The first step in model-building is to note some simple observable patterns and to establish their degree of empirical generalisation. The second step is to integrate different empirical relationships of this kind. This is where the more complex mathematical models come in.
Two examples of model-building in marketing are discussed and the empirical approach is contrasted with the prevalent tendency either to try to develop mathematical models without any prior empirical knowledge ("sonking"), or to search for statistical significance as an end in itself. The basic question in model-building remains as follows:
Take away the mathematical language and what generalised factual knowledge of the process in question still remains? If the answer is none, the mathematical symbol for that is very simple.
