Stephan Moratti moratti at MED.UCM.ES
Wed Sep 9 13:57:44 CEST 2009

Hi Eric y Nadia,

Ok, I have made a test of depsamplesF comparing results of depsamplesF and 
SPSS. I get the same results. I used data from an experiment and chose a 
timepoint and one channel that were indicated to be significant by 
dempsamplesF. In the attached TEST_dempsamplesF.mat file it is "testmat" 
(24 subjects 3 conditions). The variable "dat" is the reshaped "testmat". The 
structure "s" contains the results of depsamplesF. I added the variable 
scaledstat as the field "F" to "s" as it is the F value. The attached 
file "SPSSoutput.bmg" is a screen shot of the SPSS output entering testmat in 
a MANOVA. The results  are the same:
F(2, 22) = 6.627, p = 0.006

We can also calculate Wilk's lambda from T^2 (see page 232 formula 5-13 of 
the Johnson 1998 edition) as:

lambda = 1/(1 + s.stat/(24-1));

and get 0.624, the same as in SPSS. Maybe, as people are more used to F 
values, it would be useful to have the F(scaledstat)  value also in the "s" 
structure (and later after using timelockstats or freqstats to have a matrix 
containing the F values besides the T^2 values).

Thanks for implementing that great stuff in fieldtrip and making it accessible 
for all users !



The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also and
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SPSSoutput.bmp
Type: image/bmp
Size: 371126 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TEST_dempsamplesF.mat
Type: application/octet-stream
Size: 1588 bytes
Desc: not available
URL: <>

More information about the fieldtrip mailing list