That's bad for science. It's quite similar to adjusting your hypothesis to fit the data. Not as bad as manipulating the data, but not ethical, either.