Abstract:

In this paper I want to outline two applications of a mathematical model for consumer purchasing data. Apart from their own direct interest, these applications may serve to illustrate how a mathematical model can he useful for two somewhat different purposes, namely: A. To provide insight into a situation previously not understood - in this case the effect of particular patterns of purchasing on sampling errors; B. To allow the prediction of certain quantities instead of having to observe them directly - here the prediction of the market penetration in a longer period of time than has actually been observed. The model which will be used involves the so-called Negative Binomial distribution. Earlier work on the fit of this distribution to consumer purchasing data has already been described, but it may be helpful to summarize it briefly here before outlining the more recent developments.