Hi,
I'm reposting this query as the same happned to me! How is it possible that a 2nd level model, using simple con images such as [1 0 0 0... ] [0 1 0 0... ], shows the regressors to be non-orthogonal, when the regressors the contrasts are based on were orthogonal at the first level?
In other words in the first level I have A1B1, A2B1, A1B2, A2B2 (and other contrasts). They are orthogonal. I take each of the 4 using the simple contrast above to a 2*2 full factorial, and they are non-orthogonal. What might be happening?
(the original poster no longer receives emails on that address, we cannot tell whether her query was answered offline)
thanks! Deborah
|