Ok - if there was no rank deficiency warning, perhaps I have simply misinterpreted the design. 

Here is a previous post, where I discuss this issue in detail: 

Here is a link to some nice slides on modelling a 2x2 repeated measures (slides 78-79):

http://mumford.bol.ucla.edu/setting_up_d...
Here are some detailed instructions:
Suppose the data is order according to the following rows (2 groups, 2 time points, 2 subjects per group):
Group 1 Time 1 Subject 1
Group 1 Time 1 Subject 2
Group 1 Time 2 Subject 1
Group 1 Time 2 Subject 2
Group 2 Time 1 Subject 3
Group 2 Time 1 Subject 4
Group 2 Time 2 Subject 3
Group 2 Time 2 Subject 4
The your design matrix would look like:
1 0 0 0 1 1
0 1 0 0 1 1
1 0 0 0 -1 -1
0 1 0 0 -1 -1
0 0 1 0 1 -1
0 0 0 1 1 -1
0 0 1 0 -1 1
0 0 0 1 -1 1
Where the first four columns model the individual means of each subject, the 5th column models the main effect of time and the 6th column models the group x time interaction.
The contrast for the main effect of time is either: [0 0 0 0 1 0] or [0 0 0 0 -1 0]
The contrast for the interaction is: [0 0 0 0 0 1] or [0 0 0 0 0 -1]
The exchange block should be set as: [1 2 1 2 3 4 3 4]
Select either t-test of F-test, as appropriate.
If you want to test the main effect of group, then a repeated measures design is not particularly useful and can be eliminated. For example you can average across the two time points per subject and perform a two sample t-test, or perform a one-sample test on the difference.




Best wishes,
Andrew
  
Originally posted by Álvaro Deleglise: