A description of a method, thought by the programmer to be correct statistics, but which is not. An example is averaging together several averages of samples of different sizes. The correct way to do this is to average together all of the individual samples.