DD3 is correct because it specifies a member of a PDS using DSORG=PO (Partitioned Organization). DD4 is correct because it creates a sequential dataset with DSORG=PS (Physical Sequential). DD1 and DD2 are incorrect because KEEP is not valid for temporary/new datasets in these combinations or because of invalid parameters.