Construction of statistical models is discussed. It is shown that wrong models can lead to misleadings results.
If the model is appropriate, it is necessary to interpret the results correctly.
