Gravar-mail: Some Problems in the Statistical Analysis of Clinical Trial Data