
> across all regressors in the design matrix?
That doesn't sound correct to me. When a parameter estimate equals
zero it means that the corresponding regressor does not contribute to
explaining the data and the data is explained by the other regressors
with whatever coefficients. You can add a random regressor to any
design matrix and its coefficient will be close to zero but it doesn't
change anything about the other coefficients (or at least not much).
> *are both positive and negative parameter estimates signals that are LARGER
> than baseline, or is a negative parameter estimate LOWER/more negative-going
> than baseline and a positive parameter estimate HIGHER/more positive going?
> - this to related to EEG where signals are more or less negative and zero
> means zero!
>
That I think depends on the design matrix. If for instance you flip
the sign of your regressor, also the sign of the coefficient will
flip.
Vladimir
> thanks.
> debbie
>
>
>
|