3.1 Importance of assumptions

Let us start this chapter by recalling that the phase "asserting" is essential to a successful design. This phase must, among other things, provides guidance over the model to be used to synthesize data. To underline what is about, let us consider the following measures :  :

$$\begin{array}{rrrrrr} x& 2.8529164& -1.2502598& -3.4666558& -... ...26750& -7.5575739& -1.8178379& 9.9940873& 11.698057 \end{array}$$

If the best model is known to be a regression line, then :

$$\hat{y}=2.9974133 x+6.8773196$$

is obtained. If the best model is known to be the polynomial of least degree passing through every point, FIG. 3.1 shows that a really different result is obtained.

FIG. 3.1: Importance of assumptions