Hi Bertram, > Dear SPMers! > > When I setup a basic models two samples t-test using the group codes 'a' > and 'b' (e.g. 'aaaaabbbbb'), the GUI asks subsequently for non-sphericity > correction. If I set the group codes '0' and '1' (e.g. '0000011111'), > this > question is omitted. Is this a bug or a feature? Well, I cannot answer this, but I've just tried [111112222] and then SPM asks for sphericity correction ! It seems to be specific to the cov [0000011111] ... maybe a bug ?? Best, Cyril