What are some of the sources of bias that can affect an evaluation? How might such biases be minimized?