Recently many marketing researchers have become interested in stochastic models of behavior. This surge of interest seems to be based on three related realizations: First, predicting individual behavior without a larger error component using individual characteristics and even brand-specific attitudes and intentions data remains an elusive goal. Second, predicting individual behavior is expensive and somewhat irrelevant for marketing decisions such as pricing, advertising allocations, etc. Third, behavior may in fact have a large random component which is in principle unpredictable.