The difference between the expected value from a model and the experimental data value.