Double check casts doubt on statistics in published papers