I am receiving an error ntdtr and stdtr DOF error in a 3rd level mixed effects analysis. 

Here is what I did. I would appreciate any assistance (might also note that this particular analysis is being run on Debian using the previous version of FSL; i.e. not 4.0):

we ran a block design in which participants were engaged in a game for a long block with a short rest period at the beginning and the end of the game. 

they repeated playing the game 7 times each of which had to be a separate run/session

we analyzed each session at first level.

we then carried the 7 first level .feat directories up to a second level (fixed effects) analysis to create contrasts between game conditions for each subject.

then we tried to carry each subjects .gfeat/cope.feat to a third level mixed effects analysis to obtain the group mean for each contrast associated with the game conditions. 

When we do this the analysis fails (we tried it first with a small sample of 5 subjects to see if we were on the right track :-(. We get a ntdir and stdtr  - i have attached the .log file. 

The strange thing is that if in the third level analysis we run it as a fixed effects analysis we get good looking output; i.e. the data look roughly like what I would have predicted. 

Am I missing something?

Thanks in advance.

Best, 

Steve