There is no excuse for the shoddy practice of allowing researchers to change outcomes and goals without saying so, says Ben Goldacre.