This paper develops a model which is an extension of a class of stochastic response models first proposed by Coleman [Coleman, J. A. 1964a. Models of Change and Response Uncertainty. Prentice-Hall, Inc., Englewood Cliffs, New Jersey.]. In its present form the model applies to the binary response situation. The model postulates a population of respondents who are heterogeneous with respect to their response probabilities. A given respondent may experience changes in his response probability in the intervals between response occasions, but since the actual responses he makes do not effect his response probability, the model is zero order. The paper also presents a variant of the minimum chi square procedure which may be used to estimate and test the model. An application to consumer brand choice data demonstrates that the model is empirically viable. Potential applications to multi-wave voting intentions panels and social-psychological experiments are indicated.