Hi Davide,

The interaction column of your design matrix appears to have some errors. The interaction column should be the multiplication of the two main effects columns. For example, the interaction column should have a value of 1 for the last row of the design matrix because -1 * -1 = 1, but the current design matrix has a -1 here. 

1. You should add a column of 1's. However, another alternative is to model the repeated measures aspect of your design. If you want to model the repeated measures, you would not include a column of 1's, but instead include a new column for each subject. Each such column would have two 1's and 0's every else. In particular, you would place 1's in the two measurements for each subject and 1's elsewhere. Crucially, you will need to remove the first column (main effect of drugs) if you use repeated measures. This is because the main effect of drugs is fixed between the two conditions (does not change between conditions). So if you decide to model the repeated measures, your design matrix would have 18 columns: main effect of condition + interaction + 1 column for each subject. You can send me your new design matrix and I would be happy to have a look at it.

2. I suggest using a t-test and separately considering a contrast of -1 and 1.

3. 5000-10000 should be enough.

Andrew


Originally posted by davideit7: