Modelling with Linear Learning

Various explanations are possible for the observation of similar outcomes from repeated choices, for example, both cognitive psychologists and behaviourists find support for their viewpoints in such patterns. Pattern recognition and interpretation are necessarily model‐bound activities as, indeed, are attempts to analyse such data in terms of probabilities. In the face of many alternative procedures with their associated rationales relative simplicity of screening is important in assessing new data in terms of their structure and conformity to a model. This article outlines in a non‐technical way one such procedure which is appropriate to analyse data, particularly brand choice observations, expressed in dichotomous fashion. The recognition of such patterns and a formalisation of an underlying rationale are necessary steps in the improvement of both data‐handling technique and the furtherance of our understanding of consumer behaviour.