The effects of missing information and inferences on decision processing and evaluation.

Abstract

Many consumer choice situations may involve missing information. A situation involves missing information whenever values for one or more attributes considered relevant for the decision task are not available for one or more alternatives in the choice set. Inferences may be used by the consumer to fill-in this missing information. However, how missing information and use of inferences may effect decision processing and subsequent evaluation has not been fully examined. This paper develops a model of these effects. It is based on a cost/benefit trade-off approach to decision processing. Decision makers are expected to trade-off the effort involved in forming inferences with the increased accuracy afforded by the inferences. This trade-off is expected to motivate decision makers to switch to simplifying strategies which circumvent the need to form inferences. The type of strategy shift likely to occur is dependant on two dimensions of missing information, the amount of missing information and the overlap (or commensurability) of the missing information. Further, the model suggests a moderating role of decision processing on the amount and type of uncertainty associated with missing information alternatives. Processing which involves inferences should reduce the uncertainty associated with missing information alternatives. In this way, processing is expected to effect the ultimate evaluations of missing information alternatives relative to full-information alternatives. Moderating variables are expected to play a role in this framework in two ways. First, variables which affect the costs or benefits associated with forming inferences should impact the processing effect. Second, variables which affect the confidence with which the decision maker views their inferences and missing information alternatives should impact the evaluation effect. This model was tested via a verbal protocol experiment. The 3 x 4 x 2(4) research design tested the effects of subject product category familiarity, and amount and overlap of missing information on processing strategy use and ultimate evaluations. Manova and Partial Least Squares results support the proposed processing effects and the overall process/evaluation effect model.