A Varying-Parameter Averaging Model of On-Line Brand Evaluations

Consumer evaluations of new brands evolve over time as information is acquired. We conceptualize the extent to which evaluations are updated in terms of the weight given to new information during information integration. Based on information processing theory, we derive hypotheses regarding the weights given to new information under different processing ability conditions. We then develop a varying-parameter averaging model that captures the hypothesized moderating effects of processing ability (i.e., time pressure and knowledge) and also takes into account order effects. Scale values and weights for information items are derived by estimating the model using continuous evaluations obtained in a process-tracing experiment that allows subjects to access information that they desire in any order. Results from model estimation support the hypothesis that compared with prior evaluations new information plays a larger role in evaluations of high (vs. low) ability subjects. Estimating order effects on weights when order is endogenous, we find a recency effect such that information seen later is given a greater weight than information seen earlier. However, this recency effect is reduced as category knowledge increases. We discuss the theoretical and methodological contributions of this research.