 |
 |
Hi, Featquery uses the ppheight, which is probably causing the difference. To compare, look at the ppheight in your first level analysis for a few subjects and compare this to the min/max range that you calculated. I believe featquery averages the ppheights over all subjects, but somebody can correct me if I'm wrong. The ppheight is in the header of the design.mat file at the first level. Hopefully that will explain the difference. I prefer doing it on my own with my own scale factor, since then I know exactly how things were scaled. All-in-all, at the end of the day the scale factor you're using is simply a ruler. If I tell you a measurement in inches, you can easily transform this to mm's if you wanted. So, for example, if you had a 2s long stimulus and I want to compare it to my 1s long stimulus, I can make this conversion on my own. I just need to know what scale factor you used for your stimulus. As long as you state in your paper what you used as a scale factor, you're okay. Doing it yourself you'll always know what the scale factor was. Cheers, Jeanette On Fri, Jul 20, 2012 at 7:49 AM, Torsten Ruest <[log in to unmask]> wrote: > Heya, > > to follow up on this, I compared the results I get with the manual method > (I determined the height using matlab) with those from featquery. In the > featquery log it applies a scaling factor of ~50, while by manual > determination, following Jeanette's tutorial, I get ~20. > > I understand that the featquery scaling factor comes from the design.con > file for that contrast, and that it's hampered by the highpass filter. I > thus saved the design without the highpass filtering and it's just a little > less, ~48 (as taken from the .con file). I am not sure whether other > factors like stimulus spacing (event-related design) can account for the > massive difference, and somehow I have the feeling I am doing something > wrong > > Can someone confirm that this discrepancy is common, and the manual way is > the correct way? Note that if I apply the featquery threshold to my manual > formula, values are comparable. > > The way I manually determined the scaling factor was to follow closely the > method in Jeanette's tutorial. The contrast of interest sums up to +1 and > -1, the method I used to calculate height is identical to those in the > tutorial (only replacing values, I checked), so I am confident that this is > correct. > > > Thanks again for suggestions. > > Torsten >