Abstract
Spector recently suggested that the usefulness of the NBD model in a geographical context is limited because of its failure to take account of exogenous factors. In this paper we respond constructively to this criticism by illustrating how purchase incidence models of the NBD type may be extended to incorporate the effects of explanatory variables. This results in what might be termed purchase incidence "regression" models. These are developed in the context of likelihood methods which provide a powerful and unified approach to their estimation and testing.

This publication has 22 references indexed in Scilit: