Interpretation of response surface models

Provided that a proper experimental design has been used to establish the model parameters, and that the model adequately describes the variation of y in the experimental domain, we can use the model to evaluate the influence of each variable and assess the significance of each term in the model. This can be accomplished through statistical tests which compare the estimated parameters to estimates of the experimental error. This will answer the Which and How questions.[4] 

Will it also be possible to answer the Why questions, which implies that we give the estimated parameters a physical meaning The answer is yes, although great care should be exercised in such interpretations. 

The model parameters fBj, Bjj, Bj are partial derivatives of the theoretical response function f and as such they will have physical meanings. 

An increased yield, y, achieved through an increase in the reaction temperature, 

will show up as a positive value of Bj. This parameter is likely to be related to the temperature dependence of the kinetics of the reaction. 

---