PS:
> 3) The randomise error "Warning: You have demeaned your
> data, but at least one design column has non-zero mean"
> appears in my case although age was properly demeaned.
If you have run exactly the design.mat attached to your email, as you would have seen from my email earlier today, you should *not* use the -D option as yo have modelled the mean for each group and this would explain the error message.
If you have run a different design with just the age x group interaction, then the error message is indeed explained by the rounding problem.
Cheers,
Gwenaelle
|