Very little is to be found in specialised marketing literature on experimentation. As in the sciences in general the earliest steps forward in marketing were made as a result of using methods of observation. Data are collected either within the business concerned or at retailer or at consumer level, and these data are analysed (statistical research, market research, panels, etc.). One very seldom succeeds in this manner in eliminating all the variables that may exert an influence and one has to make use of the resources of multivariate analysis to measure the effects of the facts that one proposes to study. As in the so-called experimental sciences the experimentation should therefore make it possible under certain conditions to arrive more easily at the desired result. In order to measure the influence of the price level of a product it should be sufficient, for example, to vary the price and to measure the sales results and thereby to establish the law relating of consumption to price level.