Making Inferences About Missing Information: The Effects of Existing Information

Abstract

This article is concerned with the process by which inferences about missing information are formed. We propose that consumers use within-attribute (other-brand) processing first and, if that information is diagnostic, infer the value of the missing information with little consideration of additional information. If the within-attribute processing does not prove diagnostic, consumers then move to within-alternative (same-brand) processing. In two experiments, we find that high variation in other-brand information results in a statistically significant discount to the inferred value even with no variation in same-brand information, whereas the reverse is not true. This is supportive of an inference-formation process in which other-brand information is considered first. Further, we find that, if there is low variation in other-brand information, time and effort are reduced, attention to the attribute with the missing information is proportionately increased, and there is a shift away from alternative-based processing. These results, encompassing both traditional process measures and inference-based measures, support the process model that we advance.