EViews.com

I know this might sound stupid, but do I run the Johansen test using my original data, lagged data or on my data at first difference?
and also is it normal for the trace test give a different answer to the max-eigenvalue ?


Thanks

D