Why use a design

It is a common belief that valuable information can only be obtained from experiments which have given good results, e.g. high yields. This is a misconception. To understand why certain experimental conditions give good yields, it is essential to know also why other conditions do not. The infonnation which can be obtained in a single experimental run will depend on the results obtained in other runs. The quality of information is therefore dependent on the ensemble of all experimental runs, i.e. the experimental design. 

The idea of a factorial design is to arrange the experiments in such a way that the variation in response obtained with different settings of the experimental variables, factors, can be traced back to the variations of the factors. By proper arrangement of the factor settings it will be possible to determine influence of the variation of each factor on the response in the presence of a simultaneous variation of all other factors. It will also be possible to detennine what effects changing two or more factors simultaneously and independently will have on the variations of the response. In fact, it is possible to detennine all interaction effects, i.e. interaction effects of two factors, three factors, four factors. .., all factors. 

AU these effects can be determined independently of each other. This means, that the estimated value of any effect does not depend on the estimated value of any other effects. 

---