Thank you very much, Dr. Whittington, I never thought I can (should) put something before merge statement, now I learned something new.

Thanks

John Whittington wrote:

> At 10:04 30/03/00 -0800, Ya Huang wrote (in part): > > >data xx_h (rename=(y=ny)); > > set xx; > >by x; > >if first.x then delete; > >else output; > > > >data xx; > > merge xx(in=a) xx_h(in=b); > >by x; > >if a and not b then ny=.; /**********/ > > > >Question is why I still get ny retained in obs=5,10? > >I expected after using the "if a and not b then ny=." statement, > >ny would be missing in obs=5,6. But as you can see, > >they are still retained. Can somebody tell me why? > > Ya Huang, *any* variable which comes from a dataset which is SET, MERGEd, > UPDATed etc., is automatically RETAINed. If you want to disable that > functionality, you should put a ' ny = . ; ' satement immediately before > your merge statement. > > Kind Regards > > John > > ---------------------------------------------------------------- > Dr John Whittington, Voice: +44 (0) 1296 730225 > Mediscience Services Fax: +44 (0) 1296 738893 > Twyford Manor, Twyford, E-mail: John.W@mediscience.co.uk > Buckingham MK18 4EL, UK mediscience@compuserve.com > ----------------------------------------------------------------