 |
 |
Ok - there are various levels to this answer. It seems to be a sensible thing to do - effectively you want to know the amount amount of the data's standard deviation which is explained by a single COPE. 1) If you've only got one PE, then this is relatively trivial. the std of the design can be computed easily from design.mat (ascii file containing design timeseries). Call this sx. The std of the data is avwmaths filtered_func_data -Tstd sy then the fractional deviation explained by your PE is just avwmaths PE -mul sx -div sy Beta_norm I think, in this case, you can do this with the unwhitened data as the whitening matrix is normalised. 2) If you've got more than 1 PE, life is more complicated (and I don't think you can compute what you want with simple FSL commands ). You need to project the variance explained by all of your EVs onto a single COPE. if you assume the data is white and demeaned then and your Design is demeaned.. the std explained by your cope is: sqrt(c'*b*X*X'*b'*c)/dof c is cour contrast, b is your vector of PEs, X is your design, dof is your degrees of freedom if it's not white then sqrt(c'*b*k*X*X'*k'*b'*c)/dof k is the whitening matrix. (The whitening will change the projection) and the standard deviation of the whitened data is just std(k*Y) Dividing one by the other should give you what you want. Hate to say it, but you might need matlab!! cheers T ------------------------------------------------------------------------------- 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 ------------------------------------------------------------------------------- On Tue, 12 Aug 2003, Edward Vessel wrote: > Ok, well, here is something that might give you a feeling for the difference. > In a regression with only a single independent variable, the standardized > beta equals r (the Pearson's correlation). This would also be true (I think) > if all the variables were totally uncorrelated in a multiple regression. > > The analysis I have done is one in which I want to look at a correlation as my > statistic, and the standardized Beta coefficient is one way to report a > factor loading which takes into account that factor's covariance w/ other > factors. It would be 1 if the activity of a voxel were perfectly predictable > from that factor, and 0 if that factor had no power to predict it. So, the > reason for having it is in the _interpretation_ of the statistic. > > Typically, it is computed as: > > Beta = b * (sx / sy) > > where > > Beta: standardized regression coefficient (-1 to 1) > > b: the unstandardized regression coefficient (can take any value), which is > the 'regulular' weight, sometimes confusingly referred to as beta but isn't > standardized) - probably a pe, which is equivalent to one of my copes in this > case > > sx: standard deviation of the EV > xy: standard deviation of the data > > Does that make any sense? > > Ed > > > On Tuesday 12 August 2003 02:26 am, Tim Behrens wrote: > > Hi there - > > > > I'm not sure what this is ( or how to compute it ). > > If you can give us the equations describing this, we should be able to > > tell you how to compute it with the Feat output. > > > > What is it that you would like to describe with "Standardized Beta > > weights", that you can't describe with the original Betas and the t-stats? > > > > Sorry I'm no use > > > > T > > > > On Tue, 12 Aug 2003, Edward Vessel wrote: > > > No, I am referring not to a t or z score, but to a regression weight. > > > A standardized Beta weight is never greater than one. It is a 'factor > > > loading' so to speak. > > > > > > Ed > > > > > > On Tuesday 12 August 2003 01:16 am, Tim Behrens wrote: > > > > Hi there - by standardised beta weight, do you mean the t-statistic? > > > > > > > > You can get these post-hoc using Feat, by just running the > > > > contrast-manager (by selecting Post-Stats from the top right menu, > > > > clicking on the Post-Stats tab, and selecting "Edit Contrasts"). > > > > > > > > If you want to get them by hand, you were nearly right, you have to > > > > divide the copes by the square root of the varcopes (i.e. the standard > > > > error on the copes). > > > > > > > > If you want them to be truly sandardised (i.e. z-scores), you have to > > > > account for the degrees of freedom, you can do this with ttoz > > > > > > > > ttoz -zout zoutput varcope cope dof > > > > > > > > > > > > > > > > Hope this answers your question > > > > > > > > cheers > > > > > > > > Tim > > > > > > > > On Mon, 11 Aug 2003, Edward Vessel wrote: > > > > > Hi folks - > > > > > > > > > > How would one go about computing a standardized beta weight (in the > > > > > regression sense) from a cope or pe? > > > > > > > > > > If I am correct, the pe's are (unstandardized) regression weights > > > > > (b's). Therefore, I'd need to multiply by the standard deviation of > > > > > the predictor and divide by the standard deviation of the data. But > > > > > I am unsure which files would correspond to this. > > > > > > > > > > The varcope seems to be not just be the deviation of the predictor, > > > > > as that should be the same for all voxels. It also isn't the > > > > > standard deviation of the data, as this would be the same for all > > > > > predictors. Is it a ratio of the two? > > > > > > > > > > If that is the case, then do I just divide the cope by the varcope to > > > > > get a standardized weight? > > > > > > > > > > I'm not interested in getting percent signal change in this case ... > > > > > it is a continuously varying parameter (from 0 to 1), so I'd like to > > > > > get a beta weight (or even part correlation). > > > > > > > > > > Ed > > > > > > > > > > -- > > > > > Ed Vessel > > > > > U. of Southern California [log in to unmask] > > > > > Dept. of Neuroscience > > > > > HNB, 3641 Watt Way http://geon.usc.edu/~vessel > > > > > Los Angeles, CA 90089-2520 > > > > > (213) 740-6102 > > -- > Ed Vessel > U. of Southern California [log in to unmask] > Dept. of Neuroscience > HNB, 3641 Watt Way http://geon.usc.edu/~vessel > Los Angeles, CA 90089-2520 > (213) 740-6102 > > >