degrees of freedom

[Harry Moore]

Hi all,

I have a couple of questions regarding the degrees of freedom that are automatically 'selected' when I upload an overlay in FIVE. For post-hoc T images and main effect T images, FIVE automatically selects the error degrees of freedom (in my case 30), whereas for interaction F images it automatically selects the DF for both the model and the error (in my case 1 and 30).

I've viewed the same overlays in bspmview, which only allows you select one value for the DF - for post hocs and main effect T images it automatically selects only the DFerror, and for F images it automatically only selects the DFmodel. Bspmviewer doesn't have the same online forums and documentation that FIVE has, but I have read somewhere that when viewing an interaction F overlay in bspmview, you should only use the DFerror, not the DFmodel - although as I don't know how bspmview works, I don't know which is correct. 

So I suppose my question is why are the automatically selected DFs different in FIVE depending on whether its a post hoc, main effect or interaction overlay? (And if anyone knows how bspmview works - why does it only select single value DFs and whether for interaction overlays I should enter the DFerror or DFmodel?)

I hope I've made at least some sense, thanks for your help!

[Aaron Schultz]

I can't speak to bspmview, but I can answer your questions for FIVE.

When you run an analysis with SPM or with GLM_Flex_Fast, DF
information from the model will be entered into the description field
of the nifti header. When then open a stat image it will extract this
information to assign the DFs.

Importantly in GLM_Flex_Fast, interaction terms are always written out
as F-statistics. F-statisitcs always require two DF parameters, one
for the effect and one for the error. In the case where the the
effect df is one, the F-statistics are simply the square of the
t-statistics, but it's important for the viewer to understand what
test statistic is encoded in order to derive the proper threshold
value for a given p-value.

I hope that helps, and let me know if you have any other questions.

-Aaron

> --
> You received this message because you are subscribed to the Google Groups "Aaron's fMRI matlab tools." group.
> To unsubscribe from this group and stop receiving emails from it, send an email to fmri_matlab_to...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/fmri_matlab_tools/4d5c0df9-85a8-41c3-a46f-9ed31e51e525%40googlegroups.com.

[Harry Moore]

Hi Aaron,

I think that does help, thanks for taking the time to explain. So therefore I assume that as the DF information is written into the GLM Flex output files, there is no need to alter the DFs that are automatically generated by the viewer in FIVE.

Thanks again for your help,

> To unsubscribe from this group and stop receiving emails from it, send an email to fmri_mat...@googlegroups.com.