On Clarifying M and M
暂无分享,去创建一个
Second only to learning from one's mistakes is the benefit to be gained from those of one's friends and colleagues. Since Massy and Morrison [ 4] provide no instances of the first and few of the second, their comments on Markov and me [ 1] are less valuable than they might have been. Only a few points remain to be clarified. One such point is the stochastic or deterministic formulation of buyer behavior models. About Markov it is agreed that it would be preferable if a first-order stationary brand-switching model could be formulated stochastically, but it is also agreed that this is not really possible-"his criticisms are technically correct." Some people do however treat Markov determinz'stically, see [3]. Massy and Morrison want to dismiss this deterministic formulation as untenable on a priori grounds and seem worried because I examined it on its merits. They fear that by treating determinism with such respect I might seem to actually like it. Such a misinterpretation seems improbable; but since it is clearly possible, I recapitulate: I would accept a deterministic formulation of the Markovian brand-switching model as a working
[1] William F. Massy,et al. Comments on Ehrenberg's Appraisal of Brand-Switching Models , 1968 .
[2] Andrew Ehrenberg,et al. An Appraisal of Markov Brand-Switching Models , 1965 .
[3] T. W. Anderson,et al. Statistical Inference about Markov Chains , 1957 .