Comments on the testing of combined results